Primate Locomotion

Recent Advances
 Primate Locomotion
Recent Advances

Edited by

Elizabeth Strasser
California State University, Sacramento
Sacramento, California

John Fleagle
State University of New York at Stony Brook
Stony Brook, New York

Alfred Rosenberger
National Zoological Park
Washington, D. C.

and

Henry McHenry
University of California, Davis
Davis, California

Springer Science+Business Media, LLC

 Library of Congress Cataloglng-ln-PublIcatlon Data

Primate locomotion : r e c e n t advances / e d i t e d by E l i z a b e t h S t r a s s e r

... [ e t a l . ]
p. cm.
"Proceedings of a symposium on Primate Locomotion, held March
27-28, 1995, In Davis, Ca 1 1 f o r n l a " — T . p . v e r s o .
I n c l u d e s b i b l i o g r a p h i c a l r e f e r e n c e s and Index.
1. P r i m a t e s — L o c o m o t i o n — C o n g r e s s e s . 2. P r i m a t e s — B e h a v i o r -
-Congresses. 3. P r i m a t e s — E v o l u t i o n — C o n g r e s s e s . I. Strasser,
E l i z a b e t h . I I . Symposium on Primate Locomotion (1995 : Davis,
Calif.)
QL737.P9P7256 1998
573.7 "9198—dc21 98-38702
CIP

Proceedings of a symposium on Primate Locomotion, held March 2 7 - 2 8 , 1 9 9 5 ,

in Davis, California

ISBN 978-1-4899-0094-4 ISBN 978-1-4899-0092-0 (eBook)

DOI 10.1007/978-1-4899-0092-0

© Springer Science+Business Media New York 1998

Originally published by Plenum Press, New York in 1998
Softcover reprint of the hardcover 1st edition 1998

http://www.plenum.com

10987654321

All rights reserved

No part of this book may be reproduced, stored in a retrieval system, or transmitted in any form
or by any means, electronic, mechanical, photocopying, microfilming, recording, or otherwise,
without written permission from the Publisher
PREFACE

The study of primate locomotion is a unique discipline that by its nature is interdis-
ciplinary, drawing on and integrating research from ethology, ecology, comparative anat-
omy, physiology, biomechanics, paleontology, etc. When combined and focused on
particular problems this diversity of approaches permits unparalleled insight into critical
aspects of our evolutionary past and into a major component of the behavioral repertoire
of all animals. Unfortunately, because of the structure of academia, integration of these
different approaches is a rare phenomenon. For instance, papers on primate behavior tend
to be published in separate specialist journals and read by subgroups of anthropologists
and zoologists, thus precluding critical syntheses.
In the spring of 1995 we overcame this compartmentalization by organizing a con-
ference that brought together experts with many different perspectives on primate locomo-
tion to address the current state of the field and to consider where we go from here. The
conference, Primate Locomotion-1995, took place thirty years after the pioneering confer-
ence on the same topic that was convened by the late Warren G. Kinzey at Davis in 1965.
The 1965 conference (published in 1967) brought together students of primate behavior
(Ripley), morphology (Tuttle, Napier, Oxnard, Prost), paleontology (Simons), and others
to address a common theme and, for many, to define the study of primate locomotion as a
distinct area of research. As the papers in the current volume demonstrate, we have come a
long way in thirty years with naturalistic studies on dozens of species, new skeletal re-
mains of numerous fossil species, and an expanding battery of morphometric and experi-
mental techniques. We hope that the excitement and spirit of interdisciplinary
collaboration that characterized the 1965 conference is still evident in the field these three
decades later.
At the 1995 conference, the papers were presented in six sessions. Upon hearing the
presentations, we subsequently decided that a book divided into four sections would ac-
commodate the material better. The first section of this book (Part I: Naturalistic Behav-
ior) contains six papers describing the positional behavior of primates from all the major
extant groups. Five papers are grouped in Part II (Morphology and Behavior). These pa-
pers integrate functional studies of primate morphology with experimental studies on a
wide range of primate behaviors. They provide the critical link that enables us to recon-
struct the behavior of extinct species known only from their bones. The six papers in Part
III (Data Acquisition and Analytic Techniques) demonstrate the potential for new tech-
niques as well as the promises and problems inherent in currently used techniques. In the
final section of this book (Part IV) six papers are grouped together under the section title

v
vi Preface

Fossils and Reconstructing the Origins and Evolution of Taxa. While the origin and radia-
tion of major groups of primates have been linked to the acquisition of novel locomotor
strategies, it appears that the impetus for the development of novel locomotor behaviors is
first of all related to diet. The six papers in this section address this issue.
Given the breadth of approaches used in this book, we hope that its promise to be an
exciting update to the 1965 conference on primate locomotion holds true.
The conference Primate Locomotion-1995 was generously supported by the Na-
tional Science Foundation (SBR 9507711), The Wenner-Gren Foundation for Anthropo-
logical Research (CONF-167), and the California State University, Sacramento
Foundation (#120025). In an effort to have the highest quality papers in this book, each
manuscript was reviewed by a combination of the other contributors. the coeditors, and an
assortment of colleagues to whom we are truly grateful for their efforts. We thank Geof-
frey Kushnick for his assistance in compiling the index for this volume, Diana Norman for
the cover artwork, and Eileen Bermingham, Donna Carty, MaryAnn McCarra, and Susan
Safren for their patience in working with us on this book.

Elizabeth Strasser, Sacramento, California

John G. Fleagle, Stony Brook, New York
Alfred L. Rosenberger, Washington, D.C.
Henry M. McHenry, Davis, California

REFERENCES
Kinzey WG, editor (1967) Symposium on Primate Locomotion. Am. J. Phys. Anthropol., 26(2).
CONTENTS

Part I: Naturalistic Behavior

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
Alfred L. Rosenberger

I. Methodological Issues in Studying Positional Behavior: Meeting Ripley's

Challenge . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
Marian Dagosto and Daniel L. Gebo

2. Fine-Grained Differences within Positional Categories: A Case Study of

Pithecia and Chiropotes ........................................ 31
Suzanne E. Walker

3. Patterns of Suspensory Feeding in Alouatta palliata, Ateles geoffroyi, and Cebus

capucinus . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
David 1. Bergeson

4. Within- and Between-Site Variability in Moustached Tamarin (Saguinus mystax)

Positional Behavior during Food Procurement ...................... 61
Paul A. Garber

5. Locomotion, Support Use, Maintenance Activities, and Habitat Structure: The

Case of the Tai Forest Cercopithecids ............................. 79
W. Scott McGraw

6. The Gorilla Paradox: The Effects of Body Size and Habitat on the Positional
Behavior of Lowland and Mountain Gorillas.. . . . . . . . . . . . . . . . . . . . . . . 95
Melissa J. Remis

Part II: Morphology and Behavior

Introduction ....................... : . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109

John G. Fleagle

7. Reconstruction of Hip Joint Function in Extant and Fossil Primates 111

Laura MacLatchy

vii
viii Contents

8. Grasping Performance in Saguinus midas and the Evolution of Hand

Prehensility in Primates ........................................ 131
Pierre Lemelin and Brian W. Grafton

9. Tail-Assisted Hind Limb Suspension as a Transitional Behavior in the Evolution

of the Platyrrhine Prehensile Tail ................................. 145
D. Jeffrey Meldrum

10. Unique Aspects of Quadrupedal Locomotion in Nonhuman Primates 157

Susan G. Larson

11. Forelimb Mechanics during Arboreal and Terrestrial Quadrupedalism in Old

World Monkeys. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 175
Daniel Schmitt

Part III: Data Acquisition and Analytic Techniques

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 203
Elizabeth Strasser

12. Advances in Three-Dimensional Data Acquisition and Analysis 205

John Kappelman

13. Laser Scanning and Paleoanthropology: An Example from Olduvai Gorge,

Tanzania .................................................... 223
Leslie Aiello, Bernard Wood, Cathy Key, and Chris Wood

14. Use of Strain Gauges in the Study of Primate Locomotor Biomechanics. . . . . . . 237
Brigitte Demes

15. The Information Content of Morphometric Data in Primates: Function,

Development, and Evolution .................................... 255
Charles E. Oxnard

16. Heterochronic Approaches to the Study of Locomotion .................... 277

Laurie R. Godfrey, Stephen 1. King, and Michael R. Sutherland

17. Body Size and Scaling of Long Bone Geometry, Bone Strength, and Positional
Behavior in Cercopithecoid Primates . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 309
William L. Jungers, David B. Burr, and Maria S. Cole

Part IV: Fossils and Reconstructing the Origins and Evolution of Taxa

Introduction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 333
Henry M. McHenry

18. A{ropithecus. Proconsul. and the Primitive Hominoid Skeleton 337

Carol V. Ward
Contents ix

19. Fossil Evidence for the Origins ofTerrestriality among Old World Higher
Primates. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 353
Monte L. McCrossin, Brenda R. Benefit, Stephen N. Gitau,
Angela K. Palmer, and Kathleen T. Blue

20. Ecological Morphology of Australopithecus afarensis: Traveling Terrestrially,

Eating Arboreally ............................................. 397
Kevin D. Hunt

21. Time and Energy: The Ecological Context for the Evolution of Bipedalism 419
Robert A. Foley and Sarah Elton

22. Heel, Squat, Stand, Stride: Function and Evolution of Hominoid Feet. . . . . . . . . 435
Russell H. Tuttle, Benedikt Hallgrimsson, and Tamara Stein

23. Evolution of the Hominid Hip 449

Christopher Ruff

Index 471
 I

NATURALISTIC BEHAVIOR
INTRODUCTION TO PART I

Alfred L. Rosenberger

This first section of the book has a variety of interesting papers that deal with the
study of positional behavior in the wild and that sample all the major taxa of primates. The
first three papers explore topics ranging from methodological issues to the first, surpris-
ingly, comparative study of the positional behavior of prehensile-tailed primates. The last
three papers of this section examine the effect of habitat characteristics on positional be-
havior and come to some surprising conclusions.
In Chapter I, as our reviewers noted, Marian Dagosto and Daniel Gebo provide a
thoughtful, balanced, and judicious treatment of all the major problems and points of im-
portance concerned with the study of primate positional behavior. Their paper is both a
cautionary tale about incommensurate research designs and a sobering summary of the
state-of-the-art data base offield-workers. Not only will behaviorists find this contribution
useful, but functional morphologists interested in the design and performance of the pri-
mate postcranial skeleton also will find it to be valuable and thought-provoking on many
levels. Dagosto and Gebo's paper is appropriate as the first in this volume, since it is such
a fitting tribute to Warren G. Kinzey as well as his colleague Suzanne Ripley.
Suzanne Walker's paper (Chapter 2) addresses some current deficiencies in the defi-
nition and use of gross categories of positional behavior. She illustrates the deficiencies
with examples of positional behaviors that require finer-grained descriptions than the
gross categories of "sit" or "stand". In addition, she examines some associations between
the newly identified positional behaviors and characteristics of the supports that would be
missed using gross characterizations.
David Bergeson (Chapter 3) uses the results ofa naturalistic study of tail use in three
Costa Rican monkeys to examine a series of hypotheses about the ecological significance
of fore- and hindlimb suspensory behavior in primates with prehensile tails. In general, he
found that hypotheses put forth to explain suspensory behavior in other groups of primates
do not seem to accord with patterns of suspension using prehensile tails. Moreover, the
three platyrrhine species that he studied show different patterns of tail suspension.
In Chapter 4, Paul Garber examines both between- and within-site variation in the
positional behavior of Peruvian moustached tamarins. He found only limited between-site
and within-site variation in positional patterns: rare modes were always rare and common

3
4 A. L. Rosenberger

modes were always common. Support characteristics were found to vary more than posi-
tional behaviors. As a consequence, as one reviewer noted, in this paper Garber reinforces
the growing awareness that support choice is less constrained by morphology or ecology
than are positional behaviors.
In Chapter 5, our reviewers note that Scott McGraw's is a timely paper on the debate
of the relationship between habitat structure and locomotor behavior. McGraw examines
within-species variability among three sympatric cercopithecids to determine if the mon-
keys change their locomotion and support use during the two maintenance activities of
traveling and foraging. Like Garber (Chapter 4) he found that the pattern of support use is
more variable than are the accompanying locomotor behaviors. Furthermore, the effect of
maintenance activities on locomotor profiles is nil. As McGraw points out, the fact that
the focal species moved in the same general manner independent of differences in their re-
spective habitats has implications for the study of fossils, in that it strengthens our ability
to predict locomotor behavior from morphology.
In Chapter 6, Melissa Remis reports on the positional behavior and substrate use of
lowland gorillas and compares them to similar data for mountain gorillas. Her main argu-
ment is that the differences in positional behavior between lowland and mountain gorillas
are related to habitat differences. In contrast to lowland gorillas, mountain gorillas are
constrained from engaging in more arboreal behaviors by the distribution of food items.
As noted by Remis, her results do not correspond with the generalizations given by Garber
(Chapter 4) and McGraw (Chapter 5).
 1

METHODOLOGICAL ISSUES IN STUDYING

POSITIONAL BEHAVIOR
Meeting Ripley's Challenge

Marian Dagosto l and Daniel L. Geb02

IDepartment of Cell and Molecular Biology

Northwestern University School of Medicine
Chicago, Illinois 60611
2Department of Anthropology
Northern Illinois University
DeKalb, Illinois 60115

1. INTRODUCTION

A major goal of primate evolutionary morphology is to relate limb anatomy to as-

pects of locomotor or postural behavior or to an entire regime of positional behavior. If
well established in living forms these relationships provide the necessary framework for
reconstruction of the behavior of extinct species and to answer questions concerning the
evolution of primate locomotor systems that were the impetus for the 1965 conference
(Kinzey, 1967). Studies of this sort usually proceed by noting salient differences in behav-
ior and morphology in two or more species, establishing correlations between the two, and
attempting to explain the correlations using causal arguments derived from biomechanical
principles (e.g., Ashton and Oxnard, 1964; Fleagle, 1977a,b; Fleagle and Meldrum, 1988).
Obviously, success in such an endeavor depends on the quality of both morphological and
behavioral data sets (Ripley, 1967; Fleagle, 1979). In her contribution to the 1965 sympo-
sium, Ripley pointed out how the lack of detailed information on behavior hampered ef-
forts to assess the evolutionary importance of behaviors and to construct realistic
locomotor groupings or classifications. Morphologists tend to "underestimate the rich
complexity of locomotor behavior and to underestimate the difficulties involved in relat-
ing morphology and habitual behavior" (Ripley, 1965: 167). She especially criticized at-
tempts to define locomotor types in primates on the basis of insufficient information. At
the 1965 conference, both Ripley (1967) and Kinzey (1967) challenged primatologists to
collect more data to alleviate these problems. Ripley (1967: 149) outlined the elements of

5
6 M. Dagosto and D. L. Gebo

"an exhaustive study of the total locomotor pattern" necessary before such categorizations
would be meaningful.
Although there has been progress since 1965 we believe the study of positional behav-
ior still falls short of the goals set by Ripley and Kinzey to the point where integration of be-
havioral and morphological data is still compromised. There are several aspects of
positional behavior studies that contribute to this deficiency. (There are problems with mor-
phological studies·as well, but it is not our intent to address these here.) First, long term, de-
tailed studies of positional behavior as outlined by Ripley (1967) are still rare compared to
those of diet or social structure, and thus the base for studies built around this information is
weak. Secondly, there have been very few discussions concerning the methods of studying
positional behavior. Several techniques of data collection and analysis exist, and whether re-
sults are comparable across studies is debatable and has never been rigorously tested. Com-
parability of data is a serious issue for the type of broadly interspecific quantitative studies
that are currently being attempted (e.g., Crompton et aI., 1987; Oxnard et aI., 1990). Thirdly,
sources of intraspecific variation in behavior have only begun to be looked at, and there is
disagreement about the existence and importance of these factors.

2. RARITY OF POSITIONAL BEHAVIOR STUDIES

Previous to the 1965 symposium, assessments of positional behavior were based

largely on qualitative anecdotal reports or on observations in a captive setting (e.g., Avis,
1962; Ashton and Oxnard, 1964). To assess improvement since that time we counted up
studies that were directed at documenting positional behavior in the wild in a quantitative
fashion, or that at least contained usable quantitative accounts of positional behavior or
substrate use (Table 1). Other notable studies of positional behavior that are non-quantita-
tive or were conducted in a captive setting are listed in Table 2. Since the 1965 sympo-
sium, there have been approximately 50 field based studies (about 1.3 studies per year)
but almost half of these were undertaken within the last 10 years. There are about 50 spe-
cies on which there is some such data, although still barely a handful (maybe none) that
meet Ripley's criteria for an exhaustive study of total locomotor pattern. All major taxo-
nomic groups are represented, but there is a heavy emphasis on New World monkeys
(Alouatta has been the subject of 8 studies, Saguinus 6, and Ateles 6). By 1985 quantita-
tive data on positional behavior were known for only 29 species of primates. In contrast,
by 1985 quantitative data on group size had been collected for 111 species (Jolly, 1985),
and quantitative data on diet for 56 species (Richard, 1985).
The addition of these data has led to more realistic categorization of positional be-
havior, a subject of some concern at the 1965 conference. For example, careful field stud-
ies of Ateles, Colobus, and Presbytis by Mittermeier and Fleagle (1976), Morbeck (1974),
and Ripley (1967) resulted in a successful challenge of the usefulness of the "semi brachia-
tor" category (Mittermeier and Fleagle, 1976). At the same time the questions have
changed; there is now less emphasis on broad categorization, which may explain major
differences in morphology, but leaves us with a lot of unexplained variation in both behav-
ior and morphology within groups. We also have skeletal remains from many more fossil
primates than Simons was able to review at the 1965 conference (Simons, 1967). These
extinct species differ in more subtle aspects of morphology than those associated with ma-
jor locomotor categories, but the implication of these differences for behavior is also
something that we strive to understand. There is also renewed interest in ecomorphology
(Bock, 1990; Wainwright and Reilly, 1994); in particular relationships among habitat,
Methodological Issues in Studying Positional Behavior 7

Table 1. Studies of positional behavior in primates since 1965. Only studies that were conducted in
the wild and that are quantitative in nature are included. Studies using the same data are grouped
together. See Table 2 for other studies. The last column indicates if data on locomotion (L), posture
(P), substrate use (S; either size or orientation), or habitat (H; usually height data) are included
Years Publication Species Data
1965-1970 Richard, 1970 Ate/es geoffroyi L
A/ouatta palliata
1971-1975 Rose, 1974, 1978, 1979 C%bus guereza L,P,S
Cercopithecus aethiops
Morbeck, 1974. 1977a,b, 1979 C%bus guereza L,P,S,H
1976-1980 Fleagle, 1976, 1980 Hy/obates syndacty/us L,P,S
Hy/obates far
Mendel, 1976 A/ouatta palliata L,P,S,H
Rose, 1976, 1977 Papio anubis L,P
Cercopithecus ascanius
Cercopithecus mitis
Fleagle. 1978 Presbytis me/a/ophos L,P,H
Presbytis obscura
Mittermeier,1978 A te/es geoffi'oyi L,P.S,H
A teles paniscus
Richard, 1978 Propithecus verreauxi P,S,H
Fleagle and Mittermeier, 1980; Fleagle et aI., 1981 A teles paniscus L,S,H
A/ouatta senicu/us
Pithecia pithecia
Saguinus midas
Saimiri sciureus
Chiropotes satanas
Cebus apel/a
Garber, 1980, 1984 Saguinus oedipus L, P, S
MacKinnon and MacKinnon, 1980 Tarsius spectrum L,S,H
Susman et aI., 1980 Pan paniscus L
1981-1985 Happel, 1982 Pithecia hirsuta L,H
Crompton, 1983, 1984 Ga/ago moholi L,P,S,H
Oto/emur crassicaudatus
Gittins, 1983 Hy/obates agilis L,P,S,H
Schiin-Ybarra, 1984 A louatta seniculus L,P
Srikosamatara, 1984 Hylobates pi/eatus L,S,H
Susman, 1984 Pan paniscus L,H
Oliveira et aI., 1985 Pithecia pithecia L,H
Tuttle and Watts, 1985 Gorilla gorilla L,P
1986-1990 Cant, 1986 A touatta pigra L,P,S
Ateles geoffroyi
Crompton and Andau, 1986 Tarsius bancanus L,P,S,H
Sugardjito and van Hooff, 1986 Pongo pygmaeus L,P,H
Cant, 1987a,b Pongo pygmaeus L,P,S
Schon-Ybarra and Schon, 1987 Alouatta seniculus L,P,S
Cant, 1988 Macaca jascicularis L,P,S
Boinski, 1989 Saimiri oerstedii L,P,S,H
Wilson et aI., 1989 Eu/emur coronatus L,H
Fontaine, 1990 Ate/es geoffroyi L,S
1991-1996 Garber, 1991 Saguinus juscicollis L,S
Saguinus mystax
Saguinus geofJroyi
Hunt, 1991, 1992, 1994 Papio anubis L,P,S,H
Pan troglodytes
Doran, I 992a,b 1993a,b, 1996; Doran and Hunt, 1994 Pan troglodytes L,P,S,H
Pan paniscus
( continued)
8 M. Dagosto and D. L. Gebo

Table 1. (continued)
Years Publication Species Data
Gebo, 1992 Alouatta palliata L,P,S,H
Cebus capucinus
Bicca-Marques and Calegano-Marques, 1993, 1995 Alouatta caraya L,P
DaSilva, 1993 Colobus polykomos P
Tremble et aI., 1993 Tarsius dianae L,P,S,H
Youlatos, 1993 Alouatta seniculus L,S
Walker, 1994, 1996 Chiropotes satanas L,P
Pithecia pithecia
Cannon and Leighton, 1994 Hylobates agilis L,S,H
Macaca fascicularis
Dagosto, 1994, 1995 Varecia variegata L,P,S,H
Eulemur fulvus
Eulemur rubriventer
Propithecus diadema
Gebo et aI., 1994; Gebo and Chapman, 1995a,b Colobus badius L,P,S,H
Colobus guereza
Cercopithecus mitis
Cercopithecus ascanius
Cercocebus albigena
Garber and Preutz, 1995 Saguinus mystax L,S,H
Remis, 1995 Gorilla gorilla L,P,S,H
McGraw, 1996 Colobus polykomos L,P,S,H
Colobus badius
Colobus verus
Cercopithecus diana
Cercopithecus campbelli
Doran, 1996 Gorilla gorilla L,P,S
Walker and Ayres, 1996 Cacajao calvus L,P,S

body size, diet, and positional behavior (e.g., Fleagle and Mittermeier, 1980; the niche
metrics of Crompton et aI., 1987 and Oxnard et aI., 1990). Addressing these questions in a
broad comparative perspective often relies heavily on quantitative approaches, and there-
fore quantitative assessments of behavior.

3. METHODS OF COLLECTION AND ANALYSIS OF POSITIONAL

BEHAVIOR
Despite the increase in the number of studies since 1965, there has been surprisingly
little review and critique of methods. With the exception of Ripley's (1967) paper, very lit-
tle has been written about how to conduct such studies. What sort of data should be col-
lected, how should they be collected, and how should they be analyzed? The answers to
these questions depend of course on what we want to ask of our data, and this varies
among studies (Rose, 1979). Given the rarity of field studies, the majority of them are still
directed towards simply documenting the behavior of species, almost always with the no-
tion that these data will be useful for explaining anatomical form. Thus, we will concen-
trate our remarks on these types of studies, and on locomotor behavior in particular
(although many of the same arguments will apply to the study of posture or substrate use).
Methodological Issues in Studying Positional Behavior 9

Table 2. Studies of positional behavior that are non-quantitative or that were conducted on
animals in captivity or semi-captivity
Publication Species Data
Ripley, 1967, 1977, 1979 Presbytis entellus L,P,S,H
Walker, 1969 Perodicticus potto L,P,S
Mittermeier and Fleagle, 1976 Ateles geoffroyi L,P
Kinzey et aI., 1975 Cebuella pygmaeus S
Charles-Dominique, 1977 Arctocebus ca/abarensis S,H
Perodicticus potto
Euoticus e/egantu/us
Go/ago alieni
Go/ago demidovii
Kinzey, 1977 Callicebus torquatus S,H
Tattersall, 1977 Eu/emur fo/vus S,H
Rodman, 1979, 1991 Macaca fascicu/aris H
Macaca nemestrina
Walker, 1979 Go/ago demidovii L,S
Dykyj,1980 Nycticebus coucang S
Rollinson and Martin, 1981 Cercocebus ga/eritus L,S
Cercocebus albigena
Cercopithecus neg/ectus
Cercopithecus nictitans
Cercopithecus pogonias
Miopithecus ta/apoin
Sugardjito, 1982 Pongo pygmaeus L,P
Glassman and Wells, I 984 Nycticebus coucang L,P,S
Niemitz, 1984a,b Tarsius bancanus L,P,H
Roberts and Cunningham, 1986; Tarsius bancanus S,H
Roberts and Kohn, 1993
Gebo, 1987 Lemur. Eu/emur (7 species). Hapa/emur. L
Propithecus. Cheirogaleus (2 species). Mirza.
Microcebus. Perodicticus. Nycticebus.
Loris. Tarsius (2 species)
Tilden, 1990 Eulemur rubriventer L,S
Fontaine, 1990 Saimiri sciureus L,P,S
Curtis, 1992; Curtis and Feistner, 1994 Daubentonia madagascariensis L,P,S
Rosenberger and Stafford, 1994 Leontopithecus rosalia L
Callimico goe/dii

It is also desirable to develop more specific questions (e.g., Cant, 1992), which might re-
quire other techniques and methods not addressed in this paper.
Some aspects of methodology, such as sampling strategies, data analysis, and defini-
tions of behavioral categories directly affect comparability among studies, an important is-
sue for both broad, interspecific comparisons, and even more narrowly defined contrasts if
the behavioral data are derived from different sources. Even though both can be described
numerically, morphological and behavioral data sets are different. Most aspects of mor-
phology that we assess are relatively static over the adult life of an individual, they occur
in discrete states or are easily quantified, each individual is represented by a single score
or measurement that is generally easily repeatable, samples are characterized by well un-
derstood summary statistics, and differences among species can be tested by traditional
univariate and multivariate techniques. In other words, in morphological studies we gener-
ally have a pretty good idea what the numbers mean.
 10 M. Dagosto and D. L. Gebo

Positional behavior is a more complicated matter. For perfectly understandable prac-

tical reasons, behavior is sampled during a very brief time frame (weeks, months) so we
have a very poor idea of how plastic behavior may be during an individual's or species'
life span. Studies are also limited in space, so behavior from only a small part of the spe-
cies' range is ever sampled. The potential for behavior, substrate use, and habitat selection
to be heavily influenced by immediate circumstances, while morphological measurements
are less labile is, in fact, the impetus for ecomorphologists to discover morphological cor-
relates that can be substituted for the more intractable environmental variables (Bock,
1990; Rickelfs and Miles, 1994).
Behavioral attributes studied during these brief interludes are collected by sequential
observations on individuals, these are tallied and transformed into proportions. The nature
of the data makes it difficult to analyze statistically, and the appropriate summary statistic
or comparative test is a matter of contention (Dagosto, 1994). Even though these studies
result in hundreds or thousands of observations, should we be satisfied? Do these impres-
sive numbers reflect a deep understanding and adequate characterization of behavior, or
are they just an artifact of method (pseudoprecision). In other words, we are not sure we
understand what the numbers mean: do these data adequately epitomize the positional be-
havior of a species in the same way that osteological measurements characterize morphol-
ogy? Several aspects of data collection and analysis can affect the values published in any
single study and must be taken into account when comparing studies.

3.1. Time Samples versus Bouts

Typically, studies of positional behavior follow one of two methods, continuous

sampling (bouts; e.g., Fleagle, 1976) or instantaneous time sampling. It is usually assumed
that results (in terms of proportions of behaviors) will be comparable: behaviors that occur
more frequently than others should take up more of an animal's time. But this depends on
the duration of the behavior. Behaviors that occur infrequently but have a long duration
may take up more time than behaviors that occur more frequently but have a short dura-
tion. For example, Propithecus diadem a leaps much more often than it sits (leaping = 57%
of all bouts; sitting = 15% of all bouts), but bouts of leaping last much shorter than bouts
of sitting (sifakas spend only 3.5% of their time leaping and 64% of their time sitting).
Obviously one must be careful when comparing such numbers.
The comparability of results using bouts and time samples has rarely been tested. In
one such test, Doran (l992a) found that the resultant proportions were quite different, but
if bouts were corrected by distance traveled, the two methods yielded similar results. In
Dagosto's (1994, 1995) study of lemurs, data were also collected with both methods si-
multaneously. As in Doran's study, there were significant differences in the proportions of
behaviors measured with data from bouts and time samples; in Eulemur rubriventer and
Propithecus diadema, the differences are statistically significant (Table 3). In this case,
however, correcting bouts for distance makes the results more comparable only for E.
fulvus. In E. rubriventer, there is no great disparity in the average distance traveled per
bout for each type of locomotor behavior (Table 4), so correcting bouts for distance
changes the proportions very little, and not enough to make them equivalent to the propor-
tions calculated for time samples. Distance traveled during an event of leaping in P
diadema is greater than for quadrupedal ism or climbing, so that if bouts are "corrected"
for distance, the proportion of leaping increases relative to quadrupedalism and climbing,
making the difference between bouts and time sampling even greater (Table 3)!
 ~
~
=-g,
<:>
Table 3. Proportions of locomotor behavior in Malagasy lemurs calculated by several methods. In IND, proportions are calculated 0'
IJ<l
for each individual and the results averaged. The observed range of proportions is given below. In LAO, all observations are lumped ;;.
~
before proportions are calculated. With the bout method (Bout), proportions are based on the number of occurrences of the behavior
divided by the number of occurrences of all locomotor behaviors (N). With time samples (TS) the proportions are based on the '"'"
number of time samples during which the behavior was being performed divided by the total number of time samples of
=
'"'"6'
locomotor behavior (N). In Bout with distance (Bout(wd)), the proportions in the Bout column are "cOlTected" by the average ~
distance traveled using the mode of displacement (Table 4) following the method of Doran (1992a). P gives the statistical Co
=
'<
significance of a test of the differences in calculated proportions between the Bout and TS methods, using the :i'
IJ<l
two-sample test of Manly (1991); *, P=1.0-O.05, **P<O.05, ns = not significant.

Eulemur fulvus Eulemur rubriventer Propithecus diadema

~
o·
Bout Bout Bout ~
=
Bout TS (wd) p Bout TS (wd) P Bout TS (wd) P ~
IND
..=-o·""
N 3987 267 ns 5285 291 5696 199
.
Leap 55.7 50.6 50.9 ns 62.2 49.2 60.0 ** 88.2 75.6 92.9 **
34-72 24-73 52-74 38-60 85-91 63-96
Quad. 28.9 35.1 35.6 ns 23.1 30.3 26.3 ** 1.0 2.8 0.7 ns
(range) 17-46 11-61 13-33 14-44 0-3 0-13
Climb 12.8 13.7 12.9 ns 13.4 19.1 13.4 ** 8.9 18.2 7.1 **
(range) 7-21 0-33 10-18 11-27 6-11 4-33
Other 2.3 .5 ns 1.3 1.4 ns 1.9 3.5 ns
(range) 0-6 0-3 0-5 0-7 0-5 0-17
LAO
Leap 61.8 48.3 56.4 62.2 49.0 60.3 88.4 76.4 93.1
Quad. 25.3 39.0 31.1 23.2 32.0 26.5 0.9 2.5 0.6
Climb Il.l 11.6 1l.2 12.8 17.7 12.8 9.0 18.1 7.1
Other 1.9 l.l 1.9 1.3 1.7 3.0
 12 M. Dagosto and D. L. Gebo

Table 4. Average distance traveled (in meters) with each

type of locomotor behavior in Malagasy primates

E·fulvus E. rubriventer P'diadema

Leap 1.4 1.5 1.9
Quad. 1.9 1.8 1.3
Climb 1.5 1.6 1.5

Regardless of whether or not results can be made comparable, it must be remembered

that these data collection techniques are designed to answer different questions, so there is
no reason to expect that they will converge on the same result. Continuous recording is de-
signed to measure the frequency of occurrence of discrete events, it answers the question,
"How often does this event occur?" Time sampling is designed to measure the percent of
time devoted to states (Altmann, 1974; Martin and Bateson,1986), it answers the question
"How much time is spent in this activity?" "How much does each type oflocomotor activity
contribute to total distance traveled?" is yet another important question. Investigators need
to decide which aspect(s) of behavior they want to measure before choosing a data collec-
tion technique. Using these definitions, we would modify Doran's (1992a) initial assess-
ment that bout (without distance) overestimates the frequency of locomotor activities used
often, but for short distances; and underestimates the frequency oflocomotor activities used
rarely, but for long distances. Bout (without distance) estimates the frequencies (i.e., how
often behaviors occur) perfectly well; what it may not reflect well is the relative amount of
time spent performing the activity, nor its contribution to distance traveled. The former re-
quires time sampling or some estimate of duration and the latter requires information on av-
erage distance per bout, especially if all types of locomotor behaviors do not have similar
average distances, as was the case in the chimpanzees studied by Doran. In that case, the
distance traveled during a bout was probably highly correlated with the duration of the
event, so that correcting bouts for distance also corrected them for duration, and thus made
the resultant proportions more comparable to those estimated with time samples. In the si-
faka example, distance traveled does not correlate well with bout duration (i.e., an event of
leaping takes less time, yet covers a greater distance than an event of climbing), so that cor-
recting bouts by distance does not make them more comparable to time samples; it does,
however, yield a better picture of the contribution of each mode of locomotion to total dis-
tance traveled. It might be possible to translate time samples to bouts with a measure of time
duration, (this will be necessary for postures, as they do not have associated distances), but
since we did not collect this type of data, we cannot explore this further.
These two studies show that even data collected by the same investigator at the same
time will yield different results if different methods are employed, and means that direct
comparisons of proportions among studies using different techniques is potentially hazard-
ous. The rank order of behaviors as estimated by the two methods is comparable in the le-
mur examples, but not in Doran's study, so even this less discriminating comparison may
not always be reliable.
These examples, however, do not answer the question of which technique (if either)
gives the "right" answer, i.e., which set of proportions best reflects what actually hap-
pened during the time of the study. To answer this, one needs some way of determining
what actually happened, something which will rarely be available for a field positional be-
havior study unless all observation time is videotaped (e.g., Rosenberger and Stafford,
1994) We did, however, conduct an "experiment" in event sampling using Chicago Bulls'
Methodological Issues in Studying Positional Behavior \3

basketball games as the data source. For 10 games during the regular 1995-96 season, we
scored shot attempts using both bout and time sampling (at I minute and 2 minute inter-
vals) and calculated the proportion of shot attempts attributable to each player. In this
case, the results of bouts and time sampling should be directly comparable since there are
no differences in event duration between players. Because sitting on the couch and watch-
ing a basketball game is much easier than running around the forest chasing primates (our
respective departmental chairs were too cheap to spring for tickets for this very important
scientific endeavor so we had to watch at home) a random sample of half of the bouts ob-
served in a game was extracted for analysis. We compared the results of sampling with
what actually happened (at least the official scorer's version), since game results are
printed in the newspaper the next day. The results of this experiment are presented in Ta-
ble 5. Bout scoring differed less from what actually happened than did either I minute or 2
minute time samples in every single game and even when the events from all 10 games are
combined. Bout scoring never differed by more than more than 3% from the actual pro-
portions, while time samples differed by as much as 15%. The rank order of player shoot-
ing is the same for the top 3 positions for all sampling techniques except 2 minute time
sampling, but varies a lot for any player contributing less than 10% of shot attempts
(Spearman rank correlation does not, however, indicate any differences between the meth-
ods). The same results pertain if other events or statistics are used (scoring, 2 point versus
3 point shots, field goal percentage, etc.). The relatively poor performance of time sam-
pling is no doubt due to the small number of events scored with this method. Shot at-
tempts are frequent (more than 80 per 48 minute game for each team), but because they
have a very short duration (Mean=0.96 sec, SD=O.3 sec, N=45 ), they take up only 5% of
total game time and thus are rarely scored with a time sampling method. The same prob-
lem occurs in the study of locomotor behavior because primates spend very little time
(usually less than 10% of a day) moving. Very few events of locomotion are scored even
in many hours of field time. For example, in Dagosto's (1994) study 234 hours of observa-
tion of E. rubriventer resulted in only 291 time samples of locomotor behaviors (com-
pared to over 5000 bouts); in E.fulvus 191 hours of observation yielded 267 time samples
but 4000 bouts, and in P. diadema 250 hours produced 199 time samples and 5700 bouts.
Therefore, one might very well ask whether time sampling is a very efficient method for
collecting data on locomotor behavior even if it will ultimately converge on the correct re-
sult (Doran, 1992a; Dagosto, 1994). Those researchers particularly interested in locomo-
tion will have to ask themselves if the perceived advantages of time sampling are worth
the extra time commitment necessary to get a reasonable sample of locomotor events.
What constitutes a reasonable sample of events, or a reasonable amount of observa-
tion time, is also a question that has gone largely unaddressed. Gebo (1992; Gebo and
Chapman, 1995a) demonstrates that frequencies are labile when the sample consists of
less than 1000 bouts, but settles into a pattern of little change after 3000 observations
(about 2/3 of which are locomotor rather than postural events). (He did not, however, test
for statistical differences in behavior.) Based on our experience with lemurs and monkeys,
one can obtain 2000 locomotor bouts in about 50-100 hours of observation time; with 2
minute time sampling it might take as long as 1000 hours to get 2000 observations of lo-
comotor behavior.
On the other hand, Kawanaka (1996) calculated that 25 hours of observation is nec-
essary to accurately estimate the amount of time spent in five general behavioral catego-
ries for an individual, suggesting that less time might be adequate. This result pertains,
however, only when the whole suite of behaviors is compared using a correlation; direct
14 M. Dagosto and D. L. Gebo

Table 5. Top. Proportions of shot attempts by different players on the Chicago Bulls basketball team,
sampled during ten games scored by the bout method (Bout; a random sample of half the bouts
scored were used), one-minute time samples (I TS), and two minute time samples (2TS). These
are compared to the actual proportions determined by the official scorer (Actual). N, number of
observations recorded. The difference in the proportions between each method and the actual
record is calculated as the absolute value of the actual proportion minus the calculated
proportion. The proportions calculated with the bout method are less different from the
actual proportions than is either time sampling method. Bottom. Comparison of the
proportions of shot attempts during the whole season compared with those
from the sampling period

ABS ABS ABS

Player Actual Bout (Actual-Bout) ITS (Actual-I TS) 2TS (Actual-2TS)
N 867 415 41 16
Brown 1.4 1.7 0.3 0.0 1.4 0.0 1.4
Buechler 2.4 2.9 0.5 2.4 0.0 6.3 3.8
Caffey 0.6 0.7 0.1 0.0 0.6 0.0 0.6
Edwards 0.5 1.2 0.7 0.0 0.5 0.0 0.5
Harper 7.3 6.5 0.8 4.9 2.4 0.0 7.3
Jordan 29.8 27.0 2.8 41.5 11.7 31.3 1.5
Kerr 6.1 7.7 1.6 4.9 1.2 6.3 0.1
Kukoc 10.7 12.5 1.8 14.6 3.9 6.3 4.5
Longley 9.5 8.2 1.3 4.9 4.6 6.3 3.2
Pippen 22.4 22.4 0.0 24.4 2.0 37.5 15.1
Rodman 4.6 4.1 0.5 0.0 4.6 0.0 4.6
Simpkins 2.0 2.7 0.7 0.0 2.0 0.0 2.0
Wennington 2.9 2.4 0.5 2.4 0.4 6.3 3.4
Sum ABS(actual-sample) 11.6 35.3 47.9
Mean difJ. 0.89 2.71 3.68
Range 0-2.8 0-11.7 0.1-15.1

ABS ABS ABS ABS

Player Season (Season-Actual) (Season-Bout) (Season-I TS) (Season-2TS)
N 6997
Brown 2.7 1.4 1.1 2.7 2.7
Buechler 3.5 1.0 0.6 1.0 2.8
Caffey 2.3 1.7 1.6 2.3 2.3
Edwards 1.6 1.1 0.4 1.6 1.6
Harper 7.2 0.1 0.7 2.3 7.2
Jordan 26.4 3.3 0.5 15.0 4.8
Kerr 6.9 0.8 0.8 2.0 0.6
Kukoc 11.2 0.5 1.3 3.4 5.0
Longley 7.2 2.3 1.0 2.3 0.9
Pippen 17.4 5.0 5.0 7.0 20.1
Rodman 4.3 0.3 0.2 4.3 4.3
Simpkins 2.3 0.3 0.4 2.3 2.3
Wennington 4.9 2.0 2.5 2.5 1.3
Other 2.1 2.1 2.1 2.1 2.1
Sum ABS(actual-sample) 22.0 18.1 50.8 58.1
Mean difJ. 1.57 1.3 3.63 4.15
Range 0.1-5.0 0.2-5.0 1.0-15.0 0.9-20.1
Methodological Issues in Studying Positional Behavior 15

comparisons of proportions of time spent were not made, nor was the variance in propor-
tions among the trials reported.
In any case, these are probably minimum estimates. If the data are going to be sub-
categorized by age-sex groups, habitat, season, substrate context, activity context, etc.,
even more observations are necessary to ensure that each subcategory is adequately sam-
pled. The number of events recorded is, however, not the only important factor. Re-
searchers also need to consider the number of behavioral categories being scored, number
of days over which the data are collected (since daily variation in behavior can be enor-
mous--[Garber and Preutz, 1995]), and also the number of different individuals on which
observations are made.
It is distressingly difficult to extract information on number of observations or con-
tact hours from publications because of the variety of ways in which it is reported. Of the
studies from which we were able to determine this, only 55% are based on more than 1000
observations of locomotor behavior, and only 30% on more than 2000. Of the studies that
report the number of contact hours, only 33% are based on more than 250 hours, and 27%
are based on less than 100 hours of observation time.

3.2. Data Analysis and Statistics

In most studies all observations are lumped because they cannot be attributed to par-
ticular individuals. We see several problems with this typical approach. Sequential obser-
vations of individuals can lead to a data set in which temporal autocorrelation (when the
probability of occurrence of one event is affected by the previous event) is a problem
(e.g., Mendel, 1976). One approach to dealing with this is to have a long time interval be-
tween observations; this is the usual rationale for time sampling (but the longer the inter-
val, the longer it will take to get a reasonable sample of locomotor behaviors). Rarely,
though, has anyone tested their data to see if significant temporal autocorrelation still ex-
ists. There are statistical approaches to account for temporal autocorrelation (e.g., Altham,
1979), but these techniques have not been applied to positional behavior.
Even if several hundred or thousand temporally non-correlated observations are col-
lected, in reality these are attributable to only a small number of individuals, and thus are
still autocorrelated. We maintain that the proper unit of analysis in studies of positional
behavior is the individual animal; the multiple observations (bouts or time samples) are
crucial in that they contribute to the accuracy and precision with which the behavior of
each individual is measured, but they do not increase the number of degrees of freedom
for a statistical test. The correct sample size for determining the P value of such a data set
is thus the number of individuals studied, not the number of observations. An incorrectly
inflated sample size will result in much lower P values, more instances of statistical sig-
nificance, and therefore greater possibility of Type I errors (finding significance where
none exists) (Hurlbert, 1984; Machlis et ai., 1985). A comparable problem in morphologi-
cal studies has been the topic of much discussion (e.g., Felsenstein, 1985; Cheverud et ai.,
1989; Harvey and Pagel, 1991; Smith, 1994).
To demonstrate this problem, Dagosto (1994) analyzed data in two ways: (1) lump-
ing all observations (LAO) to calculate total species proportions and testing between spe-
cies differences with a Chi-square test versus (2) lumping the observations and calculating
proportions for each individual (IND) and testing between species differences with ordi-
nary parametric and randomization tests. For the same data LAO (N= number of observa-
tions, thousands) almost always gave a significant result while IND (N= number of
individuals studied, 12-20) almost always gave a nonsignificant result. Although there is
16 M. Dagosto and D. L. Gebo

little difference in species' "central tendency" statistics calculated by either method (cal-
culating species proportions after lumping all observations versus calculating proportions
for each individual and taking the average; see Table 3), the results of statistical tests are
greatly affected by method.
The species-wide lumping approach (LAO) has additional statistical problems. This
procedure yields a suite of categories that must be tested as a whole with a Chi-square
test, G test, or a rank order correlation; these tests do not generally allow comparison of
each category (e.g., leaping, climbing) separately. If the test is significant, most re-
searchers argue that one category or another contributes most to the total difference by
looking at raw differences in frequency or the contribution of the deviation in that cate-
gory to the Chi-square statistic, but the test itself does not bear directly on this question.
This could be accomplished by appropriately compressing categories (e.g., leaping versus
not leaping).
Assuming that individual animals are the proper units of analysis, the Chi-square
and G tests are invalid for reasons outlined by Hurlbert (1984). The pooling of data from
different individuals, which is necessary for the tests, results in: (I) loss of independence,
thus violating the underlying assumption of the tests; (2) loss of information on variability
among the individuals in each group; and, (3) weighted averages being compared if the
number of observations for each individual is different. In LAO, species are characterized
with a single number (presumably some sort of measure of central tendency), but there is
no way to estimate dispersion around that number (variance). We seem to treat behavioral
repertoires as properties of a species, rather than as an epiphenomenon of the behavior of
the individuals of the species. What the LAO approach actually measures is something
like "of all locomotor events taking place in this species, X% are leaping", when what we
really ought to measure is "how often does each individual leap", and construct an esti-
mate of species' central tendency and dispersion from a sample of individuals. The two
approaches will converge only if (I) there is no intraspecific variation in behavior or (2) if
there is intraspecific variation, no individual contributes more observations to the sample
than any other. In any case, the lack of any way to measure dispersion in LAO makes it
difficult to make interspecific (or any intergroup) comparisons as in quantitative morphol-
ogy, where interspecific variation is interpreted in the context of intraspecific variation.
If observations can be associated with individuals (lND), then proportions of behaviors
can be calculated for individuals thus solving these problems. Individual values for categories
(e.g., leaping, climbing) can be tested independently of other categories and a measure of cen-
tral tendency and the variation around it can be calculated with standard statistics (Boinski,
1989; Fontaine, 1990; Dagosto, 1994). Interspecific comparisons of behavior can proceed
much like morphological comparisons, using standard parametric, nonparametric, or ran-
domization approaches for continuous quantitative variables. We appreciate that access to in-
dividually identifiable study subjects is often practically difficult. Other approaches to
lumping data are suggested in Dagosto (1994) and Gebo and Chapman (1 995a).
An emphasis on individual behavior rather than species-wide measures is important
in other areas of endeavor within evolutionary morphology. The research strategy for
studying selection and adaptation outlined by Arnold (1983), for example, relies critically
on determining the relationship between performance and morphology of individuals.

3.3. Categories
Because data on positional behavior are always presented as proportions, the num-
bers arrived at in any study depend entirely on the number of categories used and how dif-
Methodological Issues in Studying Positional Behavior 17

ferent subcategories are lumped. This greatly affects comparability among studies. There
are two important questions regarding categorization of positional behavior. The first con-
cerns definitions: the names given to different kinds of behavior. We will not deal with
this here, suffice it to say that a common language for modes of behavior would be ex-
tremely useful (Hunt et aI., 1996). There is simply no point in comparing frequencies of
climbing among studies if climbing is defined in a different way by each investigator.
Similar confusion would arise in morphology if each researcher called the bones by differ-
ent names or defined the same measurements using different landmarks.
What constitutes a useful categorization of behavior is another important question.
For morphology-behavior associations, we presumably desire categories that capture kine-
matically meaningful differences in behavior having direct relationships with underlying
anatomical structures. This is another area where the ideal (using enough categories to cap-
ture as detailed depictions of the kinematic differences as possible) runs into practical diffi-
culties (what can actually be seen in the field; Rose, 1979; Rosenberger and Stafford, 1994).
For example, we have no doubt that what we call "leaping" in Propithecus and Eulemur is
kinematically different, or that within each species there are several kinds of kinematically
different behaviors we call leaping (some of these are discussed in Oxnard, 1984). Although
we recognize that several types of leaping occur, we are forced to lump these behaviors
when calculating frequency, because in our experience it is tremendously difficult to reli-
ably, regularly recognize these subtly (or even not so subtly) different modes, especially for
quickly occurring locomotor behaviors (e.g., did the animal land feet first, hands first, or
with all limbs simultaneously; was there an aerial phase in a quadrupedal sequence, etc.).
Thus, there will always be a certain level of unavoidable generality to an observational
study, which may compromise its usefulness to morphologists. If such subtle, yet important
kinematic characterizations are the question of interest, a significant amount of videotaping
is necessary (Rosenberger and Stafford, 1994; Terranova, 1995; Demes et aI., 1995, 1996;
but see Fontaine, 1990). Obtaining frequencies of occurrence of such categories based on
large numbers of videotaped field observations is likely to be a very demanding enterprise.
The ability to accurately apprehend subtle differences in body position is not as much of a
problem for behaviors ofiong duration, like most postures.

4. INTRASPECIFIC VARIATION

A comparison in which the data consists of observations of one species collected at

one time in one locality and of another species collected at a different time at a different
locality by another observer is a poor experimental design. One may use inferential statis-
tics and believe that one is testing for a species' effect, but in fact a number of other ef-
fects, including those from habitat structure, seasonal differences in behavior,
interobserver 'error', degree of habituation, presence and density of predators, age/sex
composition of the groups, etc. could influence the result. Such comparisons do not con-
trol for either these or for stochastic effects, and thus must be interpreted cautiously
(James, 1982; Simberloff, 1982; Hurlbert, 1984; Garland and Adolph, 1994). In accepting
the results of such "tests" or "natural experiments" we are making the assumption that
these other effects are negligible compared to the species' effect. Depending on the scale
of the question being addressed or the magnitude of the observed difference, such an as-
sumption may be a perfectly valid course of action. Nevertheless, the rather large assump-
tion of negligibility has not yet been adequately tested. It likely will be impossible to
 18 M. Dagosto and D. L. Gebo

control for all of these variables in any field study, however, we can at least attempt to as-
sess the effect of these factors in a few cases.
In the section above it was argued that interspecific comparisons are stronger when
they can be made in the light of intraspecific variation. Unfortunately, we have as yet only
a poor idea of the sources of variation in positional behavior within species and the extent
of their effects. Intraspecific variation due to age and sex have been documented in many
primates (Ripley 1967; Sugardjito and van Hooff, 1982; Crompton 1983; Cant, 1987a,b;
Boinski, 1989; Doran 1992b, 1993b; Hunt, 1994; Remis, 1995). Simple variation in be-
havior among adult individuals of the same species has rarely been assessed, however,
largely because observations are not associated with particular individuals. Positional be-
havior is known to alter with activity context (e.g., feeding, traveling, resting); studies
vary in what contexts behavior is measured and reported. In addition, variation in posi-
tional behavior that is associated with differences in habitat or season has only begun to be
documented. Understanding the causes and extent of intraspecific variation is critical for
assessing the meaning (and statistical significance) of interspecific variation. If significant
intraspecific variation exists, research designs must be constructed carefully in order to ar-
rive at accurate summary values for species. In addition, intraspecific variation is often the
key to understanding interspecific variation: "the analysis of population variation should
become a powerful weapon in the morphologist's arsenal" (Arnold, 1983: 348; see also
James, 1982; Bradshaw, 1987).
In a field study we can only measure performance, not potential-we can only see
what animals do, not necessarily all they are capable of doing (Rose, 1979; Morbeck,
1979; Gomberg et aI., 1979). An important question for the study of positional behavior is
"Does performance change due to immediate circumstances of habitat or season?" By ne-
cessity most studies of positional behavior are limited in time and space. We do not ques-
tion here whether a researcher's data accurately measures what happened in that short
period of time, but rather ask if what happened in that short period of time or in that single
place accurately characterizes long term behavior or species-wide potential.
Using our Bulls example, we compared how well our 10 game sample (collected
during December and January) reflected what happened during the total 82 game season
(October-April). Even though the Bulls had a remarkably consistent season, winning a re-
cord 72 of 82 games and with no significant player injuries, Table 5 shows that the sam-
pling methods did a better job predicting what happened in the sample games than in the
overall season. Similarly, in our studies of positional behavior, we have both found signifi-
cant intraspecific variation in frequencies of behavior associated with differences in sea-
son and habitat. Other studies have also discussed seasonal or habitat related differences
in positional behavior and substrate use, but have reached conflicting conclusions about
its presence and importance (Boinski, 1989; Crompton, 1984; Doran and Hunt, 1994; Gar-
ber and Preutz, 1995; McGraw, 1996; Remis, this volume).

4.1. Habitat

That habitat architecture can impose constraints on the positional behavior of pri-
mates has been argued by several workers, most notably Ripley (1967, 1977, 1979). Cant
(1992, p. 277) has also voiced the caution that field studies of positional behavior should
aim to control for habitat structure "in order to avoid the possibility that behavioral differ-
ences between species are artifacts of observing them in different structural contexts" (see
also Pounds, 1988, 1991).
Methodological Issues in Studying Positional Behavior 19

We approached this question by studying the same species in different habitats in

two different primate groups, which minimizes problems of interobserver differences in
data collection techniques. The following comparisons are based on overall frequencies of
locomotor behavior (all activity contexts included). Gebo and Chapman (1995b) docu-
mented the behavior of the red colobus monkey (Colobus badius) in the Kibale Forest,
Uganda. Dagosto (1995; Dagosto and Yamashita, in press) studied four species of Mala-
gasy primates, Propithecus diadema. Eulemur rubriventer, Eulemurfulvus, and Varecia
variegata at Ranomafana National Park, Madagascar.
Colobus badius was observed in primary, secondary, and pine forests. These forests
differ in two structural attributes: the size of gaps to the nearest tree, which was largest in
the secondary forest and least in the pine forest, and continuity, which measures the per-
cent of trees for which there is no gap to the nearest tree. This is largest in the pine forest
and lowest in the secondary forest. The locomotor behavior of the red colobus monkey is
significantly different in these habitats (Table 6). Leaping and quadrupedal ism are least
frequent in the primary forest and most frequent in the secondary forest. Climbing was
least frequent in the secondary forest.
At Ranomafana National Park, Eulemur and Propithecus were observed in two dif-
ferent areas, Vatoharanana and Talatakely. The Vatoharanana forest has a higher percent-
age of large trees (as measured by height, diameter at breast height, or crown diameter),
and a higher percentage of large gaps between the crowns of trees (Dagosto and
Yamashita, in press; White et aI., 1995) compared to Talatakely. All three species showed
a similar response to this habitat difference; leaping is less frequent and quadrupedal ism
and climbing more frequent at Vatoharanana. The degree of response differed, being least
in Propithecus and most in Eulemur fulvus. The difference in the species behaviors be-
tween forests may be mediated through activity budget. All species spent more time feed-
ing/foraging and less time traveling at Vatoharanana, presumably because of the larger
food patch size available there. Since leaping is more common in travel than during feed-
ing in most species of primates so far studied (Fleagle and Mittermeier, 1980; Gebo and
Chapman,1995a; Dagosto and Yamashita, in press), increasing the amount of time spent
traveling, as at Talatakely, increases the overall frequency of leaping at this site.

4.2. Season
One might expect that as the kinds of foods exploited change during the year, micro-
habitat selection or activity pattern might also change, possibly resulting in differences in
positional behavior. For the red colobus monkeys at Kibale Forest, the seasonal compari-
son was made in the primary forest only. In the dry season quadrupedal ism is more fre-
quent, and leaping and climbing less frequent (Table 6). This is associated with increased
use of the middle canopy in the wet season (Gebo and Chapman, 1995b).
Among the lemurs of Ranomafana, Propithecus is the only species that shows no
significant difference in positional behavior by season. The three lemur species (E. fulvus.
E. rubriventer, and V. variegata) show a similar pattern: leaping is more frequent and
quadrupedal ism less frequent in the dry season than in the wet season. In this case, we
were unable to identify a causal factor. Activity pattern does not differ much between sea-
sons except in Varecia, and in this species it did not predict the observed difference in lo-
comotor behavior, since leaping is more frequent in the dry season, when travel is actually
less frequent. Although there are seasonal changes in food species used in all the lemurs,
this is only sometimes correlated with changes in microhabitat as estimated by the struc-
 ....=>

Table 6. Habitat and seasonal variation in proportions of loco~otor behavior during all activities (using bouts) in Colobus badius and Malagasy lemurs

Colobus badius Eulemur fulvus Eulemur rubriventer Propithecus diadema Varecia variegata
Habitat Primary Secondary Pine P Yato Talata P Yato Talata P Yato Talata P
Leap 28.8 33.8 34.4 ** 44.2 67.7 ** 58.6 64.3 ns 85.2 88.8 **
Quad. 27.3 38.2 31.3 ** 35.6 22.2 ns 23.8 22.5 ns 1.9 0.6 ns
Climb 36.4 22.0 26.6 ** 17.1 8.6 ** 15.1 11.7 ns 9.4 8.8 **
Other 7.6 5.9 7.8 ns 3.1 3.0 ns 3.7 1.8 ns 2.3 1.8 ns

Season Wet Dry P Wet Dry P Wet Dry P Wet Dry P Wet Dry P
Leap 31.2 27.0 ns 62.7 73.3 ** 59.7 69.3 * 89.1 88.6 ns 48.0 55.5 ns
Quad. 26.6 33.7 ** 25.8 18.9 ** 27.3 18.0 * 0.5 0.8 ns 44.0 29.0 **
Climb 36.0 33.6 ns 8.5 7.1 ns 11.3 12.7 ns 8.6 8.9 ns 6.5 12.5 **
Other 6.2 5.7 ns 1.9 0.7 * 2.0 0.7 ns 1.7 1.7 ns 0.0 0.0 ns
P gives the statistical significance of a test of the differences in calculated proportions between the habitats or seasons; *. P= 1.0-.05, **P<0.05, ns = not significant. The Colobus data were tested us-
ing ANOVA, and the lemur data were tested using the two-sample test of Manly (1991). Data from Gebo and Chapman, 1995b; Dagosto, 1995; and Dagosto and Yamashita (in press).

~
1:1
co
~
g
co
Q.
=
!='
r
C')
to
go
Methodological Issues in Studying Positional Behavior 21

ture of the food trees; variation in structural attributes of resource trees does not correlate
in a regular manner with variation in positional behavior (Dagosto, 1995).

4.3. Comparison of Studies

All five species that we studied show some statistically significant variation in the
frequency of use of locomotor behaviors associated with different habitat, season or both.
Variation due to season is generally less often significant and less in magnitude than that
due to habitat. In both the Colobus and lemur studies, locomotion during travel is more
conservative (less likely to show seasonal or habitat differences) than locomotion during
feeding.
These examples indicate that studies limited in time or place may not fully capture
the expressed range of behavior of a species and could compromise comparisons between
species. For example, whether Eulemur fulvus or E. rubriventer leaps most does depend
on the site at which observations are made: at Vatoharanana E. rubriventer leaps signifi-
cantly (P<O.05) more than E. fulvus; at Talatakely, E. fulvus leaps more than E. rubriven-
ter; if data from both sites are lumped, they are indistinguishable in leaping frequency
(Dagosto, 1994). Propithecus leaps more (P<O.OO 1) than either Eulemur species regardless
of site or season although the magnitude of the difference is much less at some times and
in some places than in others (Table 6). The scale of the expected or observed differences
in behavior is important. Comparisons of closely related or behaviorally similar species
will have to be designed very carefully and be fairly exhaustive before secure conclusions
can be attained; we cannot immediately accept that the 'species effect' outweighs other
factors. In comparisons of behaviorally quite different species we may be able to relax our
skepticism, although we still might not be able to determine how much of the difference is
the result of ultimate (i.e., morphology) or proximate (habitat, season) causes.
Our results may appear to be contradictory to the conclusions reached by Garber and
Preutz (1995) in their study of Saguinus mystax in which they found no differences in the
rank order of positional behaviors despite marked structural differences in the two habitats
in which the behaviors were observed. They concluded from these data and from other
studies that positional behavior is conservative to differences in habitat. In fact, our results
are quite similar. In Colobus badius and in the Malagasy lemurs, it is also evident that
there are no radical changes in behavior; the most common behaviors are still the most
common, and rare behaviors are still rare--each species is constrained to some extent by
anatomical design. Among lemurs, as in Saguinus, the rank order of behaviors does not
differ by habitat or season, but in Colobus badius, there is a significant difference. The lat-
ter case is not surprising, nor particularly meaningful, since the three most common be-
haviors each contribute about 1/3 to the total behavioral repertoire.
Demonstrating a lack of difference in the rank order of proportions of an entire suite
of behaviors answers a different question than the one we asked, which is whether there
are meaningful contrasts in the frequency of occurrence of specific types of behaviors be-
tween habitats or seasons. We are able to demonstrate statistically significant differences
among behavioral differences and habitat, microhabitat, activity pattern, and diet and be-
lieve that these associations are crucial to understanding the transformations in behavior
and morphology among species. Of the 9 studies cited by Garber and Preutz (1995: Table
6) to indicate consistent positional repertoires in different habitats, we would interpret at
least 6 (studies on S. sciureus, C. capucinus, A. palliata, A. geoffroyi, C. guereza, and S.
mystax) to show potentially important differences in frequencies of specific types of be-
havior (although differences in methods makes direct comparisons among studies quite
22 M. Dagosto and D. L. Gebo

difficult). For example, in the Rio Blanco habitat, the frequency of quadrupedal walking
in S. mystax is 23.1% and at Padre Isla it is 35.4%, a difference that exceeds in magnitude
many of the statistically significant differences in frequency we found in our studies. In
addition, Garber and Preutz's analysis is based on positional behavior during travel only,
the context in which we found behavior to be the most conservative.
The primary difference between studies that have found no habitat differences and
those that have is a matter of the scale at which the question is asked. On a large scale
(rank order of behaviors, whole suites of behaviors compared with the G test) few differ-
ences are apparent (Garber and Preutz, 1995; McGraw, 1996; Doran and Hunt, 1994). At a
smaller scale (the frequency of occurrence of particular behaviors) many distinctions are
evident (Gebo and Chapman, 1995b; Dagosto, 1995). Morphologists must be careful to
use behavioral data and statistical tests that are appropriate to the scale of the anatomical
question being investigated. For example, Doran and Hunt (1994) find no large scale dif-
ference in the repertoire of positional behavior of chimpanzees in different habitats. There
is, however, a significant difference in the time spent in arm hanging and climbing be-
tween rainforest and woodland chimpanzees, which influences explanations for the evolu-
tion of chimp shoulder morphology (Doran, 1996).
McGraw (1996) cites other possible explanations for the lack of consensus on this
issue, including the likelihood that studies differ in their degree of habitat contrast. As
with positional behavior, there is no standard way of measuring or testing habitat differ-
ences. All investigators do not measure the same aspects of habitat structure and, since we
do not as yet understand the causal relationships among properties of habitat structure and
positional behavior, we do not know which of the variables are most important to the pri-
mates. Nonstructural variables, such as resource spacing, timing, and diversity, may also
be important, yet are rarely measured (Remis, 1995; Dagosto and Yamashita, in press). Fi-
nally, species may vary in their ability to respond to habitat or seasonal differences, thus
there may well be no generalizations about the effect of such differences.
All of these studies, including our own, suffer from the problem of 'pseudoreplica-
tion' (Hurlbert, 1994), since they consist of data collected during only one or two repli-
cates of the contrasted seasons or habitats, and thus do not provide an estimate of
variability within each season/habitat with which to assess the differences between them.
In such contrasts, stochastic effects are not controlled for, and statistical inferences are not
warranted, or are at least subject to the same sorts of assumptions outlined at the begin-
ning of this section. Better designs, perhaps including controlled experiments (e.g.,
Pounds, 1988), are needed to adequately address this issue.

5. CONCLUSION

Many studies of positional behavior are taken up with the aim of providing data use-
ful for integration with studies of morphology. At the 1965 conference, Ripley pointed out
problems with how the behavioral data then available was being used and outlined an ap-
proach to the study of positional behavior that would provide a sounder basis for these
studies. In 1965 the fundamental problem was lack of data. This has been somewhat al1e-
viated (although we would argue that we still have a long way to go), but the increase in
the number of studies of positional behavior has produced a new set of problems related to
comparability of results. This lack of comparability is likely to frustrate anyone attempting
a broadly comparative synthesis of primate locomotion. Perhaps this is why the increase in
the amount of data in the last thirty years has not resulted in any increase in synthetic
Methodological Issues in Studying Positional Behavior 23

studies. The ones that exist are limited in taxonomic scope (e.g., Fleagle and Mittermeier,
1980; Gebo and Chapman, 1995a: in both cases all the behavioral data were collected by
the same investigators) or are forced to translate qualitative and quantitative data into rela-
tive scales (Crompton et aI., 1987; Oxnard et aI., 1990). Two-species comparisons have
been successfully used to elucidate skeletal adaptations to positional behavior, in part be-
cause the behavioral data are again internally consistent since they are often collected by
the same investigator (e.g., Fleagle, I 977a,b ). However, these types of studies have their
limitations (Fleagle, 1979; Oxnard, 1979; Bradshaw, 1987; Garland and Adolph, 1994).
Broader based comparative studies are thought to provide stronger evidence for adaptation
by demonstrating multiple occurrences of evolutionary convergence (Harvey and Pagel,
1991). Since it is very unlikely that many species will be studied by the same investigator,
the issue of comparability of both the behavioral and anatomical data is an important con-
cern.
As morphologists, there are several kinds of information we would like to know
about the positional behavior of any species in order to explain form and differences in
form:
I. Repertoire: What kinds of behaviors/substrates are used? It may be possible to
obtain adequate information from captive animals. Quantification is not re-
quired, but how behaviors are defined (categorization) is critical.
2. Frequency/Time Spent: How often do kinds of events occur/how much time is
spent performing them? Answering these questions requires quantitative data.
The results may be sensitive to substrate/habitat context, so it is preferable to
conduct a field, rather than captive, study (e.g., compare results of Gebo, 1987
with Dagosto, 1994).
3. Context: To understand the impetus for the origin of different behaviors it is im-
portant to document the associations of positional behaviors with habitat types,
substrate use, season, diet, age, gender: these are the elements of Ripley's study
of total locomotor pattern. Like frequency, these aspects of positional behavior
are also best examined with quantitative data, and should be studied in the field
rather than in captivity if biological role (actual use, Bock and von Wahlert,
1965) rather than function (potential use) is the object of investigation.
4. Kinematics/ Biomechanics: How are the behaviors performed; what are the joint
angles, excursions, forces acting on the bones and joints? Some aspects of these
can be studied in the field if fairly broad categories are used, but filming (in the
field or in captivity) is necessary if more subtle distinctions are required, or if
events occur too quickly to be reliably apprehended by eye.
5. Energetics: How much energy is required to move a certain distance using a lo-
comotor behavior, or to maintain a particular posture? These can only be ad-
dressed in a laboratory setting using experimental techniques.
Field studies are only valuable for 1, 2 and 3; 4 and 5 are best studied in a captive or
laboratory setting. Frequency and context, however, can only be reliably assessed from
field data, and to be most useful, they must involve the collection of quantitative data.
Morphologists could articulate more clearly which of these types of data would be most
useful for answering the particular questions they are most interested in.
The scale of the question being asked must also be considered. General categoriza-
tions of behavior are largely based simply on repertoire, frequency is only involved at a
very low level of specificity-i.e., which kind of behavior is used the most. If relation-
ships between such general differences in behavior and morphology are all that is desired
24 M. Dagosto and D. L. Gebo

or all that we can reasonably hope to demonstrate given problems with executing adequate
experimental designs in a field situation, then one might very well ask if quantitative data
are necessary at all, and if they are not, then most of our concerns about comparability
among studies or intraspecific variation are largely irrelevant. For example, despite the
large number of quantitative studies since 1965, the basic categorization of primate loco-
motor behavior outlined by Napier and Napier (1967) and Napier and Walker (1967) has
generally survived intact, and is the basis for most anatomical and paleontological studies.
Most investigators (e.g., Oxnard, 1974) have, however, expressed some dissatisfaction
with the generality of this scheme. At such general levels we run the risk of reducing the
complexity of behavior to overly simple variables, thus missing the opportunity to investi-
gate differences in behavior and anatomy that exist within broadly defined locomotor
groups. Bock (1990) gives examples in which the oversimplification of avian morphology
by ecologists compromised the results of their ecomorphological analyses. We do not want
to similarly misrepresent behavior with the same unfortunate result.
If the establishment of more specific relationships between morphology and behav-
ior are desired, then questions about differences in performance other than repertoire be-
come more important. One might suspect that differences in frequency/time spent on
behaviors or in substrate milieu are responsible for morphological differentiation (e.g.,
Fleagle 1977a,b; Fleagle and Meldrum, 1988; Rodman, 1979; Gebo and Sargis, 1994).
Difference in frequency/time spent is the hypothesis most often proposed to explain ana-
tomical differences among fossil species, as well. Perhaps one believes that the kinematics
of the' same' behavior are different (e.g., Demes et aI., 1996), or that the energetic cost of
the same behavior is different. These more specific questions (especially those dealing
with frequency/time spent) generally require quantitative data and statistical approaches,
and thus the issue of how well the numbers are estimated becomes crucial. Simply put, if
the actual numbers are important for the question one is asking, then it is important that
we obtain good estimates of the actual numbers. Quantification without the accuracy, pre-
cision, and applicability of statistical inference gained by good experimental design is of
no more value than a qualitative description and may potentially be more misleading.
Our point is that most current data on positional behavior are collected under re-
search designs that are severely limited by the practicalities of the field situation. This
needs to be taken into account when interpreting the numbers obtained. All estimates have
associated error. Our concerns are that (l) the most commonly used method of summariz-
ing positional data does not lend itself to the calculation of error, and (2) because of the
inadequacies of experimental design the amount of error may be larger than we realize;
and, depending on the scale of the question, could have consequences for derivative stud-
ies based on these data.
Quantitative studies raise the issue of comparability of numbers (but even qualita-
tive descriptions are often difficult to compare). We have identified several factors that af-
fect interstudy comparability: (I) the definition of behaviors and the categorization
scheme used; (2) the method of data collection (bouts/time samples); (3) how data are
summarized and tested (IND versus LAO; comparisons of whole repertoires (G test, rank
correlations) versus specific behaviors); and, (4) sampling composition such as the
age/sex composition of the sample, the number of different seasons and habitats sampled,
and the behavioral contexts (e.g., traveling/feeding/resting; arboreal/terrestrial) sampled.
We are not implying that there is only one way of doing a positional behavior study.
The appropriate way to do any study depends entirely on the questions one wishes to ask.
Field studies are time consuming and expensive, therefore most researchers desire that the
results of their hard work will be useful beyond their own immediate goals. We are only
Methodological Issues in Studying Positional Behavior 25

pointing out that for this purpose, some concern for comparability of results among studies
should enter into decisions about data collection techniques, so that we can someday
achieve the goals set by Suzanne Ripley and Warren Kinzey in 1965.

ACKNOWLEDGMENTS

We would like to thank the organizers of the symposium (E. Strasser, A. Rosenber-
ger, J. Fleagle, and H. McHenry) for inviting us to contribute to this volume. Funding for
our positional behavior projects has come from The LSB Leakey Foundation, Northern Il-
linois University (DLG), the National Science Foundation and Northwestern University
(MD). We especially thank w.G. Kinzey for his advice and support of our projects. We
thank the reviewers for their helpful comments. We would also like to express our appre-
ciation to all of the contributors to the original conference for producing the stimulating
papers that have so heavily influenced the field of primate locomotion.

REFERENCES
Altham PME (1979) Detecting relationships between categorical variables observe over time: a problem of deflat-
ing a Chi-squared statistic. Appl. Stat. 28: 115-25.
Altmann J (1974) Observational study of behavior: Sampling Methods. Behaviour 49:227--67.
Arnold SJ (1983) Morphology, performance, fitness. Amer. Zool. 23:347--61.
Ashton EH, and Oxnard CE (1964) Locomotor patterns in primates. Proc. Zool. Soc. Lond, 142: 1-28.
Avis V (1962) Brachiation: the crucial issue forman's ancestry. Southwestern J. Anthro. 18:119--48.
Bicca-Marques JC, and Calegaro-Marques C (1993) Feeding postures in the black howler monkey, Alouatta
caraya. Folia Primatol. 60:169-72.
Bicca-Marques JC, and Calegaro-Marques C (1995) Locomotion of black howlers in a habitat with discontinuous
canopy. Folia Primatol. 64:55--61.
Bock WJ (1990) From Biologische Anatomie to ecomorphology. Neth. J. Zool. 40:254-77.
Bock WJ, and von Wahlert G (1965) Adaptation and the form-function complex. Evol. 19:269-99.
Boinski S (1989) The positional behavior and substrate use of squirrel monkeys: ecological implications. J. Hum.
Evol. 18:659-77.
Bradshaw AD (1987) Comparison--its scope and limits. New Phytol. 106:3-21.
Cannon CH, and Leighton M (1994) Comparative locomotor ecology of gibbons and macaques: Selection of can-
opy elements for crossing gaps. Am. 1. Phys. Anthropol. 93:505-24.
Cant JGH (1986) Locomotion and feeding postures of spider and howling monkeys: Field study and evolutionary
interpretation. Folia Primatol. 46:1-14.
Cant JGH (1987a) Effects of sexual dimorphism in body size on feeding postural behavior of Sumatran orangutans
(Pongo pygmaeus). Am. J. Phys. Anthropol. 74:143-8.
Cant JGH (1987b) Positional behavior of female Bornean orangutans (Pongo pygmaeus). Am. 1. Primatol.
12:71-90.
Cant JGH (1988) Positional behavior of long-tailed macaques (Macacajascicu1aris) in Northern Sumatra. Am. J.
Phys. Anthropol. 76:29-37.
Cant JGH (1992) Positional behavior and body size of arboreal primates: A theoretical framework for field studies
and an illustration of its application. Am. J. Phys. Anthropol. 88:273-83.
Charles-Dominique P (1977) Ecology and behavior of nocturnal primates. New York: Columbia University Press.
Cheverud JM, Wagner GP, and Dow MM (1989) Methods for the comparative analysis of variation patterns. Syst.
Zool. 38:201-\3.
Crompton RH (1983) Age differences in locomotion in the subtropical Galaginae. Primates 24:24-59.
Crompton RH (1984) Foraging, habitat structure, and locomotion in two species of Galago. In P Rodman and J
Cant (eds.): Adaptations for Foraging in Nonhuman Primates. New York: Columbia University Press, pp.
73-111.
Crompton RH, and Andau PM (1986) Locomotion and habitat utilization in free-ranging Tarsius bancanus: a pre-
liminary report. Primates 27:337-55.
 26 M. Dagosto and D. L. Gebo

Crompton RH, Lieberman SS, and Oxnard CE (1987) Morphometrics and niche metrics in prosimian locomotion:
an approach to measuring locomotion, habitat, and diet. Am. J. Phys. Anthropol. 73:149--78.
Curtis D (1992) Substrate use in captive aye-ayes Daubentonia madagascariensis. 1. Wildl. Pres. Trust 28:30-44.
Curtis DJ, and Feistner ATC (1994) Positional behavior in captive aye-ayes (Daubentonia madagascariensis). Fo-
lia Primatol. 62:155-9.
Dagosto M (1994) Testing positional behavior of Malagasy lemurs. A randomization approach. Am. J. Phys. An-
thropol. 94: 189--202.
Dagosto M (1995) Seasonal variation in positional behavior of Malagasy lemurs. Int. J. Primatol. 16:807-33.
Dagosto M, and Yamashita N (in press) Effect of habitat structure on positional behavior and support use in three
species of lemur. Primates.
DaSilva GL (1993) Postural changes and behavioral thermoregulation in Colobus polykomos: the effect of climate
and diet. Afr. J. Ecol. 31 :226-41.
Demes B, Jungers WL, Gross TS, and Fleagle lG (1995) Kinetics of leaping primates: Influence of substrate ori-
entation and compliance. Am. J. Phys. Anthropol. 96:419--30.
Demes B, Jungers W, Fleagle J, Wunderlich R, Richmond B, and Lemelin P (1996) Body size and leaping kine-
matics in Malagasy vertical clingers and leapers. 1. Hum. Evol. 31:366-388.
Doran DM (1992a) Comparison of instantaneous and locomotor bout sampling methods: A case study of adult
male chimpanzee locomotor behavior and substrate use. Am. 1. Phys. Anthropol. 89:85-99.
Doran DM (1992b) The ontogeny of chimpanzee and pygmy chimpanzee locomotor behavior: a case study of pae-
domorphism and its behavioral correlates. 1. Hum. Evol. 23: 139--57.
Doran DM (1993a) Comparative locomotor behavior of chimpanzees and bonobos: The influence of morphology
on locomotion. Am. J. Phys. Anthropol. 91:83--98.
Doran DM (1993b) Sex differences in adult chimpanzee positional behavior: The influence of body size on loco-
motion and posture. Am. J. Phys. Anthropol. 91:99--116.
Doran DM (1996) Comparative positional behavior ofthe African apes. In W McGrew, L Marchant and T Nishida
(eds.): Great Ape Societies. Cambridge: Cambridge University Press, pp. 213-224.
Doran OM, and Hunt K (1994a) The comparative locomotor behavior of chimpanzees and bonobos: species and
habitat differences. In R Wrangham, W McGrew, F de Waal, and P Heltne (eds.): Chimpanzee Cultures.
Cambridge, MA: Harvard University Press, pp. 93-108.
Dykyj D (1980) Locomotion of the slow loris in a designed substrate context. Am. J. Phys. Anthropol. 52:577-86.
Felsenstein 1 (1985) Phylogenies and the comparative method. Am. Nat. 125: 1-15.
Fleagle JG (1976) Locomotion and posture of the Malayan siamang and implications for hominoid evolution. Fo-
lia Primatol. 26:247-69.
Fleagle JG (I 977a) Locomotor behavior and muscular anatomy of sympatric Malaysian leaf monkeys Presbytis
obscura and Presbytis melalophos. Am. J. Phys. Anthropol. 46:297-308.
Fleagle JG (1977b) Locomotor behavior and skeletal anatomy of sympatric Malaysian leaf monkeys Presbytis ob-
scum and Presbytis melalophos. Yrbk. Phys. Anthro. 20:440-53.
Fleagle JG (1978) Locomotion, posture, and habitat use of two sympatric leaf-monkeys. In G Montgomery (ed.):
Ecology of Arboreal Folivores. Washington: Smithsonian Institution Press, pp. 243--51.
Fleagle JG (1979) Primate positional behavior and anatomy: naturalistic and experimental approaches. In M Mor-
beck, H Preuschoft, and N Gomberg (eds.): Environment, Behavior and Morphology: Dynamic Interac-
tions in Primates. New York: G. Fischer, pp. 313-25.
Fleagle JG (1980) Locomotion and posture. In OJ Chivers (ed.): Malayan Forest Primates. New York: Plenum
Press, pp. 191-207.
Fleagle JG, and Meldrum OJ (1988) Locomotor behavior and skeletal morphology of two sympatric pitheciine
monkeys, Pithecia pithecia and Chiropotes satanas. Am. J. Primatol. 16:227-49.
Fleagle JG, and Mittermeier RA (1980) Locomotor behavior, body size, and comparative ecology of seven Suri-
nam monkeys. Am. 1. Phys. Anthropol. 52:301-14.
Fleagle JG, Mittermeier RA, and Skopec AL (1981) Differential habitat use by Cebus ape/la and Saimiri sciureus
in central Surinam. Primates 22:361-7.
Fontaine R (1990) Positional behavior in Saimiri bolivensis and Ateles geofJroyi. Am. J. Phys. Anthropol.
82:485-508.
Garber PA (1980) Locomotor behavior and feeding ecology of the Panamanian tamarin (Saguinus oedipus geof-
froyi, Callitrichidae, Primates). Int. J. Primatol. 1:185-230.
Garber PA (1984) Use of habitat and positional behavior in a Neotropical primate Saguinus oedipus. In P Rodman
and J Cant (eds.): Adaptations for Foraging in Nonhuman Primates. New York: Columbia University Press,
pp. 112-33.
Garber PA (1991) A comparative study of positional behavior in three species of tamarin monkeys. Primates
32:219--30.
Methodological Issues in Studying Positional Behavior 27

Garber PA, and Preutz JD (1995) Positional behavior in moustached tamarin monkeys: Effects of habitat on loco-
motor variability and locomotor stability. J. Hum. Evo!. 28:411-26.
Garland T, and Adolph SC (1994) Why not to do two-species comparative studies: Limitations on inferring adapta-
tion. Physio. Zoo!. 67:797-828.
Gebo DL (1987) Locomotor diversity in prosimian primates. Am. J. Primato!. 13:271-8 I.
Gebo DL (1992) Locomotor and postural behavior in AIOl/alta palliata and Cebus capucinus. Am. J. Primatol.
26:277-90.
Gebo DL, and Chapman CA (1995a) Positional behavior in five sympatric Old World monkeys. Am. J. Phys. An-
thropo!. 97:49-76.
Gebo DL, and Chapman CA (1995b) Habitat, annual and seasonal effects on positional behavior in red colobus
monkeys. Am. J. Phys. Anthropol. 96:73-82.
Gebo DL, Chapman CA, Chapman LJ, and Lambert J (1994) Locomotor response to predator threat in red colobus
monkeys. Primates 35:219-23.
Gebo DL, and Sargis EJ (1993) Terrestrial adaptations in the postcranial skeletons of guenons. Am. J. Phys. An-
thropo!. 93:341-72.
Gittins SP (1983) Use of the forest canopy by the agile gibbon. Folia Primatol. 40: 134-44.
Glassman DM, and Wells JP (1984) Positional and activity behavior in a captive slow loris: A quantitative assess-
ment. Am. J. Primatol. 7:121-32.
Gomberg N, Morbeck M, and Preuschoft H (1979) Multidisciplinary research in the analysis of primate morphol-
ogy and behavior. In M Morbeck, H Preuschoft, and N Gomberg (eds.): Environment, Behavior, and Mor-
phology: Dynamic Interactions in Primates. New York: Gustav Fischer, pp. 5-21.
Happel R (1982) Ecology of Pithecia hirsuta in Peru. J. Hum. Evol. 11: 581-90.
Harvey P, and Pagel MD (1991) The Comparative Method in Evolutionary Biology. New York: Oxford University
Press.
Hunt KD (1991) Positional behavior in the Hominoidea. Int. J. Primatol. /2:95-118.
Hunt KD (1992) Positional behavior of Pan troglodytes in the Mahale mountains and Gombe Stream National
Parks, Tanzania. Am. J. Phys. Anthropol. 87:83-106.
Hunt KD (1994) Body size effects on vertical climbing among chimpanzees. Int. J. Primato!' 15:855-66.
Hunt K, Cant J, Gebo D, Rose M, Walker S, and Youlatos D (1996) Standardized descriptions of primate locomo-
tor and postural modes. Primates 37:363-387.
Hurlbert S (1984) Pseudoreplication and the design of ecological field experiments. Ecological Monographs
54: 187-211.
James FC (1982) The ecological morphology of birds: A review. Ann. Zoo!. Fennici 19:265-275.
Jolly A (1985) The Evolution of Primate Behavior. New York: MacMillan.
Kawanaka K (1996) Observation time and sampling intervals for measuring behavior and interactions of chimpan-
zees in the wild. Primates 37:185-196.
Kinzey WG (1967) Preface. Am. J. Phys. Anthropol. 26: 115-8.
Kinzey WG (1977) Positional behavior and ecology in Callicebus torquatus. Yrbk. Phys. Anthropol. 20:468-80.
Kinzey WG, Rosenberger AL, and Ramirez M (1975) Vertical clinging and leaping in a neotropical anthropoid.
Nature 255:327-8.
Machlis L, Dodd PWD, and Fentress JC (1985) The pooling fallacy: Problems arising when individuals contribute
more than one observation to the data set. Z. Tierpsycho!. 68:201-14.
MacKinnon J, and MacKinnon K (1980) The behavior of wild spectral tarsiers. Int. J. Primato!. 1:361-79.
Manly B (1991) Randomization and Monte Carlo Methods in Biology. New York: Chapman and Hall.
Martin P, and Bateson P (1986) Measuring Behaviour. Cambridge: Cambridge University Press.
McGraw WS (1996) Cercopithecid locomotion, support use, and support availability in the Tai Forest, Ivory
Coast. Am. J. Phys. Anthropo!. 100:507-22.
Mendel F (1976) Postural and locomotor behavior of Alouatta palliata on various substrates. Folia Primato!'
26:36-53.
Mitterrneier RA (1978) Locomotion and posture in Ateles geojJroyi and Ateles paniscus. Folia Primato!'
30:161-93.
Mitterrneier RA, and Fleagle JG (1976) The locomotor and postural repertoires of Ate/es geojJroyi and Colobus
guereza, and a reevaluation of the locomotor category semibrachiation. Am. J. Phys. Anthropol.
45:235-56.
Morbeck ME (1974) Positional behaviour in Colobus guereza: A preliminary analysis. In S Kondo (ed.): Symp 5th
Congo International Primatological Soc. Tokyo: Japan Science Press, pp. 331~3.
Morbeck ME (1977a) Leaping, bounding and bipedalism in Colobus guereza: A spectrum of positional behavior.
Yrbk. Phys. Anthropol. 20:40S-20.
28 M. Dagosto and D. L. Gebo

Morbeck ME (1977b) Positional behavior, selective use of habitat substrate and associated non·positional behavior
in free-ranging Colobus guereza (Ruppel, 1835). Primates 18:35-58.
Morbeck ME (1979) Forelimb use and positional adaptation in Colobus guereza: Integration of behavioral, eco-
logical, and anatomical data. In M Morbeck, H Preuschoft, and N Gomberg (eds.): Environment, Behavior.
Morphology: Dynamic Interactions in Primates. New York: Gustav Fischer, pp. 95-117.
Napier J, and Napier PH (1967) Handbook of Living Primates. New York: Academic Press.
Napier JR, and Walker A (1967) Vertical clinging and leaping--a newly recognised category of locomotor behav-
iour of primates. Folia Primatol. 6:204-19.
Niemitz C (I 984a) Activity rhythms and use of space in semi-wild Bornean tarsiers, with remarks on wild spectral
tarsiers. In C Niemitz (ed.): Biology of Tarsiers. Stuttgart: Gustav Fischer, pp. 85-116.
Niemitz C (1984b) Locomotion and posture of Tarsius bancanus. In C Niemitz (ed.): Biology of Tarsiers. New
York: Gustav Fischer, pp. 191-226.
Oliveira JMS, Lima MG, Bonvincino C, Ayres JM, and Fleagle JG (1985) Preliminary notes on the ecology and
behavior of the Guianan saki (Pithecia pithecia, Linnaeus 1766; Cebidae. Primate). Acta Amazonica
15:249-63.
Oxnard Cli (1974) Primate Locomotor Classifications for evaluating fossils: Their inutility and an alternative. In S
Kondo (ed.): Symp. 5th Congo Int'1. Primal. Soc. Tokyo: Japan Science Press, pp. 269-84.
Oxnard CE (1979) The morphological behavioral interface in extant primates: Some implications for systematics
and evolution. In M Morbeck, H Preuschoft, and N Gomberg (eds.): Environment, Behavior, Morphology:
Dynamic Interactions in Primates. New York: Gustav Fischer, pp. 209-28.
Oxnard CE( 1984) The Order of Man. New Haven: Yale University Press.
Oxnard CEo Crompton RH, and Lieberman SS (1990) Animal Lifestyles and Anatomies. Seattle: University of
Washington Press.
Pounds J (1988) Ecomorphology, locomotion, and microhabitat structure: Patterns in a tropical mainland Anolis
community. Ecological Monographs 58:299-320.
Pounds J (1991) Habitat structure and morphological patterns in arboreal vertebrates. In S Bell, E McCoy, and H
Mushinsky (eds.): Habitat Structure: The Physical Arrangement of Objects in Space. New York: Chapman-
Hall, pp. 109-117.
Remis M (1995) Effects of body size and social context on the arboreal activities of lowland gorillas in the Central
African Republic. Am. J. Phys. Anthropol. 97:413-33.
Richard A (1970) A comparative study of the activity patterns and behavior of Alouatta villosa and Ateles geo.f-
froyi. Folia Primatol. /2:241-63.
Richard A (1978) Behavioral variation: Case study of a Malagasy lemur. Lewisberg, Pa: Bucknell University
Press.
Richard A (\985) Primates in Nature. New York: WH Freeman.
Rickelfs RE. and Miles DE (1994) Ecological and evolutionary inferences from morphology: An ecological per-
spective. In P Wainwright and S Reilly (eds.): Ecological Morphology. Chicago: University of Chicago
Press, pp. 13-41.
Ripley S (1967) The leaping of langurs: A problem in the study of locomotor adaptation. Am. J. Phys. Anthropol.
26:149-70.
Ripley S (1977) Gray zones and gray langurs: Is the "semi-" concept seminal? Yrbk. Phys. Anthropol. 20:376-94.
Ripley S (1979) Environmental grain, niche diversification, and positional behavior in Neogene primates: An evo-
lutionary hypothesis. In M Morbeck, H Preuschoft, and N Gomberg (eds.): Environment, Behavior, and
Morphology: Dynamic Interactions in Primates. New York: Gustav Fischer, pp. 37-74.
Roberts M, and Cunningham B (1986) Space and substrate use in captive western tarsiers, Tarsius bancanus. Inl.
J. Primatol. 7:113-30.
Roberts M, and Kohn F (1993) Habitat use, foraging behavior and activity patterns in reproducing Western tar-
siers, Tarsius bancanus in captivity. Zoo Biology 12:217-32.
Rodman PS (1979) Skeletal differentiation of Macacafascicularis and Macaca nemestrina in relation to arboreal
and terrestrial quadrupedalism. Am. J. Phys. Anthropol. 51:51-62.
Rodman PS (1991) Structural differentiation of microhabitats of sympatric Macaca fascicularis and M. nemestrina
in East Kalimantan, Indonesia. Int. J. Primatol. 12:357-75.
RolJinson J, and Martin RD (1981) Comparative aspects of primate locomotion, with special reference to arboreal
cercopithecines. Symp. Zool. Soc. Lond. 48:377-427.
Rose MD (1974) Postural adaptations in New and Old World monkeys. In F Jenkins (ed.): Primate Locomotion.
New York: Academic Press, pp. 201-222.
Rose MD (1976) Bipedal behavior of olive baboons (Papio anubis) and its relevance to an understanding of the
evolution of human bipedalism. Am. J. Phys. Anthropol. 44:247-262.
Methodological Issues in Studying Positional Behavior 29

Rose MD (1977) Positional behavior of olive baboons Papio anubis and its relationship to maintenance and social
activities. Primates 18:59--116.
Rose MD (1978) Feeding and associated positional behavior of black and white colobus monkeys (Calobus guereza).
In G Montgomery (ed.): The Ecology of Arboreal Folivores. Washington, D.C.: Smithsonian Press, pp. 253-62.
Rose MD (1979) Positional behavior of natural populations: Some quantitative results of a field study of Calobus
guereza and Cercopithecus aethiops. In M Morbeck, H Preuschoft and N Gomberg (eds.): Environment,
Behavior, Morphology: Dynamic Interactions in Primates. New York: Gustav Fischer, pp. 75-93.
Rosenberger AL, and Stafford BJ (1994) Locomotion in captive Leontopithecus and Callimico: A multimedia
study. Am. J. Phys. Anthropol. 94:379--94.
Schon Ybarra MA (1984) Locomotion and postures of red howlers in a deciduous forest-savanna interface. Am. J.
Phys. Anthropol. 63:65-76.
Schon Ybarra MA, and Schon 1Il MA (1987) Positional behavior and limb bone adaptations in red howling mon-
keys (Alouatta seniculus). Folia Primatol. 49:70-89.
Simberloff D (1982) The status of competition theories in ecology. Ann. Zool. Fennici 19:241-253.
Simons EL (1967) Fossil primates and the evolution of some primate locomotor systems. Am. J. Phys. Anthropol.
26:241-54.
Smith RJ (1994) Degrees of freedom in interspecific allometry: An adjustment for the effects of phylogenetic con-
straint. Am. J. Phys. Anthropol. 93:95-108.
Srikosamatara S (1984) Ecology of pileated gibbons in South-East Thailand. In H Preuschoft, D Chivers, W
Brockelman, and N Creel (eds.): The Lesser Apes. Edinburgh: Edinburgh University Press, pp. 242-57.
Sugardjito J (1982) Locomotor behaviour of the Sumatran Orang Utan (Pongo pygmaeus abelii) at Ketambe,
Gunung Leuser National Park. Malay Nat. 1. 35:57-64.
Sugardjito J, and van Hooff JARAM (1986) Age-sex class differences in the positional behavior of the Sumatran
Orang-utan (Pango p)'gmaeus abelii) in the Gunung Leuser National Park, Indonesia. Folia Primatol.
47:14-25.
Susman RL (1984) The locomotor behavior of Pan paniscus in the Lomako forest. In R Susman (ed.): The Pygmy
Chimpanzee. New York: Plenum Press, pp. 369--93.
Susman RL, Badrian NL, and Badrian AJ (1980) Locomotor behavior of Pan paniscus in Zaire. Am. J. Phys. An-
thropol. 53:69--80.
Tattersall I (1977) Ecology and behavior of Lemurfitlvus mayottensis (Primates, Lemuriformes). Anth. Pap. Am.
Mus. Nat. Hist. 54:421-82.
Terranova C (1995) Functional morphology of leaping behaviors in galagids: Associations between landing limb
use and diaphyseal geometry. In L Alterman, G Doyle, and M Izard (eds.): Creatures of the Dark. New
York: Plenum Press, pp. 473-494.
Tilden CD (1990) A study of locomotor behavior in a captive colony of red-bellied lemurs (Eulemur rubriventer).
Am. J. Primatol. 22:87-100.
Tremble M, Muskita Y, and Supriatna J (1993) Field observations of Tarsius dianae at Lore Lindu National Park,
Central Sulawesi, Indonesia. Tropical Biodiversity 1:67-76.
Tuttle RH, and Watts DP (1985) The positional behavior and adaptive complexes of Pan gorilla. In S Kondo (ed.):
Primate Morphophysiology, Locomotor Analyses and Human Bipedalism. Tokyo: University of Tokyo
Press, pp. 261-88.
Wainwright P, and Reilly SM, eds. (1994) Ecological Morphology. Chicago: Chicago University Press.
Walker A (1969) The locomotion of the lorises, with special reference to the potto. E. Afr. Wildl. J. 7: 1-5.
Walker AC (1979) Prosimian locomotor behavior. In G Doyle and R Martin (eds.): The Study of Prosimian Behav-
ior. New York: Academic Press, pp. 543-65.
Walker SE (1994) Positional behavior and habitat use in Chiropotes satanas and Pithecia pithecia. In B Thierry, J
Anderson, J Roseder and N Herrenschmidt (eds.): Current Primatology, Vol. I: Ecology and evolution.
Strasbourg: International Primatological Society, pp. 195-202.
Walker SE (1996) The evolution of positional behavior in the Saki/Uakaris (Pithecia. Chiropotes. and Cacaiao).
In M Norconk, A Rosenberger, and P Garber (eds.): Adaptive Radiations of Neotropical Primates. New
York: Plenum Press, pp. 335-368.
Walker SE, and Ayres JM (1996) Positional behavior of the white uakari (Cacajaa calvus calvus). Am. 1. Phys.
Anthropol. 10/:161-172.
White FJ, Overdorff DJ, Balko EA, and Wright PC (1995) Distribution of ruffed lemurs (Varecia variegata) in Ra-
nomafana National Park, Madagascar. Folia Primatol. 64: 124-31.
Wilson JM, Stewart PD, Ramangason G-S, Denning AM, and Hutchings MS (1989) Ecology and conservation of
the crowned lemur, Lemur coronatus. at Ankarana, N. Madagascar. Folia Primatol. 52: 1-26.
Youlatos D (1993) Passage within a discontinuous canopy: Bridging in the red howler monkey (A/ouafla sell;ClI-
Ius). Folia Primatol. 6/:144-147.
 2

FINE-GRAINED DIFFERENCES WITHIN

POSITIONAL CATEGORIES
A Case Study of Pithecia and Chiropotes

Suzanne E. Walker

Department of Anthropology
Humboldt State University
Arcata, California 95521

1. INTRODUCTION

The study of primate locomotion and posture has advanced considerably since its
beginnings, approximately thirty years ago. Early studies were primarily concerned with
inferring the locomotor behavior of the earliest hominids, using the living great apes as
analogues (Ashton and Oxnard, 1963; Napier, 1963). A turning point for studies on loco-
motion and posture was the 1965 Primate Locomotion Symposium, organized by Warren
Kinzey. In this symposium, the importance of field studies was first emphasized, as was
the importance of distinguishing between locomotor categories based upon natural behav-
ior rather than upon skeletal anatomy or observations of zoo animals (Kinzey, 1967). At
the same time, Prost (1965) stressed the importance of a standardized system to classify
positional (i.e., locomotor and postural) behavior, which would allow for more precise
comparison between studies, and emphasized the importance of postural (in addition to lo-
comotor) behaviors in shaping the postcranium.
Many of the useful ideas and terms that are now frequently utilized in the literature
were introduced in the 60s and 70s as a result of field studies of primate positional behavior
(e.g., Ripley, 1967; Rose, 1974; Morbeck, 1976). Studies undertaken in the field are essential
in order to provide more realistic views of how primates move in their natural habitats; they
provide the basis for making inferences about environmental variables that may have pro-
vided the selective pressures that shaped primate anatomies. Ripley (1977) was instrumental
in replacing the "semi" locomotor categories with others more descriptive of actual behav-
ioral patterns (e.g., an animal cannot "semi-brachiate"). Others (e.g., Fleagle and Mittermeier,
1980) tested the predictions of Napier (1962) and Cartmill and Milton (1977) about the influ-
ence of forest structure and body size on positional behavior. Further research resulted in ex-

31
32 S. E. Walker

ami nation of more fine-grained environmental influences: that of the structure of individual
trees (Ripley, 1977), use of various tree portions (Mendel, 1976), and the location of various
classes of primate food in the forests (Crompton, 1984). Some ofthese important field studies
have emphasized a problem-oriented approach (e.g., Cant and Temerin, 1984), focusing on
the ways in which primates deal with problems presented to them by their environment and
the location of foods within the forest matrix. These and other studies have paved the way for
more sophisticated and detailed research (see Dagosto and Gebo, this volume).
Research techniques in the study of primate positional behavior are becoming in-
creasingly "fine-tuned," with researchers using new data collection techniques and more
detailed behavioral categories (e.g., Fontaine, 1990; Hunt et aI., 1996) that make for more
feasible cross-study comparisons. With ever-improving video capabilities and interfacing
computer programs (e.g., Rosenberger and Stafford, 1994), we can now attempt some of
the analyses of positional behavior of wild primates previously possible only for those in
captivity. Only recently have a rigorous set of statistical tests been applied to results of
studies of positional behavior (e.g., Boinski, 1989; Fontaine, 1990; Doran, 1992; Dagosto,
1994; Gebo and Chapman, 1995; Remis, 1995), partly because of the fact that much of the
data on positional behavior are not particularly amenable to statistical testing (e.g., Men-
del, 1976; Cant, 1988; Hunt, 1992).

1.1. The Problems

In the spirit of the Conference Primate Locomotion 1995, which marked the 30-year
anniversary of the original Primate Locomotion symposium, this paper discusses some
methodological problems in studies of positional behavior that warrant further investigation
(see also Dagosto and Gebo, this volume). These problems (described below), which be-
came apparent during the course of my comparative field study of the positional behavior of
three platyrrhine species, Pithecia pithecia, Chiropotes satanas, and Cacajao calvus, of the
tribe Pitheciini, all deal with the classification of positional behaviors into discrete catego-
ries. Pithecia and Chiropotes were studied more intensively over a longer time period than
was Cacajao; therefore, only examples from the former two species are included here.
First, establishing a positional classification and subsequently using the categories to
collect data is difficult due to two factors: some observed behaviors do not lend themselves
well to a one- or two-word descriptive category, and there are often subtle gradations of one
behavioral category into another that may result simply from a minor change in support
characteristics. For example, the behavior described as quadrupedal walk grades into
pronograde clamber (Cant, 1988) as the substrate changes from a single support to multiple
ones. Likewise, with increasing inclination of a support, quadrupedal walk grades into the
behavior caIled climb (defined below), and a sitting posture grades into a vertical cling.
Second, it is important to categorize and define behaviors in an anatomically mean-
ingful way (e.g., JoIly, 1965), so as to aIlow, for example, inferences to be drawn about the
positional behavior of fossil primates. Categories should reflect such information as limb
segment positioning and forces thought to be transmitted through limbs (Schon and SchOn,
1987), and number of cheiridia or body parts in contact with supports (e.g., Hunt, 1992).
Third, consistent terminology should be used for the categories and should be carefully
defined (Hunt et a!., 1996). For example, "climb" is a persistent problem category, defined
differently in various studies. Rose (1979) and Cartmill (1985) define climb as progression up
or down a single support with vertical or steeply sloping surfaces (the definition foIlowed in
this paper), while Fleagle and Mittermeier (1980) and Gebo (1992) define it more broadly and
include irregular gaits on mUltiple supports as weIl as bridging behaviors.
Fine-Grained Differences within Positional Categories 33

Lastly, the problem that constitutes the primary focus of this paper is with the vari-
ation exhibited within and between species in the use of traditionally defined behavioral
categories. One behavioral category may not sufficiently describe the variation that indi-
vidual conspecifics exhibit in their use of that behavior. That is, a single behavioral cate-
gory can often encompass several finer categories that differ from one another depending
upon the influences of body orientation and/or support characteristics. Conversely, vari-
ation in body orientation, support use, and morphology can result in differences in the
manner in which species exhibit the "same" behavior.
Human observers draw arbitrary boundaries between classes of behaviors. How do
we know that the boundaries that we create are biologically relevant, and how do we dis-
tinguish them? It is suggested here that we recognize this problem in the classification of
positional behavior, particularly with regard to the biological function of various posi-
tional behaviors and their influence upon anatomical features.

1.2. Qualitative Aspects of Positional Categories: Are Finer

Classifications Necessary?

Categories such as sit, quadrupedal walk, and leap provide us with the primary
means of communicating about basic positional behaviors. Each of these categories, how-
ever, can be further subdivided into more fine-grained categories. Previous workers (e.g.,
Garber, 1980) have noted the importance of these finer categories; however, in the analy-
sis of positional behavior, these behaviors are eventually pooled due to the sheer unman-
ageability of many small categories.
I propose that these finer categories of behavior be investigated for two main rea-
sons. First, we need to determine whether or not minor differences in body position or
support use influence the forces to which primate bodies are subjected. Biomechanical as-
pects of positional behavior, such as the influence of gravity, types of stresses, force appli-
cation, force absorption, and friction are highly dependent upon body orientation,
characteristics of supports used, and obviously, morphology. Second, the primary contexts
in which these behaviors are used often differ from one another, which may indicate dif-
ferent functions for them. A difference in function would presumably have a profound in-
fluence on the evolution of the animals, and therefore warrants further investigation.
In this paper, I will use my field data on Pithecia and Chiropotes to demonstrate that
limitations exist in the traditional manner of classifying positional behavior into coarse-
grained categories. By breaking down coarse-grained behavioral categories into finer-
grained ones, I have identified associations of body orientation and support characteristics
with positional behaviors that would not have been possible without these fine-grained
categories. However, further investigation of the influence of, for example, different types
of sitting behavior on postcranial anatomy is in order. Since a major goal of studies of po-
sitional behavior is to create a data base from which to interpret fossil remains, any infor-
mation that will allow us to make more detailed inferences about the behavior of extinct
species provides an important contribution.

2. STUDY SITES AND METHODS

Between 1989 and 1991 I collected data on the positional behavior of Pithecia pi-
thecia and Chiropotes satanas in Venezuela, where they are naturally allopatric. The study
34 S. E. Walker

animals were well habituated and observed on separate islands in Guri Lake (for descrip-
tion of study sites see Pemia, 1985; Parol in, 1993; Walker, 1993, 1996).
The data on positional behavior were collected using focal animal instantaneous
sampling at two minute intervals (after Garber, 1980; for a detailed description of methods
see Walker, 1993, 1996). Videotaping was utilized to assist in the initial categorization of
behaviors, and some drawings based on captured video frames are used to illustrate behav-
ioral categories in this paper.
In this study, the problem of independence of consecutive samples was first dealt
with in the field during data collection by noting the samples that involved a continuation
of behavior from the previous data point (i.e., if the focal animals had not moved at least
one body length between data points). The (presumably) non-independent samples were
then eliminated before performing statistical tests. This is similar to the post facto "pool-
ing" of instantaneous samples into artificial bouts when consecutive samples were the
same (Hunt, 1992). The latter method, however, may eliminate samples that are, in fact,
independent. The frequency data were examined for significant differences, using a Chi-
square test. Differences were deemed significant if the probability value of the Chi-square
statistic was 0.01 or less.
Below, I present a few commonly used positional behaviors of Pithecia and Chiro-
pates, and demonstrate how differences exist within coarse-grained (traditionally used)
behavioral categories due to the influence of body orientation, support characteristics and
support use, and morphological features. The results are organized into two main sections.
In the first, I present several examples that deal primarily with differences within a species
in the use of a single behavior, demonstrating the influence of various maintenance activi-
ties and support characteristics on body orientation. For some of these behaviors (i.e., sit)
quantitative data are presented; for others (i.e., quadrupedal stand and quadrupedal walk)
anecdotal observations are discussed. Data for this section are drawn from a three-month
subset of observations on Pithecia and Chiropotes. In the second section, one behavioral
example is presented that deals with differences between the two species in their use of a
positional behavior traditionally represented by a single behavioral category. This data set
is based upon fifteen months of observations on Pithecia and ten months on Chiropotes.

3. WITHIN-SPECIES DIFFERENCES IN POSITIONAL

CATEGORIES

During postures such as sit and quadrupedal stand, the orientation of the cranio-
caudal body axis can be either perpendicular or parallel to the support (Figures 1 and 2).
These two body orientations influence how the center of gravity is maintained over the
base of support, which in tum affects the use of the limbs in weight-bearing and grasping.
During perpendicular sitting (Figure 1, left), the body axis is more or less perpen-
dicular to the substrate and the back is sometimes flexed. Perpendicular sit poses few
problems for the individual in terms of balance. Body weight is balanced on the soles of
the feet, usually with the added support of the haunches. The forelimbs are often not in-
volved in weight-bearing, leaving the hands free for other activities. During parallel sitting
(Figure I, right), the body axis is aligned with the support; therefore, the base of support is
narrower and the body is subject to rolling to the side. The feet grasp the branch, with the
soles on one side and digits on the other, in front of the haunches and the haunches rest on
the support. In parallel sit, the forelimbs often playa weight-bearing role, with the hands
used to grasp the support or a nearby branch. While grasping, the pollex may oppose the
Fine-Grained Differences within Positional Categories 35

Figure 1. Sit in Pithecia. (Left) perpendicular sit, (right) parallel sit.

other digits, or grips may be taken between digits II and III, as noted previously by other
researchers (e.g" Erikson, 1957; Hill, 1960; Kay, 1990).

3.1. Influence of Different Maintenance Activities

Perpendicular and parallel sits differ in their occurrence in Pithecia during the vari-
ous activities of feeding, traveling, and resting (Table 1). Overall, the perpendicular sit is

--
Figure 2. Stand in Chiropotes. Lefi, perpendicular stand; right, parallel stand.
36 S. E. Walker

Table 1. Variation in the percentage of perpendicular vs. parallel

sit during various maintenance activities in Pithecia i
Maintenance activity N Perpendicular sit Parallel sit
Feeding 297 56.9 28.9
Traveling 33 5.9 3.9
Resting 306 37.2 67.2
'The differences in sit types are significant (/=53.814, df=2, P<O.OOI).

the more common of the two sitting postures throughout all activity contexts. Perpendicu-
lar sit is a more frequent feeding than resting posture while the opposite is true for parallel
sit, which is used more often during resting than during feeding (Table 1).
Quadrupedal standing occurs on horizontal or low-angled supports. As with sitting,
the two main types of body orientation with reference to the long axis of the support are
perpendicular and parallel, each posing distinct problems of balance. During perpendicu-
lar standing (Figure 2, left) the craniocaudal axis of the body is aligned perpendicular to
the long axis of the support, resulting in an unstable posture. The back is somewhat flexed,
and the feet often grasp the support with the hallux opposed to the remaining digits, while
the hands grasp with all digits in line. In this position a small portion of the body is di-
rectly over the base of support and an individual is subject to pitching forward or back-
ward. The parallel stand (Figure 2, right) is a relatively more stable posture (at least on
solid supports), due to its larger base of support and the ease with which the body is main-
tained directly over it. The long axis of the body is parallel to the support and the limbs
are primarily extended and adducted. Hand grips again may be taken between digits II and
III. The primary balance problem to be dealt with is rolling to either side.
Chiropotes often uses both types of stand, while Pithecia was rarely observed in per-
pendicular stand. Observations indicate that perpendicular stand often serves as a momen-
tary pause within a travel bout, and sometimes as a leap takeoff or landing position, while
parallel stand may be a longer-term posture during any of the maintenance activities of
feed, travel or rest.

3.2. Influence of Support Characteristics

3.2.1. Support Size. Pithecia uses different sized supports during perpendicular vs.
parallel sit (Table 2). For example, on the smallest supports (under 2 cm in diameter) par-
allel sit was not observed to occur while perpendicular sit was observed in 7.5% of sitting
samples (Table 2). On smaller branches, the haunches are less likely to support body

Table 2. Variation in the percentage of perpendicular vs. parallel

sit on supports of various diameters in Pithecia i
Support diameter N Perpendicular sit Parallel sit
<2cm 30 7.5 0
2-5 em 310 49.8 48.3
6-10 cm 174 28.1 26.7
>IOcm 37 4.8 7.8
Mix 2 79 9.8 17.2
'The differences in sit types are significant (X'=26.356, df=4, P<O.OOI).
'Mix, multiple supports of various diameters.
Fine-Grained Differences within Positional Categories 37

weight; the weight may be distributed entirely through the soles of the feet. During per-
pendicular sit, the hallux may be in line with other pedal digits (if the support is relatively
large) or opposed to the others (if the support is relatively small). Parallel sit occurred
more often than perpendicular sit on supports greater than 10 cm in diameter (7.8% vs.
4.8%) and on supports of mixed diameter (17% vs. 9.8%). These differences contribute to
the overall statistical significance of the data sets (Table 2).
Support size influences quadrupedal stand and quadrupedal walk in similar ways
due to problems of balance as support size decreases. On smaller supports (particularly
flexible ones), the perpendicular stand is more often observed than is the parallel stand,
due to the increased risk of rolling to the side. During both standing and walking, limb
flexion and abduction act to lower the center of gravity for increased stability on smaller
supports. The grasp taken and the degree of wrist pronation depend upon support size as
well, with pronation increasing as the central weight-bearing axis moves laterally in the
hand.

3.2.2. Support Inclination. Angled supports and multiple supports of mixed inclina-
tion are used approximately equally during both perpendicular and parallel sit. On hori-
zontal supports parallel sit occurs more often than does perpendicular sit (39.0% vs.
30.9%), whereas when on deformable supports, the opposite is true, with perpendicular sit
being more common (12.2% vs. 3.5%; Table 3). On angled supports, problems of reduced
friction can be alleviated behaviorally; for example, by the use of more than one support
for grasping.
Support inclination influences quadrupedal stand and quadrupedal walk in terms of
introducing additional problems of balance and loss of friction. With increasing support
inclination, perpendicular stand is observed less frequently than is parallel stand, and dur-
ing the latter the animal typically faces in the upward direction of the support. The ani-
mal's center of gravity shifts with increasing inclination of a support, becoming displaced
either forward or backward depending upon the direction of the angle. In order to maintain
the center of gravity over the base of support, the limbs tend to be more flexed and ab-
ducted. Friction becomes more important, since there is a potential for slippage; steep in-
clinations increase the shear force exerted between the feet and the branch (Cartmill,
1985). During quadrupedal walk, the limbs take on different roles than needed on a hori-
zontal branch, where they act as simple struts. The roles of the fore- and hindlimbs di-
verge: forelimbs are under tension while hindlimbs are used for propulsion. Therefore, in
species with more frequent use of highly angled supports, longer hindlimbs may provide
advantages in dealing with the greater demands of propulsion.

Table 3. Variation in the percentage of perpendicular vs.

parallel sit on supports of various inclinations in Pithecia l
Support inclination 2 N Perpendicular sit Parallel sit
Horizontal 214 31 39
Angled 280 44 46
Defonnable 57 12 3
Mix 81 13 12
IThe differences in sit types are significant (X 2=26.356, df=4, P<O.OO I).
2Horizontal (0-20°), angled (21-70°). deformable (under animal's body
weight), mix (supports of mixed inclinations).
38 S. E. Walker

4. BETWEEN-SPECIES DIFFERENCES IN POSITIONAL

CATEGORIES: THE SPECIAL CASE OF LEAPING

Leaping is the behavior that epitomizes qualitative differences within categories of

positional behavior, made evident, for example, by the different forms of leaping in the
prosimian vertical clingers and leapers (e.g., Oxnard, 1984) as well as various species of
callitrichids (Garber, 1991). In the genus Saguinus, differences in limb ratios correlate
with differences in the type and frequency of leaping behavior (Garber, 1991). Obvious
differences in the form of leaping and associated postcranial morphology are also apparent
in Pithecia and Chiropotes; each species has very different ways of dealing with problems
of take-off and landing forces. Below, the influence of differences in body orientation,
support characteristics, and morphology are shown to affect the biomechanical aspects of
leaping behavior.
Leaping plays an important role in the positional repertoire of both Pithecia and Chi-
ropotes, with total frequencies of 40% and 25% of travel samples, respectively. Pithecia
shares numerous behavioral and morphological specializations with the prosimian vertical
clingers and leapers (Fleagle and Meldrum, 1988; Walker, 1993, in revision), while Chiro-
pates is relatively unspecialized in its leaping behavior and associated morphology.

4.1. Influence of Body Orientation

Pithecia tends to maintain an orthograde body orientation throughout all leap
phases, even when taking off from a horizontal support (Figure 3). In virtually alJ leaps,

Preparatory Take-off Midflight Landing

Figure 3. Leap phases in Pithecia (top) and Chiropotes (bottom),

Fine-Grained Differences within Positional Categories 39

Table 4. Variation in the percentage of support diameters used for leap

take-off and landing in Pithecia and Chiropotes
Support diameter Take-off' Landing1
Pithecia Chiropotes Pitl1ecia Chiropoles
N=887 N=407 N=82 I N=366
<2 em 4.1 20.1 10.6 42.4
2-5 em 41.1 33.8 41.3 17.5
6-10 em 38 25.1 32.3 16.7
11-15 em 10.2 6.9 9.3 7.1
>15em 1.9 1.5 1.5 1.1
Mix 4.7 12.6 5 15.2
'The differences in support diameters are significant (x.2=126.24. df=5, P<O.OOI).
1The differences in support diameters are significant (x.1=222.97. df=5. P<O.OOI).

the lower body swings forward for the hindlimbs to contact the landing support first. In
longer leaps between vertical supports, the back and hindlimbs flex to pull the body into a
tuck in midleap, as in Tarsius (Peters and Preuschoft, 1984). In contrast, Chiropotes' body
orientation remains pronograde throughout take-off, midleap and landing (Figure 3). A
common feature is that both species crouch before take-off; this extreme limb flexion in-
creases the time over which propulsive extension occurs, and it potentiates the muscles to
be used in propulsion (e.g., Cavanagh, 1977). These differences between the two species
in body orientation result from differences in characteristics of the supports used for take-
off and landing, and from morphology (see below).

4.2. Influence of Support Characteristics

4.2.1. Support Size. Characteristics of the supports used by the species differ for
both takeoff and landing (Table 4). Pithecia most often uses supports between 2 and 10 cm
in diameter for takeoff, while Chiropotes also often uses those less than 2 cm (Table 4).
For landing, differences are also apparent. Most landings by Pithecia are onto landing
supports between 2 and 10 cm in diameter, while for Chiropotes the most frequent landing
supports are multiple terminal branches of less than 2 cm in diameter.

4.2.2. Support Inclination. Overall differences between the species in the inclination
of takeoff and landing supports were significant (Table 5). Pithecia prefers vertical or near
vertical supports for both takeoff and landing (Table 5), partly because of their propensity
for taking off from a clinging posture. Chiropotes, on the other hand, most often uses hori-
zontal or angled supports for take-off. Chiropotes typically travels rapidly by quadrupedal
locomotion to the terminal branches, then uses this forward momentum in leap take-off.
For landing, deformable supports are most often used by Chiropotes.

4.3. Influence of Morphology

Several of the anatomical features characteristic of specialized leapers can be distin-
guished in Pithecia, particularly at the proximal and distal femur, the vertebral column,
and the tarsal and metatarsal bones (Fleagle and Meldrum, 1988). Pithecia frequently
takes off from a stationary posture rather than during locomotion. The velocity of Pi-
thecia's leaps are greater than Chiropotes' (Walker, in revision); Pithecia's relatively
40 S. E. Walker

Table 5. Variation in the percentage of support inclinations used for

leap take-off and landing in Pithecia and Chiropotes
Take-otT~ Landing}
Pithecia Chiropotes Pithecia Chiropotes
Support inclination I N=892 N=408 N=828 N=367
Horizontal 13 24.5 8.2 14.2
Angled 26.2 29.1 28.8 19.8
Vertical 45.9 41.4 2.5
Defonnable 10.7 41.2 19.9 62.4
Mix 4.2 4.2 1.7 1.1
I Horizontal (0_20°), angled (21_70°), vertical (71_90°), deformable (under animal's
body weight), mix (supports of mixed inclinations).
~The differences in support inclinations are significant (X2=329.03, df=4, P<O.OO I).
3The differences in support inclinations are significant (/=287.49, df=4, P<O.OOI).

longer hindlimbs apply propulsive force to the takeoff substrate for a longer time than is
seen in Chiropotes. At landing, Pithecia's long hindlimbs contact the landing substrate
first, and flex to absorb the compressive forces. The relatively unspecialized (in terms of
leaping) Chiropotes deals with these forces by landing with all four limbs onto deformable
supports.
For both Pithecia and Chiropotes, we can best describe the aforementioned behavior
as leaping, but as demonstrated, the obvious qualitative and quantitative differences can-
not be ignored. In fact, the leaping specializations of Pithecia affect its entire behavioral
repertoire (Walker, 1993, in revision).

5. CONCLUSIONS
The traditional way of examining positional behavior in terms of the frequencies of
standard behavioral categories continues to be an important tool for understanding the re-
lationship between form and function. Thus, this paper in no way seeks to undermine or
reduce the importance of previous or continuing studies of functional morphology, which
have identified important correlates between form and function in numerous species (e.g.,
Fleagle, 1977; Rodman, 1979; Ward and Sussman, 1979). However, there are limitations
to the standard methods used in classifying positional behavior and it is here suggested
that we should attempt to deal with these limitations.
The use of fine-grained categories as demonstrated here has brought to light, for ex-
ample, two clear types of sitting behavior that are associated with different substrates and
with different activity contexts (e.g., resting vs. feeding). Additional studies need to be
conducted for these different behaviors in order to investigate anatomical correlates and
their biomechanical advantages. By identifying these correlates, we can strengthen infer-
ences about the positional behavior of extinct species from their fossilized remains. The
fact that the contexts in which these behaviors are used often differ from one another may
reflect a fundamental difference in biological role (Bock and von Wahlert, 1965). We need
to examine the environmental and behavioral basis for these differences, which influence
the evolution of the postcranial skeleton. Another limitation in the traditional positional
classifications, as demonstrated for leaping, is that simply using the term "leap" for the
behavior exhibited by Pithecia and by Chiropotes does not provide a detailed description
Fine-Grained Differences within Positional Categories 41

of how the behavior is conducted by the two species in terms of biomechanics, substrate
use and body orientation.
In future studies of positional behavior, greater detail in classification of behavior
should be a primary goal, and can be assisted by the use of technological advances in
video and interfacing computer programs. Following are a few suggestions in the con-
struction and use of finer-grained positional behavior classifications for field studies.
In the initial categorization of positional behavior in the field, video recording
should be used if possible in order to describe in greater detail the orientation of limbs to
trunk, body to substrate, etc., which will result in a more anatomically meaningful classifi-
cation for biomechanical analyses. Developing an ethogram in this way should take rela-
tively little additional time before data collection begins. In my own study, the finer
behavioral categories, once identified, were relatively easy to observe and record in the
field; thus, it should not be necessary to collect all data with video recorders to analyze
later.
Video recording is not only useful in the initial categorization of behaviors, but to
later convey this information in presentations and publications. With somewhat greater
emphasis on graphically presenting behaviors, such as with stills from video frames, more
efficient comparisons among species can be made, and biomechanical analyses conducted.
Short sequences of behavior could even easily be presented on the internet (Dagosto, pers.
comm.), with references in the original publication.
With continued collection of positional behavior data using carefully devised and
defined categories, and with the use of new techniques, we will continue to obtain a better
understanding of how primates most efficiently feed, travel and rest in their natural habi-
tats, and the circumstances that led to the evolution of their positional adaptations.

ACKNOWLEDGMENTS
I would like to thank the organizers of the symposium on Primate Locomotion 1995
for inviting me to participate: Drs. Elizabeth Strasser, John G. Fleagle, Henry McHenry
and Alfred Rosenberger. I give special thanks to Elizabeth Strasser, Marian Dagosto and
two anonymous reviewers for their very insightful comments on an earlier version of this
manuscript. This research was supported by grants from the National Science Foundation
(BNS 89-13349), Wenner Gren Foundation for Anthropological Research, and the Gradu-
ate Center of the City University of New York. This paper is dedicated to my husband,
whose memory continues to be an inspiration to me.

REFERENCES
Ashton EH, and Oxnard CE (1963) The musculature of the primate shoulder. Trans. ZooJ. Soc. London
29:553-{)50.
Bock WJ, and von Wahlert G (1965) Adaptation and the form-function complex. Evolution 19:269-299.
Boinski S (1989) The positional behavior and substrate use of squirrel monkeys: ecological implications. J. Hum.
EvoJ. 18:659-678.
Cant JGH, and Temerin LA (1984) A conceptual approach to foraging adaptations in primates. In PS Rodman and
JGH Cant (eds.): Adaptations for Foraging in Nonhuman Primates. New York: Columbia University Press,
pp. 249-279.
Cant JGH (1988) Positional behavior of long-tailed macaques (Macacafascicularis) in northern Sumatra. Am. J.
Phys. AnthropoJ. 76:29-37.
42 S. E. Walker

Cartmill M (1985) Climbing. In M Hildebrand, OM Bramble, KF Liem, and DB Wake (eds.): Functional Verte-
brate Morphology. Cambridge: Belknap Press, pp. 73--88.
Cartmill M, and Milton K (1977) The lorisiform wrist joint and the evolution of "brachiating" adaptations in the
Hominoidea. Am. 1. Phys. Anthropol. 47:249-272.
Cavanagh GA (1977) Storage and utilization of elastic energy in skeletal muscle. Exercise Sports Sci. Rev.
5:89-129.
Crompton RH (1984) Foraging, habitat structure and locomotion in two species of Galago. In PS Rodman and
JGH Cant (eds.): Adaptations for Foraging in Nonhuman Primates. New York: Columbia University Press,
pp.73-111.
Dagosto M (1994) Testing positional behavior of Malagasy lemurs. A randomization approach. Am. J. Phys. An-
thropol. 94: 189-202.
Doran 0 (1992) Comparison of instantaneous and locomotor bout sampling methods: a case study of adult male
chimpanzee locomotor behavior and substrate use. Am. 1. Phys. Anthropol. 89:85-100.
Erickson GE (1957) The hands of New World primates, with comparative functional observations on the hands of
other primates. Am. J. Phys. Anthropol. 15:446.
Fleagle JG (1977) Locomotor behavior and skeletal anatomy of sympatric Malaysian leaf-monkeys (Presby tis ob-
scura and Presbytis melalophos). Am. 1. Phys. Anthropol. 46:297-308.
Fleagle JG, and Meldrum OJ (1988) Locomotor behavior and skeletal morphology of two sympatric pitheciine
monkeys, Chiropotes satanas and Pithecia pithecia. Am. J. Primatol. 16:227-249.
Fleagle JG, and Mittermeier RA (1980) Locomotor behavior, body size, and comparative ecology of seven Suri-
nam monkeys. Am. 1. Phys. Anthropol. 52:301-314.
Fontaine R (1990) Positional behavior in Saimiri boliviensis and Ate1es geofJroyi. Am. J. Phys. Anthropol.
82:485-508.
Garber PA (1980) Locomotor Behavior and Feeding Ecology of the Panamanian Tamarin (Saguinus oedipus geor-
froyi, Callitrichidae, Primates). Ph.D. Dissertation, Washington University.
Garber PA (1991) A comparative study of positional behavior in three species of tamarin monkeys. Primates
32:219-230.
Gebo DL (1992) Locomotor and postural behavior in Alouatta palliata and Cebus capucinus. Am. J. Primatol.
26:277-290.
Gebo DL, and Chapman CA (1995) Positional behavior in five sympatric Old World monkeys. Am. J. Phys. An-
thropol. 97:49-76.
Hill WCO (1960) Primates: Comparative Anatomy and Taxonomy, Vol. IV. Edinburgh: Edinburgh University
Press.
Hunt KD (1992) Positional behavior of Pan troglodytes in the Mahale Mountains and Gombe Stream National
Parks, Tanzania. Am. J. Phys. Anthropol. 87:83-106.
Hunt KD, Cant JGH, Gebo DL, Rose MD, Walker SE, and Youlatos DL (1996) Standardized descriptions of pri-
mate locomotor and postural modes. Primates 37:363-387.
Jolly CJ (1965) Origins and Specialisations of the Long-faced Cercopithecoids. Ph.D. Dissertation, University of
London.
Kay RF (1990) The phyletic relationships of extant and fossil Pitheciinae (Platyrrhini, Anthropoidea). In JG
Fleagle and AL Rosenberger (eds.): The Platyrrhine Fossil Record. New York: Academic Press, pp.
175-208.
Kinzey WG (1967) Preface to symposium on primate locomotion. Am. 1. Phys. Anthropol. 26: 115-118.
Mendel F (1976) Postural and locomotor behavior of Alouatta palliata on various substrates. Folia Primatol.
26:36--53.
Morbeck ME (1976) Leaping, bounding and bipedalism in Colobus guereza: a spectrum of positional behavior.
Yrbk. Phys. Anthropol. 20:408-420.
Napier JR (1962) Monkeys and their habitats. New Scientist 295:84--88.
Napier JR (1963) Brachiation and brachiators. Symp. Zoo I. Soc. Lond. 10: 183-195.
Oxnard, C (1984) The Order of Man. New Haven: Yale University Press.
Parol in P (1993) Forest inventory in an island of Lake Guri, Venezuela. In W Barthlott, CM Naumann, K Schmidt-
Loske, and KL Schuchmann (eds.): Animal-Plant Interactions in Tropical Environments, Proc. German
Soc. Tropical Ecology, Bonn, pp. 139-147.
Pernia JE (1985) Mapa de Fisiografia y Vegetacion del Area de Inundacion de la Tercera Etapa del Embalse el
Guri - Estado Bolivar. Internal manuscript: CVG - Electrificacion del Caroni (EDELCA), Venezuela.
Peters A, and Preuschoft H (1984) External biomechanics of leaping in Tarsius and its morphological and kine-
matic consequences. In Niemitz C (ed.): Biology of Tarsiers. New York: Gustav Fisher Verlag, pp.
227-255.
Prost J ( 1965) A definitional system for the classification of primate locomotion. Am. Anthropol. 67: 1198-1214.
Fine-Grained Differences within Positional Categories 43

Remis M (1995) Effects of body size and social context on the arboreal activities oflowland gorillas in the Central
African Republic. Am. 1. Phys. Anthropol. 97:413-434.
Ripley S (1967) The leaping of langurs: a problem in the study oflocomotor adaptation. Am. J. Phys. Anthropol.
26: 149-170.
Ripley S (1977) Gray zones and gray langurs: is the "semi"- concept seminal? Yrbk. Phys. Anthropol.
20:376-394.
Rodman PS (1979) Skeletal differentiation of Macaca Jascicularis and Macaca nemestrina in relation to arboreal
and terrestrial quadrupedal ism. Am. J. Phys. Anthropol. 51:51-60.
Rose MD (1974) Postural adaptations in New and Old World monkeys. In FA Jenkins (ed.): Primate Locomotion.
San Diego: Academic Press, pp. 75-93.
Rose MD (1979) Positional behavior of natural populations: some quantitative results of a field study of Colobus
gllereza and Cercopithecus aethiops. In ME Morbeck, D Preuschoft and N Gomberg (eds.): Environment,
Behavior and Morphology: Dynamic Interactions in Primates. New York: Gustav Fischer Verlag, pp.
75-94.
Rosenberger AL, and Stafford BJ (1994) Locomotion in captive Leontopithecus and Callimico: a multimedia
study. Am. 1. Phys. Anthropol. 94:379-394.
Schon Ybarra MA, and Schon III MA (1987) Positional behavior and limb bone adaptations in red howling mon-
keys (Alollatta senicllllls). Folia Primatol. 49:70-89.
Walker SE (1993) Positional Adaptations and Ecology of the Pitheciini. Ph.D. Dissertation, City University of
New York.
Walker SE (1996) Evolution of positional behavior in the saki/uakaris (Pithecia, Chiropotes, Cacajao). In M Nor-
conk, AL Rosenberger and P Garber (eds.): Advances in Primatology: Adaptive Radiations of Neotropical
Primates. New York: Plenum, pp. 335-367.
Walker SE (in revision) Leaping behavior of Pithecia pithecia and Chiropotes satanas, and the importance ofleap-
ing in primate locomotion. Submitted to Am. J. Phys. Anthropol., 41 ms. pages.
Ward SC, and Sussman RW (1979) Correlates between locomotor anatomy and behavior in two sympatric species
of Lemur. Am. J. Phys. Anthropol. 50:575-590.
 3

PATTERNS OF SUSPENSORY FEEDING IN

ALOUATTA PALLIATA, ATELES GEOFFROYI,
AND CEBUS CAPUCINUS

David J. Bergeson

Campus Box 1114

Department of Anthropology
Washington University
St. Louis, Missouri 63130

1. INTRODUCTION

The positional behavior of a primate is not an end in itself, but is instead a means to
many ends (Andrews and Groves, 1976; Charles-Dominique, 1990; Cant, 1992). One of
the most important of these is food acquisition. Compared to other orders of mammals,
primates consume a wide variety of food resources, most of which can be subsumed under
the categories of fruit, leaves, seeds, exudates, and insect prey (Harding, 1981; Sussman,
1987). These food resources often present alternative demands to a primate consumer and
probably greatly influence patterns of substrate preference and canopy use. Although data
suggest that broad correlations do not exist between diet and positional behavior, it is
likely that specific positional behaviors are particularly important for specific resources in
the diets of many primate species (Fleagle and Mittermeier, 1980; Fleagle, 1984). The
identification of such relationships is critical if we are to interpret the positional behaviors
of extant primates, or the reconstructed positional behaviors of extinct primates. Only
when we know the ecological context of locomotor modes and feeding postures can we
begin to address questions concerning their adaptive significance (Fleagle, 1979).
It is likely that suspension has a relatively high adaptive significance in many pri-
mate species. In several primates, such as the hylobatids and atelines, suspension is a rela-
tively frequent positional behavior, particularly during feeding. The evolutionary and
biomechanical significance of frequent positional behaviors is potentially high, as there is
a relatively large opportunity for energy conservation if a frequent positional behavior is
made more efficient, and there is a relatively critical need for shaping the skeleton and
muscle to prevent fatigue or injury (Hunt, 1992). Frequency is obviously not the only cri-
terion for estimating the influence of specific positional behaviors, as those locomotor

4S
46 D. J. Bergeson

modes or postures that an animal most commonly performs do not necessarily represent
the behaviors for which the animal has been selectively adapted. Instead, physical stress is
also an important criterion: the anatomical adaptations required for the efficient perform-
ance of a particular positional mode are determined by the muscular effort required for it
and the associated stress in the musculoskeletal system (Hunt, 1992). The degree to which
suspension imparts a large amount of physical stress to the musculoskeletal system and re-
quires many anatomical specializations for efficient performance is well established (e.g.,
Stern and Oxnard, 1973; Andrews and Groves, 1976; Cartmill and Milton, 1977; Fleagle,
1983; Swartz et aI., 1989; Takahashi, 1990; Rose, 1993; Swartz, 1993; Lemelin, 1995).
An additional criterion to use when estimating the evolutionary significance of a po-
sitional behavior is its behavioral or ecological context. If the positional behavior is used
in very restricted behavioral or ecological contexts, it is possible that the positional behav-
ior has an adaptive relationship with these contexts (Clutton-Brock and Harvey, 1979;
Fleagle, 1979; Hunt, 1992). This is particularly true if the contexts have relatively high
survival values (Prost, 1965). Suspension has been suggested to be particularly common in
specific behavioral and ecological contexts and, by virtue of these associations, has been
suggested to have an adaptive relationship with these contexts. Specifically, it has been
suggested that suspension is adaptively linked with frugivory (Napier and Napier, 1967;
Andrews and Groves, 1976; Mittermeier and Fleagle, 1976; Altmann, 1989), feeding
among small branches (Napier, 1967; Fleagle and Mittermeier, 1980; Cartmill, 1985;
Cant, 1992), and feeding in the periphery of a tree crown (Avis, 1962; Napier, 1967; Car-
penter and Durham, 1969; Grand, 1972; Mendel, 1976). In this paper, I examine the effect
of diet, branch size, and crown location on the positional behavior during feeding and for-
aging of Alouatta palliata, Ateles geoffroyi, and Cebus capucinus. Specifically, I test
whether suspensory feeding and foraging positions are associated with fruit resources,
small branches, and the crown periphery.

2. METHODS

Data on Alouatta palliata, Ateles geoffroyi, and Cebus capucinus were collected
from July, 1993 to July, 1994 at Santa Rosa National Park and the La Selva Biological
Station in Costa Rica. Data were collected at two minute intervals using an instantaneous,
focal animal sampling method. Among the data collected were activity (feeding and forag-
ing), positional behavior, diet, support size and number of supports used, and location in
tree crown (Bergeson, 1996).
Feeding was defined as holding or processing a food item, and foraging was defined
as actively searching for food. With the exception of the howling monkeys at Santa Rosa,
individual monkeys were not marked and thus were identifiable only to broad age and sex
classes.
A system of 25 locomotor modes and postures was used to classify the positional be-
havior of focal animals (Bergeson, 1996). The goal of this paper is not to thoroughly docu-
ment the positional behavior of Alouatta, Ateles, and Cebus, but rather to examine the
effects of support characteristics and diet on major categories of locomotion and posture.
Thus, the rates of only the four most common positional categories are examined in this
paper: inverted bipedal, suspension, quadrupedal stand, and sit. In addition, an inclusive
"other" category was defined for the remaining 21 categories. Locomotion, which was
relatively uncommon during feeding, is included in the "other" category. The posture of
the focal animal was classified as inverted bipedal when the animal supported itself facing
Patterns of Suspensory Feeding in Alouatta palliata, Ateles geo//royi, and Cebus capucinus 47

down a branch by wrapping its tail around the branch and extending its legs. The forearms
were not used to support the body and were usually free to manipulate food items, if nec-
essary. This posture was typically used on, but was not restricted to, relatively large verti-
calor oblique branches, and is illustrated in Figure 2E of Mittermeier and Fleagle (1976).
When a focal animal used the inverted bipedal posture on horizontal branches, the legs of
the animal were extended out from the side of the branch, and the body of the animal was
perpendicular to the main axis of the branch. The inverted bipedal posture was distin-
guished from suspension by virtue of hindlimb and trunk extension. Suspension was de-
fined as a below branch position, with any combination of the hindlimbs, tail and
forelimbs used in grasping. The limbs that were acting in tension to support the animal
were recorded (e.g., 2 forelimbs and tail). In suspension, unlike the inverted bipedal pos-
ture, the hindlimbs were under tension or hung free (they were not under compression),
and the trunk was not extended out from the support. The posture of the focal animal was
classified as quadrupedal stand when the pronograde animal was supported above-branch
by compression of three or four limbs. This posture typically occurred on a relatively hori-
zontal branch or branches. Posture was classified as sit when the weight of the animal was
supported only by the ischia. Typically the trunk was orthograde, the legs were flexed, and
arms supported little or no weight.
The kind of resource items that were being eaten by the focal animal at the sample
point were recorded. These resource items included fruit, leaves, flowers, insects, vine
fruit, vine leaves, vine stems, vine flowers, pith, vertebrates, shoots or twigs, and other
(including unknown).
The diameter of the support(s) used by the focal animal was estimated in centime-
ters. When a focal animal used more than one support, usually the supports were approxi-
mately the same size. On the few occasions when the supports were not the same size, the
size of the support that supported the most weight was recorded. The diameter estimations
were later corrected using equations derived from the regression of the actual diameters of
a sample of 102 branches (as measured repeatedly by a tape) on the estimated diameters of
these branches. The following branch size categories were used in data analyses: small
(1-2 cm in diameter), medium (3-6 cm), large (7-15 cm) and very large (over 15 cm). The
number of supports used by the focal animal was recorded according to the following
categorization: Single - the focal animal was supported by one branch or support; Collat-
eral - the focal animal was supported mainly by one support, with some minor help from
one or two collateral supports; and Multiple - the focal animal was supported equally by
at least two supports of identical characteristics.
Data were collected on the crown use of focal animals by dividing trees into four
concentric crown areas. Each crown area corresponded to one fourth of the distance from
the trunk to the periphery of the tree crown. Crown area one included the trunk and the in-
nermost one-fourth of the branch lengths. Crown areas two and three were more periph-
eral than crown area one. Crown area four was the terminal branch region, and
corresponded to the distal one-fourth of the tree crown. Crown location was estimated
without reference to support size; that is, although branches in the crown periphery were
generally small, it was possible for an animal to use large branches and still be in crown
area four.
Results are presented in this paper as the proportion of all two minute samples in
which each positional behavior pattern occurred. For the purposes of data analyses, all in-
stantaneous samples from each species were pooled to form three data sets corresponding
to Alouatta, Ateles, and Cebus positional behavior samples. Between-site variation in posi-
tional behavior was low compared to interspecific variation, and the data patterns dis-
48 D. J. Bergeson

cussed in this paper occurred at both Santa Rosa and La Selva. Over one hundred hours of
data were collected on each species, including 797 Alouatta feeding and foraging samples,
764 Ateles feeding and foraging samples, and 1192 Cebus feeding and foraging samples.
The statistical significance of data patterns was evaluated using randomization pro-
cedures. Unlike conventional parametric methods, randomization tests do not make any
assumptions about the distribution of the population from which the data were collected.
Instead, they create distributions of the test statistic from permutations of the observed
data (e.g., Manley, 1991; Potvin and Roff, 1993; Dagosto, 1994). Significance levels are
determined by the percentage of test statistic values equal to or more extreme than the
value from the original data set. I used Manley's (1991) two sample randomization test
where the test statistic used in the randomization loops, i.e., the mean differences between
groups, is equivalent to the Students t statistic. When examining interspecific differences
in categorical data (such as locomotion or posture), the use rate of each category was the
test statistic in the randomization loops. In this project, 1000 permutations were used to
estimate the probability value (P). I use these tests to determine the probability that the ob-
served differences in rates occur if no relationship exists between the variables being
tested. If the observed difference occurs less than one time out of 100 (i.e., P<O.Ol), then I
consider the difference to be statistically significant.
In this paper, I chose not to evaluate the context specificity of positional behaviors
using the percent samples of a positional behavior that occur in a particular area. Instead, I
use the percentage of feeding and foraging context samples that is achieved using a par-
ticular positional behavior as a measure of context specificity. For example, when address-
ing the question of whether suspension is associated with the crown periphery, I do not
calculate the percent of suspension samples that occur in the crown periphery, and com-
pare this with the percent of suspension samples that occur in the crown interior. This
comparison is affected by the distribution of feeding in the tree crown: most suspension
occurs in the tree periphery because most feeding occurs in the tree periphery. Instead, I
examine the association of suspension and crown use by calculating the rate of suspension
in different areas of the crown. If the rate of suspension in the crown periphery is signifi-
cantly higher than the rate of suspension in other parts of the tree crown, then I would
conclude that this difference in rates is not random. I would further conclude that suspen-
sion is likely a positional strategy that is particularly suited to solve the problems associ-
ated with feeding and foraging in the crown periphery.

3. RESULTS

3.1. Diet
Previous studies have indicated that Alouatta, Ateles, and Cebus differ significantly
in their patterns of resource use (e.g., Chapman, 1987). Similar patterns were exhibited by
the primates in this study (Table I). Alouatta relied primarily on leaves, although it also
frequently consumed fruit and flowers. The primary resource in the diet of Ateles was
fruit, which was supplemented by flowers, vine resources, and leaves. Like Ateles, Cebus
relied on fruit resources throughout the year. Unlike Ateles, however, Cebus fed heavily
on invertebrate prey, often descending to the ground in search of insects in the leaf litter.
Cebus also was observed feeding on vertebrate prey such as mice, birds, and young coati.
It is clear that feeding and foraging were postural activities in all three species (Ta-
ble 2). In Alouatta, feeding and foraging were achieved primarily above-branch, in a sit-
Patterns of Suspensory Feeding in A/ouatta palliata, Ate/es geofJroyi, and Cebus capucinus 49

Table 1. Diet of Alouatta palliata, Ateles geoffroyi, and Cebus

capucinus at Santa Rosa National Park and La Selva Biological
Station. Data are percentages of each species diet
Diet Alouatta Ateles Cebus
Fruit 25.7 59.9 43.4
Leaves 46.9 7.7 0.0
Flowers 22.2 12.4 0.0
Vine fruit 1.9 10.4 0.9
Vine leaves and stems 2.9 8.5 0.0
Invertebrate fauna 0.0 0.0 49.2
Other 0.4 l.l 6.5

ting or quadrupedal standing position. Below branch suspension occurred in almost 13%
of feeding and foraging samples. In Ateles, suspension and sitting were the most important
postures used when feeding and foraging, although the inverted bipedal posture and quad-
rupedal stand were also frequently used. Like Alouatta, Cebus usually fed or foraged by
sitting or standing quadrupedally; however, Cebus differed from Alouatta in that it used
suspensory postures relatively rarely.
The form of suspensory feeding and foraging postures differ greatly between howl-
ing monkeys and spider monkeys: the former almost always use tail only suspension
(40.9% of suspensory feeding samples) or tail plus two legs suspension (34.4%), while
spider monkeys are capable of suspending from any combination of arms, legs, and tail,
particularly tail only suspension (40.7% of suspensory feeding samples) and one arm plus
tail (35.9%).
None of the species used suspension more often when feeding and foraging on fruit
than when feeding and foraging on other resources (Table 3). Nonetheless, in all three spe-
cies different positional behaviors were used to acquire different dietary items. In
Alouatta, the rate of suspension was highest when feeding on leaves and was lowest when
feeding on fruit (Table 4). In addition, the rate of quadrupedal standing was significantly
higher when it fed on fruit than on other items. Ateles also modified its positional behavior
with respect to resource items, but again, the rate of suspension was not higher when feed-
ing on fruit (Table 4). Instead, the most significant trends were associated with feeding on
flowers: Ateles had a lower rate of suspension and a higher rate of sitting. In Cebus, sus-
pension was relatively rare when feeding and the rate was not affected by resource type
(Table 4). The rate of the inverted bipedal posture was significantly higher when feeding

Table 2. The positional behaviors used while feeding and foraging.

Data are percent total feeding samples in each species

Alouatta Ateles Cebus

Positional behavior N=735 N=700 N=1127
Inverted bipedal posture 8.6 13.1 7.0
Quadrupedal stand 13.5 12.6 25.1
Sit 56.8 25.7 39.2
Suspension 12.7 33.3 2.7
Other 8.4 15.3 26.0
50 D. J. Bergeson

Table 3. Rates of suspensory feeding and foraging on

fruit versus all other resources
Other
Species Fruit resources P
Alouatta 8.5 14.3 0.043
Ateles 33.0 32.5 0.936
Cebus 2.8 2.7 1.000

on fruit than on other resources and the rate of sitting was relatively low when feeding on
invertebrate fauna.

3.2. Branch Size

In all three species, smaller supports were the focus for feeding and foraging (Ta-
ble 5). In Alouatta, over sixty percent of feeding and foraging samples occurred on small
supports, and almost thirty percent occurred on medium supports. Feeding and foraging
rarely occurred on large or very large supports. In Aleles, over half of feeding and forag-
ing samples occurred on small supports, and many occurred on medium supports. Cebus
displayed a pattern of support use similar to that of Alouatta.

Table 4. The effect of diet on the positional behavior. Data are percent total positional behavior
samples while feeding or foraging on each resource

Vines leaves Invert.

Species Positional behavior Fruit Leaves Flowers Vine fruit & stems fauna Other
A louatta , N=200 N=385 N=159
Inverted bipedal 10.5 7.4 7.5
Quadrupedal stand 23.5 9.4 9.4
Sit 49.5 58.4 63.5
Suspension 8.5 16.5 9.4
Other 8.0 8.3 10.2
Ateles' N=397 N=53 N=85 N=69 N=63
Inverted bipedal 10.6 \3.2 16.9 17.4 22.2
Quadrupedal stand 14.1 3.8 13.5 15.9 9.5
Sit 24.2 32.1 40.4 14.5 19.0
Suspension 33.0 41.5 19.1 33.3 44.4
Other 18.2 9.4 10.1 18.9 4.9
Cebus 3 N=457 N=519 N=60
Inverted bipedal 10.5 5.4 3.3
Quadrupedal stand 21.9 28.7 19.7
Sit 42.0 33.1 65.6
Suspension 2.8 2.5 1.6
Other 22.8 30.3 9.8
'Significant Differences: Inverted Bipedal - none; Quadrupedal Stand - Fruit vs. Leaves (P<O.OO I). Fruit vs. Flowers (P<O.OO I);
Sit - none; Suspension - Fruit vs. Leaves (P=O.003).
2Significant Differences: Inverted Bipedal - none; Quadrupedal Stand - none; Sit - Flowers vs. Vine Fruit (P<O.OOI). Flowers vs.
Vine Leaves and Stems (P<O.OO I). Flowers vs. Fruit (P=O.002); Suspension - Fruit vs. Flowers (P=O.005). Leaves vs. Flowers
(P=O.005). Flowers vs. Vine Leaves and Stems (P=O.002).
3Significant Differences: Inverted Bipedal - Fruit vs. Other (P=O.OO I). Fruit vs. Invert. Fauna (P=O.007); Quadrupedal Stand -
none; Sit - Invert. Fauna vs. other (P<O.OO I); Suspension - none.
Patterns of Suspensory Feeding in Alouatta palliata, Ateles geoffroyi, and Cebus capucinus 51

Table 5. Support sizes used during feeding and foraging.

Data are percentages of total feeding samples
for each species
Alouatta Ateles Cebus
Branch size N=703 N=602 N=999
Small 63.3 58.2 64.2
Medium 27.3 33.8 25.9
Large 8.4 6.9 7.8
Very Large 1.0 1.1 2.1

Only in Alouatta was suspension significantly more common on small branches than
on larger branches (Table 6). Branch size affected the positional behavior of Alouatta differ-
ently than it affected the positional behavior of Ateles or Cebus. Unlike Mendel (1976), I
did not find that howling monkeys used suspension almost exclusively when feeding on
small supports. Instead, sitting remained the dominant howling monkey feeding posture,
even on very small branches (Table 7). Nonetheless, the rate of suspension was higher on
small branches than on larger branch sizes. The distributions of branch sizes used by
Alouatta during above-branch feeding (quadrupedal stand and sit) and suspension suggest
that this is because not only was suspension preferred on small branches, but also suspen-
sion was avoided on large branches (Figure 1, top). Alouatta rarely used suspension when
on branches over ten centimeters in diameter. However, the distributions of branch sizes
used during above and below branch feeding differed greatly in the small branch range, and
confirm the association of suspension with small branches. In Alouatta, the mean branch
size used during suspension was significantly smaller than the mean branch size used during
above-branch feeding. The mean branch sizes are relatively unaffected by extreme values,
as the mean branch size used during suspension was significantly smaller than the mean
branch used during above-branch feeding, even after excluding two extreme values over 15
cm. In Alouatta, sitting was used at different rates on different branch sizes. Alouatta sat
more often on medium and large branches than on small and very large branches, where
suspension and the inverted bipedal posture were used, respectively (Table 7).
In Ateles, the rate of suspension was much lower on large and very large branches
than it was on small or medium branches (Table 7). This pattern could be caused by either a
preference for suspension on small branches or an avoidance of suspension on larger
branches. The data in Figure I (bottom) suggest that the latter explanation is more likely
correct than the former. The fact that the distributions are very similar in the small branch
range suggests that in Aleles, suspension does not necessarily allow greater use of small

Table 6. Rates of suspensory feeding on small branches

versus other branches. Percent feeding and foraging
samples in branch sizes in which suspension was used

Small All other

Species branches branches P
Alouatta 17.9 5.2 <0.001
Ateles 31.4 31.2 1.000
Cebus 3.0 I.7 0.303
52 D. J. Bergeson

Table 7. The effect of branch size on the positional behavior while feeding and
foraging. Data are percent total positional behavior samples on each
branch size category
Species Positional behavior Small Medium Large Very large
A louatta , Inverted bipedal 7.2 8.1 10.3 71.4
Quadrupedal stand 15.2 9.2 8.6 28.6
Sit 50.8 68.7 70.7 0.0
Suspension 18.0 4.9 6.9 0.0
Other 8.8 9.1 3.5 0.0
Atelei Inverted bipedal 11.2 13.4 22.0 42.9
Quadrupedal stand 15.3 8.9 4.9 0.0
Sit 24.2 31.2 39.0 57.1
Suspension 31.4 35.2 17.1 0.0
Other 17.9 11.3 17.0 0.0
Inverted bipedal 5.9 9.3 15.8 14.3
Quadrupedal stand 27.4 17.5 19.7 47.6
Sit 37.6 48.6 43.4 28.6
Suspension 3.1 2.0 1.3 0.0
Other 26.0 22.7 19.8 9.5
'Significant Differences: Inverted Bipedal - Small vs. Very Large (P<O.OO I), Medium vs. Very Large
(P<O.OO I), Large vs. Very Large (P<O.OO I); Quadrupedal Stand - none; Sit - Small vs. Medium
(P<O.OO I), Small vs. Large (P=O.006), Small vs. Very Large (P=O.OO t), Medium vs. Very Large
(P<O.OOI), Large vs. Very Large (P<O.OOI); Suspension - Small vs. Medium (P<O.OOI).
2Significant Differences: Inverted Bipedal - Small vs. Very Large (P=O.OO I), Medium vs. Very Large
(P=O.002); Quadrupedal Stand - none; Sit - Small vs. Very Large (P=O.004); Suspension - none.
3Significant Differences: Inverted Bipedal - Small vs. Large (P=O.006); Quadrupedal Stand - Small vs.
Medium (P=O.003), Medium vs. Very Large (P=O.003); Sit - Small vs. Medium (P=O.004); Suspen-
sion - none.

branches, and instead, more precisely, above-branch postures allow greater use of very large
branches. In Ateles, the mean branch size used during suspension was smaller than the mean
branch size used during above-branch feeding (quadrupedal stand and sit). This measure of
context specificity is potentially misleading, however, as the mean branch size used during
above-branch feeding is greatly affected by four samples of branch sizes over 15 cm. If
these four samples are excluded, the difference in means is no longer significant. One likely
reason why suspension was not associated with small branches in Aleles is their use of mul-
tiple branches. When data only from single branch samples are analyzed, the rate of suspen-
sion was much higher on small branches (Table 8). Aleles also displayed other trends in
positional behavior with respect to branch size. Although the data in Table 7 are suggestive
that the rate of the inverted bipedal posture increased with branch size category, only the
contrasts between small and very large branches and between medium and very large
branches were significant. Sitting was more common on very large than on small branches.
Suspensory feeding and foraging was relatively rare in Cebus, and was not affected
by branch size (Table 7). Quadrupedal standing was common on both small and very large
branches, and the rate of sitting was not as high on small branches as on medium
branches. As in Alouatta and Ateles, the inverted bipedal posture was strongly associated
with larger branches.

3.3. Crown Location

In all three species, the crown periphery was the most common site of feeding and
foraging (Table 9). The atelines were particularly restricted to the crown periphery when
Patterns of Suspensory Feeding in Alouatta palliata, Ateles geoffroyi, and Cebus capucinus 53

90

80
III
CII
Q. 70
E 60
'"
In
50
C
CII
u
~ 40
CII
Q.
30
20
10
0
$25 2.6 - 4 .5 4.6 - 6.5 6.6 -10.5 10.6 -15.5 15.6 - 20.5 >20.5

Branch Size (cm)

60

III 50
CII
Q.
E 40
'"
In
C 30
CII
~
CII
Q. 20

10

0
$ 25 2.6 - 4.5 4 .6 - 6 .5 6 .6 10.5
- 10.6-15.5 15.6 - 20.5 >205

Branch Size (cm)

Figure L Branch sizes of above-branch postures and suspension. Black bars. above-branch postures. Hatched
bars, suspension. (Top) Howling monkeys: mean above-branch branch size = 3.1 cm, mean suspension branch size
= 2.2 cm, P<O.OOI. (Bottom) Spider monkeys: mean above-branch branch size = 3.4 cm, mean suspension branch
size = 2.8 cm, P=0.028.

Table 8. The relationship between suspension and branch

size on single branches. Data are percent suspension
samples in each branch size category

Branch size A/oualta Ate/es Ceblls

1-2 cm 28.6 75.4 5.5
3-6 cm 6.4 42.0 1.3
7-15 cm 8.7 22.2 1.6
>15cm 0.0 0.0 0.0
54 D. J. Bergeson

Table 9. Crown use during feeding and foraging. Data

are percent total feeding samples in each crown area
Alouatta Ateles Cebus
Crown Area N=769 N=716 N=1019
Area I 2.5 5.0 17.6
Area 2 3.1 5.7 13.2
Area 3 13.7 14.9 18.4
Area 4 80.7 74.4 50.8

feeding and foraging, as Alouatta and Ateles fed and foraged in the periphery in over 80%
and 70% of samples, respectively. Cebus fed and foraged over a much greater portion of
the tree crown than did Ateles or Alouatta, although the periphery was nonetheless the
most common site of feeding and foraging.
Although for each species crown area did affect the positional behaviors used to ac-
quire resources, in none of the species did the rate of suspensory feeding positions in-
crease in the crown periphery (Table 10). A more detailed analysis of the effect of crown
location on the positional behavior of feeding and foraging suggests different patterns. In
Alouatta, suspensory feeding did not increase in the periphery, and the inverted bipedal
posture was closely associated with the trunk region (Table II). Ateles displayed similar
patterns, as the rate of quadrupedal standing but not suspension was elevated in the pe-
riphery (Table II). Sitting was most common in crown area 2, and the inverted bipedal
posture was closely associated with crown area 1. The positional behavior of Cebus during
feeding and foraging was not as affected by crown position as was the positional behavior
of the atelines (Table II). Nonetheless, one significant trend that was found in the atelines
was also found in Cebus: during feeding and foraging the rate of the inverted bipedal pos-
ture was significantly higher in the trunk region.

4. DISCUSSION

4.1. Positional Behavior and Diet

The hypothesis that suspension is adaptively linked with frugivory has traditionally
been supported by data from the hylobatids: data collected on the positional behavior of
gibbons and siamangs suggest that suspensory feeding positions are more common when
they feed on fruit than when they feed on leaves (Chivers, 1974; Fleagle, 1976). In other
apes that frequently use suspensory postures, however, suspension is not particularly asso-

Table 10. Rates of suspensory feeding in crown area 4

versus all other crown areas. Data are percent total
feeding and foraging samples in each area
in which suspension used

Species Crown area 4 AII other areas p

Alollatta 13.8 9.2 0.158
Ateles 30.7 37.5 0.107
Cehlls 3.5 2.2 0.334
Patterns of Suspensory Feeding in Alouatta palliata, Ateles geofJroyi, and Cebus capucinus 55

Table 11. The effect of crown location on the positional behavior while feeding and
foraging. Data are the percent total positional behavior samples in each crown area
Species Positional behavior Crown area I Crown area 2 Crown area 3 Crown area 4
Alollatta' Inverted bipedal 63.2 13.0 6.9 6.9
Quadrupedal stand 10.5 0.0 10.9 14.6
Sit 15.8 47.8 69.3 55.9
Suspension 5.3 17.4 7.9 13.8
Other 5.2 21.8 5.0 8.8
Ateles1 Inverted bipedal 41.7 23.7 13.7 10.1
Quadrupedal stand 11.1 2.6 3.9 15.3
Sit 8.3 42.1 27.5 26.0
Suspension 36.1 28.9 41.2 30.7
Other 2.8 2.7 13.7 17.9
Ceblls 3 Inverted bipedal 19.4 3.8 4.3 5.8
Quadrupedal stand 20.0 25.2 20.5 27.4
Sit 41.1 42.0 49.7 35.5
Suspension 2.3 3.1 1.6 3.5
Other 17.2 25.9 23.9 27.8
'Significant Differences: Inverted Bipedal - I vs. 2 (P<O.OOI), I vs. 3 (P<O.OOI), I vs. 4 (P<O.OOI); Quadru-
pedal Stand - none; Sit - I vs. 3 (P<O.OO I), I vs. 4 (P<O.OO I), 3 vs. 4 (P=O.007); Suspension - none.
1Significant differences: Inverted Bipedal-I vs. 3 (P<O.OOI), I vs. 4 (P<O.OOI); Quadrupedal Stand - 3 vs. 4
(P=O.OOI); Sit - I vs. 2 (P<O.OOI); Suspension- none.
3Significant Differences: Inverted Bipedal - I vs. 2 (P<O.OOI), I vs.3 (P<O.OOI), I vs.4 (P<O.OOI); Quadru-
pedal Stand - none; Sit - 3 vs. 4 (P=O.002); Suspension - none.

ciated with frugivory. Cant (1987), for example, found that female Pongo pygmaeus used
suspensory feeding postures in 59% of leaffeeding samples and 43% offruit feeding sam-
ples. Hunt (1992) found that the rates of suspensory feeding in Pan troglodytes was about
ten percent for both fruit and leaf feeding samples. Less is documented about the relation-
ship between suspensory feeding and resource use in New World monkeys, although
Bicca-Marques and Calegaro-Marques (1993) found a small difference in the rate of sus-
pension by Alouatta caraya when it fed on fruit and leaves (18% vs. 14%).
In this study, the positional behavior of food acquisition was affected by resource
use in each species. The effect of resource use was not as strong as was the effect of
crown location or support size. This reinforces suggestions that the positional behavior
of arboreal primates is not directly determined by resources, but instead it is affected in-
directly by the location of these resources in the tree crown and the characteristics of the
specific supports that contain these resources (e.g., Fleagle, 1984; Walker, 1993, 1996).
Although there were trends in each species, in none of the species was suspension associ-
ated with fruit. Both Alouatta and Ateles preferred to sit while feeding on flowers, which
were obtained relatively rarely in suspensory positions. Bicca-Marques and Calegaro-
Marques (1993) also found that Alouatta caraya preferentially used sitting postures when
feeding on flowers. One possible explanation for this relationship between sitting and
feeding on flowers is the increased manipulation required when harvesting flowers.
Flowers occur in the same location as the fruits that later replace them. But when feeding
on flowers, the primates in this study inspected and manipulated the flowers much more
than they inspected and manipulated fruit when feeding on it. The tightest association be-
tween positional behavior and resource in Cebus was between the inverted bipedal pos-
ture and fruit.
Mittermeier and Fleagle (1976) emphasized that suspended postures are important
feeding adaptations for spider monkeys. By suspending itself by its tail alone, or by its tail
56 D. J. Bergeson

and any combination of its hindlimbs, a spider monkey can increase its feeding sphere
150% of that available when sitting or standing quadrupedally. Suspension is much more
frequent in the atelines than in other platyrrhines, and it is clear that the feeding sphere of
ate lines is increased by the possession of a prehensile tail. This increase, however, is asso-
ciated with all parts of the tree crown, and is not restricted to the periphery of the tree.
Similarly, this increase is not solely associated with frugivory, but is associated with all re-
sources. Prehensile tail use, like suspensory feeding, is common in all parts of the tree
crown, and occurs frequently regardless of resource type (Bergeson, 1996). The prehensile
tail is not only used for suspensory feeding in the crown periphery, but instead serves sev-
eral biological roles in each species. These roles include increasing the feeding sphere via
suspensory postures, increasing the ability to balance on small branches, freeing the hands
during feeding activities, increasing the ability to feed as well as forage in the crown pe-
riphery, increasing the ability to feed and move on vertical supports, and increasing the
ability to cross gaps between trees (Bergeson, 1996).

4.2. Positional Behavior and Branch Size

The mechanics of arboreal movement dictate that small branches present unique
problems of access, and it is reasonable to expect arboreal primates to modify their posi-
tional behavior in order to compensate for these problems. The stability of an animal lo-
cated above a branch is largely determined by the ratio of the size of the animal to the
size of the branch (Napier, 1967; Cartmill, 1985). On small branches, small displace-
ments by an arboreal primate can lead to frequent pitching and toppling, because small
movements cause its center of mass to move lateral to the branch surface. Because these
problems of stability on small branches can be avoided if an arboreal animal hangs be-
neath a branch, many researchers have predicted that suspensory positions should be
more common on small branches (e.g., Napier, 1967; Fleagle and Mittermeier, 1980;
Cant, 1992).
Suspension is not the only way for an arboreal primate to solve the problems ofbal-
ance that are associated with moving and feeding on small branches. Another way is to use
more than one branch to support above-branch movement (e.g., Cartmill and Milton,
1977; Fleagle and Mittermeier, 1980; Cartmill, 1985; Fleagle, 1985). By distributing its
body weight over many branches, an arboreal primate has greater control over pitching
and rolling movements. In addition, the weight of a large arboreal primate is less likely to
break several small branches than one small branch.
Given that suspension and multiple branch use are alternate solutions to the same
problems, which solution is adopted more frequently by Alouatta, Ateles, and Cebus? Data
in this study suggest that multiple branches are used more often than suspension to solve
problems associated with small branch use. In all three species, feeding on small branches
was most often achieved using quadrupedal, above-branch positions on multiple branches
(Table 12). Suspension on a single branch was relatively rare. In fact, Ateles used single
branches in less than 25% of feeding samples on small branches. The single branch model
used to predict the influence of support size on positional behavior (e.g., Napier, 1967) is
heuristically useful, but often disregards more common solutions.
Although spider monkeys use suspension more often than do howling monkeys,
howling monkeys are more restricted to small branches when using suspension than are
spider monkeys (Figure I). One possible explanation for this restriction is the limitation
imposed by hindlimb suspension on howling monkeys, and the relative versatility of fore-
limb suspension in spider monkeys. While spider monkeys frequently hang from their
Patterns of Suspensory Feeding in A/ouatta palliata, Ate/es geoffroyi, and Cebus capucinus 57

Table 12. Positional solutions on small branches during feeding. Data

are percent total positional behavior samples on small branches
Positional solutions Alouatta Ateles Cebus
Above-branch (Single Br.) 18.6 5.7 30.3
Above-branch (Multiple/Collateral Br.) 61.5 51.9 66.5
Suspension (Single Br.) 7.6 18.6 1.7
Suspension (Multiple/Collateral Br.) 12.3 23.8 1.5

arms, howling monkeys almost always hang from their legs and/or tail. It is possible that
the long arms and mobile shoulder of Ateles not only allow spider monkeys to use suspen-
sory positions more often than howling monkeys, but also enable spider monkeys to use a
wider variety of supports during suspension. On medium branches, the rate of suspension
in spider monkeys is over six times higher than the rate of suspension in howling monkeys
(Table 7). Perhaps this is because the maximum diameter of a branch that a howling mon-
key can efficiently grasp with its foot is smaller than the maximum diameter of a branch
that a spider monkey can efficiently hang from with its forelimbs.

4.3. Positional Behavior and Crown Location

It is clear that the characteristics of tree branches vary with respect to the branch's
position in the tree crown. Although trees exhibit a wide variety of crown shapes and
branching patterns, the structures of most trees are variations of the same pattern: large,
vertical, inflexible boles give rise to outwardly-radiating smaller, obliquely-inclined flex-
ible branches. Branches at the crown periphery are relatively very small, flexible, and are
oriented at many angles (e.g., Hom, 1971; Garber, 1980; Bertram, 1989). As the charac-
teristics of tree branches differ with respect to crown position, it is reasonable to expect
that the positional behavior of arboreal animals is affected by crown position (Ripley,
1967). In an influential paper, Grand (1972) focused on the mechanical problems associ-
ated with moving and feeding in terminal branches. Grand pointed out that by suspending
under a branch, a gibbon increases its feeding sphere 100% over that achieved in an
above-branch position. In addition, the deformation of terminal branches brings fruit
closer to a below-branch gibbon, and moves fruit farther away from an above-branch ma-
caque. Other authors have suggested that suspensory positions are particular useful and
common in the crown periphery (e.g., Avis, 1962; Napier, 1967; Carpenter and Durham,
1969; Grand, 1972; Mendel, 1976). Fleagle (1980) has pointed out that sitting while feed-
ing is a common approach to terminal branch feeding, and should probably be viewed as a
functionally different, but not necessarily a more efficient, method of foraging. Nonethe-
less, suspensory feeding has been associated with the crown periphery to the extent that it
has been assumed or implied that large primates cannot obtain food in the crown periphery
unless they hang while feeding (e.g., Hollihn, 1984).
Although location in the tree crown significantly affected the positional behavior of
each species in this study, the rate of suspensory feeding was not higher in the tree periph-
ery than in other areas of the tree. Instead, the tightest association was between the in-
verted bipedal posture and crown area one (the trunk region). The inverted bipedal
posture, characterized by an upside down body posture, hindlimb extension, and high ten-
sion prehensile tail use, was frequently used by each species when feeding in the trunk re-
gion. The inverted bipedal posture is possible only in prehensile-tailed animals, and is
 58 D. J. Bergeson

likely an important component of the manipulative, destructive, visually-oriented foraging

strategy of capuchin monkeys. It is significant that while in an inverted bipedal posture
the hands remain free and are thus able to manipulate food items. This freedom of the
hands may playa significant role in the foraging of Cebus, allowing capuchins to more
easily access palm fruit near the trunk, and allowing them to more efficiently rip apart and
search branches for invertebrate prey.

5. SUMMARY

Suspension has been suggested to be particularly common in certain behavioral and

ecological contexts, and by virtue of these associations, has been suggested to have an
adaptive relationship with these contexts. Specifically, it has been suggested that suspen-
sion is adaptively linked with frugivory, with feeding among small branches, and with
feeding in the periphery of a tree crown. In general, the data presented in this paper do not
support these suggestions. In all three species, suspension was common regardless of re-
source, and was not used more often when feeding and foraging on fruit. Branch size did
affect suspension in Alouatta, in that suspension was more common on small branches
than on large branches, but this pattern was not exhibited in Ateles or Cebus. Finally,
while in each species crown position did affect the positional behaviors used to acquire re-
sources, in none of the species did the rate of suspensory feeding positions increase in the
crown periphery.
Although data on the frequencies of particular positional behaviors are important,
these alone do not tell us what specific problems are solved by each positional behavior
(e.g., Cant, 1992), or the ecological consequences of specific positional behaviors (e.g.,
Charles-Dominique, 1990; Garber, 1992; Hunt, 1992). Instead, data on the context of posi-
tional behaviors are critical if we are even to begin interpreting the positional behavior of
extant primates (e.g., Garber, 1992; Hunt, 1992), or reconstructing the positional behavior
of extinct primates (e.g., Fleagle, 1976, 1983; Cartmill and Milton, 1977; Rose, 1993). This
paper is one example of how quantitative data on the specific contexts of positional behav-
iors can be used to identify new ecological relationships (e.g., the inverted bipedal posture
and trunks), confirm or reject proposed ones (e.g., suspension is associated with small
branches in howling monkeys, but not spider or capuchin monkeys), or identify functional
relationships between anatomy, behavior, and environment (e.g., forelimb suspension in spi-
der monkeys allows them to use a greater variety of branches than does hindlimb suspen-
sion in howling monkeys).

ACKNOWLEDGMENTS

Funding for this project was provided by: the L.S.B. Leakey Foundation, the Organi-
zation for Tropical Studies, Sigma Xi, and National Science Foundation Grant SBR-
9307631. I am grateful to the National Park Service of Costa Rica for allowing me to
work in Santa Rosa National Park from 1993 to 1994, and am grateful to the La Selva
Biological Station, Roger Blanco, Rodrigo Morera Avila, Orlando Vargas, and Sue Ber-
geson for their help with this project. I thank B.W. Bergeson, J. Fleagle, E. Strasser, and
four anonymous reviewers for their constructive comments on an earlier version of this
paper.
Patterns of Suspensory Feeding in A/ouatta palliata, Ate/es geojJroyi, and Cebus capucinus 59

REFERENCES
Altmann S (1989) The monkey and the fig: A socratic dialogue on evolutionary trends. Am. Sci. 77:256--263.
Andrews P, and Groves C (1976) Gibbons and brachiation. Gibbon and Siamang 4: 167-218.
Avis V (1962) Brachiation: The crucial issue for man's ancestry. Southw. J. Anthropol. 18: 119-148.
Bergeson DJ (1996) The positional behavior and prehensile tail use of Alauaua pa/liata, Ateles geaffi·ayi. and Ce-
bus capucinus. Ph.D. Dissertation, Washington University.
Bertram JEA (1989) Size-dependent differential scaling in branches: The mechanical design of trees revisited.
Trees 4:241-253.
Bicca-Marques J, and Calegaro-Marques C (1993) Feeding postures in the black howler monkey, Alauatta caraya.
Folia Primatol. 60: 169-172.
Cant JGH (1987) Positional behavior of female Bomean orangutans. Am. J. Primatol. 12:71-90.
Cant JGH (1992) Positional behavior and body size of arboreal primates: A theoretical framework for field studies
and an illustration of its application. Am. 1. Phys. Anthropol. 88:273--283.
Carpenter C, and Durham N (1969) A preliminary description of suspensory behavior in nonhuman primates. Pro-
ceedings ofthe Second International Congress Primatology, Atlanta, GA Volume 2. New York: Karger. pp.
147-154.
Cartmill M (1985) Climbing. In M Hildebrand, D Bramble, K Liem, and D Wake (eds.): Functional Vertebrate
Morphology. Cambridge: Belknap Press. pp. 73--88.
Cartmill M, and Milton K (1977) The lorisiform wrist joint and the evolution of "brachiating" adaptations in the
Hominoidea. Am. 1. Phys. Anthropol. 47:249-272.
Chapman CA (1987) Flexibility in diets of three species of Costa Rican primates. Folia Primatol. 49:90--105.
Charles-Dominique P (1990) Ecological adaptations related to locomotion in primates: An introduction. In F Jouf-
froy, M Stack and C Niemitz (eds.): Gravity, Posture, and Locomotion in Primates. Firenze: Editrice II
Sedicesimo, pp. 19-31.
Chivers DJ (1974) The siamang in Malaya: A field study ofa primate in tropical rain forest. Contrib. Primatol.
4:1-331.
Clutton-Brock TH, and Harvey PH (1979) Comparison and adaptation. Proc. R. Soc. Lond. B 205:547-565.
Dagosto M (1994) Testing positional behavior of Malagasy lemurs: A randomization approach. Am. J. Phys. An-
thropol. 94: 189-202.
Fleagle JG (1976) Locomotion and posture of the Malayan siamang and implications for hominoid evolution. Fo-
lia Primatol. 26:245--269.
Fleagle JG (1979) Primate positional behavior and anatomy: Naturalistic and experimental approaches. In ME
Morbeck, H Preuschoft, and N Gomberg (eds.): Environment, Behavior, and Morphology: Dynamic Inter-
actions in Primates. New York: Gustav Fischer, pp. 313--325.
Fleagle JG (1980) Locomotion and posture. In DJ Chivers (ed.): Malayan Forest Primates. New York: Plenum
Press, pp. 191-207.
Fleagle JG (1983) Locomotor adaptations of Oligocene and Miocene hominoids and their phyletic implications. In
R Ciochon and R Corruccini (eds.): New Interpretations of Ape and Human Ancestry. New York: Plenum
Press, pp. 301-324.
Fleagle JG (1984) Primate locomotion and diet. In D Chivers, B Wood, and A Bilsborough (eds.): Food Acquisi-
tion and Processing in Primates. New York: Plenum Press, pp. 105--117.
Fleagle JG (1985) Size and adaptation in primates. In W Jungers (ed.): Size and Scaling in Primate Biology. New
York: Plenum Press, pp. 1-19.
Fleagle JG, and Mittermeier RA (1980) Locomotor behavior, body size, and comparative ecology of seven Suri-
nam monkeys. Am. J. Phys. Anthropol. 52:301-314.
Garber PA (1980) Locomotor behavior and feeding ecology of the Panamanian tamarin. Ph.D. Dissertation, Wash-
ington University.
Garber PA (1992) Vertical clinging, small body size, and the evolution offeeding adaptations in the Callitrichinae.
Am. J. Phys. Anthropol. 88:469-482.
Grand TI (1972) A mechanical interpretation of terminal branch feeding. J. Mammal. 53: 198--20 I.
Harding R (1981) An order of omnivores: Nonhuman primate diets in the wild. In R Harding and G Teleki (eds.):
Omnivorous Primates: Gathering and Hunting in Human Evolution. New York: Colombia University Press,
pp.191-209.
Hollihn U (1984) Bimanual suspensory behavior: Morphology, selective advantages and phylogeny. In H
Preuschoft, D Chivers, W Brockelman, and N Creel (eds.): The Lesser Apes. Edinburgh: Edinburgh Uni-
versity Press. pp. 85--95.
Hom HS (1971) The Adaptive Geometry of Trees. Princeton: Princeton University Press.
60 D. J. Bergeson

Hunt KD (1992) Positional behavior of Pan troglodytes in the Mahale mountains and Gombe Stream National
Parks, Tanzania. Am. 1. Phys. Anthropol. 87:83-105.
Lemelin P (1995) Comparative and functional myology of the prehensile tail in New World monkeys. 1. Morph.
224:1-18.
Manly BF1 (1991) Randomization and Monte Carlo Methods in Biology. New York: Chapman and Hall.
Mendel FC (1976) Postural and locomotor behavior of Alouatta palliata on various surfaces. Folia Primatol.
26:36-53.
Mittermeier RA, and Fleagle JG (1976) The locomotor and postural repertoires of Ateles geoffroyi and Colobus
guereza, and a reevaluation of the locomotor category semibrachiation. Am. 1. Phys. Anthropol.
45:235-256.
Napier JR (1967) Evolutionary aspects of primate locomotion. Am. J. Phys. Anthropol. 27:333-342.
Napier JR, and Napier P (1967) A Handbook of Living Primates. New York: Academic Press.
Potvin C and Roff D (1993) Distribution-free and robust statistical methods: Viable alternatives to parametric sta-
tistics? Ecology 74: 1617-1628.
Prost JH (1965) A definitional system for the classification of primate locomotion. Am. Anthropol. 67: 1198-1214.
Ripley S (1967) The leaping of langurs: A problem in the study of locomotor adaptation. Am. 1. Phys. Anthropol.
26:149-170.
Rose MD (1993) Locomotor anatomy of Miocene hominoids. In D Gebo (ed.): Postcranial Adaptation in Nonhu-
man Primates. DeKalb: Northern Illinois University Press, pp. 252-272.
Stem JT, and Oxnard C (1973) Primate Locomotion: Some Links with Evolution and Morphology. Basel: Karger.
Sussman RW (1987) Species-specific dietary patterns in primates and human dietary adaptations. In W Kinzey
(ed.): The Evolution of Human Behavior: Primate Models. Albany: State University of New York Press,
pp.151-179.
Swartz SM (1993) Biomechanics of primate limbs. In D Gebo (ed.): Postcranial Adaptation in Nonhuman Pri-
mates. DeKalb: Northern Illinois University Press, pp. 5-42.
Swartz SM, Bertram JEA, and Biewener AA (1989) Telemetered in vivo strain analysis of locomotor mechanics of
brachiating gibbons. Nature 342:270-272.
Takahashi LK (1990) Morphological basis of arm-swinging: Multivariate analyses of the forelimbs of Hylobates
and Ateles. Folia Primatol. 54:70-85.
Walker SE (1993) Positional adaptations and ecology of the Pitheciini. Ph.D. Dissertation, City University of New
York.
Walker SE (1996) The evolution of positional behavior in the saki/uakaris (Pithecia, Chiropotes and Cacajao). In
M Norconk, A Rosenberger, and P Garber (eds.): Advances in Primatology: Adaptive Radiations of Neot-
ropical Primates. New York: Plenum Press, pp. 335-367.
 4

WITHIN- AND BETWEEN-SITE VARIABILITY IN

MOUSTACHED TAMARIN (SA GUINUS MYSTAX)
POSITIONAL BEHAVIOR DURING FOOD
PROCUREMENT

Paul A. Garber

Department of Anthropology
109 Davenport Hall
University of Illinois
Urbana, Illinois 61801

1. INTRODUCTION

Despite evidence of marked seasonal changes in diet, day range, and patterns of habitat
utilization in several primate species, little is currently known regarding intraspecific variabil-
ity in locomotor behavior. In general, morphologists have tended to emphasize the constraints
that anatomy places on positional behavior and have classified primate locomotion into a
small set of broad categories such as leaping, suspension, and arboreal or terrestrial quadru-
pedal ism. These categories have descriptive and functional value, but they typically have
failed to include both the specific environmental context in which the behavior takes place
(e.g., size, orientation, and rigidness of the support; location within the tree crown, etc.), or
the relationship of positional behavior to the activity pattern of the animal (e.g., differences in
patterns of posture and locomotion when exploiting particular food types). If one assumes that
animals exploit substrates that allow them to move efficiently through the arboreal canopy,
then there should be consistent associations between postural and locomotor behavior, sup-
port type, and activity pattern (Prost, 1965). How limited or how variable these associations
are within a species is determined by physiologicalfactors influencing body size and the en-
ergetic and mechanical costs of movement (Tuttle and Watts, 1985; Cant, 1987; Steudel,
1990, 1996; Doran, 1993); environmental factors affecting the structure of the arboreal can-
opy, and the spatial relationship of support types to each other and to the food resources ex-
ploited (Crompton, 1984; Garber, 1984, 1993a; Grand 1984; Pound, 1991; Cant, 1992;
Bergeson, 1996); and social factors affecting individual spacing and differential access to re-
sources of particular age, sex, or status classes (Hunt, 1992; Remis, 1995) (Figure 1).

61
62 P. A. Garber

Positional Behavior

/""
Substrate preference ...._ -....~

t
Activity Pattern

Energetic + Mechanical costs of Movement

/
Physiological Factors Habitat Factors

--- ------
Locomotor anatomy Structural diversity of the
Ontogenetic changes in body arboreal canopy
weight and limb proportions Spatial relationship of support
Body size dimorphism types to each other, to gaps in
Reproductive condition the canopy, and to food resources
Nutritional requirements

Social Factors

Access of individuals of particular age,

/
sex, or status classes to resources
Group size

Figure 1. Schematic representation of factors that influence within species variability in positional behavior.

Recent field studies indicate that individuals of the same species may exhibit
marked variation in their positional repertoire (Gebo and Chapman, 1995a,b; Dagosto,
1995). For example, Gebo and Chapman (1995a) report that in Colobus badius the fre-
quencies of positional behaviors such as quadrupedal ism, leaping and climbing vary con-
siderably between years, during different seasons of the same year, and particularly in
forests of different structure. In addition, Cant (1987) offers evidence of significant in-
tersexual differences in positional behavior in adult orangutans. Male orangutans are at
least twice as heavy as females orangutans. Based on observations of one fully adult male
and at least 3 adult females exploiting 2 large Ficus trees, he concluded that males used
larger supports and more above-branch postures whereas females were more frequently
found on smaller branches and employed suspensory postures (Cant, 1987). In Cant's
study, sex differences in body size and preferential access to parts of the tree crown were
associated with differences in food harvesting techniques. Similarly, Remis (1995) reports
significant differences in selection of support size and the use of suspensory postures be-
tween female and male lowland gorillas. In other species, however, patterns of positional
behavior may vary minimally during different seasons of the year (Richard, 1978; Da-
gosto, 1995), at different field sites (see Table 6 in Garber and Pruetz, 1995), or between
individuals of larger and smaller body sizes (Gebo and Chapman, 1995b). If specific
knowledge of the ecological and anatomical factors that influence positional behavior in
living primates is a pre-requisite for reconstructing positional behavior and habitat utiliza-
tion in extinct primates, then field studies must identify those aspects of the positional rep-
ertoire that are highly sensitive to differences in forest structure and resource distribution
Within- and Between-Site Variability in Moustached Tamarin 63

from those that are more conservative and occur in relatively equal frequencies across a
species range. In this regard our goal is to link morphology to ecology, or what has been
called 'ecomorphology', through patterns of behavior (Ricklefs and Miles, 1994). The
morphological design of a species results in a range of locomotor patterns. Regardless of
whether this range is narrow or wide, however, it is bounded and we must be able to iden-
tify these boundaries in order to evaluate the relationship between musculoskeletal anat-
omy, positional behavior, and the manner in which a species exploits its environment.
In this study I examine intraspecific variability in positional behavior during feeding
and foraging in two groups of moustached tamarins (Saguinus mystax) inhabiting forests
that differ in vertical stratification, crown shape, and tree density. Saguinus mystax is one
of the largest callitrichine species. Body weight of wild-trapped adult males averages 562
gm and body weights of adult females averages 583 gm (Garber and Pruetz, 1995). Differ-
ences in body weight, body length (crown to rump), armspan, and legspan between males
and females are less than 5% (Table 1). The absence of dimorphism in this species elimi-
nates the potential confounding effects of body weight and size in understanding behav-
ioral variability.
Previous studies of locomotion in moustached tamarins indicate that travel is domi-
nated by slow and rapid forms of quadrupedal progression, and by leaping (Garber, 1991).
In this species, positional behavior during travel appears to be highly conservative. Garber
and Preutz (1995) found only minor differences in patterns of substrate preference and lo-
comotion among 2 groups of moustached tamarins living in forests of different structure.
Less, however, is known regarding variation in the tamar in positional repertoire during
feeding and foraging. In several species of Old World primates (Gebo and Chapman,
1995a,b) and Malagasy prosimians (Dagosto, 1995), there is evidence of greater variabil-
ity in locomotor and postural behavior during feeding than during traveling. Given the
conservative nature of travel in Saguinus mystax, I examine evidence of intraspecific vari-
ability in positional behavior and substrate preference during food procurement.

2. METHODS

The positional behavior and feeding ecology of moustached tamarins were studied at
two field sites in the Amazon Basin of northeastern Peru. The Padre Isla field site is a pro-
tected biological reserve located approximately 10 km north of the city ofIquitos (3°44'S,
73°14'W). The island is comprised of a series of forested areas separated by narrow la-
goons. The study group contained 5 adults (2 males and 3 females) and its home range
was dominated by areas of low secondary forest, gallery forest, and cultivated fields (Gar-
ber et aI., 1993). Saguinus mystax is the only species of nonhuman primate on Padre Isla.
Data at this field site were collected from June through December 1990. This period en-
compassed the major dry season (July through September) and the early wet season (Octo-
ber through December).
The second field site, Rio Blanco, is a relatively undisturbed primary and secondary
rainforest, located 110 km southeast ofIquitos (4°05'S, 72° lO'W). At this field site, mous-
tached tamarins form mixed species troops with saddle-back tamarins (Saguinus fuscicol-
lis). Data presented on the diet and positional behavior of Rio Blanco moustached
tamarins (group size ranged from 10-15 individuals) were collected from members of a
mixed species troop from October through December 1984 (data were also collected on
this group from March through September, but are presented elsewhere [Garber 1993b D.
The home range of the study troop was dominated by high forest (51 %), low seasonally
64 P. A. Garber

inundated forest (25%) and palm swamps (12%) (Norconk, 1986). In addition to mous-
tached and saddle-back tamarins, at least 8 other primate species are found at this field site
(Cebuella pygmaea, Callicebus moloch, Aotus sp., Cacajao calvus, Cebus apella, Cebus
albifrons, Pithecia aequatorialis, and Saimiri sciureus). Information on the ecology of this
area has been published elsewhere (Norconk, 1986; Garber, 1993b). Identical methods of
data collection and identical definitions of food items consumed, forest architecture, posi-
tional behavior, activity patterns (feeding, foraging), and branch size and orientation were
used at each site.
Data on the tamarin diet were collected using a 2-minute focal animal time sampling
technique. Food items were classified as ripe or fleshy fruits, legumes (a dry fruit contain-
ing several seeds enclosed in a pod), floral nectar (the tamarins were not observed to con-
sume flower petals or ovaries), exudates (plant gums and saps that exude from the trunks
of trees), and insect prey. Despite their small body size, tamarins are easily observed when
feeding on plant resources. They are sloppy feeders and commonly drop husks and flower
parts onto the ground, facilitating identification of the plant part consumed. Insect feeding
presents considerably greater problems of identification. Moustached tamarins often visu-
ally scan vegetation searching for exposed prey, or glean insects attached to leaf surfaces.
Once sighted, arthropods are captured rapidly using a stalk-and-pounce technique (Garber,
1993a; Peres, 1992). In the present study differences in insect foraging techniques and as-
sociated positional behaviors were identified, but no attempt was made to distinguish be-
tween the different types of insects exploited.
Information on forest architecture was collected in order to assess how site-specific
differences in the tree density, canopy continuity, and branching patterns affect tamar in
feeding and foraging activities. Forest architecture was studied by randomly selecting 5 x
5 m vegetation quadrates in each group's home range. The number, height, and circumfer-
ence of all trees greater than 2 m in stature and ~2 cm in diameter at breast height (dbh),
as well as the number of lianas in each quadrat were recorded. Approximately 2% (61
quadrats) of the vegetation in the 7 ha home range of the tamarin study group on Padre
Isla and 1% (200 quadrats) of the vegetation in the 40 ha home range of the tamarin group
at the Rio Blanco was sampled (Garber and Pruetz, 1995).
Quantitative information on positional behavior in adult moustached tamarins also
was collected using a 2-minute instantaneous focal animal point sampling technique.
Climbing was defined as "progression along continuous supports using various combina-
tions of three or more limbs" (Fleagle and Mittermeier 1980:249). Grasping is a postural
activity associated with prehension using any combination of two or more limbs. Sitting
was scored when an animal was stationary while supported on its haunches. Limb place-
ment was variable. Hindlimb grasp (or grasping with hindlimbs only) is a form of below
branch feeding in which the forager is oriented vertically downward, supported solely by
the prehensile grasp of its hind feet which are loaded in tension. Clinging is a postural ac-
tivity in which the clawlike nails of at least three limbs are embedded into the bark. Cling-
ing typically occurs on moderate to large vertical or sharply inclined supports. Bipedal
crouch represents a postural behavior in which both hindlimbs are in compression and
supporting the animal's full body weight. The knees are flexed, both arms are extended.
This posture was observed when an animal was attempting to pull a large food item, still
attached to the stem, to its mouth to feed. Locomotor behaviors were defined as follows:
Quadrupedalism was defined as pronograde progression in which all four limbs were in
compression. Typically the tamarins adopted a diagonal sequence-diagonal couplet gait
during both quadrupedal walking and quadrupedal running. Leaping is associated with
hindlimb propulsion that enables the forager to navigate gaps in the canopy. Leaping can
Within- and Between-Site Variability in Moustached Tamarin 65

occur from a stationary position, slow quadrupedal progression, or from rapid quadrupedal
bounding. During foraging, leaps typically involved movement between small, discontinu-
ous supports. Less frequently did moustached tamarins leap from trunk-to-trunk in the un-
derstorey. Scansoriallocomotion occurs on moderate to large vertical or sharply inclined
supports. Movement is facilitated by embedding their claw-like nails into the bark.
Foraging was defined as localized movements within the crown of a tree associated
with food search and pursuit. Feeding was defined as handling or ingesting prey items. For
example, if a tamarin was moving from one tree branch to another and scanning a nearby
substrate or moving to another part of the same tree crown that bore ripe fruit, it was
scored as foraging. In contrast, if an individual was turning over leaves, holding a fruit,
extracting gum from a tree wound, or masticating a food item, it was scored as feeding.
Each activity record contained information on: (1) characteristics of the substrate (a)
height in the canopy, estimated in 5 meter intervals above the ground; (b) branch angle,
classified as horizontal (0-15°), oblique (16-74°), and vertical (75-90°); and (c) branch
diameter, classified as small (:5:5 cm), medium (6-10 cm), and large (>10 cm); (2) posi-
tional behavior; and (3) activity pattern. A total of 3,283 samples of positional behavior,
substrate utilization, and height in the canopy during feeding and foraging were recorded
for moustached tamarins on Padre Isla, and 2,531 samples of positional behavior during
feeding and foraging were recorded for moustached tamarins at the Rio Blanco site.
At the Padre Isla site, we were unable to follow the study group over the course of
an entire day. The data are presented in terms of a profile of positional behavior, substrate
utilization, and height in the canopy across twelve 2-week periods. At the Rio Blanco site,
the study group was followed from dawn until dusk. The unit of study was a complete
tamarin day, and for each day a profile of positional behavior, substrate utilization, and
height in the canopy was compiled. Data are presented for 15 complete days over 3 con-
secutive I-month periods. During each I-month period at the Rio Blanco, and each 2-
week period on Padre Isla, between 4.5 and 20 hrs of quantitative data on feeding and
foraging were recorded (total of 109 hours of quantitative data on Padre Isla moustached
tamarins and 84 hours of quantitative data on Rio Blanco moustached tamarins). Observa-
tion hours were calculated by assuming that each feeding and foraging bout occurred
throughout the entire 2-minute sampling period and then multiplying the number of feed-
ing and foraging samples by 2 minutes.
Between-site and within-site comparisons of positional behavior and substrate pref-
erence were accomplished by compiling matrices of biweekly frequencies (Padre Isla) and
daily frequencies (Rio Blanco) for each activity, and then converting these scores into a
ranked format. Pearson's product-moment correlation coefficients were calculated on
ranked data for all combinations of samples (i.e., biweekly sample 1 with biweekly sample
2, biweekly sample 1 with biweekly sample 3, etc. or day I with day 2, day I with day 3,
day 2 with day 3, etc.). These values were normalized and converted to Z-scores accord-
ing to a hypergeometric distribution calculated using Fisher's Exact Test. The distribution
of Z-scores between samples within a site and the distribution of Z-scores between-sites
was compared (Students t-test, two-tailed probability) to test for differences in patterns of
positional behavior across sample periods at the same site as well as between sites.
A Chi-square goodness of fitness test was used to compare the forest height profile
of each forest site with the patterns of vertical ranging exhibited by the tamarins. A Stu-
dents t-test (two-tailed probability) was used to examine inter-site differences in forest ar-
chitecture.
Body weights and body measurements presented in Table 1 were collected by the
author from several groups of wild-caught and released tamarins during the course of this
66 P. A. Garber

Table 1. Body weights and body measurements of wild-trapped adult female and
male moustached tamarins (Saguinus mystax)'
Body weight Body length Armspan Legspan
Sex N (g) (cm) (cm) (cm)
Female 2 48 583 ± 63 24.4 ± 0.8 31.3±1.1 37.9 ± 1.4
Male 46 562 ± 59 24.6 ± 0.8 31.2±0.8 37.8 ± 1.4
I Data are from Garber (1991) and Garber and Pruetz (1995).

"Measurements from nonpregnant, pregnant, and lactating females.

and other studies (Garber and Pruetz, 1995). These data were analyzed using a Students t-
Test (two-tailed probability).

3. RESULTS

3.1. Diet
At each field site, moustached tamarins exploited a similar set of resources (Ta-
ble 2). Based on time spent feeding and foraging, the diet of the Padre Isla study group in-
cluded 56.1% insect prey, 25.4% legumes, and 16.1% fleshy fruits. Floral nectar and plant
exudates accounted for the remaining 2.3% of the diet. Virtually all observations of leg-
ume feeding on Padre Isla involved trees of the genus lnga (Leguminosae). At least 3 spe-
cies of Inga (Inga edulis, Inga alba, and lnga sp.) were common in the study troop's home
range.
At the Rio Blanco site, insects (44.4%), fleshy fruits (39.9%), and legumes (7.3%)
accounted for 92% of feeding time (Table 2). In this forest, fleshy fruits were available
throughout most of the year (Garber, 1993b) and were the principle source of readily
available energy. The exploitation of legumes was more seasonal than on Padre Isla, but
did account for 24.5% of total plant feeding time during the late wet season months of
May and June (these results are reported in Garber, 1993b). At the Rio Blanco field site,
nectar accounted for 6% of feeding time. Overall, the tamarin diet at the Rio Blanco was
more diverse and species-rich than the tamarin diet on Padre Isla.

Table 2. Dietary patterns of moustached tamarins

at two study sites in northeastern Peru'

Food type Padre Isla Rio Blanco

Insects 56.1 44.4
Fleshy Fruits 16.1 39.9
Legumes 25.4 7.3
Nectar 1.1 6.0
Exudates 1.2 2.2
I Datafrom Padre Isla are based on 3283 instantaneous focal
animal point samples from June through November, 1990.
Data from Rio Blanco are based on 2531 instantaneous fo-
cal animal point samples from October through December,
1984.
Within- and Between-Site Variability in Moustached Tamarin 67

3.2. Forest Structure

Data collected on forest structure provide evidence of significant inter-site differ-
ences. The number of trees per hectare (1 170/ha at Padre Isla compared to 4240/ha at Rio
Blanco; t=14.5, df=259, P<O.OOOI) and the number of Ii an as per hectare (640/ha at Padre
Isla compared to 1280/ha at Rio Blanco; t=3.2, df=259, P<O.OOI) were significantly fewer
at Padre Isla than on Rio Blanco. On Padre Isla, trees were smaller and characterized by
crowns in which growth in height exceeded growth in width. In these trees, main highway
branches tended to be vertical or oblique rather than horizontal in orientation. Over 63%
of the trees sampled were less than 5 meters in height, and only 6.6% were over 15 meters
in height. At the Rio Blanco site, taller trees with large horizontal boughs and wide
spreading crowns were more common. In this forest, 20% of sample trees exceeded 15
meters in height. A vertical height profile of both forests indicate that Rio Blanco tamarins
exploited a forest of significantly greater stature (X 2 = 148.6, df=4, P<O.OO 1; Figure 2) and
a more continuous canopy than did Padre Isla tamarins.

3.3. Vertical Use of Canopy Levels

Figure 2 provides a comparison of the height profile of each forest with tamarin ver-
tical ranging activities during feeding and foraging. Differences between these distribu-
tions are significant (X 2 =8403, df=4, P<O.OOOI), and indicate that Padre Isla moustached
tamarins exhibited a clear preference for exploiting resources in the mid to upper levels of
the forest canopy. For example, despite the fact that trees between 11-15 meter in height
accounted for only 4.6% of the trees sampled at Padre Isla, 33.5% of feeding and foraging
occurred in this level of the canopy.
An identical pattern of tamarin vertical height preference was found at the Rio
Blanco study site. In this forest, tall trees were also a focus of tamarin feeding activities

Padre Isla
Rio Blanco
A. 80 B.
40
0 Tree Height Profile

60 • Feedi ngIForaging Height

30
....c
....c
~
u
. 40 ..
~
~
20
~
20 10

0 0
0-5 6-10 11- 15 16·20 20+ 0-5 6-1 0 11-15 16-20 20+
Height (meters) Height (meters)

Figure 2. Comparison of the vertical height profile of the study group's home range and the feeding and foraging
height preferences of that group. (A) Data for Padre Isla are based on height estimates of 194 trees in the home
range of the study group. (B) Data for the Rio Blanco represent height estimates of 1198 trees in the home range
of this study group.
68 P. A. Garber

Table 3. Percentages and ranked order of positional behaviors during

feeding and foraging)
Padre Isla Rio Blanco
Positional behavior 0/0 Rank % Rank
Climb/Grasp 33.5 I 43.5
Sit 32.5 2 26.5 2
Hindlimb grasp 15.1 3 8.2 4
Quadrupedalism 11.5 4 12.9 3
Cling 3.0 5 2.1 6
Leap 2.1 6 4.1 5
Bipedal crouch 1.8 7 1.6 7
Scansorial 0.3 8 0.6 8
I Datafrom Padre Isla are based on 3183 records of positional behavior. Data from the
Rio Blanco are based on 2531 records of positional behavior.

(Figure 2). Trees exceeding 15 meters in height accounted for 20% of the forest profile but
fully 65% of feeding and foraging time (X 2 =3953; df=4, P<O.OO 1).

3.4. Positional Behavior

During food procurement, climbing/grasping and sitting dominated the moustached
tamarin positional repertoire. Table 3 presents a profile and description of locomotor and
postural behaviors adopted by S. mystax during feeding and foraging. On Padre Isla,
climbing and grasping using a combination of at least 3 limbs accounted for 33.5% of lo-
comotor and postural activities (Table 3). Sitting was the next most common posture
(32.5%), followed by grasping with hindlimbs only (15.1%), quadrupedal progression
(11.5%), and a range of relatively infrequent behaviors such as clinging, jumping, crouch-
ing bipedally, and scansoriallocomotion. These latter 4 positional behaviors accounted for
only 7.2% of feeding and foraging activities.

Table 4. Variation in the percentage of feeding and foraging

time spent in positional behaviors by Saguinus mystax)

Positional category Padre Isla Rio Blanco

Climb/Grasp 34.0 ± 7.0 43.5 ± 7.8
21.6 - 47.0 34.3 - 61.4
Sit 32.5 ± 8.6 26.5 ± 5.9
17.7-47.7 14.4 - 33.3
Hindlimb grasp 15.1 ± 7.7 8.2 ± 5.8
8.0 - 33.5 0-24.3
Quadrupedalism 11.5±3.2 12.9 ± 5.7
7.2 - 16.2 2.7 - 24.9
Cling 3.0 ± 1.6 2.1 ± 1.8
0.5 - 5.4 0-6.7
Leap 2.1 ±1.5 4.1±2.3
0.7 - 5.8 0.8 - 7.5
Bipedal crouch 1.8 ± 1.0 1.6 ± 1.9
0.8 - 3.9 0-7.4
Scansorial 0.3 ± 0.3 0.6 ± 1.0
0-0.9 0-3.5
I Data are mean ± I S.D. (top row). range (bottom row).
Within- and Between-Site Variability in Moustached Tamarin 69

At the Rio Blanco field site, moustached tamarins exhibited a similar positional rep-
ertoire to that used by the Padre Isla group (Table 3). In this taller and less disturbed rain-
forest, climbing and grasping continued to be the most common positional behaviors,
accounting for 43.5% of food procurement activities. Sitting (26.5%), quadrupedal pro-
gression (12.9%), and grasping solely with hindlimbs (8.2%) represented the next 3 most
common positional activities.
Overall there was marked similarity in the relative percentage and pattern of posi-
tional activities between sites. Between-site differences in moustached tamarin positional
behavior were largely the result of the greater use of combinations of fore- and hindlimb
grasping/climbing at the Rio Blanco, and a greater use of grasping with hindlimbs only on
Padre Isla. If climbing/grasping and grasping with hindlimbs only are combined into a sin-
gle prehensile positional category, then during feeding and foraging these behaviors ac-
counted for 48.6% of the positional repertoire on Padre Isla and 51.7% at the Rio Blanco.
Even when grasping with hindlimbs only is scored as a distinct postural category, a rank-
ing of 8 postural and locomotor behaviors (Table 3) indicates limited between-site differ-
ences in the moustached tamarin positional repertoire.

3.5. Temporal Variation in Patterns of Positional Behavior

Given limited evidence for major between-site differences in the relative percent-
ages of each positional behavior, is there evidence of significant levels of daily or bi-
weekly variation among group members inhabiting the same site? Table 4 provides
information on within-site variation in postural and locomotor behavior. The data from Pa-
dre Isla indicate that across any 2 sample periods the frequencies of a particular postural
or locomotor behavior could vary by a factor of 2-4. For example, during sample period 2,
sitting accounted for 17.7% of the positional repertoire, whereas during sample period 12,
sitting occurred 47.7% of the time. Similarly, grasping with hindlimbs was observed 8.0%
of the time during a given sample period and 33.5% in another sample period. A similar
range of variation in posture and locomotion between sample periods was found at Rio
Blanco (Table 4). These data can be interpreted as evidence of substantial within-site vari-
ation in positional behavior.
In contrast, analyzing the tamarin positional repertoire in terms of a ranking of pos-
tural and locomotor activities for each sample period offers support for limited within-site
variability. An analysis using ranked data focuses on patterns of behavior and relative fre-
quencies rather than absolute frequencies. The rankings for each of the twelve 2-week pe-
riods for the Padre Isla tamarins are presented in Table 5. These data indicate that certain
positional behaviors, such as grasping/climbing and sitting dominated feeding and forag-
ing throughout the study, whereas other modes of posture and locomotion were always ob-
served to occur in moderate frequency or infrequently. This pattern was highly consistent
between two-week periods. For example, a comparison of Pearson's product-moment cor-
relation coefficients calculated on ranked data for all combinations of biweekly samples
fails to indicate any significant differences in positional rankings between sample periods
1--6 and sample periods 7-12 (t=1.2, df=14, P=0.23).
A similar analysis of correlation coefficients based on,patterns of positional behavior
during feeding and foraging at the Rio Blanco field site also revealed evidence of limited
within-site variability. As indicated in Table 5, climb/grasp was the most common posi-
tional behavior during each sample day. Sitting ranked second during 12 of 15 full day fol-
lows and third on the remaining 3 sample days. Quadrupedal progression or grasping with
hindlimbs only were ranked third on the remaining 12 days. On no sample day did leap-
70 P. A. Garber

1
Table 5. Ranking of the positional repertoire in Saguinus mystax during feeding and foraging

Padre Isla Rio Blanco

Positional
behavior Rank 1-2 Rank 3-4 Rank 5-6 Rank 7-8 Rank 1-2 Rank 3-4 Rank 5-6 Rank 7-8
Climb/Grasp 11 I 0 0 15 0 0 0
Sit 9 3 0 0 12 3 0 0
Hindlimb grasp 3 9 0 0 I 10 2 2
Quadrupedal ism 0 12 0 0 2 12
Cling 0 0 10 2 0 0 11 4
Leap 0 0 8 4 0 12 2
Bipedal crouch 0 0 4 8 0 1 13
Scansorial 0 0 0 12 0 0 2 13
I Data from Padre Isla represent 12 2-week periods. Data from Rio Blanco represent 15 full day follows. The bold values serve to
emphasize the consistent rank of that behavior in the tamarin positional repertoire.

ing, clinging, scansorial locomotion, or bipedal standing rank among the top 4 positional
behaviors. A comparison of patterns of positional behavior across sample periods (Days
1-7 vs. Days 8-15) indicated no significance rank differences in 8 locomotor and postural
categories used by moustached tamarins at the Rio Blanco (t==1.1, df==20, P==O.27).

3.6. Association between Insect Foraging and Positional Behavior

Figure 3 presents a between-site comparison of insect capture techniques used by
moustached tamarins. These techniques are (1) gleaning of prey attached to small
branches and leaf surfaces; (2) visual scanning for exposed and mobile prey which, once
located, are quickly seized; and (3) exploitation of insects refuging or concealed on trunks
and other large vertical supports. Each capture technique is associated with an alternative
set of positional behaviors. Gleaning involves active searches and manipulation of leaves
using climbing, grasping, and leaping locomotion. During visual scanning, tamarins typi-
cally adopt a sitting posture or stand quadrupedally and move their heads from side to side
and/or up and down searching for prey. Once detected, prey are pursued using rapid quad-
rupedal movement and seized or pinned by the hands. The capture of trunk refuging in-
sects involves inspection of crevices in the bark during vertical clinging and traveling on

60

50

....I::
~ 30
~
20

10
Figure 3. Between-site comparison of insect foraging
o techniques in Saguinus mystax. The three capture tech-
Gleaning Visual Trunk niques: gleaning, scanning and striking, and trunk foraging
from leaves scanning foraging
are described in the text.
Within- and Between-Site Variability in Moustached Tamarin 71

trunks using scansoriallocomotion. As indicated in Figure 3, 38.6% of insect foraging by

Padre Isla tamarins was associated with gleaning nonmobile or attached prey from leaf
substrates. In this forest, tamarins relied heavily on visual scanning and pouncing to ac-
quire insect prey (55.7%). In contrast, gleaning nonmobile or attached prey was the pri-
mary insect foraging technique observed in Rio Blanco moustached tamarins and
accounted for over 55.3% prey captures (Figure 3). Foraging for prey on large vertical
trunks occurred infrequently at each site. It remains uncertain whether site-specific differ-
ences in insect foraging techniques reflect differences in prey type and prey availability, or
differences in capture rates associated with vegetation density and prey visibility. These
differences in foraging techniques may, however, help to account for much of the inter-site
variability in tamarin positional behavior.

3.7. Substrate Preference

Resources exploited by moustached tamarins were located principally on small
obliquely-oriented supports in the mid- and upper-levels of the tree crown. At the Padre
Isla field site, 55.4% of foraging substrates were small in diameter and 78% were angled
supports (Figure 4). Medium-sized supports accounted for 38.7% of the positional reper-
toire, and large supports were used infrequently (5.7%).
A similar pattern of support size was recorded for the Rio Blanco study group. At
this site, 68.8% of feeding and foraging time occurred on small branches (Figure 4). Me-
dium-sized supports accounted for 23.3% of food gathering activities, and large diameter
supports 8.8%.
There was, however, evidence of considerable between-site variation in the fre-
quency with which branches of different angles or orientation were utilized. Although
obliquely angled branches were exploited most frequently at each field site, observations
of the Rio Blanco tamarins indicated that horizontal and vertical substrates were used in
considerably higher frequency, accounting for 33.7% and 25.2% of feeding/foraging ac-
tivities respectively (Figure 4). On Padre Isla, vertically oriented branches were rarely vis-

A. 80
B.
80
Padre Isla

60 Rio Blanco Padre I la

60
....c ....C Rio Branco
...'"'
Q/
...
~
If 40 If 40

20 20

o o
small medium large horizontal oblique vertical
Branch Size Branch Angle

Figure 4. Between-site comparison of substrate utilization in moustached tamarins during feeding and foraging.
Comparative data are presented on support size (a) and support orientation (b). See text for definitions of branch
size and branch orientation categories.
72 P. A. Garber

ited and accounted for only 6.3% of substrates exploited. Similarly, use of horizontal sup-
ports on Padre Isla was approximately one-half that recorded at the Rio Blanco (1S.6% vs.
33.7; Figure 4).

3.8. Temporal Variability in Patterns of Substrate Utilization

A ranking of the data on branch size and branch angle offers additional evidence for
consistency in patterns of substrate use in moustached tamarins (Table 6). On Padre Isla,
small supports dominated feeding and foraging in 9 of 12 sample periods. Medium sup-
ports were the most common substrate exploited in 3 samples and the second most com-
mon in 9 samples. Large branches ranked last in support use in each of the 12 2-week
periods. Use of branch angles by Padre Isla moustached tamarins followed a similar pat-
tern. In 12 of 12 2-week sample periods obliquely-oriented supports were the most com-
monly used platform for feeding and foraging. In 10 of 12 sample periods, horizontal
supports were the second most common feeding and foraging substrate. Vertical supports
were rarely used.
Daily variation in use of supports of small, medium, and large size by Rio Blanco
moustached tamarins was also highly conservative. Despite the fact that on any given day,
small branches accounted for between SS-71 % of the substrates exploited and large
branches accounted for 2-19%, a ranking of these data indicate a consistent pattern. Small
branches were selected in highest frequency on each of the IS sample days. Medium
branches ranked second on 14 of IS days and tied for second on the remaining day. Large
branches were exploited least frequently on all days (except for one day in which these
supports were tied for last).
In contrast to the pattern observed on Padre Isla, Rio Blanco moustached tamarins
were significantly more variable in their use of branches of different orientation (t=3.S,
df=6S, P<O.OO 1). Daily values for the use of horizontal supports at the Rio Blanco ranged
from 31-4S%. Use of oblique supports ranged from 32-S4%, and for vertical supports
these values ranged from 6-34%. A comparison of the daily rankings indicate that hori-
zontal supports ranked highest or tied for highest on 46% (7/1S) of the sample days,
oblique supports ranked highest or tied for highest on 66% (lO/1S) of the sample days, and
vertical supports tied for the highest rank on 6% (1/1S) of the sample days. Daily variation

Table 6. Ranking of the positional repertoire in Saguinus mystax during feeding

and foraging)

Padre Isla Rio Blanco

Substrate Rank I Rank 2 Rank 3 Rank I Rank 2 Rank 3
Small 9 3 0 IS 0 0
Medium 3 9 0 0 IS 0
Large 0 0 12 0 14
Oblique 12 0 0 10 4
Horizontal 0 10 2 7 8 0
Vertical 0 2 12 0 14
'Data from Padre Isla represent 12 2-week periods. Data from Rio Blanco represent 15 full day fol-
lows. Column totals for the Rio Blanco may not always equal 15. This is the due to several cases of
tied ran kings. When two categories were tied for the same rank, they were both assigned to that rank.
The emboldened values serves to emphasize the consistent rank of those behaviors in the tamarin po-
sitional repertoire.
Within- and Between-Site Variability in Moustached Tamarin 73

in support orientation was associated with differences in the time spent exploiting particu-
lar food types. At this site there was a significant negative correlation in the frequency of
time spent on oblique branches each day and percent time feeding and foraging on ripe
fruit on that day (R=-O.56, P=O.029). Given the relatively large size of many fruits eaten
by moustached tamarins (Garber, 1986), these primates frequently may select above-
branch feeding postures on low angled platforms in an attempt to increase stability and
free the hands to hold and manipulate food items.

4. DISCUSSION

Tamarins are small bodied New World monkeys that exploit a diet composed princi-
pally of insects, ripe fruits, legumes, floral nectar, and plant gums. These resources are lo-
cated in different parts of the canopy, and access to each requires alternative patterns of
postural and locomotor behavior. For example, several species of tamarins exploit re-
sources (exudates, insects, and small vertebrate prey) that are located on tree trunks in the
forest understorey (Garber, 1992, 1993a). Trunk foraging appears to be an important ele-
ment of callitrichine feeding ecology and is facilitated by the evolution of elongated, later-
ally compressed, claw like nails on all manual and pedal digits excluding the hallux.
Claw like nails enable these small bodied monkeys to cling to and travel on large vertical
and sharply angled supports that are too large to be spanned by their tiny hands and feet.
In contrast, when exploiting fruits, legume pods, and more mobile insect prey, tamarins
concentrate their foraging activities on a network of small-to-medium sized branches lo-
cated in the periphery of the tree crown (Garber, 1984, 1993a).
In this study, data are presented on within- and between-site variability in moustached
tamarin positional behavior during food procurement. Tamarins must make behavioral ad-
justments in response to daily, weekly, and monthly changes in the availability and distribu-
tion of food resources, as do all foragers. These adjustments involve decisions regarding
what to eat and where to look for food, and include the selection of particular microhabitats
to exploit within their range, the selection of particular trees or food patches within these
microhabitats, and the selection of particular foraging substrates or zones within the crown
of a feeding tree. Given differences in the architecture of tree crowns (Horn, 1974; Halle et
ai., 1978) and associations between certain food types and the arboreal substrates on which
they are found (Garber, 1984; Crompton, 1984; Boinski, 1989), how variable or flexible is
positional behavior during feeding? In taxa such as Propithecus verreauxi (Richard 1978),
Propithecus diadema (Dagosto, 1995), and Eulemur fulvus (postural behavior; Dagosto,
1995) there is little evidence of seasonal variation in positional behavior associated with
diet. In other species such as Galago senegalensis (Crompton, 1984), Otolemur crassi-
caudatus (Crompton, 1984), Colobus badius (Gebo and Chapman, 1995a), Eulemur ru-
briventer (Dagosto, 1995), and Varecia variegata (Dagosto, 1995) such differences are more
evident. In these more 'variable' species, the degree to which flexibility in positional behav-
ior is determined principally by differences in forest structure, differences in foraging strate-
gies, and/or differences in locomotor anatomy remains unclear.
Despite exploiting forests that differed significantly in tree density, canopy structure,
canopy height, liana density, plant species composition, and presence of potential primate
competitors (Garber and Pruetz, 1995), patterns of positional behavior in the 2 mous-
tached tamarin study groups were highly conservative. A rank order correlation of 8 pos-
tural and locomotor behaviors at each field site revealed marked similarity in the relative
breakdown of positional activities. In addition, the rankings of positional behavior within
74 P. A. Garber

each site varied minimally between individual days and over the course of several weeks.
Overall, variability in the moustached tamarin positional repertoire did not differ signifi-
cantly between field sites. This occurred despite the fact that the Rio Blanco moustached
tamarins formed cohesive and stable mixed species troops with saddle-back tamarins
(Saguinusfuscicollis). On Padre Isla, saddle-back tamarins were absent. Several studies of
mixed species troops of moustached tamarins and saddle-back tamarins (Norconk, 1986;
Garber, 1991; Peres, 1991) indicate clear differences in patterns of positional behavior and
vertical use of the canopy. Data presented in the present study support the contention that
the presence or absence of saddle-back tamarins at these study sites had little effect on po-
sitional behavior in moustached tamarins.
On Padre Isla, climbing/grasping ranked first or tied for first on 8 of 12 biweekly sam-
ples and ranked second in 3 of the remaining samples. Sitting ranked second in 6 of 12 sam-
ple periods and was first or tied for first in 4 biweekly samples. Small supports were used in
greatest frequency in 9 of the 12 biweekly samples and ranked second in each of the 3 re-
maining observation periods. A pattern of limited behavioral variability was also docu-
mented at the Rio Blanco. At this site, climbing/grasping was the most frequent positional
behavior during each of 15 full day follows. Sitting was the next most frequent posture,
ranking second on 12 of 15 days. Small supports were the most commonly used feeding and
foraging platform on each sample day and medium supports were the second most common
on 14 of 15 days. Overall, patterns of positional behavior and substrate preference that were
commonly used by moustached tamarins on Padre Isla were also commonly used by mous-
tached tamarins at the Rio Blanco. Rare behaviors on Padre Isla were rare at the Rio Blanco.
There was evidence of more pronounced between-site variation in support orienta-
tion and foraging and feeding height. The home range of the Padre Isla study group was
characterized by a young and highly disturbed forest in which tree crowns were shorter,
and exhibited a branching pattern dominated by vertically and obliquely-oriented boughs.
During feeding and foraging almost 80% of the substrates used by moustached tamarins
on Padre Isla were oblique in orientation. In contrast, the Rio Blanco field site was charac-
terized by older, taller, and more mature forest trees, many of which exhibited wide-
spreading crowns and a large number of major horizontal branches. Rio Blanco
moustached tamarins more commonly fed and foraged at heights above 15 meters and
were less consistent in their use of obliquely-oriented branches. Although oblique
branches were the most frequently utilized support type (41 %) by tamar ins at this field
site, horizontal branches were visited at a frequency equal to or greater than oblique sup-
ports on 46% of sample days. Differences in canopy architecture, tree heights, and the
availability of support types probably account for these results.
Marked consistency in patterns of posture and locomotion between moustached
tamarins living in distinct forest types suggests that, in this species, locomotor anatomy
acts as a primary constraint on patterns of behavior and habitat utilization. The major be-
tween-site differences in positional behavior were associated with a greater frequency of
sitting (32.5 vs. 26.5) and grasping with hindlimbs only postures (15.1 vs. 8.2) on Padre
Isla and more climbing/grasping activities using a combination of fore- and hindlimb pre-
hension (43.5 vs. 33.5) at the Rio Blanco. These differences appear to reflect site-specific
variations in foraging strategies and patterns of dietary emphasis rather than high levels of
behavioral variability.
Hindlimb grasping is a form of below branch feeding in which the forager is·ori-
ented vertically downward, supported solely by the prehensile grasp of its hind feet. In
adopting this posture, the hands are free to manipulate and seize plant food and animal
prey, but play no direct role in weight bearing. As indicated in Table 7, at both field sites
Within- and Between-Site Variability in Moustached Tamarin 7S

Table 7. Association between diet and hindlimb

grasp in Saguinus mystax'
Food type Padre Isla Rio Blanco
Legumes 36.3 21.0
Fleshy Fruits \5.7 7.7
Invertebrate 7.2 6.2
Exudates 1.7
IThese data are interpreted as follows: During 36.3% of all
observations of legume feeding at Padre Isla, the tamarins
were found to adopt a grasping with hindlimbs only pos-
ture. Other positional behaviors, such as sitting (23.1 %)
and grasping with a combination of fore- and hindlimbs
(29%) were also common when feeding on legumes at Pa-
dre Isla.

grasping with hindlimbs only occurred most frequently when feeding and foraging on leg-
ume pods (36% of legume feeding at Padre Isla and 21 % of legume feeding at the Rio
Blanco). These pods are presented at the distal ends of terminal branches and often hang
vertically down. Many of the pods have a fibrous or tough outer husk. The tamarins use
their incisors and canines to pierce and strip the husk and then pry the pod open with both
hands in order to feed. The high concentration of Inga trees and the greater reliance on
legumes as a dietary staple by tamarins on Padre Isla help to explain between-site differ-
ences in the frequency of grasping with hindlimbs only postures. For example, if Rio
Blanco tamarins maintained their same pattern of positional behavior but increased their
level of legume feeding to a value comparable to that of Padre Isla tamarins, then group
differences in the frequency of hindlimb grasping postures would be reduced to only 3%.
Moustached tamarins spent 44-56% of feeding and foraging time exploiting insect
prey. In terms of insect foraging techniques, they are described "as leaf-gleaners, captur-
ing and flushing mobile prey directly from the midstorey foliage" (Peres, 1991: 127). Or-
thopteran insects such as katydids, grasshoppers, and stick insects are the most common
prey items consumed (Peres, 1991; Nickle and Heymann, 1996). In the present study,
moustached tamarins were observed to rely on three primary methods of prey capture. The
relative frequencies of each capture method varied between field sites. Moustached tam a-
rins on Padre Isla relied more heavily on visual scanning followed by quadrupedal attack
(pounce) to obtain prey (55.7%). Visual scanning typically occurred from a sitting posi-
tion. At the Rio Blanco, gleaning sedentary and attached prey from leafy substrates was
the most common insect capture technique (55.3%). This involved leaf inspection and ma-
nipulation and was associated with climbing, grasping, and leaping behavior. The degree
to which differences in prey capture techniques reflect site-specific ecological differences
in the availability and abundance of prey types remains unclear. Nevertheless, shifts in the
frequency of prey capture techniques did have a direct effect on positional behavior. Dur-
ing insectivory, sitting accounted for 38% of the positional repertoire of tamarins on Padre
Isla, but only 18% of the positional repertoire of tamarins at the Rio Blanco.
In this paper relationships between diet, habitat utilization, and positional behavior
in wild moustached tamarins are described. These relationships were consistent among
members of 2 different groups inhabiting 2 different rainforest sites in Amazonian Peru.
Given that another tamarin species, Saguinus juscicollis, is characterized by a somewhat
greater range of behavioral variability (Garber and Davis, 1996), what factors are likely to
playa primary role in limiting the expression of behavioral variability in S. mystax? Fig-
76 P.A.Garber

ure 1 outlines a set of physiological, social, and habitat variables that serve to constrain or
broaden the way in which individuals within a species exploit their environment. In terms
of the present study, differences in habitat such as forest structure and resource availability
were found to have a direct, but limited effect on patterns of positional behavior. No at-
tempt was made to measure age or sex-specific behavioral patterns. However, given mini-
mal body size dimorphism (Table 1), infrequent within-group aggression (Heymann, 1996;
Garber, 1997), and high levels of group cohesion and social cooperation, individual differ-
ences in patterns of habitat utilization are probably small. This remains to be verified em-
pirically.
How might locomotor anatomy influence the expression of locomotor behavior? As
suggested by Ricklefs and Miles (1994: 16) "The structural complexity of morphological de-
sign may determine the breadth of performance by an individual." In this regard, it is possi-
ble that a species characterized by a higher degree of individual variation in positional
anatomy (i.e., age or sex related differences in ontogenetic development or body size) might
exhibit a potential for greater variability in positional behavior than would a species charac-
terized by greater morphological consistency. In a recent study of positional anatomy in cal-
litrichines, Garber and Davis (1996) identified a complex of traits of the upper limb that
exhibited significantly greater levels of individual variation in saddle-back tamarins (S fus-
cicollis) than in moustached tamarins (S mystax). Several of the skeletal traits exhibiting in-
creased variance were part of a single morphological complex and functionally interpreted
as being consistent with field data indicating that saddle-back tamarins also exhibited
greater variability in positional behavior than do moustached tamarins (Garber and Davis,
1996). It is possible that changes in morphological design or phenotypic variability of these
skeletal elements have a disproportionate effect on the range of locomotor and postural be-
haviors exhibited. Analogous relationships between anatomical variability/stability and be-
havioral variability/stability in other primate lineages need to be explored.
In conclusion, in order to identify relationships between morphology, ecology, and
patterns of behavior in living primates (ecomorphology) studies of positional behavior and
positional anatomy need to include information on within-species variability. Although
most traits in a population are characterized by "low levels of phenotypic variation"
(Travis, 1994:99), the presence or absence of plasticity in certain traits may have an im-
portant adaptive function. In the case of several primate species (see review by Garber and
Pruetz, 1995) including prosimians, monkeys, and apes there is evidence of marked be-
tween-site consistency in patterns of positional behavior. Within these taxa, regardless of
whether the data are ranked or tallied by frequency, postural and locomotor behaviors that
were rare at one site were rare at another site. Similarly, postural and locomotor behaviors
that were common at one site were also common at another site. For other primate taxa,
seasonal, site-specific, ontogenetic, and sex-based differences in positional behavior may
be more extreme. The challenge to studies of locomotion and posture therefore, is not only
to establish functional relationships between behavior and morphology, but in addition to
identify those elements of the positional repertoire that are stable from those that are more
variable and determine how each is influenced by habitat, physiology, the energetic and
mechanical cost of movement, and the social environment (Figure I).

ACKNOWLEDGMENTS

This study was conducted with the permission and assistance of the Proyecto Pe-
ruano de Primatologia "Manuel Moro Sommo" and the Instituto Veterinario de Investi-
Within- and Between-Site Variability in Moustached Tamarin 77

gaciones Tropicales y de Altura (IVITA). We thank Dr. Enrique Montoya Gonzales, Ex-
ecutive Director of the Proyecto Peruano de Primatologia, Filomeno Encarnacion, and
Carlos Ique for their support in this project. Assistance in the field was provided by Jill
Pruetz, Walter Mermao and Eriberto Mermao. Helpful comments on an earlier draft of this
manuscript were provided by Dr. David Bergeson, Dr. Elizabeth Strasser, and an anony-
mous reviewer. Input on statistical analyses of the data were provided by Dr. Steven
Leigh. Funds to conduct this research were provided by the National Geographic Society,
the National Science Foundation, and the William and Flora Hewlett Foundation. As al-
ways I wish to thank Sara and Jenni for being Sara and Jenni.

REFERENCES
Bergeson D1 (1996) The Positional Behavior and Prehensile Tail Use of Alouaua palliata, Ateles geoffroyi, and
Cebus capucinus. Ph.D. Dissertation, Washington University.
Boinski S (1989) The positional behavior and substrate use of squirrel monkeys: ecological implications. J. Hum.
Evo!. 7:659-{)78.
Cant JGH (1987) Effects of sexual dimorphism in body size on feeding postural behavior of Sumatran orangutans
(Pongo pygmaeus). Am. 1. Phys. Anthropo!. 74: 143-148.
Cant 1GH (1992) Positional behavior and body size of arboreal primates: a theoretical framework for field studies
and an illustration of its application. Am. 1. Phys. Anthropo!. 88: 273-283.
Crompton RH (1984) Foraging, habitat structure, and locomotion in two species of Galagos. In PS Rodman and
JGH Cant (eds.): Adaptations for Foraging in Nonhuman Primates. New York: Columbia University Press,
pp. 73-1 II.
Dagosto M (1995) Seasonal variation in positional behavior of Malagasy lemurs. Int. J. Primato!' 16:807-834.
Doran DM (1993) Sex differences in adult chimpanzee positional behavior: The influence of body size in locomo-
tion and posture. Am. J. Phys. Anthropo!. 9/:99-115.
Fleagle JG, and Mittermeier RA (1980) Locomotor behavior, body size, and comparative ecology of seven Suri-
nam monkeys. Am. J. Phys. Anthropo!. 52:301-314.
Garber PA (1984) Use of habitat and positional behavior in a Neotropical primate, Saguinus oedipus. In PS Rod-
man and 1GH Cant (eds.): Adaptations for Foraging in Nonhuman Primates. New York: Columbia Univer-
sity Press, pp. 113-133.
Garber PA (1986) The ecology of seed dispersal in two species of callitrichid primates (Saguinlls mystax and
Saguinusfllscicollis). Am. 1. Primato!' 10: 155--170.
Garber PA (1991) A comparative study of positional behavior in three species of tamarin monkeys. Primates
32:219-230.
Garber PA (1992) Vertical clinging, small body size, and the evolution of feeding adaptations in the Callitrichinae.
Am. J. Phys. Anthropo!. 88:469--482.
Garber PA (1993a) Feeding ecology and behaviour of the genus Sagllinus. In AB Rylands (ed.): Marmosets and
Tamarins: Systematics, Ecology and Behaviour. Oxford: Oxford University Press, pp. 273-295.
Garber PA (I 993b) Seasonal patterns of diet and ranging in two species of tamarin monkeys: Stability versus vari-
ability. Int. J. Primato!' 14: 1-22.
Garber PA (1997) One for all and breeding for one: Cooperation and competition as a tamarin reproductive strat-
egy. Evo!. Anthropo!. N: 135--147.
Garber PA, Pruetz JD, and Issacson J (1993) Patterns of range use, range defense, and intergroup spacing in mous-
tached tamarin monkeys (Saguinus mystax). Primates 34: 1-22.
Garber PA, and Pruetz 1D (1995) Positional behavior in moustached tamarin monkeys: effects of habitat on loco-
motor variability and locomotor stability. J. Human Evo!. 28:411-426.
Garber PA, and Davis LC (1996) Intraspecific variability in anatomy and positional behavior in two tamarin spe-
cies. Am. 1. Phys. Anthropo!. Supplement 22: 110-111 (abstract).
Gebo DL, and Chapman CA (1995a) Habitat, annual, and seasonal effects on positional behavior in red colobus
monkeys. Am. J. Phys. Anthropo!. 96:73-82.
Gebo DL, and Chapman CA (1995b) Positional behavior in five sympatric Old World monkeys. Am. J. Phys. An-
thropo!. 97:49-76.
Grand TI (1984) Motion economy within the canopy: Four strategies for mobility. In PS Rodman and JGH Cant
(eds.): Adaptations for Foraging in Nonhuman Primates. New York: Columbia University Press, pp. 54-72.
78 P. A. Garber

Halle F, Oldemann RA, and Tomlinson PB (1978) Tropical Trees and Forests: An Architectural Analysis. New
York: Springer-Verlag.
Heymann EW (\ 996) Social behavior of wild moustached tamarins, Saguinus mystax, at the Estaci6n Biol6gica
Quebrada Blanco, Peruvian Amazonia. Am. J. Primatol. 38: 101-113.
Hom HS (1974) The Adaptive Geometry of Trees. New Jersey: Princeton University Press.
Hunt KD (1992) Social rank and body size as determinants of positional behavior in Pan troglodytes. Primates
33:347-357.
Nickle DA, and Heymann, EW (\ 996) Predation on Orthoptera and other orders of insects by tamarin monkeys,
Saguinus mystax mystax and Saguilllls fuscicollis nigrifrons (Primates: Callitrichidae), in north-eastern
Peru. 1. Zool. Lond. 239:799--819.
Norconk MA (1986) Interactions between species in a Neotropical forest: mixed-species troops of Sagllinus mys-
tax and Saguinusfilscicollis (Callitrichidae). Ph.D. Dissertation, University of California, Los Angeles.
Peres CA (\ 991) Ecology of mixed-species groups of tamarins in Amazonian terra firme forests. Ph.D. Disserta-
tion, University of Cambridge.
Peres, CA (1992) Prey-capture benefits in a mixed-species group of Amazonian tamarins, Saguinus fuscicollis and
S. mystax. Behav. Ecol. Sociobiol. 31: 339--347.
Pound J (\ 991 ) Habitat structure and morphological patterns in arboreal vertebrates. In S Bell, E McCoy, and H
Mushinsky (eds.): Habitat Structure: The Physical Arrangement of Objects in Space. New York: Chapman
and Hall, pp. 109--119.
Prost J (1965) A definitional system for the classification of primate locomotion. Am. Anthropol. 67: 1198-1214.
Remis M (1995) Effects of body size and social context on the arboreal activities oflowland gorillas in the Central
African Republic. Am. J. Phys. Anthropol. 97:413-434.
Ricklefs RE, and Miles DB (1994) Ecological and evolutionary inferences from morphology: An ecological per-
spective. In PC Wainwright and SM Reilly (eds.): Ecological Morphology: Integrative Organismal Biol-
ogy. Chicago: University of Chicago Press, pp. 13-41.
Richard AF (1978) Behavioral Variation: Case Study of a Malagasy Lemur. Lewisburg, PA: Bucknell University
Press.
Steudel K (1990) The work and energetic cost of locomotion I. The effects of limb mass distribution in quadru-
peds. J. expo BioI. 154:273-285.
Steudel K (1996) Limb morphology, bipedal gait, and the energetics of hominid locomotion. Am. J. Phys. Anthro-
pol. 99:345-356.
Travis J (1994) Evaluating the adaptive role of morphological plasticity. In PC Wainwright and SM Reilly (eds.):
Ecological Morphology: Integrative Organismal Biology. Chicago: University of Chicago Press, pp.
99--122.
Tuttle RH, and Watts DP (1985) The positional behavior of adaptive complexes in Pan gorilla. In S Kondo (ed.):
Primate Morphophysiology, Locomotor Analyses and Human Bipedalism. Tokyo: University of Tokyo
Press, pp. 261-288.
 5

LOCOMOTION, SUPPORT USE, MAINTENANCE

ACTIVITIES, AND HABITAT STRUCTURE
The Case of the Tai Forest Cercopithecids

w. Scott McGraw
Department of Anatomy
New York College of Osteopathic Medicine
New York Institute of Technology
Old Westbury, New York 11568

1. INTRODUCTION

Information on support use is routinely collected in locomotor field studies to estab-

lish predictive relationships between substrates and locomotor behavior (e.g., Fleagle et
ai., 1981). Nevertheless, the extent that the expression of specific locomotor activities
(e.g., climbing, leaping, quadrupedal ism) depends on particular support types remains un-
clear. While it is doubtful that we will be able to satisfactorily address certain adaptive
questions as they pertain to habitat (e.g., in what structural context did brachiation
evolve?), it is at least possible to examine how locomotor activities vary in architecturally
dissimilar contexts. Such investigations could aid in explaining how the architectural
properties of arboreal habitats (i.e., support characteristics) are responsible for promoting
and/or limiting specific positional behaviors.
The data necessary for this type of investigation are scarce, stemming largely from
the difficulty in accurately quantifying those habitat features relevant to locomotion, viz.,
the relative abundance, size and inclination of different support types. Indeed, analyzing
the structural environment as it pertains to even a single locomotor behavior (e.g., climb-
ing) is a complex, multivariate problem (Cartmill, 1985). Of course, most primates have
multiple behaviors in their locomotor repertoires. As Stern and Oxnard (1973: I) observe,
"primates stand, sit, lie, walk, run, hop, leap, climb, hang, swing, swim, and engage in
other activities too numerous to mention. They may do these things often or rarely,
quickly or slowly, with agility or clumsiness, on the ground or in the trees (or, with swim-
ming, in the water), on thick branches or thin ones, on vertical, oblique or horizontal sup-
ports, with all appendages or only some." Thus, establishing a predictive relationship

79
80 W.S. McGraw

between any of the above activities and some habitat characteristic would be difficult
enough if habitat were a constant variable, yet many species have wide distributions
throughout structurally dissimilar habitats (e.g., Gartlan and Brain, 1968; Lawes, 1990)
and even single populations commonly range throughout architecturally heterogeneous
forests (e.g., McGraw, 1994; Moreno-Black and Maples, 1977).
The fact that a single species may be found in structurally different environments
could have profound implications for interpreting locomotor data, particularly if the fre-
quencies of locomotor activities are significantly affected by the habitats in which they oc-
cur. If locomotor behavior reflects proximate responses to the environment rather than
evolved or morphology driven tendencies (Pounds, 1991), then establishing relationships
between anatomy and behavior will be made more difficult unless behavior is observed in
each of an animal's possible habitats.
Recognizing this potential source of variation, a number of workers have examined
how locomotor behavior is influenced by differences in arboreal architecture (Clennon
and Gebo, 1996; Dagosto, 1992; Doran and Hunt, 1994; Fleagle and Mittermeier, 1980;
Garber and Pruetz, 1995; Gebo and Chapman, 1995; Kinzey, 1976; McGraw, 1996a; Orn-
dorff, 1996). The results are not conclusive. In the study by Garber and Pruetz (1995)
populations of Saguinus mystax exhibited virtually identical locomotor profiles in two
structurally distinct Peruvian forest types. These authors found that while locomotor pro-
files changed little between habitats, "patterns of support preference were more vari-
able ... and appeared to reflect site-specific differences in forest architecture (1995:411)."
In another study, Gebo and Chapman (1995) examined the positional behavior of Colobus
badius inhabiting three regions of Uganda's Kibale Forest with different logging histories.
These authors observed a number of significant differences in the frequencies that quadru-
pedalism, climbing, and leaping occurred in primary, secondary and pine forest and they
related this variability to the size and abundance of gaps between adjacent trees.
Fleagle and Mittermeier (1980) examined the locomotion of seven monkeys ranging
throughout edge, liana, mountain savanna and high forest within the Raleighva\len-
Voltzberg Nature Reserve in Central Surinam. Although locomotor variation between for-
est types per se was not their primary focus (nor were the structural properties of the
habitats quantified), these authors found that with the exception of leaping in liana forest,
"correlations between other locomotor patterns and utilization of forest types are very
low ... "( 1980:311). While these data do not necessarily imply locomotor changes between
habitats, they do demonstrate the conservative nature of locomotion in a diverse group of
monkeys.
Kinzey (1976) studied a group of titi monkeys (Callicebus torquatus torquatus)
ranging throughout a number of Peruvian vegetation zones. The structural properties of
these zones were not quantified; however, Kinzey (1976:468) described each zone in de-
tail and concluded that, "patterns of locomotion and posture differed among different
vegetations." In particular, elevated frequencies of vertical leaping were observed primar-
ily in forest areas characterized by short, densely packed trees with few palms and lianas.
In a more recent study, Doran and Hunt (1994) compared the locomotion of one
chimpanzee popUlation at two sites (Gombe Stream and Mahale Mountains, Tanzania),
two chimpanzee subspecies at three sites (Tai Forest, Gombe and Mahale) and two chim-
panzee species at four sites (Tai, Gombe, Mahale and Zaire's Lomako Forest). Doran and
Hunt demonstrated, among other things, that when the degree of arboreality is accounted
for, no significant differences in locomotion were observed between the Tai and the Tanza-
nian chimpanzee subspecies. These findings are remarkable considering the differences
between habitats: Gombe is characterized as a thicket/woodland or semi-deciduous forest,
Locomotion, Support Use, Maintenance Activities, and Habitat Structure 81

the Mahale Mountain site is a closed forest or woodland (Collins and McGrew, 1988)
while the Tai site consists of lowland, evergreen forest (Boesch and Boesch, 1983). Al-
though the structural properties of the sites were not systematically quantified, there is lit-
tle doubt that differences exist. Indeed, differences in the "openness" of the East and West
African forest canopies have been used to explain variation in other adaptations, most no-
tably in the degree and manner that monkeys are hunted by Tai and Gombe chimpanzees
respectively (Boesch, 1994).
This author recently presented data on the overall locomotor profiles (combined
traveling and foraging) of five cercopithecid monkeys ranging in the Ivory Coast's Tai
Forest (McGraw, 1996b). Despite significant variation in the relative number of boughs,
branches and twigs at the same heights in two forest types, the locomotor profile of each
species did not change with habitat. Further, in four of the five species, locomotor consis-
tency was maintained due to monkeys selecting the same set of supports in each of two
forest types.
The evidence for whether primates alter their locomotion in response to the struc-
tural characteristics of different forest types is, therefore, equivocal. Data from both the
Neotropics and the Paleotropics suggest that New and Old World primates can exhibit
conservative locomotor behavior even if their associated support use changes (Doran and
Hunt, 1994; Garber and Pruetz, 1995). On the other hand, the opposite case can be made
in light of the findings of Kinzey (1976) and Gebo and Chapman (1995), which imply
that, in some instances, locomotor frequencies do change in response to habitat heteroge-
neity. Clearly, further analysis is warranted.
The first study on monkey locomotion in the Tai Forest (McGraw, I 996b) examined
overall locomotor behavior; the locomotor profiles of five cercopithecids were considered
without controlling for maintenance activities. Differences that could exist between loco-
motion during traveling compared to locomotion duringfeeding (foraging) may, therefore,
have been obscured. Various authors have shown that primates not only change their loco-
motor behavior during traveling and foraging respectively, but that they also choose dif-
ferent support types during these activities (Cant, 1987; Doran, 1993; Fleagle and
Mittermeier, 1980; Gebo, 1992; Hunt, 1992; Rose, 1977). If true, then these behavioral ad-
justments should be accounted for. This paper expands the original analysis by treating lo-
comotion and support use during traveling separately from locomotion and support use
during foraging across two forest types. Results are compared with those from the pre-
vious Tai study as well as similar analyses on other primates. Finally, the implications of
these data and the problems associated with collecting information on the availability of
supports are discussed.

2. METHODS
The methods in this study, summarized below, are described in detail in McGraw
(1 996b ). Data were collected in the Ivory Coast's Tai Forest, approximately twenty-five
kilometers from the Liberian border. The core study area, divided into 100 x 100 m cells,
covers approximately 2 and 112 km2 of moist, evergreen rain forest. Within this area, two
forest subtypes are recognized: disturbed and undisturbed forest. Disturbance refers to any
phenomenon, man-made or natural, which disrupts the vertical profile of the canopy.
These phenomena, which include tree falls, may begin as "holes" in the forest extending
through all canopy layers, which later become filled with thicker, colonizing, secondary
growth (Brokaw, 1982). In the Tai Forest, canopy gaps such as these are important ele-
82 W. S. McGraw

ments of the habitat profile because, as Jans et al. (1993:258) report, "forest turnover time
is long (ca 240 years) compared to other tropical moist forests, resulting in a less dynamic,
fine textured mosaic consisting of small eco-units." In some areas, disturbance is also as-
cribed to two major trails and an abandoned road that border the core study area. In addi-
tion, there is evidence that a few trees were felled within the study grid as recently as ten
years ago.
Three cercopithecid species are discussed in this analysis: Colobus polykomos (west-
ern black and white colobus), Colobus badius (red colobus) and Cercopithecus diana (di-
ana monkey). Behavioral data were taken on adult females at three minute time points and
the same individual was not sampled within 15 minutes of itself to assure independence of
data points (see Dagosto, 1994; Mendel, 1976; Walker, 1993). Adult females were chosen
because there are more females in each species' social group; the potential sampling pool
is thus greatly increased because the observer has more independent data points to chose
from (see Janson, 1984, 1990). Every three minutes, the locomotor activity, maintenance
activity (traveling or foraging), support type, height in forest (estimated in m) and location
(grid cell) of an adult female from a given species was recorded (see McGraw, 1996b).
Five locomotor activities were recognized: quadrupedal walking, quadrupedal running,
leaping, climbing, and arm swinging (after Fleagle, 1977; Figures 1-3). Traveling is de-
fined as directed, usually uninterrupted movement between major food sources and/or
sleeping trees. Foraging is locomotion during feeding usually, though not always, confined
to single or contiguous trees. Supports were classified as one of three types: boughs (large
supports usually greater than 10 cm in diameter in which grasping with hands or feet is
not possible), branches (medium sized supports between 2 and 10 cm in diameter permit-

Figure 1. Colobus badius (red colobus monkey) leaping between twigs.

Locomotion, Support Use, Maintenance Activities, and Habitat Structure 83

Figure 2. Colobus polykomos (western black and white colobus monkey) running on a bough.

ting grasping by hands and feet) and twigs (small flexible terminal branches usually less
than 2 cm in diameter) (after Fleagle, 1976 and Fleagle and Mittermeier, 1980). To deter-
mine if intraspecific locomotor differences exist between the two forest types, G-tests (R x
C contingency tests) with William's correction were performed on raw data (Sokal and
Rohlf,1981)_
A canopy survey was conducted to quantify the relative abundance of different sized
substrates at different heights in each forest type. Each 100 x 100 m grid cell sampled
within the core study area was designated as either undisturbed, if the area showed no vis-
ible signs of structural disruption at any height, or disturbed, if tree falls or human activity
were sufficient to influence the architecture' of the forest. Using a calibrated spotting
scope and rangefinder, the vertical profile of the canopy was sampled at six points within
each grid cell. Sampling involved counting the number of different sized supports around
a focal support (within a three m sampling field) at descending ten m intervals. In total,
36,000 square m of forest were sampled. The number of supports in each forest type was
compared using a three-factor (support type, height, forest type) log linear model (Sokal
and Rohlf, 1981).
To determine if monkeys chose the same general supports in each of the two forest
types, the interaction of three variables (chosen support type, height, forest type) was ex-
amined simultaneously. A significant result from the interaction of these variables indi-
cates that a monkey used different supports in each forest type. A non-significant result

• Architecture refers to the forms of stems and their derivatives (e.g., branches) as opposed to canopy physiog-
nomy, canopy organization and canopy texture (see Parker, 1995)_
84 w. S. McGraw

Figure 3. Cercopithecus diana (diana monkey) climbing a trunk.

Table 1. Distribution of support types (twig, branch, bough) in undisturbed and disturbed
forests'

Twig Branch Bough Total

Height interval Undist. Dist. Undist. Dist. Undist. Dist. Undist. Dist.
>40 85 48 29 7 114 55
31-40 429 314 200 156 58 \3 687 483
21-30 582 748 264 427 127 101 973 1276
11-20 626 1526 295 503 108 35 1029 2064
0-10 607 1031 374 827 26 \3 1007 1871
Total 2329 3667 1162 1920 319 162 3810 5749
'Support type data are summed frequencies at 10 m height intervals. A three factor (forest type, height interval,
support type) log-linear model of support differences analysis is significant: Interaction, G(Williams) = 70.328
[Critical -/ (0.05) = 15.5].
Locomotion, Support Use, Maintenance Activities, and Habitat Structure 85

from the three-way interaction does not necessarily indicate similar support use and an ad-
ditional test of conditional independence is required, i.e., does use of support types (factor
A) at different heights (factor B) differ given two forest types (factor C)?

3. RESULTS

Table 1 summarizes the results of the canopy survey. Two general points characterize
both forest types: twigs are the most abundant support type at every level and boughs are
(by far) the least common. There are important differences, however, between the forest
types. Disturbed forest is generally denser at lower levels than undisturbed forest. Disturbed
forest contains more (1) twigs, (2) branches, and (3) total stems, both overall and within
each 10 m interval. This forest type has fewer boughs, particularly at lower levels. For ex-
ample, between II and 20 m, disturbed forest has 2/3 fewer boughs as undisturbed forest
despite having twice as many total supports in the same 10m interval. These differences, as
revealed by the Williams-corrected G value (70.328), are highly significant.
Table 2 details each species locomotor profile in both forest types. In general, the lo-
comotor behavior of these monkeys conforms to the patterns seen among other primates
(e.g., Cant, 1988; Fleagle and Mittermeier, 1980; Gebo and Chapman, 1995): leaping and
quadrupedal running are more common during traveling, and climbing is more common
during foraging. The major exception is leaping in Colobus polykomos; although the black
and white colobus leaped less often during traveling than foraging in undisturbed forest, it
leaped more often during traveling than foraging in disturbed forest.

Table 2. Locomotor behavior during traveling and foraging in undisturbed and

disturbed forests I

Undisturbed forest Disturbed forest

Locomotor
Species behavior Traveling Foraging Traveling Foraging
Colobus polykomos2 Arm swing
Climb 7.9 21.9 6.9 23.6
Leap 7.9 12.5 16.5 13
Run 41.3 15.6 40 14.4
Walk 42.9 50 36.6 49
Colobus badius 3 Arm swing 3.7 6.2 2.5 4.2
Climb 12 24.5 12.3 19.8
Leap 19.4 14 21.6 15.6
Run 8.7 3.5 9.7 5.9
Walk 56.2 51.8 53.9 54.5
Cercopithecus diana 4 Arm swing 0.1
Climb 2.8 20.5 9.1 21.6
Leap 15.7 8.6 17.4 10
Run 27.8 8 20.5 7.8
Walk 53.7 62.9 53 60.5
I For each species. the values in the traveling and foraging columns are the percentage of locomotor time
spent engaged in the five locomotor behaviors.
2There is no significant difference in locomotion between the forest types during traveling (G[Williams]=
3.7) and during foraging (G[Williams]=O.OS). For both activities the critical ./ value is 7.S.
3There is no significant difference in locomotion between the forest types during traveling (G[Wiliiams] =
1.7) and during foraging (G[Williams]= 5.2). For both activities the critical·/ value is 9.5.
'There is no significant difference in locomotion between the forest types during traveling (G[Williams] =
5.5) and during foraging (G[Williams]= 1.2). For both activities the critical ./ value is 7.8.
86 W.S. McGraw

Table 3. Support use while traveling and foraging in undisturbed and disturbed
forests by Colobus polykomos
Undisturbed forest Disturbed forest
Height interval' Support Type' Traveling) Foraging4 Traveling) Foraging4
31-40 m Twig 15.8 27.3 17.6 30.4
Branch 26.3 27.3 11.8 27.7
Bough 57.9 45.4 70.6 41.9
21-30m Twig 20 33.3 21.2 28.7
Branch 28 33.3 26.4 37.3
Bough 52 33.3 52.4 34
11-20 m Twig 21 38.5 39.4 46.4
Branch 31.6 46.2 31.7 41.1
Bough 47.4 15.3 28.9 12.5
'For each height interval, the values in the traveling and foraging columns are the percentage of locomotor
time spent on each support type. Supports used at heights greater than 40 m and between 0 and 10m were
not compared across forest types due to insufficient sample sizes.
'The test for conditional independence during travel was insignificant (/=5.8) as was the test for condi-
tional independence during foraging (/=0.43). The critical/value is 12.6.
3A three factor (forest type, height interval, support type) log-linear model of support differences analysis is
insignificant: Interaction, G(Williams) = 4.3 [Critical X' (0.05) = 9.5].
4A three factor (forest type, height interval, support type) log-linear model of support differences analysis is
insignificant: Interaction, G(Williams) = 0.26 [Critical X' (0.05) = 9.5].

The critical question is whether frequencies of locomotion during either traveling or

foraging change across habitat types. In every instance, the test of inter-forest locomotion
indicates that treating these behaviors separately reveals little difference; for no species is
there a significant change in the locomotor profile during either traveling or foraging
when changing forest types. This does not necessarily imply that individual locomotor ac-
tivities are always constant across habitats. Two cases stand out, both involving movement
during traveling: C. polykomos leaped over twice as often in disturbed forest compared to
undisturbed forest and C. diana climbed over three times as much in disturbed forest as in
undisturbed forest. Nevertheless, when locomotor behaviors are considered together (i.e.,
locomotor profile), monkeys did not move in significantly different ways.
Tables 3-5 show the frequencies with which each support type in each height inter-
val is used by each species across the two forest types during traveling and during forag-
ing. Results of both the 3-way interaction and the test of conditional independence are
also indicated. Treating support use during traveling and foraging separately does not re-
veal a consistent trend, unlike the pattern observed when overall locomotion is examined
(McGraw, 1996b). The analysis reveals that Colobus polykomos is the only species that
uses the same supports in both forest types-regardless of the activity; both the 3-way in-
teraction and the test for conditional independence of support use during traveling and for-
aging yielded non-significant results. Despite an overall lack of significance, two cases in
which there was a sizable difference in the frequency with which a support was used (both
involving traveling) deserve mention. Between 31 and 40 m, C. polykomos used 12.7%
fewer boughs and 14.5% more branches in undisturbed forest than in disturbed forest. In
addition, between 11 and 20 m, C. polykomos used 18.4% fewer twigs, but 18.5% more
boughs in undisturbed compared to disturbed forest.
For Colobus badius (Table 4), analysis reveals that support use during traveling does
not differ between forest types: the largest observed disparity was a 10% difference in the
frequency that twigs were used in undisturbed forest (14.9%) compared to disturbed
Locomotion, Support Use, Maintenance Activities, and Habitat Structure 87

Table 4. Support use while traveling and foraging in undisturbed and disturbed forests
by Colo bus badius
Undisturbed forest Disturbed forest
Height interval' Support type 2 Traveling 3 Foraging 4 Traveling 3 Foraging4
31-40 m Twig 14.9 44.9 24.9 25.2
Branch 16.8 31.9 17.9 30.3
Bough 68.3 23.2 57.2 44.5
21-30 m Twig 24.8 39.3 27.6 35.8
Branch 26.3 33.3 24.6 35.8
Bough 48.9 27.4 47.8 28.4
II-20m Twig 37.8 46.9 34.1 50.0
Branch 31.1 32.8 34.1 32.4
Bough 31.1 20.3 31.8 17.6
'For each height interval, the values in the traveling and foraging columns are the percentage of locomotor
time spent on each support type. Supports used at heights greater than 40 m and between 0 and 10m were
not compared across forest types due to insufficient sample sizes.
lThe test for conditional independence during travel was insignificant (X 2=8.7) although the test for condi-
tional independence during foraging was significant at the 0.05 significance level (X2= 16.1). The critical
Xl values are 12.6 for traveling and 15.6 for foraging.
3 A three factor (forest type, height interval, support type) log-linear model of support differences analysis is
insignificant: Interaction, G(Williams) = 4.3 [Critical Xl (0.05) = 9.5].
4A three factor (forest type, height interval, support type) log-linear model of support differences analysis is
insignificant: Interaction, G(Williams) = 0.26 [Critical Xl (0.05) = 9.5].

(24.9%). On the other hand, the significant result from the test of conditional inde-
pendence indicates that during foraging, this monkey chooses different supports in the two
forest types. In particular, between 31 and 40 m, C. badius uses 20% more twigs and 21 %
fewer boughs in undisturbed compared to disturbed forest.
The diana monkey alters its support use the most (Table 5). Significant results from
the tests of conditional independence reveal that C. diana chooses different supports dur-
ing both traveling and foraging in both forest types. In many instances, the disparity in use
of a support type at a particular height interval between forest types is substantial and
highly significant (Table 5). For example, when diana monkeys travel between 21 and 30
m, they use almost 30% more boughs in undisturbed forest than in disturbed forest. This
difference is magnified even further during foraging: between 21 and 30 m, 43% more
boughs were used in undisturbed forest compared to disturbed forest.

4. DISCUSSION
The results of this analysis, together with those from the first study on Tai monkey
locomotion (McGraw, 1996b) indicate that the locomotion of three cercopithecid mon-
keys, whether examined overall (combined traveling and foraging) or while controlling
for maintenance behaviors (traveling versus foraging) does not change in different forest
types. Although individual locomotor activities may vary, the combined locomotor profile
of each monkey is statistically similar in each forest type. As such, these findings support
the general conclusions of Doran and Hunt (1994), Fleagle and Mittermeier (1980) and
Garber and Pruetz (1995); namely, that locomotor behavior is conservative with respect to
the habitat in which it occurs.
88 w. S. McGraw

Table 5. Support using while traveling and foraging in undisturbed and disturbed
forests by Cercopithecus diana
Undisturbed forest Disturbed forest
Height interval i Support type 2 Traveling3 Foraging4 Traveling3 Foraging4
31-40 m Twig 23.5 37.4 43.4 40.8
Branch 29.5 6.3 29.2 34.8
Bough 47 56.3 27.4 24.4
21-30 m Twig 20.4 20.5 37.8 45.3
Branch 26.6 13.6 38.8 31.5
Bough 53 65.9 23.4 23.2
11-20 m Twig 37.5 39.7 48.1 47.5
Branch 22.5 36.5 37.3 42.8
Bough 40 23.8 14.6 9.7
i For each height interval, the values in the traveling and foraging columns are the percentage of locomotor
time spent on each support type. Supports used at heights greater than 40 m and between 0 and 10m were
not compared across forest types due to insufficient sample sizes.
'The test for conditional independence during travel was significant at the 0.05 significance level (X'=31.9)
as was the test for conditional independence during foraging (X'=45.4). The critical X' value for both ac-
tivities is e 12.6.
3 A three factor (forest type, height interval, support type) log-linear model of support differences analysis is
insignificant: Interaction, G(Williams) = 1.9 [Critical X' (0.05) = 9.5].
4A three factor (forest type, height interval, support type) log-linear model of support differences analysis is
insignificant: Interaction, G(Williams) = 5.3 [Critical X' (0.05) = 9.5].

The data also indicate that patterns of support choice (i.e" selective use and avoid-
ance of specific substrate types) during traveling and foraging in different forest types are
more varied and complex, When support use during traveling and foraging are examined
individually, each of the three monkey species responded in a different manner across the
two forest types, Although no monkey changed its locomotion during either maintenance
activity, one (Cercopithecus diana) used different supports during both activities, one
(Colobus badius) used different supports during foraging but not during traveling and one
(Colobus polykomos) used the same supports during both maintenance activities in both
forest types, This contrasts with the author's earlier study (McGraw, 1996b) in which it
was shown that locomotor consistency is maintained across structurally distinct habitats
because all monkeys select the same types of supports despite differences in their relative
availabilities, Rather, these data demonstrate that, depending on whether a monkey is find-
ing food within a tree or traveling greater distances between feeding or sleeping sites, the
supports available can affect which supports a monkey chooses,
Although the importance of controlling for different maintenance activities is clear
(Cant, 1987; Doran, 1993; Fleagle and Mittermeier, 1980; Gebo, 1992; Gebo and Chap-
man, 1995; Hunt, 1992; Rose, 1977), the more interesting issue is determining why some
monkeys change their support use when others do not. For example, why might searching
for food cause a change in support use between forests?
The fact that two (Colobus badius and Cercopithecus diana) of the three species
chose different supports during foraging (Tables 4 and 5) leads one to suspect that inter-
habitat differences in the location of preferred food items was likely responsible for this
variation, While these data, per se, are not yet available, the behavior of Colobus badius
and Cercopithecus diana in the Tai Forest is now becoming more widely known (Holen-
weg et ai., 1996; Noe and Bshary, 1997; Wachter et ai., 1997). If the spatial arrangement
of food items differed between habitats, then focusing on the relationship between inter-
Locomotion, Support Use, Maintenance Activities, and Habitat Structure 89

forest phenology and the supports used while moving within feeding patches could ex-
plain why these species (in contrast to Colobus polykomos) changed support types during
foraging in the two forest types. In addition, attention should be given to the properties
and locations of the supports containing preferred food items, not merely those that are
used to get there. For example, it would be extremely useful to know how often a monkey
feeds on an item attached above, below or next to the support(s) bearing the animal's
weight. What criteria are used by monkeys to get into the best foraging positions? As
Fleagle (\ 984: 109) noted, "Describing and quantifying the location and abundance of pri-
mate food resources in the habitat by other then retrospective observations remains one of
the major gaps in all studies of primate foraging strategies." Any compelling explanation
of how monkeys manage to maintain consistent locomotor profiles while changing their
support use during foraging (e.g., Colobus badius and Cercopithecus diana) should con-
tain this information.
Why is support use during traveling less variable between forest types than support
use during foraging? Two of the three species (Colobus polykomos and C. badius) chose
the same substrate types for traveling in both forest types (Tables 3 and 4). This supports
the idea (as others have argued before) that primates are generally conservative in their
travel pathways and that routes chosen for uninterrupted movement are fairly constant and
oft-frequented. Indeed, travel paths may be chosen because they provide the safest, most
direct route on preferred supports (Cant, 1992). Colobus polykomos and C. badius are both
large monkeys. Because large monkeys are more likely to prefer relatively larger supports
(Fleagle and Mittermeier, 1980), the number of preferred supports are limited since the
largest supports are always the least abundant (Table 1). These facts may, therefore, re-
quire that large monkeys select the same supports even if different forests vary in the
number of each support type. This appears to be the case for Colobus polykomos and C.
badius. Smaller monkeys, because of the branch to body size ratio, are able to use a wider
variety of available supports and, moreover, to change their support use in response to
more abundant substrates. Thus, Cercopithecus diana (a small monkey) is able to travel on
different supports in different forest types while still using the same locomotor behaviors.
Taken together, these results certainly argue for the conservative nature of locomotion,
but they also emphasize what we still do not know: Are animals that do not change their sup-
port use (e.g., Colobus polykomos) more "constrained" to chose a fixed support profile (e.g.,
Hughes et aI., 1995)? If so, what are the constraining factors? What mechanisms allow a pri-
mate such as the diana monkey to change the supports it moves on, yet still maintain a consis-
tent locomotor profile? What is the appropriate time frame (e.g., daily, weekly, seasonally)
over which to sample locomotor change? These questions are the basis for a wide range of
captive and field studies, all of which can contribute to the larger goal of understanding the
causes of locomotor variability in order to better establish predictive relationships.
The most obvious start is to gather additional information on how other primates re-
spond to differences in habitat. Before we can determine where the balance of evidence
lies, far more field data are necessary for inter- and intra-specific comparisons. This is par-
ticularly important when one considers that the authors of no two studies discussed in this
paper have quantified the habitat in the same way nor controlled for the same behavioral
or habitat variables. For example, is trunk density (Garber and Pruetz, 1995), distance be-
tween canopy gaps (Cannon and Leigton, 1994a; Gebo and Chapman, 1995), or the
number of different-sized supports in each canopy level (McGraw, 1996b) the most appro-
priate habitat variable to consider when studying the locomotion of arboreal animals?
How do the hypotheses we test and the animals we study (e.g., galagos versus gorillas)
dictate the architectural characteristics we focus on (see Emmons, 1995)?
 90 w.s. McGraw

Clearly, identifying the critical structural features is only one of the important tasks
facing primatologists. Accurately quantifying these properties will undoubtedly be even
more challenging. Support size (Fleagle and Mittermeier, 1980; Garber, 1980; Gebo,
1992), inclination (Cant, 1986; Fontaine, 1990; Gebo and Chapman, 1995; Rose, 1978),
compliance (Crompton et aI., 1993, Demes et aI., 1995), density (Cannon and Leighton,
1994b; Garber and Pruetz, 1995; Gebo and Chapman, 1995), and spatial patterning (Kin-
zey, 1976; Mendel, 1976; Morbeck, 1977; Ripley, 1967) all likely play important roles in
support choice and associated locomotor behaviors. This and other studies have reported
information on overall support density, but quantifying the inclination and compliance of
supports in addition to those used during locomotion (i.e., those that were avoided) re-
quires far more rigorous methods. Such data are vital, because as Rose (1978:256-257)
noted, "the degree to which particular branches were used must depend to a certain ex-
tent on the frequency with which branches of different types occur within the habitat."
This requires placing data on used-supports within the broader statistical context of those
supports that were ignored. For example, in an exhaustive survey of locomotion and sup-
port use in Saimiri and Ateles, Fontaine (1990:503) concluded that, "although size con-
straints might lead one to expect major contrasts between the two species in their use of
supports, observed use rate differences for specific support types were minor relative to
the strong similarities. Both Saimiri and Ateles appeared to make full use of available
supports." While this may be true, to convincingly demonstrate support preference (or to
assess the selective use of any habitat feature), it is necessary to have an independent as-
sessment of support availability and the properties of those supports that were not se-
lected. This kind of structural analysis will therefore require a more comprehensive
research protocol.
Various methods are currently available to measure such diverse architectural vari-
ables as canopy structure (Cannon and Leighton, 1994b; Halle et aI., 1978; Parker, 1993;
Schaik and Mirmanto, 1985), spatial heterogeneity of tree types (Chazdon, 1996; Condit,
1996; Hart et aI., 1989; Lemos and Strier, 1992; Lieberman et aI., 1985; Uhl and Murphy,
1981), tree-fall gap dynamics (Brokaw, 1985; Jans et aI., 1993; Martinez-Ramos et aI.,
1988), and vertical stratification of foliage and supports (Beadle et aI., 1982; Ford and
Newbold, 1971; Hubbell and Foster, 1986; Hutchinson et aI., 1986; Malcolm, 1995;
Parker et aI., 1989, 1992). Future studies on locomotion and habitat structure will hope-
fully integrate these and/or other data to better understand the determinants of variability
in positional behavior.
Finally, the solutions to some problems, perhaps best pursued under captive situ-
ations, will be immensely valuable. For example, what are the critical limits of support in-
clination and flexibility that constrain support use for animals of different sizes? When
does quadrupedal ism on oblique supports become climbing? How does this gradation of
behaviors affect the quantification of what we see in the field? What effect does differen-
tial limb mobility have on support choice by generalized arboreal quadrupeds of varying
body sizes? Are the volar pads of some cercopithecids more or less suited to deal with the
frictional properties of different kinds of supports? If yes, what implications does this
have for locomotion in architecturally heterogeneous habitats?
The majority of evidence discussed in this paper indicates that for some species at
least, locomotion is conservative relative to the habitat in which it occurs. On the other
hand, all studies demonstrate that patterns of support use can, in some cases, reflect site-
specific differences. Our ability to develop predictive relationships between behavior,
morphology and habitat therefore depends on the extent that we understand those factors
that cause locomotion to vary as well as the processes that constrain it.
Locomotion, Support Use, Maintenance Activities, and Habitat Structure 91

5. SUMMARY AND CONCLUSIONS

John Napier (1967:333) remarked that "locomotor adaptations have provided the prin-
ciple milestones along the evolutionary pathway of primates from the Eocene to the Pleisto-
cene." This statement echoed earlier arguments of Wood Jones (1916) and Le Gros Clark
(1934), that differences in fossil and living primate post-crania could be explained largely by
differential adaptations to arboreal living. To this end, the relationship between forest archi-
tecture and positional behavior has figured prominently in the evolutionary scenarios offered
to explain the divergence of many primate taxa (e.g., Andrews and Aiello, 1984; Avis, 1962;
Cartmill, 1985; Grand, 1972, 1984; Napier, 1967; Napier and Walker, 1967; Ripley, 1967;
Rose, 1984; Stern and Oxnard, 1973; Temerin and Cant, 1983). Such scenarios are based un-
questionably on a detailed understanding of the morphology and behavior of living primates
as well as the habitats in which these animals live. It is therefore curious that there have been
relatively few attempts to demonstrate how the locomotion of extant primates varies with or is
constrained by elements of forest architecture (but see Cannon and Leighton, 1994a; Doran
and Hunt, 1994; Garber and Pruetz, 1995). This information is critical because while morpho-
logical studies may show what behaviors are possible, comparative field studies document the
range of probable behaviors as well as the contexts in which they occur.
This study has attempted to document the range of behaviors and contexts in which
they occur by providing additional evidence on how some primates respond to habitat het-
erogeneity. The data reveal that although individual locomotor behaviors may vary, the lo-
comotor profiles of three monkey species, during traveling and foraging respectively, do
not change in two forest types; no species demonstrated a significant change in its loco-
motor profile during either maintenance activity. Patterns of support use are more vari-
able, however. Specifically, Colobus polykomos did not change its support use during
either maintenance activity, Cercopithecus diana used different supports in both forest
types during both activities and Colobus badius used the same supports during traveling,
but different supports during foraging in each forest type. Based on these results, the fol-
lowing conclusions can be made:
1. Locomotion is more conservative than support use; i.e., patterns of support use
will change before accompanying locomotor behaviors do.
2. Controlling for maintenance activities does not change the fact that locomotor
profiles are constant across different habitats: the locomotion of Colobus polyk-
omos, C. badius and Cercopithecus diana does not change in two forest types
whether it is examined overall (e.g., combined traveling and foraging)
(McGraw, 1996b) or individually (traveling compared to foraging).
Tai Forest monkeys move in the same general manners independent of differences in
their respective habitats. This fact should strengthen our ability to make predictive rela-
tionships between behavior and morphology. Nevertheless, until more detailed studies can
identify the reasons why some monkeys change their accompanying support use while
others do not, then establishing predictive relationships between behavior and habitat
(e.g., different support types) will prove more difficult.

ACKNOWLEDGMENTS

Support for fieldwork was provided by SUNY at Stony Brook, Sigma Xi, Max
Planck Institute Fur Verhaltensphysiologie, and the National Science Foundation. I thank
 92 W.S.McGraw

Ronald Noe and Bettie Sluijter for permission to work in the Tai Forest and the assistants
of the Tai Forest Monkey Project for helping collect the botanical data. The comments of
Elizabeth Strasser, John Fleagle, Randy Susman, Diane Doran, Brigitte Demes and three
anonymous reviewers greatly improved this paper.

REFERENCES
Andrews P, and Aiello L( 1984) An evolutionary model for feeding and positional behavior. In DJ Chivers, BA
Wood and A Bilsborough (eds.): Food Acquisition and Processing in Primates. New York: Plenum Press,
pp.42~66.
Avis V (1962) Brachiation: The crucial issue for man's ancestry. Southwest. 1. Anthropo!. 18: 119-148.
Beadle CL, Talbot H, and Jarvis PG (1982) Canopy structure and leaf area index in a mature Scots pine forest.
Forestry 55: 105-123.
Boesch C (1994) Hunting strategies of Gombe and Tai Chimpanzees. In RW Wrangham, WC McGrew, FBM de
Waal, and P Heltne (eds.): Chimpanzee Cultures. Cambridge: Harvard University Press, pp. 77-9\.
Boesch C, and Boesch H (1983) Optimization of nut cracking with natural hammers by wild chimpanzees. Behav-
ior 314:265-286.
Brokaw NVL (1982) The definition of tree fall gap and its effect on measures of dynamics of forests. Biotropica
/4:158-160.
Brokaw NVL (\985) Gap-phase regeneration in a tropical forest. Ecol. 66:682-687.
Cannon CH, and Leighton M (1994a) Comparative locomotor ecology of gibbons and macaques: Selection of can-
opy elements for crossing gaps. Am. J. Phys. Anthropo!. 93:505-524.
Cannon CH, and Leighton M (1994b) A field method for making sterile morphological descriptions of tropical
trees and lianas using a reliability measure. Trop. Biod. 2:276-279.
Cant JGH (1986) Locomotion and feeding postures of spider and howling monkeys: Field study and evolutionary
implications. Folia Primatol. 46:1-14.
Cant JGH (1987) Positional behavior of female Bornean orangutans (Pongo pygmaells). Am. J. Primatol.
/2:71-90.
Cant JGH (1988) Positional behavior of long-tailed macaques (Macacafasciclilaris) in northern Sumatra. Am. J.
Phys. Anthropol. 76:29-37.
Cant JGH (\ 992) Postural behavior and body size of arboreal primates: A theoretical framework for field studies
and an illustration of its application. Am. J. Phys. Anthropol. 88:273--283.
Cartmill M (1985) Climbing. In M Hildebrand, D Bramble, K Liem, and D Wake (eds.): Functional Vertebrate
Morphology. Cambridge: Belknap Press, pp. 73--88.
Chazdon RL (1996) Spatial heterogeneity in tropical forest structure: Canopy palms as landscape mosaics. Tree
11:8-9.
Clennon lA, and Gebo DL (1996) Positional behavior in Ceblls caplicinlis. Am. J. Phys. Anthropol. Supplement
22:86-87 (abstract).
Collins DA, and McGrew WC (\ 988) Habitats of three groups of chimpanzees (Pan troglodytes) in western Tanza-
nia compared. l. Hum. Evo!. 17:553--74.
Condit R (\996) Defining and mapping vegetation types in mega-diverse tropical forests. Tree 1/:4-5.
Crompton RH, Sellers WI, and Gunther MM (\ 993) Energetic efficiency and ecology as selective factors in the
saltatory adaptation of pro simian primates. Proc. R. Soc. Lond. (BioI.) 254:41-45.
Dagosto M (\ 992) Effect of habitat structure on positional behavior and substrate use in Malagasy lemurs. Am. 1.
Phys. Anthropo!. Supplement 14: 167 (abstract).
Dagosto M (1994) Testing positional behavior of Malagasy lemurs: a randomization approach. Am. J. Phys. An-
thropol. 94: 189-202.
Demes B, Jungers WL, Gross TS, and Fleagle JG (1995) Kinetics of leaping primates: Influence of substrate ori-
entation and compliance. Am. J. Phys. Anthropo!. 96:41~29.
Doran DM (1993) Sex differences in adult chimpanzee positional behavior: The influence of body size on locomo-
tion and posture. Am. J. Phys. Anthropol. 91:99-116.
Doran DM, and Hunt KD (1994) Comparative locomotor behavior of chimpanzees and bonobos: Species and
habitat differences. In RW Wrangham, WC McGrew, FBM de Waal, and P HeItne (eds.): Chimpanzee Cul-
tures. Cambridge: Harvard University Press, pp. 93-108.
Emmons LE (1995) Mammals of rain forest canopies. In MD Lowman and NM Nadkarni (eds.): Forest Canopies.
New York: Academic Press, pp. 199-223.
Locomotion, Support Use, Maintenance Activities, and Habitat Structure 93

Fleagle JG (1976) Locomotion and posture of the Malayan siamang and implications for hominoid evolution. Fo-
lia Primatol. 26:245-269.
Fleagle JG (1977) Locomotor behavior and skeletal anatomy of sympatric Malaysian leaf-monkeys (Presby tis ob-
scura and Presby tis melalophos) Yrbk. Phys. Anthropol. 20:440-453.
Fleagle JG (1984) Primate locomotion and diet. In DJ Chivers, BA Wood, and AL Bilsborough (eds.): Food Acqui-
sition and Processing in Primates. New York: Plenum Press, pp. 105-117.
Fleagle JG, and Mittermeier RA (1980) Locomotor behavior, body size, and comparative ecology of seven Suri-
nam monkeys. Am. J. Phys. Anthropol. 52:301-314.
Fleagle JG, Mittermeier RA, and Skopec A (1981) Differential habitat use by Cebus apel/a and Saimiri seiureus in
Central Surinam. Primates 22:361-367.
Fontaine R (1990) Positional behavior in Saimiri boliviensis and Ateles geoffroyi. Am. J. Phys. Anthropol.
82:485-508.
Ford ED, and Newbold PJ (1971) The leaf canopy of a coppiced deciduous woodland. I. Development and struc-
ture. J. Ecol. 59:843-862.
Garber PA (1980) Locomotor behavior and feeding ecology of the Panamanian tamarin (Saguinus oedipus geo.f-
froyi, Callitrichidae, Primates). Int. J. Primatol. I: 185-201.
Garber PA, and Pruetz JD (1995) Positional behavior in moustached tamarin monkeys: Effects of habitat on loco-
motor variability and locomotor stability. J. Hum. Evol. 28:411-426.
Gartlan JS, and Brain CK (1968) Ecology and social variability in Cercopithecus aethiops and C. mitis. In P Jay
(ed.): Primates: Studies in Adaptation and Variability. New York: Holt, Rhinehart and Winston, pp.
253-292.
Gebo DL (1992) Locomotor and postural behavior in Alouatta palliata and Cebus capueinus. Am. J. Primatol.
26:277-290.
Gebo DL, and Chapman CL (1995) Positional behavior in five sympatric Old World monkeys. Am. J. Phys. An-
thropol. 97:49-76.
Grand TI (1972) A mechanical interpretation of terminal branch feeding. J. Mammal. 53: 198-201.
Grand TI (1984) Motion economy within the canopy: Four strategies for mobility. In P Rodman and JGH Cant
(eds.): Adaptations for Foraging in Nonhuman Primates. New York: Columbia University Press, pp. 54-71.
Halle F, Oldeman RAA, and Tomlinson PB (1978) Tropical Trees and Forest: An Architectural Analysis. New
York: Springer-Verlag.
Hart TB, Hart JA, and Murphy PG (1989) Monodominant and species-rich forests of the humid tropics: Causes for
their co-occurence. Am. Nat. 133:613-633.
Holenweg A, Noe R, and Schabel M (1996) Waser's gas model applied to associations between red colobus and
diana monkeys in the Tai National Park, Ivory Coast. Folia Primatol. 67: 125-136.
Hubbell SP, and Foster RB (1986) Canopy gaps and the dynamics of a Neotropical forest. In MJ Crawley (ed.):
Plant Ecology. Oxford: Blackwell, pp. 77-95.
Hughes JJ, Ward D, and Perrin MR (1995) Effects of substrates on foraging decisions by Namib Desert gerbils. J.
Mammal 76:638-645.
Hunt KD (1992) Positional behavior of Pan troglodytes in the Mahale Mountains and Gombe Stream National
Parks, Tanzania. Am. J. Phys. Anthropol. 87:83-105.
Hutchinson BA, Matt DR, McMillen RT, Gross LJ, Tajchman SJ and Norman JR (1986) The architecture of de-
ciduous forest canopy in eastern Tennessee, U.S.A. J. Ecol. 74:635--646.
Jans L, Poorter L, van Rompaey RSAR, and Bongers F (1993) Gaps and forest zones in tropical moist forest in
Ivory Coast. Biotropica 25:258-269.
Janson CH (1984) Female choice and mating systems of the brown capuchin monkey Cebus apella (Primates: Ce-
bidae). Z. Tierpsychol. 65: 177-200.
Janson CH (1990) Social correlates of individual spatial choice in foraging groups of brown capuchin monkeys,
Cebus apella. Anim. Behav. 40:910-921.
Kinzey WG (1976) Positional behavior and ecology of Callicebus torquatus. Ybk. Phys. Anthropol. 20:468-480.
Lawes MJ (1990) The distribution of the samango monkey (Cercopithecus mitis erythrarchus Peters, 1852 and
Cercopithecus mitis labiatus I. Geoffroy, 1843) and forest history in southern Africa. J. Biogeography
17:669-680.
Le Gros Clark WE (1934) Early Forerunners of Man. London: Bailliere, Tindall and Cox.
Lemos de Sa R, and Strier KB (1992) A preliminary comparison of forest structure and use by two isolated groups
of woolly spider monkeys, Brachyteles arachlloides. Biotropica 24:455-459.
Lieberman M, Lieberman D, Hartshorn GS and Pevalta R (1985) Small-scale altitudinal variation in lowland
tropical forest vegetation. J. Ecol. 73:505-516.
Malcolm JR (1995) Forest structure and the abundance and diversity of neotropical small mammals. In MD Low-
man and NM Nadkarni (eds.): Forest Canopies. New York: Academic Press, pp. 179-197.
94 W.S.McGraw

Martinez-Ramos M, Alvarez-Buylla E, Sarkukhan J, and Pinero D (I 988} Treefall age determination and gap dy-
namics in a tropical forest. 1. Ecol. 76:700-716.
McGraw WS (I 994} Census, habitat preference and polyspecific associations of six monkeys in the Lomako For-
est, Zaire. Am. J. Primatol. 34:295-307.
McGraw WS (l996a) The Positional Behavior and Habitat Use of Six Monkeys in the Tai Forest, Ivory Coast.
Ph.D. Dissertation. State University of New York at Stony Brook.
McGraw WS (l996b) Cercopithecid locomotion, support use, and support availability in the Tai Forest, Ivory
Coast. Am. J. Phys. Anthropol.lOO:507-522.
Mendel F (1976) Postural and locomotor behavior of Alouatta palliata on various substrates. Folia Primatol.
26:36-53.
Morbeck ME (1977) Positional behavior, selective use of habitat substrate and associated non-positional behavior
in free-ranging Colobus guereza (Ruppel, 1835). Primates 18:35-58.
Moreno-Black G, and Maples WR (1977) Differential habitat utilization of four Cercopithecidae in a Kenyan for-
est. Folia Primatol. 27:85-107.
Napier JR (1967) Evolutionary aspects of primate locomotion. Am J. Phys. Anthropol. 27:333-341.
Napier JR, and Walker A (1967) Vertical clinging and leaping - a newly recognized category of locomotor behav-
ior of primates. Folia Primatol. 6:204-219.
Noe R, and Bshary R (1997) The formation of red colobus-diana monkey associations under predation pressure
from chimpanzees. Proc. R. Soc. London, B. 264:253-259.
Orndorff KA (1996) Positional behavior of Cebus capucinus in a Costa Rican rainforest: Comparisons with pub-
lished data collected in a Costa Rican dry forest. Am. 1. Phys. Anthropol. supplement 22: 180-181 (ab-
stract).
Parker GG (1993) Structure and dynamics of the outer canopy of a Panamanian dry forest. Selbyana /4:51.
Parker GG (1995) Structure and microclimate of forest canopies. In MD Lowman and NM Nadkarni (eds.): Forest
Canopies. New York: Academic Press, pp. 73-106.
Parker GG, O'Neil JP, and Higman D (1989) Vertical profile and canopy organization in a mixed deciduous forest.
Vegetatio 89: 1-12.
Parker GG, Hill SM, and Kuehnel LA (1992) Access to the upper forest canopy with a large tower crane. Bios-
cience 42:664-670.
Pounds JA (1991) Habitat structure and morphological patterns in arboreal vertebrates. In S Bell, E McCoy, and H
Mushinsky (eds.): Habitat Structure: The Physical Arrangement of Objects in Space. New York: Chapman
and Hall, pp. 109-119.
Ripley S (1967) The leaping of langurs: A problem in the study of locomotor adaptation. Am. J. Phys. Anthropol.
26: 149-170.
Rose MD (1977) Positional behavior of olive baboons (Papio anubis) and its relationship to maintenance and so-
cial activities. Primates /8:59-116.
Rose MD (1978) Feeding and associated positional behavior of black and white colobus monkeys (Colobus
guereza). In GG Montgomery (ed.): The Ecology of Arboreal Folivores. Washington DC: Smithsonian In-
stitution, pp. 253-262.
Rose MD (1984) Food acquisition and the evolution of positional behavior: The case of bipedalism. In DJ Chivers,
BA Wood, and A Bilsborough (eds.): Food Acquisition and Processing in Primates. New York: Plenum
Press, pp. 509-524.
Schaik CP van, and Mirmanto E (1985) Spatial variation in the structure and litterfall of a Sumatran rain forest.
Biotropica 17: 196-205.
Sokal RR, and FJ Rholf (1981) Biometry: The Principles and Practice of Statistics in Biological Research. New
York: WH Freeman and Company.
Stern JT, and Oxnard CE (l973) Primate locomotion: Some links with evolution and morphology. Primatologia
4:1-93.
Temerin LA, and Cant JGH (l983) The evolutionary divergence of Old World monkeys and apes. Am. Nat.
122:335-351.
Uhl C, and Murphy PG (1981) Composition, structure and regeneration of a terra firma forest in the Amazon basin
of Venezuela. Trop. Ecol. 22:219-237.
Wachter B, Schabel M, and Noe R (l997) Diet overlap and poly-specific associations of red colobus and diana
monkeys in the Tai National Park, Ivory Coast. Ethology 103:514-516.
Walker SE (1993) Positional Adaptations and Ecology of the Pitheciini. Ph.D. Dissertation, City University of
New York.
Wood Jones F (1916) Arboreal Man. London: Edward Arnold.
 6

THE GORILLA PARADOX

The Effects of Body Size and Habitat on the Positional
Behavior of Lowland and Mountain Gorillas

Melissa 1. Remis

Department of Sociology and Anthropology

Purdue University
West Lafayette, Indiana 47907-1365

1. INTRODUCTION

Gorillas posses general ape adaptations for climbing yet the best studied gorillas in
the Virungas depart from the arboreal, frugivorous pattern seen among the other apes. Ter-
restriality and folivory among gorillas have traditionally been viewed as consequences of
their large body size. This study aims to separate the influences of body size from those of
habitat on tree climbing by gorillas by comparing the results of 27 months of study of
western lowland gorillas (Gorilla gorilla gorilla) at Bai Hokou, Central African Republic
(Remis, 1994, 1995) to published reports on the positional behavior of mountain gorillas
(Gorilla gorilla beringei) at Karisoke, Rwanda (Tuttle and Watts, 1985; Doran, 1996).
The first European descriptions of gorillas (Gorilla gorilla gorilla) by explorers and
scientists in West Africa emphasized their arboreal habits (Purchas, 1625; Savage and Wy-
man, 1847; DuChaillu, 1861; Gregory and Raven, 1937). Nevertheless, subsequent explo-
ration and scientific study of the terrestrial mountain gorilla (Gorilla gorilla beringei) in
the Virungas (Akeley, 1923; Bingham, 1932; Kawai and Mizuhara, 1959; Emlen and
Schaller, 1960) overshadowed the West African work and the early accounts were dis-
missed as fanciful exaggerations based on unreliable reports from local people.
Detailed field studies of the terrestrial and herbivorous mountain gorilla (Schaller,
1963; F ossey and Harcourt, 1977; Watts, 1985) have influenced scientific notions of the
ecological significance of large body size among primates (e.g., Clutton-Brock and
Harvey, 1977). Body size is one of the few attributes of animals that can be estimated
from fossil remains. We want to be able to interpret its significance for the behavior ofliv-
ing and fossil species. Before we can understand the impact of body size on the lifestyles

95
96 M.J. Remis

of gorillas or other species, we must separate its effects from those of habitat on diet and
arboreality.
Early descriptions of gorillas by anatomists (Savage and Wyman, 1847; Owen,
1859; Sanford, 1862; Morton, 1922) emphasized morphological characters associated with
arboreality. Yet, the natural historian Akeley (1923) considered the arboreal aspects of go-
rilla morphology to be vestigial, since the adult gorillas he observed in the Virungas did
not spend a significant amount of time in trees. It is not surprising that the Virunga gorillas
are terrestrial, given the dwarf nature of the montane forest, the reduced canopy cover and
the scarcity of tree foods (Richards, 1957; Schaller, 1963). It is surprising, however, that
Akeley (1929) ignored the significance of the gorilla's specialized habitat in explaining
their terrestrial adaptation. Instead, he invoked body size as the causal factor, as did others
(e.g., Jones and Sabater-Pi, 1971).
Akeley (1929), Schaller (1963) and Fossey (1983) each painted a vision of mountain
gorillas as terrestrial gentle giants. Meanwhile, until recently, the earlier accounts of goril-
las in lowland regions using trees for feeding and nesting were discounted as anomalies,
or as unscientific and invalid reports. Although they have proved difficult to study, the ma-
jority of gorillas (Gorilla gorilla gorilla) live in lowland tropical forests in west and cen-
tral Africa. These tropical forests contain a rich complement of fauna and flora and a much
higher diversity of arboreal substrates and foods, particularly fruits, than montane forest
habitats (Richards, 1957; Tutin and Fernandez, 1984).
Recently, studies in tropical forests have demonstrated that lowland populations are
quite distinct from mountain gorillas. Work at Lope, Gabon, several sites in the Congo and
the Central African Republic has shown that western lowland gorillas are both more arbo-
real and frugivorous than mountain gorillas (Fay, 1989; Fay et aI., 1989; Williamson et aI.,
1990; Remis, 1994, 1995, in press; Nishihara, 1995; Carroll, 1996). This suggests that
mountain gorilla behavior reflects an adaptation to a specialized habitat rather than a body
size based species-specific pattern of obligate terrestriality.

1.1. Ape Anatomy and Behavior: The Gorilla Paradox

Gorillas are the largest living primates, with lowland males weighing, on average,
169 kg and females 71 kg (Willoughby, 1978; Jungers and Susman, 1984). Body size
ranges for male mountain and lowland gorillas overlap. Female mountain gorillas may be
somewhat larger than western lowland females (mountain gorilla males average 159 kg,
females 98 kg, but data are few (Jungers and Susman, 1984).
Some authors have suggested that the adaptive radiation of the African apes incorpo-
rated wide variation in body size but little morphological or ecological diversity (Shea,
1983; Jungers and Susman, 1984). The African apes are frugivorous, arboreal climbers
and engage in some suspensory behavior (Fleagle, 1988; Hunt, 1991), except the special-
ized mountain gorilla. Morphological adaptations for terrestrial knucklewalking among
gorillas (Lewis, 1969; Tuttle, 1969; Tuttle and Watts, 1985), co-occur with features that
aid tree climbing (Keith, 1923; Schultz, 1963; Washburn, 1973; Fleagle et aI., 1981).
The discrepancy between general gorilla morphology for climbing and their terres-
trial adaptation has traditionally been explained as a result of phyletic history coupled
with a recent body size increase among gorillas relative to chimpanzees (Akeley, 1929;
Schaller, 1963; Tuttle and Watts, 1985). Nevertheless, researchers have recently begun to
acknowledge behavioral differences between subspecies that help us to understand the re-
lationship between gorilla morphology and behavior (Fleagle, 1988; Watts, 1990; Hunt,
1991 ).
The Effects of Body Size and Habitat on the Positional Behavior of Lowland and Mountain Gorillas 97

The terrestrial and herbivorous mountain gorilla can now be seen to depart from the
arboreal, suspensory and frugivorous pattern common among other apes, including low-
land gorillas. Behavioral differences between the subspecies of gorillas are correlated with
anatomical differences. Mountain gorillas have relatively shorter arms and less divergent
big toes than lowland gorillas (Straus, 1930; Schultz, 1934). They have more highly
crested molar surfaces than the lowland forms (Uchida, 1992). Systematic data on the ar-
boreal habits and positional behavior of lowland gorillas dispel the notion of a paradox be-
tween overall gorilla morphology and behavior and emphasize the similarities between
lowland gorilla positional behavior and that of chimpanzees.
Positional behavior encompasses locomotor activities, postural modes, and manipu-
lation of the environment (Prost, 1965) and is fundamental to lifetime reproductive suc-
cess (Zihlman, 1992). Many recent positional behavior studies have built on the early
work of Ripley (1967, 1979), Kinzey (1976), Morbeck (1976, 1979) and Rose (1977,
1979) that emphasized the interaction between individuals and their environment (Garber,
1984; Cant, 1992; Dagosto and Gebo, this volume). A new focus for the study of posi-
tional behavior has been assessing intraspecific variability (Garber and Preutz, 1995;
Gebo and Chapman, 1995; McGraw, 1996). In addition, as Hunt (1992) and Remis (1995)
have emphasized, the social realm also influences positional behavior.
Positional behavior studies of the African apes have upheld predictions concerning
the relationship of body size to arboreal behavior. Large animals should face difficulties
with balance, substrate size and the energetics of climbing and these should affect their
frequency of climbing, size of substrates used and positional modes (Grand, 1972;
Cartmill and Milton, 1977; Hunt, 1994). For example, the frequency of arboreal behaviors
among pygmy chimpanzees and chimpanzees may vary with body size (Doran, 1993a,b).
Arboreality among the great apes is also correlated with the amount of fruit in the diet,
and lowland gorillas that eat a lot of fruit are more arboreal than mountain gorillas that do
not (Williamson et aI., 1990; Remis, 1994).
The influence of habitat and resulting ecological adaptation on the degree of arboreal-
ity among gorillas can be assessed by comparing the positional behavior of the western low-
land and mountain subspecies. Comparisons of age and sex differences in posture and
locomotion make it possible to examine the effects of body size on arboreal competence
while controlling for both phylogeny and habitat (Cant, 1987a,b, 1992). Among all of the
great apes, the large body size of males may constrain their positional behavior relative to
females and juveniles. Specifically, males may spend more time on the ground or at lower
heights of trees. When higher in trees, they may suspend more frequently, and spend less
time in the small or terminal branch milieu than smaller animals (e.g., Goodall, 1977; Tuttle
and Watts, 1985; Cant, 1987b, 1992; Doran and Hunt, 1994; Remis, 1995). Of course, in ad-
dition to body size, substrate use is affected by patch size, the number and types of supports
available in a given tree, foraging party size, and social rank (Hunt, 1992; Remis, 1995).

2. METHODS

Data from 27 months of research on the positional behavior of western lowland gorillas
at the Bai Hokou Study Site in the Central African Republic were compared to published re-
ports on the positional behavior of the mountain gorillas at Karisoke, Rwanda collected by
Watts (Tuttle and Watts, 1985) and Doran (1996). The Bai Hokou study site occupies lowland
rainforest (463 m altitude) and receives, on average, 1400 mm of rain a year. Karisoke is situ-
ated in a montane rainforest (2680-3700 m altitude) and receives 1800 mm of rain per year.
98 M.J. Remis

During the Bai Hokou study, instantaneous samples of activities (percent time feed-
ing and resting), diet, positional behavior, and substrate use were collected at one minute
intervals on focal animals. Ad-libitum data were collected opportunistically on the use of
trees at first encounter during contacts (see Remis, 1995 for greater detail). These data are
compared to Tuttle and Watts' (1985) report of the activities and postural behavior of
mountain gorillas at Karisoke. It is not possible to statistically analyze these comparisons
because, although the data sets are comparable, there is no published information about
the age/sex distribution of Tuttle and Watts' data. The Bai Hokou data are also compared
to Doran's (1996) report of locomotion among the gorillas at Karisoke. Methods used to
collect locomotor data during Doran's study of gorillas at Karisoke (Doran, 1996) were
similar to those used at Bai Hokou, and the resulting data sets are analyzed by two-tailed,
G tests of Independence (Sokal and Rohlf, 1981).

3. RESULTS

3.1. Gorilla Foraging Patterns

Lowland gorillas live in a spatially and temporally complex habitat. The tree canopy
in lowland forests in West and Central Africa averages about 40 m or more and canopy
foods (leaves, flowers and fruit) are plentiful. Not surprisingly, lowland gorillas consume
many foods arboreally, especially fruit, and their daily and home ranging patterns reflect
heterogeneity in the temporal and spatial distribution of foods (Table 1).
At Karisoke, trees are short (many are less than 7 m in height) and there are minimal
food items available in trees to the gorillas (Fossey, 1983). Therefore, mountain gorillas
have fewer incentives and opportunities to climb trees than do lowland gorillas. Most
mountain gorilla foods are obtained terrestrially and are perennially available (Watts,
1984). The Karisoke gorillas do not consume the leaves of trees and eat only one kind of
tree fruit (Fossey, 1983; Watts, 1984). When mountain gorillas climb trees to eat vines or
epiphytes, they are often on large fallen trees and boughs only a few meters off of the
ground.

3.2. Tree Climbing among Gorillas

The gorillas at Bai Hokou were never fully habituated. As a consequence, the data
are biased towards arboreal sightings (only 15% of the data are of terrestrial sightings) and
preclude an accurate reconstruction of percent time spent in trees. While lowland gorillas

Table 1. Gorilla foraging patterns

Bai Hokou 1 Karisoke'

Number food types 236 75
Number fruit species eaten 77 3
Percent food types obtained in trees 79 13
Day range 2.3 km 0.57 km
Annual range 18.1 km' 9.4 km'
Average height in trees >20m <7m
1Bai Hokou lowland gorilla data from Remis (1994, 1995).
'Karisoke mountain gorilla data from Watts (1984, 1991) and Doran (1996).
The Effects of Body Size and Habitat on the Positional Behavior of Lowland and Mountain Gorillas 99

probably spend at least 20% of their time in the trees, they are probably less arboreal than
chimpanzees that live in similar habitats. Adult mountain gorillas spend only 3% of their
time in the trees (Tuttle and Watts, 1985).
Sex differences in arboreal substrate use among lowland gorillas are suggested by
ad-libitum data on first encounters of gorillas (in trees: females 56% of samples (n= 143),
males 24% of samples (n=239». Males were also recorded in the trees less frequently than
females during the dry season when fruit is rare and during fruit-poor wet seasons, (instan-
taneous interval sampling, dry season in trees: females 91 % of samples (n=89), males 64%
of samples (n= 188), wet season in trees: females 89% of samples (n=369), males 83% of
samples (n=1053); fruit-poor wet season 1995: females 95% of samples (n=555), males
58% of samples (n=377». In addition, females use small branches and the peripheries of
trees more than do males (Remis, 1995). In all seasons, silverbacks differ from other indi-
viduals in the frequency of nesting in trees (non-silverback nests in trees: 21 % of nests
(n= 1020), silverback nests in trees: 4% (n=211». Similarly, the amount of time spent in
the trees is correlated with body size among male and female mountain gorillas and east-
ern lowland gorillas (Goodall, 1977; Tuttle and Watts, 1985; Doran, 1996). At Karisoke,
females spend 7% time above ground, males only 2% (Doran, 1996).

3.3. Arboreal Activity Budgets

Adult lowland gorilla arboreal activity budgets during posture are divided between
feeding and resting (Bai Hokou group adults, arboreal postural activities: feed = 64%, rest
= 35%, n = 777). At Bai Hokou, lone male arboreal activity budgets differ from those of
group animals. They rest even more and feed less than other individuals (feed = 29%, rest
= 70%, n = 475). In contrast, adult mountain gorillas spend almost all of their arboreal
time feeding on bark, leaves of vines or galls, and come out of the trees to rest (Karisoke
adults, arboreal postural activities: feed = 91 %, rest = 9%, n = 1700 hours).

3.4. Terrestrial Positional Behavior

In general, lowland and mountain gorillas have similar terrestrial positional behavior
profiles. Both lowland and mountain gorillas travel on the ground by knucklewalking and
more rarely run quadrupedally or bipedally during play or display (Tuttle and Watts, 1985;
Remis, 1994). While feeding on the ground, lowland gorillas at Bai Hokou stand more than
do mountain gorillas. Lowland gorilla females stand even more than males and travel more
frequently between food patches. Mountain gorillas appear to be able to squat and reach a
large number of terrestrial foods from one location (Tuttle and Watts, 1985; Table 2).

3.5. Arboreal Feeding Postures

All gorillas spend the majority of their time in trees in postural rather than locomo-
tor modes: sitting or squatting are common and suspensory postures are rare. Adult low-
land gorillas suspend more frequently than mountain gorillas and juveniles of both
subspecies suspend more than larger bodied animals (Table 3). Sex differences in arboreal
posture exist among lowland gorillas. Group males sit and recline more than females, and
squat less. They spend slightly less time in bipedal or suspensory postures, contrary to pre-
dictions (Cant, 1987b). Females also differ significantly from lone males, who squat less
and recline more than females. When feeding arboreally, mountain gorillas squat and use
100 M.J. Remis

Table 2. Terrestrial feeding postures

Bai Hokou l Karisoke 2
Females Males
Posture N=19 N=93 Females Males
Sit 63.2 52.7 71.4 60.9
Squat 9.7 25.5 36.1
Biped 3.2 0.1 0.2
Quad. stand 36.8 31.2 2.7 2
Lie 3.2 0.4 I
I Bai Hokou lowland gorilla data from Remis (1995).
2Karisoke mountain gorilla data from Tuttle and Watts (1985).

tripedal postures more and sit less than lowland gorillas, probably as a result of a low den-
sity of arboreal foods.

3.6. Arboreal Locomotion

Gorillas use arboreal locomotion when climbing in or out of food trees and between
feeding sites within trees. Gorillas generally travel between trees by descending and knuck-
lewalking on the ground. For both lowland and mountain gorillas, most arboreal locomotion
is quadrupedal walking and climbing, although sample sizes are small. Although there are
no significant sex differences in the arboreal locomotor patterns of lowland gorillas, there
are significant differences between subspecies and between male and female mountain go-
rillas (Table 4). Lowland gorillas vertical climb and scramble more than mountain gorillas.
Lowland females engage in more bipedal locomotion than lowland males while mountain
males walk bipedally in trees more than do females of their subspecies.

4. DISCUSSION

Wherever gorillas have been studied in lowland rain-forests, which contain a high
number of arboreal substrates and tree foods, they use trees at a much higher frequency
than do mountain gorillas. In fact, lowland gorillas have a positional repertoire as varied

Table 3. Arboreal feeding postures

Bai Hokou l Karisoke 2 N= 1700 hrs
Juveniles Females Males
Posture N=26 N=269 N=366 Juveniles Females Males
Sit 34.6 62 54.6 1.2 0.8
Squat 3.8 27 36.1 84.1 84.6 63.1
Suspend 3.8 3 1.3 2.8 0.3
Biped 7.7 3.4 0.8 0.8 0.6
Triped 3.8 1.5 1.1 8.3 9.1 14.3
Quad. stand 42 3.8 1.9 1.9 3.7 22.6
Lie 3.8 0.4 0.3 0.9 0.8
I Bai Hokou lowland gorilla data from Remis (1995).
2Karisoke mountain gorilla data from Tuttle and Watts (1985).
The Effects of Body Size and Habitat on the Positional Behavior of Lowland and Mountain Gorillas 101

Table 4. Arboreal locomotion

Bai Hokou i Karisoke~

Females Group Males Lone Males Females Males

Locomotor activity N=95 N=38 N=41 N=118 N=35
Quadruped. walk 23.2 21.5 22 47.9 68.2
Climb 40 42.1 54 45 21.9
Scramble 24 23.7 17
Suspend 4.2 7.9 5.2 5.6
Biped 4.2 2.6 1.4 4.2
Acrobat 4.2 7.9 7 0.5

Tests for similarit/ G P df

Bai Hokou gorillas (BH)
Females vs. Group Males 2.4 ns 5
Group Males vs. Lone Males 1.2 ns 4
Karisoke gorillas
Females vs. Males 21.2 0.001 3
BH vs. Karisoke gorillas
Females vs. Females 13.1 0.001 3
Males vs. Males 15.1 0.001 3
iSai Hokou data from Remis (1994,1995).
2Karisoke data from Doran (1996).
3G, G Test; P, probability; df, degrees of freedom; ns, not significant.

as those of the other great ape species. In contrast, arboreal activities comprise a small
proportion of mountain gorilla behaviors. The impoverished positional repertoire of moun-
tain gorillas reflects their reliance on ground foods and the scarcity of arboreal substrates
and foods in their habitat. At Bai Hokou, canopy cover reaches 100% and both logged and
unlogged forest areas contain, on average, 505-516 trees/ha (Carroll, 1996). In contrast, at
Karisoke, even in the densest lower altitude "mixed" or Hagenia sp. forest areas, canopy
cover is less than 50% and mean tree density is only 101.3 trees/ha. In addition, most trees
produce wind-dispersed rather than fleshy fruits (Fossey, 1983; Watts, pers. comm.).
Lowland gorillas have more active terrestrial foraging patterns than do mountain go-
rillas, which are likely a response to habitat differences in food distribution. The lower
density of herbaceous foods in lowland forest, even in areas of previous disturbance, rela-
tive to montane forest may cause lowland gorillas to move further between patches of
herbs than mountain gorillas (Rogers and Williamson, 1987; Watts, 1987; Malenky et al.,
1994; Remis, 1994). Further, many of the differences in arboreal postures between low-
land and mountain populations are related to habitat differences in substrate availability.
While lowland gorillas use the small substrates found at great heights in the canopy, the
mountain subspecies make greater use of large low-lying boughs and fallen trunks.
The locomotor repertoire of lowland gorillas resembles that of chimpanzees and
orangutans. Within trees, lowland gorillas scramble/clamber more than smaller bodied
chimpanzees in a similar habitat. Bai Hokou gorillas scramble/clamber almost as much as
orangutans, more than might be expected given the differences in locomotor anatomy be-
tween the two (Sigmon, 1974; Cant, 1987a; Table 5). As large animals, both gorillas and
orangutans use scrambling/clambering behaviors to distribute their body weight over mul-
tiple substrates, especially in the periphery of trees. Among these large apes, scram-
ble/clamber often occurs during suspension by the forelimbs because feet can reach, grasp
and support weight on lower branches, resulting in quadrumanous locomotion (Cant,
102 M.J. Remis

Table 5. Arboreal locomotion of the great apes

Tai' Bai Hokou Kutai'
Pan Females Pan Males Gorilla Females Gorilla Males Pongo Females
Locomotor activity N=122 N=103 N=95 N=38 N=4360
Quadruped 30.3 11.7 23.2 21.5 12
Climb 52.4 60.2 40 42.1 31.3
Scramble/Clamber 4.9 1.9 24 23.7 39.4
Suspend 7.4 5.8 4.2 7.9 11.8
Biped 0.8 5.8 2.6 4.2
Misc. 4.1 14.6 4.2 7.9 5.6
'Data from Doran and Hunt (1994).
'Data from Cant ( 1987b).

1987b). Chimpanzees employ more quadrupedal ism than larger apes and large males of all
species use tree-swaying and bridging to transfer between trees. Contrary to predictions
(Cant, 1987b, 1992), lowland gorillas do not appear to engage in significantly less arbo-
real suspensory locomotion than chimpanzees (Cant, 1987b; Doran and Hunt, 1994; Re-
mis, 1995).
The large size of gorillas may set limits on their arboreal behavior and many of the
body size based predictions outlined at the outset of this paper are supported. Despite this,
while lowland gorillas are more arboreal than the mountain subspecies, lowland gorilla
males do not appear to weigh less than the mountain variety. Large male gorillas of all
subspecies use trees, but they appear to do so less frequently than other (smaller) apes.
Chimpanzees spend between 33--68% of their time in trees (Doran and Hunt, 1994), and
degree of arboreality varies with habitat.
At Bai Hokou, differences in body size between male and female gorillas were asso-
ciated with differences in substrate use (size, height, section and level) and modes of posi-
tional behavior. For example, females' use of smaller substrates at Bai Hokou resulted in
sex differences in arboreal posture (Remis, 1995). As among smaller and less dimorphic
chimpanzees, however, the most challenging results of the Bai Hokou study emerged out-
side the context of body size predictions and involved the effects of season, social context
and patch size on substrate use (Remis, 1995; also Hunt, 1992).
At Bai Hokou, the season and the presence of other individuals, their number, and
sex affect the size of the patch used and the choice of substrates once in a tree. Female go-
rillas use the peripheral parts of trees while feeding on fruit in the rainy season. All goril-
las remain in the core of trees on large and secure substrates when feeding on leaves and
bark in the dry season. In all seasons, bisexual foraging parties climb larger trees (contain-
ing a broader array of available substrates) than smaller single-sex parties or lone indi-
viduals, group males use more peripheral parts of trees than lone males, and females with
males use more peripheral parts of trees than those in single sex parties (Remis, 1995).
Lone males climb more, and scramble and suspend less than males in groups. Hence, the
relationships between body size, diet, social context and positional behavior can be de-
scribed as part of a multidimensional system by which an organism gets access to food.
This comparison of lowland and mountain gorilla positional behavior highlights the
importance of the effects of habitat on intraspecific differences in positional behavior. The
terrestrial repertoires of lowland and mountain gorillas are similar. Feeding is the primary
impetus for climbing among all gorillas. Marked habitat differences in temporal and spa-
tial food distribution, and the density and availability of substrates appear to be responsi-
The Effects of Body Size and Habitat on the Positional Behavior of Lowland and Mountain Gorillas 103

ble for the major differences in arboreal and terrestrial foraging profiles between subspe-
cies. Lowland gorillas spend more time in trees than mountain gorillas, obtaining a much
larger proportion of their foods arboreally. This analysis suggests that the extreme habitat
differences between gorilla popUlations may prevent maintenance of a "gorilla-typical" lo-
comotor pattern (unlike the tamarins studied by Garber and Preutz (1995) or the monkeys
studied by McGraw (1996) at the Tai Forest,). In addition, there are likely taxonomic dif-
ferences in intraspecific plasticity of locomotor profiles (McGraw, 1996).
For most primates studied, habitat structure, food availability, availability of supports
of different sizes, season, and social context shape positional behavior and substrate use
(Dagosto, 1992; Hunt, 1992; Gebo and Chapman, 1995). We need to take these factors into
account when attempting broad range intraspecific or interspecific comparisons. To fully
understand the effects of body size on arboreal behavior of African apes we need to quantify
habitat differences and availability of supports between populations and to conduct studies
of animals in similar habitats eating similar resources. Future study of sympatric chimpan-
zees and gorillas in lowland forests, eating the same foods, may provide the best compari-
sons for understanding the ways in which the large body size of gorillas may constrain their
degree of arboreality and shape interspecific differences in arboreal positional behavior.

ACKNOWLEDGMENTS

The Bai Hokou study was completed under the auspices of the Central African Min-
istries of Forests and Waters and Scientific Research. The support of Fulbright-lIE, Na-
tional Science Foundation (with A. Richard), National Geographic Society, Wenner-Gren
Foundation for Anthropological Research, World Wildlife Fund, L.S.B. Leakey Founda-
tion, and the Mellon Fund are gratefully acknowledged. This research could not have been
completed without the hard work of Louise Dion, Etienne Ndolongbe, Wonga Emile and
Mokedi Priva, the logistical assistance of the Dzanga-Sangha Dense Forest Reserve Staff
or the mentoring of Alison Richard. I would also like to acknowledge Warren Kinzey for
helping to provide me with the opportunity to study positional behavior. David Watts
kindly provided unpublished data on the trees at Karisoke. The comments of Elizabeth
Strasser and three other reviewers improved this paper.

REFERENCES
Akeley CE (1923) In Brightest Africa. New York: Doubleday.
Akeley MLJ (1929) Carl Akeley's Africa. The Account of the Akeley-Eastman-Pomeroy African Hall Expedition
of the American Museum of Natural History. New York: Blue Ribbon.
Bingham, HC (1932) Gorillas in a native habitat. Carnegie Inst. Wash. Publ. 426:1-66.
Cant JGH (1987a) Positional behavior of female Bornean orangutans (Pongo pygmaeus). Am. J. Primatol.
12:71-90.
Cant JGH (1987b) Effects of sexual dimorphism in body size on feeding postural behavior of Sumatran orangutans
(Pongo pygmaeus). Am. J. Phys. Anthropol. 74: 143-148.
Cant JGH (1992) Positional behavior and body size of arboreal primates: A theoretical framework for field studies
and an illustration of its application. Am. J. Phys. Anthropol. 88:273-283.
Carroll RW (1988) Relative density, range extension, and conservation potential of the lowland gorilla (Gorilla
gorilla gorilla) in the Dzanga-Sangha region of southwestern Central African Republic. Mammalia 52(3):
311-323.
Carroll RW (1996) Feeding ecology of lowland gorillas (Gorilla gorilla gorilla) in the Dzanga-Sangha Dense For-
est Reserve of the Central African Republic. Ph.D. Dissertation. Yale University, New Haven, Connecticut.
 104 M.J. Remis

Cartmill M, and Milton K (1977) The lorisiform wrist joint and the evolution of "Brachiating" adaptations in the
Hominoidea. Am. J. Phys. Anthropol. 47:249-272.
Clutton-Brock TH, and Harvey PH (1977) Species differences in feeding and ranging behavior in primates. In TH
Clutton-Brock (ed.): Primate Ecology. London: Academic Press, pp. 557-579.
Dagosto M (1992) Effect of habitat structure on positional behavior and substrate use in Malagasy lemurs. Am. J.
Phys. Anthropol. 14:167.
Dagosto M, and Gebo DL (199?) Methodological issues in studying positional behavior: Meeting Ripley's chal-
lenge. In E Strasser, JG Fleagle, AL Rosenberger, and HM McHenry (eds.): Primate Locomotion: Recent
Advances. New York: Plenum Press, pp. 5-29.
Doran DM (I 993a) Sex differences in adult chimpanzee positional behavior: The influence of body size on loco-
motion and posture. Am. J. Phys. Anthropol. 91:99-115.
Doran DM (1993b) Comparative locomotor behavior of chimpanzees and bonobos: The influence of morphology
on locomotion. Am. J. Phys. Anthropol. 91:83-98.
Doran DM (1996) The comparative positional behavior of the African apes. In WC McGrew and T Nishida (eds.):
Great Ape Societies. Cambridge: Cambridge University Press, pp.213-224.
Doran DM, and Hunt KD (\994) Comparative locomotor behavior of chimpanzees and bonobos: Species and
habitat differences. In RW Wrangham, WC McGrew, FB de Waal, and PG Heltne (eds.): Chimpanzee Cul-
tures. Cambridge: Harvard University Press, pp. 93-106.
Du Chaillu P (1861) Exploration and Adventures in Equatorial Africa. London: John Murray.
Emlen JT, and Schaller GB (1960) Distribution and status of the mountain gorilla (Gorilla gorilla beringei) 1959.
Zool. 45:41-52.
Fay JM (1989) Partial completion of a census of the western lowland gorilla (Gorilla g. gorilla (Savage and Wy-
man» in southwestern Central African Republic. Mammalia 53(2):203-215.
Fay JM, Agnagna M, Moore J, and Oko R (1989) Gorillas (Gorilla gorilla gorilla) in the Likouala Swamp Forests
of north central Congo: Preliminary data on populations and ecology. Int. J. Primatol. 10(5):477-486.
Fleagle JG (1988) Primate Adaptation and Evolution. San Diego: Academic Press.
Fleagle JG, Stern JT, Jungers WL, and Susman RL (\ 981) Climbing: A biomechanical link with brachiation and
with bipedalism. Symp. Zool. Soc. Lond. 48:359-375.
Fossey D (1983) Gorillas in the Mist. Boston: Houghton Mifflin Co ..
Garber PA (1984) Use of habitat and positional behavior in a neotropical primate, Saguinus oedipus. In PS Rod-
man and JGH Cant (eds.): Adaptations for Foraging in Nonhuman Primates: Contributions to an Organis-
mal Biology of Pro simians, Monkeys and Apes. New York: Columbia Univ. Press, pp. 112-133.
Garber PA, and Pruetz JD (1995) Positional behavior in moustached tamarin monkeys: Effects of habitat on loco-
motor variability and locomotor stability. J. Hum. Evol. 28:411-426.
Gebo DL, and Chapman CA (1995a) Positional behavior in five sympatric Old World Monkeys. Am. J. Phys. An-
thropol. 97(1)49-76.
Gebo DL, and Chapman CA (1995b) Habitat, annual, and seasonal effects on positional behavior in red colobus
monkeys. Am. J. Phys. Anthropol. 96:73-82.
Goodall AG (1977) Feeding and ranging in Kahuzi gorillas. In TH Clutton-Brock (ed.): Primate Ecology: Studies
of Foraging and Ranging in Lemurs, Monkeys and Apes. London: Academic Press. pp. 450-479.
Grand TI (1972) A mechanical interpretation of terminal branch feeding. J. Mammal. 53(1): 19&--20 I.
Gregory WK, and Raven HC (1937) In Quest of Gorillas. New Bedford: Darwin Press.
Hunt KD (1991) Positional behavior in the Hominoidea. Int. J. Primatol. /2(2):95-118.
Hunt KD (1992) Social rank and body size as determinants of positional behavior in Pan troglodytes. Primates
33(3):347-357.
Hunt KD (\994) Body size effects on vertical climbing among chimpanzees. Int. J. Primatol. 15(6):855-865.
Jones C, and Sabater-Pi PJ (1971) Comparative ecology of Gorilla gorilla (Savage & Wyman) and Pan troglo-
dytes (Blumenbach) in Rio Muni, West Africa. Bibl. Primatol. 13: 1-96.
Jungers WL, and Susman RL (1984) Body size and skeletal allometry in African apes. In Susman RL (ed.): The
Pygmy Chimpanzee. Evolutionary Biology and Behavior. New York: Plenum Press, pp. 131-178.
Kawai M, and Mizuhara H (1959) An ecological study of the wild mountain gorilla (Gorilla gorilla beringei): A
report of the Japan Monkey Center second gorilla expedition, 1959. Primates 2: 1-42.
Keith A (1923) Man's posture: Its evolution and disorder. Brit. Med. J. 1.451--{)72.
Kinzey WG (1976) Positional behavior and ecology in Callicebus torquatus. Yrbk. Phys. Anthropol. 20:468-481.
Lewis OJ (1969) The hominoid wrist joint. Am. J. Phys. Anthropol. 30: 251-268.
McGraw WS (1996) Cercopithecid locomotion, support use, and support availability in the Tai Forest, Ivory
Coast. Am. J. Phys. Anthropol. 100(4):507-522.
The Effects of Body Size and Habitat on the Positional Behavior of Lowland and Mountain Gorillas 105

Malenky RK, Kuroda S, Vineberg EO, and Wrangham RW (1994) The significance of terrestrial herbaceous foods
for bonobos, chimpanzees, and gorillas. In RW Wrangham, WC McGrew, FB de Waal, and PG Heltne
(eds.): Chimpanzee Cultures. Cambridge: Harvard University Press, pp. 59-75.
Morbeck ME (1976) Leaping, bounding and bipedalism in Colobus guereza: A spectrum of positional behavior.
Yrbk. Phys. Anthropol. 20:408-420.
Morbeck ME (1979) Forelimb use and positional adaptation in Colobus guereza: Integration of behavioral, eco-
logical and anatomical data. In ME Morbeck, H Preuschoft, and N Gomberg (eds.): Environment, Behavior
and Morphology: Dynamic Interpretations. New York: Gustav-Fischer, pp. 95-118.
Morton DJ (1922) The evolution of the human foot. Am. 1. Phys. Anthropol. 5:305-336.
Nishihara T (1995) Feeding ecology of western lowland gorillas in the Nouabale-Ndoki National Park, Congo.
Primates 36(2): 151-168.
Owen R (1859) On the Gorilla (Troglodytes gorilla, Sav.). Proc. Zool. Soc. Lond. 381: 1-23.
Prost J (1965) A definitional system for the classification of primate locomotion. Am. Anthropol.
67(5): 1198-1214.
Purchas S (1625) The Strange Adventures of Andrew Battell. In Purchas, His Pilgrims. Contayning a History of
the World, in Sea Voyages and Land Travells (4th ed.). London: William Stansby for Henry Fetherstand.
Remis MJ (1994) Feeding Ecology and Positional Behavior of Western Lowland Gorillas (Gorilla gorilla gorilla)
in the Central African RepUblic. Ph.D. Dissertation. Yale University.
Remis MJ (1995) The effects of body size and social context on the arboreal activities of western lowland gorillas
in the Central African Republic. Am. J. Phys. Anthropol. 97:413-433.
Remis MJ (in press) Western lowland gorillas (Gorilla gorilla gorilla) as seasonal frugivores: Use of variable re-
sources. Am. J. Primatol. 43:N-N.
Richards PW (1957) Tropical Rain Forest. Cambridge: Cambridge University Press.
Ripley S (1967) The leaping of langurs: A problem in the study of locomotor adaptation. Am. 1. Phys. Anthropol.
26: 149-170.
Ripley S (1979) Environmental grain, niche diversification and positional behavior in Neogene primates. An evo-
lutionary hypothesis. In ME Morbeck, H Preuschoft, and N Gomberg (eds.): Environment, Behavior and
Morphology: Dynamic Interpretations. New York: Gustav-Fischer, pp. 37-74.
Rogers ME, and Williamson EA (1987) Density of herbaceous plants eaten by gorillas in Gabon: Some prelimi-
nary data. Biotropica 19(3):278-281.
Rose MD (1977) Positional behavior of olive baboons (Papio anubis) and its relationship to maintenance and so-
cial activities. Primates 18(1):59-116.
Rose MD (1979) Positional behavior of natural populations and some quantitative results of a field study of
Colobus guereza and Cercopithecusaethiops. In ME Morbeck, H Preuschoft, and N Gomberg (eds.): Envi-
ronment, Behavior and Morphology: Dynamic Interpretations. New York: Gustav-Fischer, pp. 76-93.
Sanford LJ (1862) The gorilla: Being a sketch of its history, anatomy, general appearance and habits. Am. J. Sci.
and Arts. xxiii: 1-17.
Savage TS, and Wyman J (1847) Notes of the external characters and habits of Troglodytes gorilla. a new species
of orang from the Gaboon River. Boston 1. N. H. V(IV):417-443.
Schaller GB (1963) The Mountain Gorilla: Ecology and Behavior. Chicago: University of Chicago Press.
Schultz AH (1934) Some distinguishing characters of the mountain gorilla. 1. Mammal. 15:51-61.
Schultz AH (I 963 ) Relations between the lengths of the main parts of the foot skeleton in primates. Folia Prima-
tol.l:150--171.
Shea BT (1983) Size and diet in the evolution of African ape craniodental form. Folia Primatol. 40:32-68.
Sigmon BA (1974) A functional analysis ofpongid hip and thigh musculature. 1. Hum. Evol. 3:161-185.
Sokal RR, and Rohlf S1 (1981) Biometry. New York: W.H. Freeman and Company.
Straus M (1930) On the foot musculature of the highland gorilla (Gorilla beringei). Quart. Rev. BioI.
5(3):261-317.
Tutin CEG, and Fernandez M (1984) Nationwide census of gorilla (Gorilla g. gorilla) and chimpanzee (Pan t.
troglodytes) populations in Gabon. Am. 1. Primatol. 6:313-336.
Tuttle RH (1969) Quantitative and functional studies on the hands of the Anthropoidea. J. Morph.)28(3}:309-364.
Tuttle RH, and Watts DP (1985) The positional behavior and adaptive complexes of Pan (Gorilla). In S Kondo
(ed.): Primate Morphophysiology, Locomotor Analyses and Human Bipedalism. Tokyo: Univ. of Tokyo
Press, pp. 261-288.
Uchida A (1992) Intra-species variation among the great apes: Implications for taxonomy of fossil hominoids.
Ph.D. Dissertation, Harvard University.
Washburn S (1973) Primate field studies. In GH Bourne (ed.): Nonhuman Primates and Medical Research. New
York: Academic Press, pp. 467-485.
106 M.J. Remis

Watts DP (1984) Composition and variation of mountain goriIia diets in the Central Virungas. Am. J. Primatol.
7:323--356.
Watts DP (1987) Effect of mountain gorilla foraging activities on the productivity of their food plant species. Afr.
J. Ecol. 25:155-163.
Watts DP (1990) Ecology of gorillas and its relation to female transfer in mountain gorillas. Int. J. Primatol.
11 (1):21-46.
Willoughby D (1978) All About Gorillas. New York: A.S. Barnes and Co.
Williamson EA, Tutin CEG, Rogers ME, and Fernandez M (1990) Composition of the diet of lowland gorillas at
Lope Gabon. Am. 1. Primato!. 2/:265-277.
Zihlman AL (1992) Locomotion as a life history character: The contribution of anatomy. J. Hum. Evo!.
22:315-325.
 II

MORPHOLOGY AND BEHAVIOR

INTRODUCTION TO PART II

John G. Fleagle

Throughout its development the study of primate locomotion has been characterized
by its interdisciplinary focus combining ecological and behavioral observations with
physiological and morphometric studies to understand both how and why primates use dif-
ferent patterns of locomotion and posture. In addition, studies of primate locomotion are
increasingly involving experimental approaches that allow researchers to examine behav-
ior, form and function in a controlled situation and go beyond descriptive and correlative
analyses. The papers in this section demonstrate both the interdisciplinary and experimen-
tal approaches to the study of primate locomotion. Moreover, by clarifying the link be-
tween bony morphology and physiological function they provide ever widening windows
into the behavior of extinct primates.
Laura MacLatchy (Chapter 7) uses 3-D imaging of hindlimb bones and in vivo ma-
nipulations of living animals to examine hip joint function in extant and fossil prosimians.
She finds that patterns of joint surface distribution on the acetabulum and head of the femur
accord well with measurements of joint mobility in living animals. Thus imaging of joint
surfaces in fossils can be used to reconstruct patterns oflimb excursion during locomotion.
Pierre Lemelin and Brian Grafton (Chapter 8) conducted experimental studies of
hand use in golden-handed tamarins and squirrel monkeys in conjunction with kinematic
analysis and morphometric studies to examine the functional consequences of nails and
claws on primate digits. They found relatively few differences in the abilities of clawed
tamarins and nail-bearing squirrel monkeys to grasp most food items and concluded that
claws do not necessarily hinder grasping function. Likewise they found little evidence for
a clear link between grasping ability and digital proportions.
In Chapter 9, Jeffrey Meldrum addresses an important, but rarely examined problem
in functional morphology-the intermediate steps in the evolution of complex morpho-
logical features. He uses both morphological and behavioral observations to identify a se-
ries of behaviors among extant primates that seem to foreshadow the evolution of fully
prehensile tails in platyrrhines.
Susan Larson (Chapter 10) uses kinematic analysis of limb excursions and elec-
tromyography of shoulder muscles to examine unique features of primate quadrupedal lo-
comotion. She finds that compared with other vertebrates, which have been reported to be

109
110 J. G. Fleagle

very conservative and uniform in forelimb muscle activity, primates are unique in using
more forelimb protraction and in the recruitment of shoulder muscles during quadrupedal
gaits.
Daniel Schmitt (Chapter 11) uses kinematic and force analysis to compare primate
quadrupeds on terrestrial and arboreal supports. He finds that the primates use more flexed
limbs on arboreal supports, but surprisingly this does not generate higher joint reaction
forces as predicted and hypothesizes that primates are using compliant gaits during arbo-
real locomotion.
 7

RECONSTRUCTION OF HIP JOINT FUNCTION

IN EXTANT AND FOSSIL PRIMATES

Laura MacLatchy

Department of Anthropology
Boston University
Boston, Massachusetts 02215

1. INTRODUCTION

The strepsirhines engage in a variety of locomotor behaviors, including vertical

clinging and leaping, active arboreal quadrupedal ism and deliberate climbing and suspen-
sion. In each of these locomotor modes, hip abduction plays a role in enabling an animal
to move in the complex, three-dimensional arboreal realm (Figure 1). For instance, hip ab-
duction characterizes clinging postures prior to leap take-off in galagos, indriids and
Lepilemur (Anemone, 1990; Demes et aI., 1996; Demes, pers. com.). Abducted hip pos-
tures also occur among above-branch quadrupeds such as cheirogaleids and many le-
murids who primarily walk, run and leap, but who also climb and engage in hindlimb
suspension (Oxnard et aI., 1990), both of which require moderate hip abduction. In turn,
lorisids, such as the potto, frequently use highly abducted hip postures and even on the
ground progress slowly with limbs abducted (Walker, 1969, 1979; Oxnard et aI., 1990).
Since hip abduction is such an important and variable component of hindlimb locomotion
in extant small-bodied primates, the ability to reconstruct this behavior in small-bodied
fossil primates should prove useful in determining their overall locomotor profile.
Investigations of the mammalian hip joint have emphasized the importance of relat-
ing joint morphology to ranges of movement, both to understand function and to recon-
struct locomotion in extinct forms. This line of research has been largely limited to
making qualitative comparisons of joint shape, and indirectly inferring joint mobility (e.g.,
Elftman, 1929; Walker, 1974; McHenry, 1975; Fleagle, 1976; Jenkins and Camazine,
1977; Harrison, 1982; Dagosto, 1983; Stern and Susman, 1983; Fleagle and Meldrum,
1988; Gebo, 1989; Anemone, 1990; Ward, 1991). Several excellent studies have investi-
gated pelvic and femoral morphology in small bodied primates (e.g., McArdle, 1981;
Fleagle and Anapol, 1992; Dagosto and Schmid, 1996), but none have focused on mobility
perse.

111
112 L. MacLatchy

Figure 1. Examples of strepsirhine primates using hip ab-

duction. (a) during preparation for leaping in Galago (after a
figure in Charles-Dominique, 1977); (b) during suspension
in Perodicticus (after a photograph in Napier and Napier,
1967); (c) during climbing downwards in Microcebus (after
a drawing by Nash in Mittermeier et aI., 1994); (d) during
bridging in Loris (after a drawing in Schulze and Meier,
1995 and a photograph in Charles-Dominique, 1977).

In the case of anthropoids, however, two recent studies (MacLatchy, 1996;

MacLatchy and Bossert, 1996) have focused on mobility (as well as loading patterns) at
the hip joint and generated new analytical techniques that can be used to study hip func-
tion in strepsirhines. In these studies, three-dimensional modeling and quantification of
acetabular joint surface distribution were used to infer joint loading. More orthograde spe-
cies, like Homo and Pongo, were found to have expanded cranial lunate surfaces com-
pared to more pronograde species, such as Pan and Old World monkeys, reflecting the
cranial direction of hip joint reaction forces in orthograde species. Analogous patterns may
be found among strepsirhines. Both Galago and Lepilemur frequently engage in behaviors
in which the trunk is oriented vertically, such as vertical clinging and leaping, vertical
climbing and bipedal hopping on the ground (Walker, 1974, 1979; Oxnard et al., 1990),
resulting in hip joint forces that would be pointed into the cranial part of the acetabulum
(e.g., as has been demonstrated in humans (Dalstra and Huiskes, 1995)), causing a rela-
tively expanded cranial lunate surface. Walker (1979:553) reports that mouse lemurs
" ... run about quadrupedally in a small branch milieu ... when running the body is horizon-
tal and aligned with the support." The generally horizontal body orientation of Microcebus
may generate stresses on the dorsal aspect of the acetabulum, causing it to expand. The
potto, with a varied locomotor repertoire that includes pronograde quadrupedal ism, be-
low-branch suspension, cantilevering, spiraling, bridging and climbing (Walker, 1969 and
1979; Charles-Dominique, 1977; Table I), might be expected to experience more uniform
Reconstruction of Hip Joint Function in Extant and Fossil Primates 113

Table 1. Extant and fossil primate specimens used in analyses. AMNH, American Museum of
Natural History; DUPC, Duke University Primate Center; MCZ, Museum of Comparative Zoology;
PM, Peabody Museum; UMMP, University of Michigan Museum of Paleontology
Positional behavior
N of in vivo (reconstructed for fossil species)
Species Skeletal specimens specimens from DUPC and/or morphological affinities
Microcebus murinlls AMNH 174385, 174415, arboreal quadruped, frequent
174424, 174430, 174431, leaping and hopping and
174499, 185627; MCZ occasional use of cantilever
44843 posture (Walker, 1974)
Cheirogaleus major AMNH 100640; MCZ none available so 2 arboreal quadruped; runs and
(animation only) 44946,44952 Cheirogaleus scurries; leaps rarely (Walker,
medius used 1974)
Galago senegalensis AMNH 86502, 86503, 2 arboreal: leaps frequently;
86504, 187356, 187357, quadrupedal running and
185358; PM 8535, 8537, climbing (Walker, 1979;
10595 Oxnard et aI., 1990)
Perodicticlls potto AMNH 52698, 52702, 2 arboreal: slow quadrupedal
52717,86898;MCZ climbing and below branch
25831 suspension; limbs habitually
abducted (Walker, 1969, 1979;
Charles-Dominique, 1977)
Lepilemur mustelinus AMNH 170553, 170555, none available so 2 arboreal: leaps frequently using
170560, 170561, 170562; Lemur fulvus predominantly vertical
MCZ 25407, 44925, collaris used supports; walks slowly or hops
44926,44931 bipedally on ground (Oxnard et
aI., 1990)
Notharctus tenebrosus AMNH 11474, 129382 active arboreal quadruped;
leaping ability; pelvic
morphology similar to
Lepilemur (Gregory, 1920;
Fleagle and Anapol, 1992)
Hemiacodon gracilis AMNH 12613 femur similar to Galago and
Tarsius; pelvis similar to
Galago and Lemur (Simpson,
1940; Covert, 1995); femur
resembles Microcebus
(Cartmill, 1972)
Omomys carteri UMMP98604 elongated talus and calcaneus
suggest some leaping; hip
morphology resembles Galago
(Dagosto, 1993, 1996;
Anemone, 1995)

acetabular loading than the other strepsirhines and so have a more uniformly distributed
lunate surface.
The studies on anthropoids (ibid) also related acetabular shape to hip mobility. The
hip socket was more shallow on its dorsal aspect in apes than in monkeys, permitting a
wider range of abduction in the former. It is therefore expected that among strepsirhines,
those species with the greatest mobility (e.g., Perodicticus) will have shallower hip sockets.
Mobility was further assessed by MacLatchy (1996) using three-dimensional com-
puter animation of hip joint elements to estimate species' maximum range of abduction, an
important ability among climbing primates. Apes were found to have greater ranges of ab-
 114 L. MacLatchy

duction than monkeys, who in tum had more mobile hips than humans. The technique was
used to infer mobility in the hominid Australopithecus afarensis, which was found to have
the same level of mobility as modem humans. A similar mobility gradient should exist
among strepsirhines who have varied locomotor patterns (Table 1). The slow climbing and
suspensory potto, for example, is expected to have greater ranges of hip abduction than
typical above-branch quadrupeds.
Once a comparative context has been established among extant strepsirhines, the hip
function of fossil primates can be considered. For the fossil primates Notharctus tenebro-
sus, Omomys carteri and Hemiacodon gracilis, selection of a phylogenetically limited ex-
tant comparative sample is difficult because the relationships of the notharctines and
omomyines to living species are not entirely clear (Beard et al., 1988; Dagosto, 1988;
Fleagle, 1988; Martin, 1993; Kay et ai., 1997). Nevertheless, extant strepsirhines are a rea-
sonable choice because their behavior spans several locomotor modes, because they have
been described as good analogs for the fossil taxa under scrutiny, and because their hip
joint elements resemble those of the fossil species (Table 1).
Each of the fossil taxa possesses characteristics of "hindlimb-dominated arboreal lo-
comotion" as outlined by Martin (1990), including overall adaptations for arboreal leaping.
Notharctus is a notharctine adapid from the middle Eocene of North America. It was a diur-
nal folivore (Fleagle, 1988) and its positional repertoire has been variously characterized as
both an active arboreal quadruped and a vertical clinger and leaper (Gregory, 1920; Napier
and Walker, 1967; Gebo, 1985; Szalay and Dagosto, 1980; Dagosto, 1993). Omomys and
Hemiacodon are two genera of omomyine omomyids, also from the middle Eocene of North
America. Both were probably nocturnal and ate insects and fruit (Covert, 1995). As with
Notharctus, suggestions that these species typically used vertical postures (e.g., Napier and
Walker, 1967) have been challenged, although leaping is thought to comprise a significant
portion of their locomotor repertoire (Dagosto, 1993; Covert, 1995). Locomotor reconstruc-
tions of the three taxa have focused on whether they are best characterized as active arbo-
real quadrupeds or as vertical clingers and leapers, in part because of the role these
categories played in early attempts to describe the locomotor mode of the ancestral primate
(Napier and Walker, 1967; Cartmill, 1972). In recent years, the former characterization has
become most prevalent (Gebo, 1985; Fleagle, 1988; Dagosto, 1993; Covert, 1995).
In the present study, the emphasis will be on reconstructing the abductory capabili-
ties of the hip and how this helps to characterize the overall adaptations of the hindlimb.
This will be accomplished using three-dimensional modeling techniques that enable an
analysis of acetabular joint surface shape, which is used to infer excursion and loading
patterns at the hip, and direct estimates of hip joint posture and maximum abduction using
computer animation. In addition, hip abduction will be assessed in vivo in order to evalu-
ate the accuracy of the computer animation.

2. MATERIALS AND METHODS

2.1. Specimens
The extant sample is a phylogenetically diverse one (2 cheirogaleids, 2 lorisoids and
1 lemuroid; Table 1). While this does not affect comparisons of direct measurements of
mobility, inferences concerning acetabular morphology are unconstrained by phylogeny.
As discussed above, this is because the relationships of the notharctines and omomyines to
living species are unclear. In addition, use of Magnetic Resonance Imaging (MRI) im-
Reconstruction of Hip Joint Function in Extant and Fossil Primates 115

posed time constraints and reduced the number of museum collections that could be used,
and so limited the number of taxa and specimens that could be examined.
Acetabular morphology was analyzed in four extant taxa (Mieroeebus, Galago,
Perodietieus and Lepilemur); these four and one additional taxon (Cheirogaleus major)
were analyzed by computer animation (Table 1), while five taxa were studied in vivo.
Only three Cheirogaleus major specimens were imaged, but they are included in order to
compare estimated mobility to in vivo data collected for Cheirogaleus medius. No Lepile-
mur are present in captivity, so in vivo estimates could not be obtained for this genus, and
Lemur was studied instead. While the computer/in vivo species matches are not exact, they
do allow comparisons to be made of the two methods of quantifying hip abduction.
The sacrum is necessary to orient the pelvis in three dimensional space, but no asso-
ciated sacra were available for the fossil specimens. For Notharetus, unassociated sacra of
two other Notharetus specimens (AMNH 11475, 11752) were used; for the omomyids, a
Galago senegalensis sacrum was used because of size and gross morphological similari-
ties between Galago and omomyid pelves.

2.2. Imaging and Image Analysis

Specimens were imaged on an 8.45 Tesla MRI system (Bruker Instruments, Billerica,
MA) as described in MacLatchy and Bossert (1996). In the x and y directions, the image
resolution varied between 0.08 mm (for the small Microeebus specimens) and 0.18 mm (for
the larger Perodietieus specimens) and was twice as large (i.e., between 0.17 mm and 0.37
mm) in the z direction. For the femur, contiguous images were taken from the top of the
greater trochanter (or the top of the femoral head, whichever was higher) to a level one-third
down the length of the femoral shaft. For the pelvis, contiguous images over the entire
acetabular and adjacent areas (including the ischium and pubis) were taken.
Three-dimensional images were reconstructed from digitized MRI images
(MacLatchy and Bossert, 1996). Three-dimensional coordinates were interactively se-
lected and coded according to anatomical region to facilitate modeling. The subchondral
bone of the femoral head, the femoral shaft, the fovea capitis, the lunate surface, the is-
chium and the acetabular fossa were all differently coded. The long axes of the ischium
and femoral shaft were determined by finding the line that bisected the center of mass of
two cross-sections through the ischium and two cross-sections through the femoral shaft
(MacLatchy and Bossert, 1996).
A mid-sagittal plane was created for each three-dimensional model of a pelvis.
Three three-dimensional coordinates are required to define a plane. As described in
MacLatchy (1996), two x, y, z coordinates were located on the pubic symphysis, which
coincided with the midplane of the animal. The location of the third coordinate was deter-
mined by first measuring the perpendicular distance from the tip of the ischial spine to the
mid-sagittal plane (as determined by the midplane of the sacrum) while the pelves and sa-
crum were articulated. In the three-dimensional reconstruction, the coordinate on the is-
chial spine was then projected by the measured distance (in the x-plane) to supply the
third x, y, z coordinate. The plane was used to determine femoral posture during computer
simulations of joint movement.

2.3. Modeling of Lunate Surface

Lunate surface distribution relates to both joint force patterns and mobility. Expan-
sion of articular surfaces is thought to occur in areas normal to the greatest or most fre-
116 L. MacLatchy

quent joint forces (Latimer et ai., 1987; Ward, 1991), hence the distribution of acetabular
articular surface may reflect the loading environment of the joint. In the case of mobility,
greater coverage of the femoral head by a deep acetabulum restricts hip movement, while
less coverage by a shallow acetabulum permits a wider range of excursion (Elftman, 1929;
Jenkins and Camazine, 1977; Ruff, 1988; Ward, 1991).
A detailed description of the modeling procedure is in MacLatchy and Bossert
(1996). The basis for the analysis was to fit a linear least squares best-fit sphere to com-
puter reconstructions of the acetabulum. Only the coordinates defining the lunate surface
were fit to a sphere; the acetabular fossa was not included since it is not part of the weight
bearing surface of the joint.
In order to quantify the distribution pattern of the subchondral bone, two axes were
set up within the best-fit sphere (Figure 2). Axis 1 was defined as the line between the
center of the best-fit sphere and the center of the acetabular fossa. Axis 2 was on a plane
perpendicular to Axis 1 and went through the center of the best-fit sphere; the location of
Axis 2 on this plane was determined by projecting a line parallel to the long axis of the is-
chium onto the plane. Hence, these reference axes were consistent from specimen to

Figure 2. Modeling of lunate surface. (a) Lateral

view of a right acetabulum, showing Axes I and 2
and the cranial data points. Caudal to the left, dorsal
to the top. Angles from Axis 2 are measured in a
clockwise direction from the caudal end (0°) of Axis
2. (b) Ventral view of an acetabulum, showing the
best fit sphere and the same cranial data points as in
a. The location of the acetabular rim and junction be-
tween the acetabular fossa and the lunate surface are
quantified by measuring the angular displacement
from Axis I at specific angles from Axis 2 (i.e., at
caudal, dorsal and cranial locations). The data point
on the rim of the acetabulum (Point A) has an angular
displacement from Axis I of 100° and an angular dis-
Axis 2 placement from Axis 2 of 165°. The data point at a
junction of the articular surface and the acetabular
fossa (Point 8) has an angular displacement from
Axis I of 40° and an angular displacement from Axis
2 of 165°. The total angular extent of the articular
surface at this cranial location is determined by sub-
tracting the minimum angle from Axis I from the
maximum angle from Axis I (100° - 40° =60°).
Reconstruction of Hip Joint Function in Extant and Fossil Primates 117

specimen. Once the axes were defined, the locations of data points on the articular surface
could be localized by plotting the angle from Axis I (the angle subtended by the line from
a point on the articular surface to the center of the sphere and Axis I) against the angle
from Axis 2 (the angle subtended by the line from a point on the articular surface to the
center of the sphere and Axis 2) for data points on the articular surface. The angle from
Axis 1 gives information about how far up the sides of the best fit sphere (or the walls of
the acetabulum) a given point is located. The highest angles from Axis 1 occur at the
acetabular rim while the lowest occur at the lunate surface/acetabular fossa junction. The
angle from Axis 2 is indicative of where a given point is located relative to the circumfer-
ence of the acetabulum.
The difference between the highest and lowest angles from Axis I at cranial, dorsal
and caudal locations on Axis 2 were used to assess the relative size of the lunate surface
wall in order to relate it to regional loading patterns. The caudal (or ischial) end of Axis 2
was designated as 0° and angles were measured in a clockwise direction. Dorsal measure-
ments were taken at 75° from Axis 2, cranial measurements at 165°, and caudal measure-
ments at 345°. The maximum angle from Axis I of coordinates at these same cranial,
dorsal and caudal locations on Axis 2 were used to assess relative femoral coverage (or the
maximum distal extent, relative to the fossa, of the lunate surface). Regional variation in
coverage should have different functional implications, with cranial coverage limiting me-
dial rotation, dorsal coverage limiting abduction and caudal coverage limiting lateral rota-
tion (MacLatchy and Bossert, 1996).
All pelves were analyzed in this way, and the means and standard deviations of the
angles from Axis I at the three locations were calculated for each species.

2.4. Quantification of Femoral Posture

Jenkins and Camazine (1977) quantified the location ofthefovea capitis in carnivores
and found that ambulatory carnivores have foveae that are located more dorsoposteriorly
than do cursorial carnivores; they related this to a more abducted femoral posture. This
study uses foveal location to infer femoral posture by quantifying the orientation of the
femoral shaft relative to the midsagittal plane when the fovea is centered in the acetabular
fossa. The orientation of the femoral shaft was standardized by flexing it at an angle of 90°
to the long axis of the ischium (Figure 3). The 90° angle was chosen because it represents
the approximate degree of flexion relative to the ischium of the femur in a standing or mod-
erately flexed clinging posture as determined from X-rays and observations of articulated
skeletons (Peabody Museum and Museum of Comparative Zoology, Harvard University).
A linear least squares best-fit sphere was fit to computer reconstructions of the sub-
chondral bone of the femoral head (excluding the fovea capitis). The 3-dimensional model
of the femur was placed in its corresponding hip socket so that the centers of curvature of
the best-fit spheres for the head and acetabulum were coincidental. This procedure elimi-
nated the need to assign a value for cartilage thickness, but assumes that cartilage is
evenly distributed, as in studies that estimate range of movement by manually opposing
the subchondral surfaces of osteological specimens and moving them throughout their
range of congruence (e.g., Latimer et aI., 1987; Swartz, 1989).
The midpoint of all of the fovea capitis coordinates was placed opposite the mid-
point of all of the coordinates of the acetabular fossa. In this position, the angle that the
long axis of the femoral shaft made with the midsagittal plane (in the coronal plane) was
measured. This angle was deemed the "neutral posture" of the femur, or the degree of ab-
duction of the femur in a hypothetical stance phase.
 118 L. MacLatchy

Neutral femoral orientation

(flexed at an angle of 90 degrees
to the long axis of the ischium)

Figure 3. Lateral view of a right femur and innominate showing the three angles of flexion of the femoral shaft
during abduction simulations.

A sphere closely approximates the shape of the femoral head and the acetabulum for
all the extant species except the Galago femoral head, which has a posterior femoral ar-
ticular surface that resembles a cylinder (Walker, 1974; McArdle, 1981; Anemone, 1990;
MacLatchy, 1993). A larger portion of the femoral head extended beyond the surface of
the best-fit sphere in Galago than in any other species, so the possibility existed that a
portion of the head might contact the acetabulum during rotation of the femur about the
center of the superimposed best-fit spheres. The surface of the femoral head was thus care-
fully monitored in simulations.

2.5. Computer Simulation of Maximum Abduction

Maximum hip joint abduction is determined by simulating joint movement with
three-dimensional computer reconstructions of the femur and pelvis. The centers of curva-
ture of the best fit spheres of the femoral head and acetabulum were made coincidental.
Abduction simulations were conducted from a flexed position, in which the long axis of
the femoral shaft was flexed at an angle of 135° to the long axis of the ischium, and an ex-
tended position, in which the femoral axis was parallel to the long axis of the ischium
(Figure 3). The abduction simulations could have been performed at virtually any degree
of flexion or extension. The 135° flexed position was chosen to represent femoral orienta-
tion during climbing during forward progression, while the extended position was chosen
to represent a more acrobatic, postural position for the femur.
At least three factors may constrain abduction as performed in the computer simula-
tions. First, the fovea and its attached ligament normally remain within the confines of the
acetabular fossa in order to prevent the ligamentum teres from becoming wedged between the
femoral head and the lunate surface (Jenkins and Camazine, 1977). Second, bony constraints
such as the greater trochanter or femoral neck may contact the acetabular rim or wall and halt
movement, and third, movement may be limited by the femoral subchondral bone margin.
The cineradiographic study by Jenkins and Camazine demonstrated that during normal gait in
Reconstruction of Hip Joint Function in Extant and Fossil Primates 119

carnivores, the edge of the acetabular articular surface overlaps slightly with non-articular ar-
eas of the femoral head in some species. Hence, in this analysis, acetabular surface was al-
lowed to overlap with non-articular parts of the femur until one of the other constraints was
met (see below) and movement had to cease. Overlap, when it occurred, was very small.
The simulations proceeded as follows: a femur was oriented at the appropriate de-
gree of flexion relative to the ischium (0° or 135°) and then abducted and rotated (while
maintaining the degree of flexion relative to the ischium) until the greater trochanter ap-
proached the pelvis, or the fovea capitis threatened to overlap with the lunate surface, or
both. When abduction was limited by the greater trochanter, the femur was abducted until
the greater trochanter just contacted the pelvis, and then adducted 5° in the coronal plane
to provide a standardized way to account for soft tissue between the trochanter and pelvis.

2.6. In Vivo Quantification of Maximum Abduction

In vivo assessments of joint excursion are rare (e.g., Turnquist, 1983, 1985; Demes
et aI., 1996) and these data are the first on maximum abduction of the hip joint in any pri-
mate species.
Five species of strepsirhines (Table I) were studied at the Duke University Primate
Center. Two individuals of each species were anesthetized with iso-fluorane, with the ex-
ception of Microcebus where only one individual could be used. While anesthetized, the
legs of the animals were passively manipulated and the maximum angles of abduction in
highly flexed and completely extended postures were measured. These postures approxi-
mately mimicked the angles offemoral flexion (0° and 135° to the ischium) in the computer
simulations. X-rays of the various postures were taken in order to determine joint configura-
tion, but the X-rays were in the dorsal-ventral plane and so the angle of the femoral shaft
relative to the ischium in the parasagittal plane could not be precisely replicated.
Turnquist (1983) found that the range of flexion-extension of the joints of caged
patas monkeys was greater than that of free-ranging patas monkeys, and attributed it to
lack of muscle tone and concomitant lack of resistance to passive manipulation. All of the
animals used in this study were caged, but it is unlikely that this affects hip abduction,
since the primary constraints to abduction are thought to be osteological.

2.7. Statistical Comparisons

Interspecific statistical comparisons among means were made using the least signifi-
cant difference (LSD) test statistic for planned comparisons (Sokal and Rohlf, 1981), since
species were chosen because of their locomotor diversity and it was assumed a priori that
the functional parameters would differ. Results were considered significant if the P-values
were <0.05. The LSD tests were performed using the PowerPC version of the Statview
Statistical package.

3. RESULTS

3.1. Lunate Surface Morphology

The total angular extent of the lunate surface was determined in order to assess sus-
pected regional differences in hip joint loading. The cranial lunate surface wall is more ex-
tensive than the dorsal and caudal walls in all species (Figure 4, top). Interspecific
 120 L. MacLatchy

0
Cranial
Nl>H N
o I I
.6.
~
Dorsal
ONH
10 .6. I

~
Caudal I
20 30 40 50 60 70 80 90 100

Angle determined by subtracting the minimum

angular displacement from Axis I from the
maximum angular displacement from Axis I at
three locations on the lunate surface (cranial.
dorsal and caudal) (measured in degrees)

Angular extent: 85 - 35 =50 degrees N Notharctus

H Hemiacodon
o Omomys
h. Microcebus
o Galago
Cranial <> Perodicticus
o Lepilemur

Dorsal

Caudal I I I
50 60 70 80 90 100 110 120 130

Angle subtended by Axis I and a line

from the acetabular rim to the center of
the best-fit sphere (measured in degrees)

85

Figure 4. (Top) Interspecific mean comparisons of regional extent of the lunate surface as determined by subtract-
ing the minimum angular displacement from Axis I (measured in degrees) from the maximum angular displace-
ment from Axis I at three locations on the lunate surface (cranial, dorsal and caudal). Data points are means +/-
one standard deviation. A horizontal line over extant species means indicates that there is no difference in means
as determined by Fisher's LSD (P=O.05). (Bottom) Interspecific mean comparisons of regional femoral coverage
as determined by maximum angular displacement from Axis I at three locations on the lunate surface (cranial,
dorsal and caudal). Same conventions as in Top. Note that the dorsal rim of the Omomys acetabulum is slightly
eroded and therefore the total extent of the dorsal acetabular waH and the maximum extent of the dorsal acetabular
waH are probably underestimates.
Reconstruction of Hip Joint Function in Extant and Fossil Primates 121

differences show few distinct trends, with the notable exception of the dorsal acetabular
surface, which is more extensive in Microcebus than in the other extant species. The fossil
species group with Galago, Perodicticus and Lepilemur, with the exception of one
Notharctus specimen with a degree of dorsal expansion that is greater than in Galago,
Perodicticus and Lepilemur, though not as great as in the committed quadruped Microce-
bus.
The maximum extent of the lunate surface reflects the amount of coverage that the
acetabulum provides for the femoral head. In the four extant species, the cranial aspect of
the lunate surface extends significantly farther up the sides of the best-fit sphere than do
the dorsal and caudal surfaces, which are similar (Figure 4, bottom). Hence this parameter
and that assessing lunate surface expansion described above both indicate that across spe-
cies, there is a preponderance of joint surface on the cranial aspect of the lunate surface.
The quadrupedal Microcebus has more dorsal and cranial coverage than do Galago,
Perodicticus, and Lepilemur. Notharctus, Hemiacodon and Omomys have a pattern of cov-
erage like that of Galago, Perodicticus and Lepilemur with less cranial and dorsal cover-
age than Microcebus.

3.2. Ranges of Abduction

Careful monitoring of the Galago hip joint during computer simulations revealed
that the femoral head and acetabulum never contacted each other during simulations, de-
spite the fact that the best-fit sphere used to oppose the two joint components only loosely
approximated the Galago femoral head. Presumably, contact was avoided because the ar-
ticular cartilage is thick. A comparison of hip joint X-rays for Galago and Perodicticus in-
deed reveals a relatively wide interval between subchondral bone surfaces and therefore
thick cartilage in the Galago (Figure 5). The spherical femoral modeling is thus deemed
adequate for the purpose of the computer simulations.
Little variation was observed in neutral posture among the strepsirhines, with all
species tending to have moderately abducted neutral postures (Figure 6, top). Microcebus

Figure S. X-rays of hip joints of (A) Galago and (8) Perodicticus. Femora are flexed and abducted. Arrows are
pointing to the caudal aspect of the acetabulum. Note the relatively "looser" fit of the femoral head in the hip
socket of the galago compared to the potto.
 122 L. MacLatchy

Fossils
Microcebus
Galago
Perodicticus
Lepilemur
I I I J I I
-30 -20 -10 0 10 20 30 40 50 60 70
Angle of femoral shaft relative to mid-sagittal plane
in a neutral posture (degrees)

Fossils
Microcebus
Cheirogaleus
Galago
Perodicticus
Lepilemur 1--0---1
Lemur ** Figure 6. (Top) Angle of femoral shaft with
the midsagittal plane when the fovea is cen-
I I I I I I tered in the acetabular fossa and the femoral
o 10 20 30 40 50 60 70 80 90 100 shaft is flexed at an angle of 90° to the ischial
Maximum angle of abduction during flexion (degrees) ramus. This measure is described in the text
as the "neutral posture". Same conventions as

*0 ---- NN
Fossils H in Figure 4. (Middle) Maximum angle of ab-
Microcebus ~
duction during flexion. Bars are means +/-
Cheirogaleus * I-D-l * one standard deviation for computer simula-
tions, and stars represent in vivo trials. (Bot-
Galago ~ * tom) Maximum angle of abduction during
Perodicticus
extension. Bars are means +/- standard devia-
Lepilemur 1--0--1 tions for computer simulations, and stars rep-
Lemur
* * resent in vivo trials. No standard deviation
bars are shown for Perodicticus because the
" I I I
computer program evaluates angles up to 90°,
o 10 20 30 40 50 60 70 80 90 100
and all angles were greater than or equal to
Maximum angle of abduction during extension (degrees) 90°.

has a more abducted neutral posture than Galago but otherwise no significant differences
were observed (Table 2). In a neutral posture, two Notharetus specimens have a mean an-
gle of abduction of 24°, similar to the other strepsirhines. Hemiaeodon has a similar de-
gree of abduction in a neutral posture (23°), but Omomys has a less abducted neutral
posture (9°) than the other fossil species and all the extant species except Galago.
Results for mean abduction during flexion and extension demonstrate very good
concordance between the computer simulations and the in vivo data (Figure 6). There were
significant interspecific differences in the angles of maximum abduction during flexion,
with the slow climber Perodietieus having a greater range of abduction than all other spe-
cies (Figure 6, middle). Among the remaining extant species, only Cheirogaleus and Mi-
eroeebus differ significantly (Table 2). The two Lemur fulvus specimens, like the three
Eocene primates, have maximum angles of abduction that are similar to all but Mieroee-
bus at one extreme and Perodietieus at the other.
During abduction in the flexed posture, it was sometimes possible to rotate the head
of the femur so that the fovea was no longer opposite the fossa, but was inferolateral to the
acetabular notch; in vivo this would result in the ligamentum teres being dragged into the
 :1:1
"'"
=
~
.,=
=
:?
0'
=
....
=
::c
oS'
...
=
a
'!l
=
=
:?
Table 2. Comparisons of properties between species. All units are in degrees 0'
Microcebus Cheiroga/eus Galago Perodicticus Lepilemur
=
5'
t"l
Variable Mean SD Mean SD Mean SD Mean SD Mean SD Interspecific comparisons' ..;-
Cranial extent of lunate surface 84.4 13.5 73.8 18.4 72.3 7.2 60.9 16.3 MIL .,::.
Dorsal extent of lunate surface 57.3 14.5 34.2 16.6 41.0 9.5 30.9 8.1 MIG, M/P, MIL C.
=
Caudal extent of lunate surface 45.7 16.4 33.7 16.2 40.7 7.7 38.6 6.0 '!l
~
Cranial femoral coverage 119.9 13.2 103.6 13.9 100.7 6.5 91.5 14.5 MIG, M/P, MIL ~
Dorsal femoral coverage 89.9 8.3 67.2 16.3 74.6 7.5 68.4 10.4 MIG, MIP, MIL ~
Caudal femoral coverage 74.7 15.0 65.9 9.7 71.9 8.9 65.5 8.4 S·
Neutral femoral posture 32.1 12.8 14.2 12.5 27.8 15.7 25.2 10.9 MIG
.,
~
Abduction during flexion 38.8 4.0 50.1 10.5 45.9 10.9 73.0 10.3 45.6 8.2 MIC, M/P, CIP, G/P, P/L
Abduction during extension 40.9 13.7 58.2 2.2 52.0 5.7 90.0 0.0 37.7 9.0 MIC, M/P, CIP, CIL, GIP, GIL, P/L
'Only those species with significantly different means (P<O.05) as determined by Fisher's LSD test are listed; all other pairings are not significantly different. Species are represented by the tirst letter of
the genus name.

-........
124 L. MacLatchy

notch, but not over the articular surface. This orientation was regularly achieved in the
potto specimens, which had wide, cranioventrally oriented acetabular notches. Other
strepsirhines possess a wide notch, but their femoral necks and/or greater trochanters
tended to inhibit abduction before the fovea could position itself inferolateral to the
acetabular notch.
Pottos have greater hip abduction during femoral extension than do the other strep-
sirhines (Figure 6, bottom). There is greater variability among the other extant taxa during
extension than in flexion, with Cheirogaleus and Lemur having greater mobility than mouse
lemurs and Lepilemur, while Galago is intermediate. Greater variability was also observed
among the fossil species: the 2 Notharctus specimens have mobilities that exceed all but
potto. Hemiacodon falls within a single standard deviation of the means of Microcebus,
Galago and Lepilemur, while Omomys is similar to Microcebus and Lepilemur.
Repeated measures of angle of abduction in the same individual yielded estimates
that differed by 3.20 (or 6.5 % difference) on average (ten trials), substantially better than
the 5-10 0 differences reported using manipulated osteological specimens (Swartz, 1989).

4. DISCUSSION

4.1. Lunate Surface Morphology

4.1.1. Extant Species. Analysis of acetabular morphology discriminates between Mi-

crocebus on the one hand, and Galago, Perodicticus and Lepilemur on the other. The rela-
tively expanded cranial lunate surfaces combined with unexpanded dorsal lunate surfaces
in Lepilemur and Galago seem reasonable given the frequent vertical behaviors, espe-
cially leaping, exhibited by these taxa (Table 1). Likewise, the more expanded dorsal ar-
ticular surface (and inferred higher dorsal loads) in the quadrupedal Microcebus compared
to Galago, Perodicticus and Lepilemur agrees with inferences made from locomotor data
(Table 1). Microcebus also practices leaping, but lands on all fours after leaping, unlike
Lepilemur and Galago senegalensis, which frequently land only on their hind feet
(Walker, 1979; Oxnard et aI., 1990).
Somewhat surprisingly, the acetabular morphology of the potto resembles that of
Lepilemur and especially Galago in suggesting primarily cranially-directed loads. Al-
though pottos use vertically (and diagonally) oriented substrates (Charles-Dominique,
1971), their varied locomotor repertoire was expected to generate relatively uniform
acetabular loading. It may be that the acetabular morphology in this species is phyletic
baggage from a leaping ancestor (Walker, 1969), and that joint reaction forces from slow
climbing have been insufficient to cause significant acetabular remodeling. Other com-
parative studies of lorisid pelves have documented that morphological variation between
galagos and lorises is not as great as might be expected given their divergent locomotor
modes (McArdle, 1981; Fleagle and Anapol, 1992).
It is also possible that the loading environment is not adequately assessed by a con-
sideration of "pronograde" vs. "orthograde" positional behavior, although studies on an-
thropoids have shown they are strongly correlated (MacLatchy and Bossert, 1996;
MacLatchy, 1996). Studies of the human acetabulum have demonstrated that there are sig-
nificant relationships between orthogrady and both joint force direction (cranial) and the
locations of the highest density cortical and trabecular bone (cranial) (Dalstra et ai., 1993;
Dalstra and Huiskes, 1994). It is possible, however, that the distribution of acetabular ar-
ticular surface may reflect force transmission patterns as much as habitual orientation of
Reconstruction of Hip Joint Function in Extant and Fossil Primates 125

the pelvis. Hip joint loads being transferred to the sacroiliac joint must pass through at
least the lateral aspect of the cranial acetabular region, hence, regardless of trunk orienta-
tion, the cranial acetabulum may be under greater overall stress than the dorsal or caudal
lunate surfaces. This would also explain why Microcebus, despite having more dorsal
acetabular articular surface than the other species, also has more articular surface on the
cranial aspect of the acetabulum than elsewhere. These issues could best be resolved with
experimental data. One could insert pressure sensitive film (e.g., as in Wang et al., 1995)
into the acetabula of cadaver specimens and determine contact areas and pressures about
the hip socket under different femoral loads. The loading environment can also be recon-
structed from mechanical testing of cadaver pelves using strain gauges to measure stress
distributions, and through finite element modeling (e.g., Dalstra et aI., 1995).
Lunate surface morphology was used to infer mobility as well as loading. Microce-
bus murinus has more femoral coverage than do the other primates for both the cranial and
dorsal surfaces. Greater cranial coverage would inhibit medial rotation while greater dor-
sal coverage would inhibit abduction, and, indeed, the maximum range of abduction of
Microcebus tended to be somewhat lower than those of most of the other species. The
slightly reduced abductory capabilities of Microcebus may be related to its small size.
Jenkins has proposed that for small mammals, both terrestrial and arboreal substrates pose
similar challenges (1974). These substrates necessitate a versatile locomotor pattern and
significant abductory capabilities (Jenkins, 1974); however, to a mouse lemur, the small
branch environment may not be as "discontinuous" as arboreal substrates are for larger
primates, and may generally require less extremes of hip abduction in order to move be-
tween supports.

4.1.2. Fossil Species. The distribution of the acetabular articular surfaces of the
three fossil species are like those of Galago, Perodicticus and Lepilemur. The cranial lu-
nate surface is expansive (particularly in one Notharctus specimen and in Hemiacodon),
with less articular surface on the dorsal and caudal aspects. One Notharctus individual,
however, has an expanded dorsal lunate surface that approaches, but does not reach, the
level of expansion seen in Microcebus. Pronounced cranial expansion could be indicative
of vertically oriented behaviors, but this association is speculative since the potto, a spe-
cies that frequently uses horizontal and diagonal supports as well as vertical ones (Char-
les-Dominique, 1971), has a similar acetabular morphology. Little dorsal femoral
coverage in all three species suggests that the potential for hip abduction was substantial.

4.2. Ranges of Abduction

4.2.1. Extant Species. In a neutral posture all of the small bodied primates in this
study have femoral shaft angles that are equal to or greater than the angles of abduction
found in anthropoid primates (MacLatchy, 1996). The angles are similar to the degree of
abduction in a stationary stance for non-cursorial small mammals such as the opossum,
echidna and hamster (35-40°), the rat and ferret (25°), and the tree shrew (30°), and, with
the exception of Galago, greater than values for more cursorial mammals such as the cat
and fox (10°) (Jenkins, 1971; Jenkins and Camazine, 1977). The leapers (Galago and
Lepilemur) and above-branch quadruped (Microcebus) do not have limbs that are more ad-
ducted in a neutral posture than does Perodicticus, despite the potto's frequently abducted
limbs (Walker, 1969).
Simulations of maximum hip abduction agree well with in vivo measures. This is
particularly significant because it suggests that the bony constraints used in the simula-
126 L. MacLatchy

tions, namely fovea and fossa locations and sizes, and greater trochanter height, are the
primary determinants of maximum abduction. This is not to deny the importance of soft
tissue features in joint function, but structures like the acetabular labrum are probably
more important in providing joint stability, hence their absence from the models does not
seem to affect the predictive power of the three-dimensional animation.
All of the strepsirhines, except the potto, had similar maximal ranges of abduction
during flexion. Abduction during flexion regularly occurs during climbing, as when an
animal reaches for a distant support, as well as during various feeding and resting pos-
tures. It is also associated with vertical clinging postures prior to leaping in some primates
(Anemone, 1990; Demes et a!., 1996).
More interspecific variability in mean maximal abduction during extension is evi-
dent. A similar study on anthropoids (MacLatchy, 1996) found that chimpanzees and Old
World monkeys do not differ very much in maximal abduction during flexion, but during
extension, they have different maximal angles of abduction. An abducted, extended femo-
ral posture is probably rarer than an abducted, flexed femur in all but the potto whose legs
are habitually abducted and rotated even in extension. Abduction during extension prob-
ably only occasionally occurs during active climbing, since it would require that an animal
back up (e.g., descend tail first along a vertical support), or maintain a grasp on a support
while the femur is extended, and is probably more likely to be limited to postural behav-
iors. It is hypothesized that the ability to both extend and abduct the femur is a charac-
teristic of primates capable of varied climbing postures, the potto and other lorisids being
extreme examples (Oxnard et a!., 1990; Schulze and Meier, 1995; Figure 1). The extreme
mobility of the potto hip is partly enabled by the wide acetabular notch, which allows con-
siderable repositioning of the ligamentum teres. Although not quantified in this study, the
apparent marked elevation of the femoral head above the neck may also playa role in ena-
bling hip abduction, similar to the way in which relatively small neck size in orangutans
has been proposed to facilitate abduction by providing more articular contact area on the
superior aspect of the femoral head (Ruff, 1988).

4.2.2. Fossil Species. In a neutral posture, the two Notharctus specimens and Hemia-
codon had a relatively abducted femur like most of the strepsirhines; Omomys' femoral
posture was more adducted, but fell within the broad range of values exhibited by Galago.
For all three fossil species, maximum angles of abduction during flexion fell within
the middle of the range of the extant sample and indicate significant abductory, and there-
fore climbing, capabilities. The two Notharctus specimens distinguished themselves by
having a degree of abduction during extension that was equal to or greater than all but the
potto, suggesting that this species may have been capable of more versatile climbing pos-
tures than the other extant strepsirhines and the omomyids. Omomys has a lower degree of
abduction during extension than Hemiacodon; combined with postural data this suggests
that Omomys may have used less abducted hindlimb postures.

4.3. Femoral Morphology

Additional information is supplied by the morphology of the femoral head of
Hemiacodon and Omomys, which are similar in appearance to those of Tarsius and Galago
(Simpson, 1940; Napier and Walker, 1967; Walker, 1974; MacLatchy, 1993; Anemone,
1995; Dagosto and Schmid, 1996), and which are characterized by a flattened posterior
portion and a spherical anterior portion. This femoral morphology has somewhat inappro-
priately been described as "cylindrical" (Walker, 1974); among living primates, only
Reconstruction of Hip Joint Function in Extant and Fossil Primates 127

Ga/ago and Tarsius have "cylindrical" femoral heads (e.g., Walker, 1974; McArdle, 1981;
Anemone, 1990). The "cylindrical" characterization stems from the posterior and superior
extension of the articular surface onto the neck, and the observation that the neck is rela-
tively thick and oriented perpendicular to the shaft (Walker, 1974). This characterization,
however, ignores the proximal portion of the femoral head, which consists of an anteriorly
offset hemispherical cap.
The pronounced extension of the subchondral bone on to the back of the neck has been
interpreted to be a postural adaptation to vertical clinging (Anemone, 1990, 1995). In the
clinging posture, the posterior part of the femoral head is in contact with the caudal articular
surface of the acetabulum and the femora are strongly flexed, abducted and laterally rotated.
When a femur is placed in the acetabulum (either by hand or in the computer) to simulate a
vertically clinging posture (the close-packed position (Figure 5)) and then extended, it is clear
that the caudal aspect of the acetabulum articulates smoothly with the relatively flat posterior
surface of the femoral head. Hence, during flexion and extension the posterior aspect of the
femur is effectively loaded as a cylinder. Rather than a postural adaptation, I suggest that the
posterior flattening of the head may provide stability during flexion/extension, as in cursorial
mammals such as bovids whose femoral heads are flattened on both the anterior and posterior
aspects (Kappelman, 1988). While this femoral morphology is found in tarsiers (Anemone,
1990), Miocene loris ids (Walker, 1970) as well as Hemiacodon and Omomys (Dagosto, 1993;
Anemone, 1995; Dagosto and Schmid, 1996), other clinging and leaping primates, notably
the indriids (Demes et aI., 1996) and Lepilemur have more spherical femoral heads. Demes et
al. (1996) have suggested that the "cylindrical" head shape may provide necessary stability
during the higher substrate forces generated by smaller species during leaping. This hypothe-
sis thus implies a significant leaping adaptation in the fossil primates that bear this morphol-
ogy: Hemiacodon, Omomys and Miocene lorisids.·
There is no reason to suppose that Galago (or Hemiacodon or Omomys) is inordinately
limited in its range of hip movement by its femoral head morphology. Most of the proximal
hemispherical cap is not in contact with the acetabulum during flexion and extension. Be-
cause the anterior cap is spherical, it will permit substantial medial rotation and adduction
from the laterally rotated, abducted position of the femur during flexion (MacLatchy, 1993).
The composite femoral morphology suggests ajoint with stability through flexion and exten-
sion but also a wide range of rotation and abduction/adduction (Gebo, 1989).
The femoral features of the omomyines are congruent with a Lepilemur or Galago-
like locomotor repertoire that includes both leaping and climbing. Other workers have
found that some aspects of the Hemiacodon skeleton, such as the ischial morphology
(Fleagle and Anapol, 1992) and the tarsal bones (Simpson, 1940) lack Galago-like spe-
cializations for leaping. In the case of Omomys, Covert (1995) notes that the distal femur
is anteroposteriorly deep, supporting leaping adaptations, while Dagosto (1993) interprets
foot bones to be elongated and rather cheirogaleid-like, but without the extreme lengthen-
ing seen in Tarsius or Galago. Thus, there is a rather interesting parallel between Omomys
and Hemiacodon and Miocene lorisids: all have the "cylindrical" femoral head, which
may be associated with leaping, but lack the Galago-like elongation of the tarsals (Simp-
son, 1940; Dagosto, 1993; Walker, 1970). This implies leaping kinematics that probably

* Walker's (1970:254) initial characterization of the Miocene lorisids was that they had " ... a vertical clinging and
leaping locomotion like that of modem galagids." Gebo (1989:362) subsequently described the Miocene species
as having " ... a more generalized locomotor pattern than the vertical clinging and leaping characteristic of
galagos ... ". Femoral features such as a straight femoral shaft and an anteroposteriorly high and mediolaterally
narrow distal femur, however, argue for leaping adaptations.
128 L. MacLatchy

differed substantially from the extant "foot-powered jumpers" (sensu Gebo and Dagosto,
1988) that have a "cylindrical" femoral head morphology. Hence, the shared features of
the Hemiacodon, Omomys and Galago hip joints probably imply gross similarities related
to their locomotor mode, but not an equivalency in the biomechanics of their leaping.

5. SUMMARY

One of the most useful consequences of a three-dimensional approach to studying

joints is that the separate elements can be brought together within the computer in order to
investigate how they might have functioned as a unit. Hip posture and mobility can be di-
rectly quantified, instead of relying on intermediate proxies to infer skeletal excursion.
Based on preliminary comparisons of the computer simulations and in vivo ranges of move-
ment in strepsirhines, bony morphology seems to be the primary constraint to joint mobility
in the hip. This bodes well for behavioral reconstruction of hip use in fossil species.
Differences in abduction during extension may be an indicator of differential ability to
assume versatile suspensory, bridging and/or reaching postures, while abduction during
flexion may be a more general indicator of arboreality, except in species with extremely mo-
bile hips like Perodicticus. The potto had a greater range of maximum excursion than the
other strepsirhines despite the fact that its acetabular anatomy and its neutral posture (and
therefore its foveal morphology) did not differ from them. This extreme mobility may be en-
abled by the wide acetabular notch and the elevation of the femoral head above the neck.
Acetabular morphology does not distinguish between loading environments or dif-
ferential mobilities in Galago, Perodicticus and Lepilemur. Galago, Perodicticus and
Lepilemur all have expanded cranial lunate surfaces relative to the rest of the acetabulum.
In all but Perodicticus, this is supported behaviorally by the frequent use of vertical pos-
tures. In contrast, the quadrupedal Microcebus also has an extensive dorsal lunate surface,
indicative of relatively higher or more frequent dorsal loads. The maximum extent of the
lunate walls was used to gauge femoral coverage, which is related to hip mobility. Rela-
tively lower cranial and dorsal acetabular walls in Galago, Perodicticus and Lepilemur
supports greater hip mobility than in Microcebus.
The three fossil species exhibit a similar acetabular morphology, one that was not
characterized by substantial dorsal loading and, therefore, does not suggest restriction to
committed quadrupedalism. Mobility was moderate although Notharctus had a greater
range of maximum hip abduction than did the omomyids, and so possibly had a more ver-
satile repertoire of hindlimb postures. Femoral head morphology in the two omomyids is
proposed to be related to providing stability during flexion and extension without compro-
mising mobility, and suggests leaping and climbing adaptations.

ACKNOWLEDGMENTS

I am grateful to A.W. Crompton and Maria Rutzmoser at the Museum of Comparative

Zoology, and to Guy Musser and Ross MacPhee at the American Museum of Natural His-
tory for allowing me to study specimens under their care. Gregg Gunnell (University of
Michigan) and John Alexander (AMNH) kindly permitted me to study their unpublished
fossil material. I thank Ken Glander for permission to conduct research at the Duke Univer-
sity Primate Center, Patricia Feeser (DUPC) for performing the anesthetizations and Ted
Wheeler (Duke University Medical Center) for taking the X-rays. Ultra-detailed X-ray film
Reconstruction of Hip Joint Function in Extant and Fossil Primates 129

was provided free of charge by Kodak. Thank-you to Edward Cheal (VA Hospital, West
Roxbury, MA) for use of the digitizing program "DIGIT', and especially to Debbie Burstein
and PV Prasad (both of the Beth Israel Hospital, Boston, MA) for assistance with the MRI.
Three-dimensional computer programs were written by William Bossert. Figures 1 and 2
were prepared by Luci Betti. Valuable comments were provided by William Bossert,
Brigitte Demes, Farish Jenkins, Jr., David Pilbeam, three anonymous reviewers and, espe-
cially, Elizabeth Strasser. Generous funding was provided by the National Science Founda-
tion (SBR 9300671), the Wenner-Gren Foundation and the Mellon Foundation.

REFERENCES
Anemone RL (1990) The VCL hypothesis revisited: Patterns offemoral morphology among quadrupedal and sal-
tatorial prosimian primates. Am. J. Phys. Anthropol. 83:373-393.
Anemone RL (1995) The hip of Omomys carteri, a North American omomyine primate from the Middle Eocene.
Am. 1. Phys. Anthropol. Suppl. 20:58.
Beard KC, Dagosto M, Gebo DL, and Godinot M (1988) Interrelationships among primate higher taxa. Nature
331:712-714.
Cartmill, M (1972) Arboreal adaptations and the origin of the Order Primates. In R Tuttle (ed.): The Functional
and Evolutionary Origin of Primates. Chicago: Aldine, pp. 97-122.
Charles-Dominique P (1977) Ecology and Behaviour of Nocturnal Primates. London: Duckworth.
Covert HH (1995) Locomotor adaptations of Eocene primates: Adaptive diversity among the earliest prosimians.
In L Alterman, GA Doyle, MK Izard (eds.): Creatures of the Dark: The Nocturnal Prosimians. New York:
Plenum Press, pp 495-509.
Covert HH and Jolley LA (1995) The forelimb of Omomys carteri, a North American primate from the middle Eo-
cene. Am. J. Phys. Anthropol. Suppl. 20:79.
Dagosto M (1983) Postcranium of Adapis parisiensis and Leptadapis magnus (Adapiformes, Primates). Adapta-
tiona I and phylogenetic significance. Folia Primatol. 4/:49-101.
Dagosto M (1988) Implications of postcranial evidence for the origin of euprimates. J. Hum. Evol. 17:35-56.
Dagosto M (1993) Postcranial anatomy and locomotor behaviour in Eocene primates. In DL Gebo (ed.): Postcra-
nial Adaptation in Nonhuman Primates. DeKalb: Northern Illinois University Press, pp. 199-220.
Dagosto M, and P Schmid (1996) Proximal femoral morphology of omomyiform primates. J. Hum. Evol.
30:29-56.
Dalstra M, and Huiskes R (1995) Load transfer across the pelvic bone. J. Biomech. 28:716-724.
Dalstra M, Huiskes R, Odgaard A, and van Erning L (1993) Mechanical and textural properties of pelvic trabecu-
lar bone. J. Biomech. 26:523-535.
Demes B, Jungers WL, Wunderlich RE, Richmond BG, Fleagle JG, and Lemelin P (1996) Body size and leaping
kinematics in Malagasy vertical clingers and leapers. 1. Hum. Evol. 31:367-388.
Elftman HO (1929) Functional adaptations in the pelvis of marsupials. Bull. Am. Mus. Nat. Hist. 58: 189-232.
Fleagle JG (1976) Locomotor behaviour and skeletal anatomy ofsympatric Malaysian leaf monkeys (Presby tis ob-
scura and Presbytis melalophos). Yrbk. Phys. Anthropol. 20:441-453.
Fleagle JG (1988) Primate Adaptation and Evolution. New York: Academic Press.
Fleagle JG, and Anapol FC (1992) The indriid ischium and the hominid hip. J. Hum. Evol. 22:285-305.
Fleagle JG, and Meldrum DJ (1988) Locomotor behaviour and skeletal morphology of two sympatric pithecine
monkeys, Pithecia pithecia and Chiropotes satanas. Am. J. Primatol. 16:227-249.
Gebo DL (1989) Postcranial adaptation and evolution in Lorisidae. Primates 30:347-367.
Gebo DL (1985) The nature of the primate grasping foot. Am. J. Phys. Anthropol. 67:269-277.
Gebo DL, and Dagosto M (1988) Foot anatomy, climbing, and the origin of the Indriidae. 1. Hum. Evol.
/7:135-154.
Gregory WK (1920) On the structure and relations of Notharctus, an American Eocene primate. Mem. Am. Mus.
Nat. Hist. 3:49-243.
Harrison T (1982) Small-bodied apes from the Miocene of East Africa. Ph.D. Dissertation, University College of
London.
Harvey PH, Martin RD, and Clutton-Brock TH (1987) Life histories in comparative perspective. In BB Smuts, DL
Cheney, RM Seyfarth, RW Wrangham, and TT Struhsaker (eds.): Primate Societies. Chicago: University of
Chicago Press, pp. 181-196.
 130 L. MacLatchy

Jenkins FA, Jr. (1971) Limb posture and locomotion in the Virginia opossum (Didelphis marsupialis) and in other
non-cursorial mammals. 1. Zool., Lond. 165:303-315.
Jenkins FA, Jr. (1974) Tree shrew locomotion and primate arborealism. In FA Jenkins, Jr. (ed.): Primate Locomo-
tion. New York: Academic Press, pp. 85-115.
Jenkins FA, Jr., and Camazine SM (1977) Hip structure and locomotion in ambulatory and cursorial carnivores. 1.
Zool., Lond. 181:351-370.
Kappelman J (1988) Morphology and locomotor adaptations of the bovid femur in relation to habitat. J. Morphol.
198: 119--130.
Kay RF, Ross C, and Williams B (1997) Anthropoid origins. Science 275:797-804.
Latimer B, Ohman JC, and Lovejoy, CO (1987) Talocrural joint in African hominoids: Implications for Australopi-
thecus afarensis. Am. J. Phys. Anthropol. 74:155-175.
MacLatchy LM (1993) Three dimensional quantification of hip morphology in fossil primates. 1. Vert. Paleontol.
13:48a.
MacLatchy LM (1996) Another look at the australopithecine hip. J. Hum. Evol. 31:455-476.
MacLatchy LM, and Bossert W (1996) An analysis of the articular surface distribution of the femoral head and
acetabulum in anthropoids, with implications for hip function in Miocene hominoids. J. Hum. Evol.
31:425-453.
McArdle JE (1981) Functional morphology of the hip and thigh of the loriformes. Contrib. Primatol. 17: 1-132.
McHenry HM (1975) A new pelvic fragment from Swartkrans and the relationship between the robust and gracile
australopithecines. Am. J. Phys. Anthropol. 43:245-262.
Martin RD (1990) Primate origins and evolution: A phylogenetic reconstruction. Princeton: Princeton University
Press.
Martin RD (1993) Primate origins: Plugging the gaps. Nature 363:223-234.
Mittermeier RA, Tattersall I, Konstant WR, Meyers DM, Mast RB, and Nash S (1994) Lemurs of Madagascar.
Washington: Conservation International.
Napier JR ,and Napier PH (1967) A Handbook of Living Primates. New York: Academic Press.
Napier JR, and Walker AC (1967) Vertical clinging and leaping, a newly recognized category of locomotory be-
haviour among primates. Folia Primatol. 6: 180-203.
Oxnard CE, Crompton RH, and Lieberman SS (1990) Animal Lifestyles and Anatomies. Seattle: University of
Washington Press.
Ruff CB (1988) Hindlimb articular surface allometry in Hominoidea and Macaca, with comparisons to diaphyseal
scaling. J. Hum. Evol. 17:687-714.
Simpson GG (1940) Studies on the earliest primates. Bull. Am. Mus. Nat. Hist. 77: 185-212.
Schulze H, and Meier B (1995) Behavior of captive Loris tardigradus nordicus: A qualitative description, includ-
ing some information about morphological bases of behavior. In L Alterman, GA Doyle, and MK Izard
(eds.): Creatures of the Dark: The Nocturnal Prosimians. New York: Plenum Press, pp 221-249.
Sokal RR, and Rohlf FJ (1981) Biometry. San Francisco: W.H. Freeman and Company.
Stem JT, and Susman RL (1983) The locomotor anatomy of Australopithecus afarensis. Am. J. Phys. Anthropol.
60:279--317.
Swartz SM (1989) The functional morphology of weight bearing: Limb joint surface area allometry in anthropoid
primates. J. Zool., Lond. 218:441-460.
Szalay FS, and Dagosto M (1980) Locomotor adaptations as reflected on the humerus of Paleogene primates. Fo-
lia Primato!' 34: 1-45.
Turnquist JE (1983) Influence of age, sex and caging on the joint mobility in the patas monkey (Erythrocebus
patas). Am. J. Phys. Anthropol. 61:211-220.
Turnquist JE (1985) Passive joint mobility in patas monkeys (Erythrocebus patas): Rehabilitation of caged ani-
mals after release into a free-ranging environment. Am. J. Phys. Anthropol. 67:1-5.
Walker A (1969) The locomotion of the lorises, with special reference to the potto. 1. E. Afr. Wildlife 7: 1-5.
Walker A (1970) Post-cranial remains of the Miocene Lorisidae of East Africa. Am. J. Phys. Anthropol.
33:249--262.
Walker A (1974) Locomotor adaptations in past and present prosimian primates. In FA Jenkins, Jr. (ed.): Primate
Locomotion. New York: Academic Press, pp. 349--381.
Walker A (1979) Prosimian locomotor behaviour. In GA Doyle, and R D Martin (eds.): The Study of Prosimian
Behaviour. London: Duckworth, pp. 543-565.
Wang C, Cheng C, Chen C, Lu C, Hang Y, and Liu T (1995) Contact areas and pressure distributions in the sub-
talar joint. J. Biomech. 28:269-279.
Ward CV (1991) Functional anatomy of the lower back and pelvis of the Miocene hominoid Proconsul nyanzae
from Mfangano Island, Kenya. Ph.D. Dissertation, Johns Hopkins University, Baltimore.
 8

GRASPING PERFORMANCE IN SA GUINUS

MIDAS AND THE EVOLUTION OF HAND
PREHENSILITY IN PRIMATES

Pierre Lemelin l and Brian W. Grafton2

1Department of Anatomy
Northeastern Ohio Universities College of Medicine
Rootstown, Ohio 44272
2School of Biomedical Sciences
Division of Biological Anthropology
Kent State University
Kent, Ohio 44242

1. INTRODUCTION

The Order Primates is characterized by a unique suite of cranial and postcranial fea-
tures that may have evolved for visual predation on insects or exploitation of fruits in a
small-branch milieu [see Cartmill (1992) for recent review]. Compared to more general-
ized mammals like tree shrews, primates possess a relatively larger brain with a more de-
veloped visual area and more reduced olfactory bulbs, orbits that are more approximated
and more convergent with one another, and grasping hands and feet that bear nails instead
of claws (Cartmill, 1970, 1972, 1974a,b, 1992; Le Gros Clark, 1959, 1963; Martin, 1986;
Wood Jones, 1916).
With regard to the grasping extremities of primates, the anthropological literature
has usually linked enhanced grasping abilities of the hands and feet of primates with the
lack of claws. For example, Le Gros Clark (1959: 174) argued that "Compared with claws
these [nails] provide a much more efficient grasping mechanism for animals which find it
necessary to indulge in arboreal acrobatics, for by their greater pliability they can be
adapted with much more precision to surfaces of varying shape, size, and texture". In an
important contribution on primate hands, Napier (1993: 40) stated the following: "Claws
are not compatible with prehensility because of the mechanical obstruction of claws over-
growing the fingertips ... " In the same vein, Hershkovitz (1970) suggested that claws de-
generated in primates to enable the extremities to achieve full phalangeal flexure when
gripping. This character association between grasping abilities of the extremities and lack

131
132 P. Lemelin and B. W. Grafton

of claws can also be found in many introductory textbooks of physical anthropology.

Among the evolutionary trends that distinguish primates from the rest of the mammals,
Nelson and colleagues (1992: II3) wrote that primates are characterized by "Flexible
hands and feet with a good deal of prehensility (grasping ability)" and added that "This
feature is associated directly with the lack of claws and retention of five digits."
The ability to manipulate objects is inherent to the definition of hand prehensility.
According to Napier (1961: 116-117), "A convergent hand can be termed prehensile when
the digits approximate in such a manner that an object may be grasped and held securely
against external influences (e.g., gravity) that may be tending to displace it." Tree shrews,
many rodents and some carnivores use both hands together to bring food objects to the
mouth, whereas primates can accomplish the same action with only one hand (Bishop,
1964; McClearn, 1992; Napier, 1961, 1993; Polyak, 1957). Again, the argument can be
made that claws prevent prehensile grips while an animal is reaching and gripping objects
with only one hand.
Tamarins represent an appropriate group of primates to test this hypothesis because
of their hand anatomy and ecology. Like all callitrichids, they have hands that bear sharp
claws (Hershkovitz, 1977), which are likely to have evolved from a platyrrhine ancestor
with nails (Ford, 1980; Rosenberger, 1977; Thorndike, 1968). Field data revealed that
tamarins are eclectic feeders that mainly eat ripe fruits and nectar, but also eat arthropods,

Figure 1. Saguinus midas (drawing courtesy of Stephen

D. Nash). The fingers and toes of the golden-handed tama-
rin sport sharp claws.
Grasping Performance in Saguinus midas and the Evolution of Hand Prehensility in Primates 133

especially orthopteran insects such as grasshoppers and crickets (Garber, 1980, 1984,
1988, 1992, 1993; Mittermeier and van Roosmalen, 1981; Snowdon and Soini, 1988;
Soini, 1987; Sussman and Kinzey, 1984; Terborgh, 1983). Tamarins display different in-
sect-catching strategies, but all of them appear to involve a great deal of hand use (Garber,
1993; Terborgh, 1983).
The putative association between presence of claws and the inability to achieve pre-
hensile grips implies that tamarins have limited capabilities of grasping and manipulating
using only one hand. This study aimed to test that implication in the golden-handed tam a-
rin, Saguinus midas (Figure 1). Manual grasping behavior was observed in tamarins for
several types of food objects under experimental conditions. For each type of food object,
the ability of tamarins to accomplish single-handed grips was quantified and compared to
that of a squirrel monkey (Saimiri sciureus), a platyrrhine with nails. Then, finger propor-
tions were compared between callitrichids, squirrel monkeys, and a variety of clawed
mammals for which some behavioral data on their manual grasping abilities were avail-
able. These comparisons were made in order to identify morphological features of the
hand that may be linked with potential differences in grasping behavior.

2. MATERIALS AND METHODS

Three different types of food objects (i.e., blueberries, live crickets, and live meal-
worms) were presented one by one on a platform placed just outside the bottom of enclo-
sures housing three adult male and one adult female golden-handed tamarins, or one adult
male squirrel monkey. During the experiments, live crickets were kept in a cooler filled
with ice so that they did move but not hop out of reach of the animals when positioned on
the platform. The mesh of the enclosure wire was narrow enough so that only one fore-
limb could be used to reach for a food object placed on the 9 cm diameter platform. Each
subject was able to reach the platform easily using protraction of one forelimb.
For all three categories of food objects, a total of 606 grips for tamarins and 60 grips
for the squirrel monkey were recorded on videotape using a Panasonic AG-450 Super
VHS camcorder or Sony 8 mm camcorder equipped with a lOX zoom lens and mounted
on a tripod. The high-speed shutter (between 11125 and 1/1000 sec) of the camcorder was
used when possible during filming (30 frames/sec). This option was valuable for obtaining
still frames with less motion blur. Each videotape was viewed and analyzed with a video
cassette recorder plugged into a monitor. The frame-by-frame playback option allowed the
classification of grips into successful and unsuccessful events for each food object cate-
gory. A successful grip involved a reach and a secure grab of the food object, completed
by retrieval toward the mouth for ingestion. A grip was considered unsuccessful when an
animal reached for a food object, but either failed to grab it securely while touching it, or
simply dropped it while bringing the food object to the mouth.
The ability to grasp and retain a food object with only one hand should be reflected in
the proportions of the manual ray elements. Shorter metacarpals relative to the digits, longer
digits relative to the metacarpals, or a combination of both provide some of the mechanical
requirements to achieve prehensile grips. In animals using simultaneous convergence of all
the digits during grasping, such proportions allow the digital portion of the ray to encircle
and to secure an object against the palm by flexion of both metacarpophalangeal (MP) and
proximal interphalangeal (IP) joints. Following Napier (1993), a phalangeal index was com-
puted to obtain an estimate of the degree of prehensility of each finger. Lengths of the meta-
carpals, proximal phalanges, and middle phalanges for rays II through V were measured on
134 P. Lemelin and B. W. Grafton

skeletal specimens of callitrichids (Saguinus midas, Saguinus geoffroyi, Leontopithecus

rosalia, and Callithrixjacchus), squirrel monkeys (Saimiri sciureus), tree squirrels (Sciurus
carolinensis), tree shrews (Tupaia glis), and procyonids (Procyon totor and Potosflavus).
The phalangeal index corresponds to the sum of the length of the proximal and intermediate
phalanges times one hundred and divided by the length of the corresponding metacarpal.
The terminal phalanges were excluded in the calculation of the phalangeal indices for sev-
eral methodological reasons. The distal phalanges of many skeletal specimens were either
missing, disarticulated or covered with dry skin. Moreover, since most skeletal specimens
measured for this study possess pointed and elongated distal phalanges in the shape of a
claw, it was difficult to assess their true length contributing to the prehensility of the hand.

3. RESULTS

3.1. Behavioral Data

3. J. J. Comparison of Grasping Behavior and Abilities between Saguinus midas and

Saimiri sciureus. Tamarins and squirrel monkeys used similar prehensive patterns referred
to as "whole-hand control" by Bishop (1964). Such prehensive patterns, which are charac-
teristic of pro simians and most New World monkeys (Bishop, 1964; Costello and Fra-
gaszy, 1988), involved no independent movement of the digits and objects were picked up
in a manner analogous to the human "power grips" described by Napier (1956, 1993). Al-
though the prehensive patterns of the primates under investigation were fairly uniform, the
positions of the hand while holding an object or prehensile grips (Bishop, 1964) varied ac-
cording to the size and shape of the object held.

Figure 2. Series of video frames ofa single grasping event involving a blueberry in Saguinus midas.
Grasping Performance in Saguinus midas and the Evolution of Hand Prehensility in Primates 135

Blueberries were gripped in different manners by tamarins and the squirrel monkey
despite using a similar "whole-hand control" prehensive pattern. In tamarins, the hand
reached with the fingers flexed 90Q at the proximal IP joints (Figure 2a) and dropped from
above the food object for its retrieval (Figure 2 b,c). In some instances, blueberries were
swept from the side instead of from above. Also, blueberries were gripped occasionally by
a scissor-grip between the sides of the second and third digits. In the squirrel monkey, the
hand reached for blueberries in a similar manner with the fingers flexed 90 Q at the proxi-
mal IP joints. Unlike tamarins, however, the thumb of the squirrel monkey played a more
important role when gripping blueberries off the platform. When the fingers encircled a
blueberry by flexion of both MP and proximal IP joints, the thumb was pressing laterally
on the food object. As a result, the volar surface of the thumb was oriented in a different
plane from that of the fingers. Although the thumb was involved extensively during grasp-
ing, there was no independent movement of the digits that characterizes precision gripping
as in Cebus apella (Costello and Fragaszy, 1988).
Blueberries were gripped successfully in a proportion of 9 to 1 in both tamarins and
the squirrel monkey (Figure 3). In tamarins, a total of 188 grasping events involving only
one hand for blueberries was counted. This number was much lower for the squirrel mon-
key with a total of 31 events. The percentages of successful grips, however, were similar
for both taxa (89.9% for tamarins and 90.3% for the squirrel monkey) (Figure 3).
Tamarins and the squirrel monkey caught live crickets using grips reminiscent of
those reported for small-bodied prosimians (Bishop, 1964; Lemelin, 1996a,b; Niemitz,
1984) and Callithrix jacchus (Rothe, 1971). In most cases, the hand reached toward the
cricket with the digits extended and adducted at the MP joints (Figure 4a). A few centime-
ters from the cricket, the hand opened with a very rapid movement of abduction of all the
digits at their MP joints, including the thumb (Figure 4b). Thus, just before contact with
the cricket, the digits were widely spread and covered the greatest area possible. At con-
tact, all the digits converged and flexed at their MP and proximal IP joints, encircling the
prey for retrieval (Figure 4c,d). Retrieval of the object toward the mouth was completed
by flexion and supination of the forearm.
Like blueberries, crickets were gripped successfully in a proportion of 9 to 1 in both
tamarins and the squirrel monkey (Figure 3). In tamarins, 157 single-handed grips involv-
ing crickets were counted and 89.9% of these grips were successful. The proportion of

S. midas S. sciureus S. midas S. sciureus S. midas S. sciureus

n=188 n=31 n= 157 n=7 n=261 n=22
%100~---------------'-----------------r--------------~
90
80
70
60
50
40
30
20
10
o
Blueberries Crickets Mealworms

Figure 3. Proportions of successful (black) and unsuccessful (shaded) grips for three types of food objects in
Saguinus midas and Saimiri sciureus.
136 P. Lemelin and B. W. Grafton

Figure 4. Series of video frames of a single grasping event involving a live cricket in Saguinus midas.

successful grips for the squirrel monkey was similar (85.7%). This result should be inter-
preted with caution, however, because of the very small sample size.
The most notable differences in grasping behavior between tamarins and the squirrel
monkey were observed when mealworms were presented. In tamarins, the hand reached
toward the mealworm with all the digits extended and adducted at the MP joints (Fig-
ure 5a). Prior to contact with the mealworm, the digits were more adducted at their MP
joints as for crickets. Upon contact, the meal worm was brought toward the palm by sev-
eral successive and rapid movements of flexion and extension of all the digits at both MP
and IP joints (Figure 5b,c). Despite firm contact between the tips and claws of the central
digits and the palm, meal worms had a tendency to slip away during retrieval (Figure 5d).
On many occasions, successful grips involving meal worms were completed using scissor-
grips between the sides of digits II and III or digits III and IV. In the squirrel monkey, the
hand reached for mealworms in a manner similar to that for crickets with the digits ex-
tended and abducted at the MP joints. At contact, the fingers flexed at both proximal IP
and MP joints to encircle the mealworm. As for blueberries and crickets, the thumb was
positioned laterally on the food object so that the volar surface of the thumb was oriented
in a different plane from that of the fingers. Unlike tamarins, the thumb played a more im-
portant role in the squirrel monkey when gripping and only one movement of flexion and
extension of all the digits was necessary to retrieve successfully a mealworm from the
platform.
Not surprisingly, the percentages of successful grips for mealworms differed be-
tween tamarins and the squirrel monkey (Figure 3). With a total of 261 grips, tamarins
only had a 62.8% success rate for mealworms. In contrast, the squirrel monkey had an
86.4% success rate for the same type of food object for a total of 22 grips.
Grasping Performance in Saguinus midas and the Evolution of Hand Prehensility in Primates 137

Figure 5. Series of video frames of a single grasping event involving a live meal worm in Saguinus midas.

3.2. Morphometric Data

3.2.1. Comparisons of Hand Proportions and Grasping Behavior between

Callitrichids, Saimiri, and Nonprimate Mammals. Saguinus midas and other 'callitrichids
can be distinguished from Saimiri sciureus not only by having claws instead of nails, but
also by different hand proportions. Squirrel monkeys possess higher phalangeal indices for
the central digits (III and IV) compared to those of each callitrichid taxon (pairwise com-
parisons using a Mann-Whitney V-test, P < 0.05; Figure 6; Table 1). Among callitrichids,
L. rosalia has significantly lower phalangeal indices for all rays (Mann-Whitney V-test, P
< 0.01; Figure 6; Table 1). This can be explained by the more important contribution of
the metacarpals in total hand length in L. rosalia compared to Callithrix and Saimiri (Jouf-
froy et aI., 1991). Of all anthropoids, Leontopithecus possesses the longest hands relative
to forelimb length (Jouffroy et aI., 1991) and compared to other callitrichids and Saimiri,
Leontopithecus captures prey by probing through narrow holes or dense foliage
(Hershkovitz, 1977; Rylands, 1989). Relatively long hands, independently of their degree
of prehensility, may be more useful for probing behavior.
For most rays, primates and several nonprimate mammals form two distinctive clus-
ters (Figure 6). Raccoons (Procyon) have the lowest phalangeal indices of all taxa consid-
ered in the comparative sample (Figure 6; Table 1). McCleam (1992) mentioned that P
lotor uses both forepaws to hold small food items despite its notorious ability to locate and
handle food items with the hands. In contrast, its close relative the kinkajou (Potos jlavus)
commonly relies on single-handed grips to hold small food objects like wild figs
(McCleam, 1992). Interestingly, kinkajous are well-differentiated from raccoons in terms
of hand proportions (Figure 6; Table 1). For all rays, kinkajous have higher phalangeal in-
138 P. Lemelin and B. W. Grafton

Palos ffavus II
Procyon fofor [Ij
Tupaia glis - _a:!-
SciuflJS carofinensis
Saimiri sciureus
Saguinus midas
V
Saguinus oedipus
Caf/ilhrix jacchus

-
Leonto ilhecus rosalia

C1D-- IV
·CIJI~-

r.

-G - -
-QD-
III

-a:J- Figure 6. Box-plot charts of the phalangeal

--c:E:J- indices for manual rays II through V in non-
ID
II]
primate mammals, Saimiri, and callitrichids.

~
-=- II
The diamond represents the mean, the vertical
line the median, the left and right comers of
--m- the rectangle the 25th and 75th percentile, and
~

c:::B:I
the left and right ends of the horizontal line
-a:::J- the 10th and 90th percentile. The shaded rec-
40 60 80 100 120 140 160 180% tangles represent platyrrhine taxa and open
ones non primate mammals. See Table I for
Phalangeal Indices sample sizes.

Table 1. Descriptive statistics of the phalangeal indices for manual rays II through V of
nonprimate mammals, Saimiri, and callitrichids J

Taxon Ray II Ray III Ray IV Ray V

Potosflavus 121.71 120.49 118.55 119.93
(N=3) 101.83--141.59 104.66-136.31 110--127.1 115.62-124.24
Procyon lotOl}. 84.03 79.22 81.48 92.95
(N=\o, II) 81.93--86.13 78.09--80.35 80.04--82.92 90.85-95.04
Tupaia glis 99.84 90.2 96.92 114.6
(N=IO) 97.77-101.92 88.62-91.78 94.85-99 108.38-120.81
Sciurus carolinensis 151.89 139.22 142.86 156.95
(N=IO) 147.6-156.19 137.63-140.8 140.23-145.49 152.27-161.63
Saimiri seillreus 132.39 145.6 155.17 147.94
(N=5) 124.46-140.32 140.07-151.12 147.57-162.77 141.65-154.24
Saguillus midas 128.91 129.94 136.34 133.11
(N=8) 125.6-132.23 125.41-134.48 126.47-146.21 128.18-138.05
Saguilllls oedipus 128.03 131.12 130.44 129.19
(N=8) 120.54--135.51 126.62-135.63 126-134.88 125-133.37
Callithrix jacchus 134.72 134.56 136.85 141.97
(N=8) 129.74--139.7 128.38-140.74 132.17-141.52 136.33--147.61
Leolltopitheclls rosalia 116.84 116.11 117.34 109.59
(N=7) 111.55-122.12 112.56-119.67 111.66-123.01 102.6-116.58
'Values under the columns are (top) arithmetic mean and (bottom) the 95% confidence intervals around the
mean.
1Forray II N=IO, for rays III-V N=II.
Grasping Performance in Saguinus midas and the Evolution of Hand Prehensility in Primates 139

dices, thus longer digits relative to the corresponding metacarpals, compared to raccoons
(Mann-Whitney U-test, P < 0.01). Tree shrews (Tupaia glis) cluster in between raccoons
and kinkajous, except for ray V (Figure 6). Like raccoons, tree shrews rarely pick up food
objects with single-handed grips and both hands are used whenever possible (Bishop,
1964). Unlike other nonprimate mammals considered here, tree squirrels (Sciurus
carolinensis) have similar or higher phalangeal indices to those of callitrichids and Saimiri
(Figure 6; Table I). The digits of Sciurus contribute close to 65% of total hand length
compared to 55% or lower for Saimiri and Callithrix (Jouffroy et aI., 1991). Behavioral
data on grasping abilities of tree squirrels are very scanty. Typically, they use both hands
to hold food items, especially when adopting a bipedal stance (Koprowski, 1994).

4. DISCUSSION

4.1. Presence of Claws and Prehensile Abilities

The results of the behavioral experiments demonstrate that despite the presence of
claws on their hands, tamarins are as competent as squirrel monkeys to achieve single-
handed (i.e., prehensile) grips when picking up food objects from the platform. The nailed
digits of squirrel monkeys may be more efficient to grasp smaller and slippery objects
such as mealworms. Nonetheless, the comparable success rates for blueberries and crick-
ets between tamarins and squirrel monkeys do not support the general notion that claws
hinder the ability of grasping food objects with only one hand.
If claws do not impede the ability to use prehensile grips, one wonders why most
primates have nails. Cartmill (1974a) demonstrated that claws are superior to nails and vo-
lar skin in most arboreal situations, except for small-branch supports. Field studies by
Garber (1980, 1984) and Garber and Sussman (1984) on the positional behavior of
Saguinus geojJroyi, however, showed that fine-branch supports (less than 3 cm in diame-
ter) are preferred when foraging for fruits and insects. Similarly, a large portion of the lo-
comotor repertoire of S. midas during feeding involves quadrupedal movement on
small-diameter supports (Fleagle and Mittermeier, 1980; Thorington, 1968). Among non-
primate mammals, Caluromys, a South American didelphid marsupial that sports sizable
claws except on the hallux, also relies preferentially on small and terminal branches while
moving and foraging for fruits and insects high in the canopy (Charles-Dominique et aI.,
1981; Rasmussen, 1990). These data suggest that claws may not hinder the ability of an
animal gathering fruits and chasing insects on fine-branch supports.
Another possible explanation for nailed extremities may lie in the histological struc-
ture of the glabrous skin of primates compared to other mammals. In his essay estab-
lishing the bases of the arboreal theory of primate origins, Wood Jones (1916: 158)
stressed that "Not only does the hand come to take over the crude grasping functions of
the teeth and jaws, but in gradual stages it slowly but surely usurps the delicate tactile du-
ties of the muzzle." Le Gros Clark (1959: 285) championed the arboreal theory by reiterat-
ing that" ... in the evolution of the Primates, the more primitive tactile organs represented
by the vibrissae have been gradually replaced by the development of the more delicately
informative tactile pads on the terminal phalanges of the digits." Despite the pitfalls of the
arboreal theory in explaining the adaptiveness of unique primate features (Cartmill, 1970,
1972, 1974b, 1992), Wood Jones and Le Gros Clark's observations were important: com-
pared to most mammals, primates appear to rely extensively on the tips of their digits to
acquire information about their environment.
140 P. Lemelin and B. W. Grafton

Comparative studies from Krause (1860) and Winkelmann (1963, 1965) indicated
that the glabrous skin of primates possesses a unique type of cutaneous sense organ called
Meissner's corpuscle. Meissner's corpuscles are rapidly adapting mechanoreceptors [i.e.,
the neuronal response occurs as long as the stimulus is moving (Johansson, 1979)] located
in the dermal papillae of the glabrous skin and playa vital role in tactile acuity and dis-
crimination (Johansson, 1978, 1979; Munger and Ide, 1988; Vallbo and Johansson, 1978).
Studies of humans revealed that the proportion of Meissner's corpuscles is highest at the
tip of the digits compared to other areas of the hand (Johansson, 1979; Johansson and
Vallbo, 1976, 1979) and that the number of Meissner's corpuscles varies with age or the
frequency of manual labor (Cauna, 1959; Quilliam and Ridley, 1971). In contrast, the
glabrous skin of most nonprimate mammals has a type of cutaneous sensory organ analo-
gous to Meissner's corpuscles called "mammalian end organ" (Winkelmann, 1963, 1965).
Although morphological differences have been described between mammalian end organs
and Meissner's corpuscles (Winkelmann, 1965), there is little comparative neurophysi-
ological data to support possible tactile differences between these two types of cutaneous
receptors. From a comparative viewpoint, it is also interesting to note that Meissner's cor-
puscles have been reported for the glabrous skin of Didelphis virginiana (Winkelmann,
1964, 1965) and mouse toe pads (Ide, 1976).
From this evidence, it is possible that early in their evolutionary history, primates
may have improved tactile faculties at the tip of their digits in response to specific manual
tasks by increasing apical pad area and consequently the number of Meissner's corpuscles.
It is also possible that this putative increase in area may have triggered morphological
changes from a claw to a nail. No present data exist, however, to support these statements.
Further study of the histological and molecular structure of the digital volar skin of pri-
mates and other mammals, as well as the relationship between the presence/absence of
claws, nail size and apical pad area, could prove fruitful in clarifying the problem of the
lack of claws in primates.

4.2. Hand Proportions and Prehensile Abilities

The comparison of ray proportions between Saguinus midas and other clawed mam-
mals also provides insights in sorting some of the morphological characteristics that co-
vary with hand prehensility. For example, Tupaia possesses shorter proximal and middle
phalanges relative to the length of the metacarpals compared to Saguinus (Figure 7). If Tu-
paia uses solely one hand, a food object is secured only marginally between the palm and
the digits. Again, behavioral data from Bishop (1964) show that tree shrews have diffi-
culty in achieving prehensile grips and usually use both hands to hold food objects. In
contrast, Saguinus has longer proximal and middle phalanges relative to the length of the
metacarpals (Figure 7). When the flexed digits converge toward the palm, they can com-
pletely encircle food objects, thus providing a secure single-handed grip. The behavioral
data presented above are evidence for the ability of S. midas to accomplish single-handed
grips. Similarly, the differences in hand proportions between two closely related pro-
cyonid taxa parallel their reported differences in manual prehensile abilities. Raccoons
have shorter fingers relative to the palm and usually hold small objects between both
hands (McCleam, 1992). In contrast, kinkajous have longer digits relative to the palm and
hold small fruits using single-handed grips (McCleam, 1992).
The case of the tree squirrels also demonstrate that factors other than relative digit
length were equally important in the evolution of manual prehensility in primates. From
the strict point of view of hand proportions, we would expect tree squirrels to behave like
Grasping Performance in Saguinus midas and the Evolution of Hand Prehensility in Primates 141

Tupaia Saguinus

Figure 7. Differences in hand proportions between Tupaia and Saguinus. In tamarins, the proximal and middle
phalanges (in gray) are longer relative to the metacarpals (in black) compared to tree shrews. [Modified from
Hershkovitz (1977) with permission from The University of Chicago Press]

Saguinus. Comparative studies from Garber (1980) and Garber and Sussman (1984)
clearly indicated that Saguinus and Sciurus are behaviorally and ecologically distinct. Un-
like tamarins, tree squirrels (S granatensis) avoid fine branches (less than 3 cm in diame-
ter) during travel and foraging, and transport food objects in their mouth to larger and
more stable branches for consumption (Garber, 1980; Garber and Sussman, 1984). From a
kinematic point of view, it is possible that tree squirrels have more limited flexion when
the fingers converge toward the palm, thus preventing prehensile grips, compared to pri-
mates with similar hand proportions.
Tree squirrels are also very different from tamarins and other primates in the anat-
omy of their visual system. Although tree squirrels possess an enlarged cortical repre-
sentation of the binocular portion of their field of vision (Hall et aI., 1971; Kaas et aI.,
1972), they have much less convergent and frontated orbits compared to primates, espe-
cially anthropoids (Cartmill, 1970, 1972, 1974b; Ross, 1995, 1996). Allman (1977: 27)
emphasized the importance of more frontally directed orbits to improve" ... the optical
quality of the retinal images for the central part of the visual field located in front of the
anima!..." [italicized words are by Allman). This is especially useful for predatory ani-
mals relying on precise eye-hand coordination to catch moving prey (Allman, 1977). A re-
cent study by Servos et al. (I 992) demonstrated that humans perform significantly better
 142 P. Lemelin and B. W. Grafton

when reaching and grasping objects with one hand under binocular vs. monocular viewing
conditions. On that basis, we suggest that elongation of the digits relative to the palm was
probably a critical factor in the evolution of manual prehensility in early primates, in com-
bination with other factors like the development of musculoskeletal features that promoted
digital movement as well as neural specializations that enhanced vision quality, hand con-
trol, and tactile faculties of the digits.

ACKNOWLEDGMENTS

We would like to thank Drs. Alfred Rosenberger, John Fleagle, Henry McHenry, and
Elizabeth Strasser for inviting us to contribute a chapter for this book. Dr. Patricia Wright
gave us permission to film the tamarins, Dr. Patricia McDaniel and Alan Sironen from the
Cleveland Metroparks Zoo granted us permission to film the squirrel monkey under their
care, and Joanne Labate and the caretakers of the Cleveland Metroparks Zoo provided in-
valuable help during the behavioral experiments. The curators and staffs of the American
Museum of Natural History (New York), Anthropologisches Institut und Museum der Uni-
versitat Zurich-Irchel (Zurich), Cleveland Museum of Natural History (Cleveland), Field
Museum of Natural History (Chicago), Museum of Comparative Zoology, Harvard (Cam-
bridge), National Museum of Natural History (Washington, D.C.), and Nationaal
Natuurhistorisch Museum (Leiden) gave much appreciated help and access to their skele-
tal collections. Stephen Nash kindly provided the artwork for Figure I and The University
of Chicago Press gave permission to use copyrighted material for Figure 7. Finally, we are
grateful to Dr. Paul Garber and especially to an anonymous reviewer who gave very thor-
ough and useful reviews on earlier versions of this manuscript. This research was sup-
ported by grants from the National Science Foundation (SBR-9318750), Sigma-Xi, The
Scientific Research Society, and the Doctoral Program in Anthropological Sciences,
SUNY at Stony Brook to PL.

REFERENCES
Allman J (l977) Evolution of the visual system in the early primates. In JM Sprague, and AN Epstein (eds.): Pro-
gress in Psychobiology and Physiological Psychology. Vol. 7. New York: Academic Press. pp. I-53.
Bishop A (1964) Use of the hand in lower primates. In J Buettner-lanusch (ed.): Evolutionary and Genetic Biology
of Primates, Vol. 2. New York: Academic Press, pp. 133-225.
Cartmill M (1970) The orbits of arboreal mammals: A reassessment of the arboreal theory of primate evolution.
Ph.D. Dissertation, University of Chicago.
Cartmill M (1972) Arboreal adaptations and the origin of the order Primates. In RH Tuttle (ed.): The Functional
and Evolutionary Biology of Primates. Chicago: Aldine-Atheton, pp. 3-35.
Cartmill M (I 974a) Pads and claws in arboreal locomotion. In FA Jenkins (ed.): Primate Locomotion. New York:
Academic Press, pp. 45-83.
Cartmill M (l974b) Rethinking primate origins. Science 184:436-443.
Cartmill M (1992) New views on primate origins. Evo!. Anthropo!. I: I 05-111.
Cauna N (1956) Nerve supply and nerve endings in Meissner's corpuscles. Am. J. Ana!. 99:315-350.
Charles-Dominique P, Atramentowicz M, Charles-Dominique M, Gerard H, Hladik A, Hladik CM, and Prevost
MF (1981) Les mammiferes frugivores arboricoles nocturnes d'une foret guyanaise: Inter-relations pi antes-
animaux. Rev. Ecol. (Terre Vie) 35:341-435.
Costello MB, and Fragaszy DM (1988) Prehension in Cebus and Saimiri: I. Grip type and hand preference. Am. J.
Primato!' 15:235-245.
Ford SM (1980) Callitrichids as phyletic dwarfs, and the place of the Callitrichidae in Platyrrhini. Primates
21:31-43.
Grasping Performance in Saguinus midas and the Evolution of Hand Prehensility in Primates 143

Fleagle JG, and Mittermeier RA (1980) Locomotor behavior, body size, and comparative ecology of seven Suri-
nam monkeys. Am. J. Phys. Anthropol. 52:301-314.
Garber PA (1980) Locomotor behavior and feeding ecology of the Panamanian tamarin (Saguinus oedipus geof
froyi, Callitrichidae, Primates). Int. J. Primatol. 1:185--201.
Garber PA (1984) Use of habitat and positional behavior in a neotropical primate, Saguinus oedipus. In P Rodman
and JGH Cant (eds.): Adaptations for Foraging in Nonhuman Primates. New York: Columbia University
Press, pp. 112-133.
Garber PA (1988) Diet, foraging patterns, and resource defense in a mixed species troop of Saguinus mystax and
Sagllinus./ilscicol/is in Amazonian Peru. Behaviour 105:18--34.
Garber PA (1992) Vertical clinging, small body size, and the evolution of feeding adaptations in the Callitrichinae.
Am. J. Phys. Anthropol. 88:469-482.
Garber PA (1993) Feeding ecology and behaviour of the genus Saguinus. In AB Rylands (ed.): Marmosets and
Tamarins. Oxford: Oxford University Press, pp. 273-295.
Garber PA, and Sussman RW (1984) Ecological distinctions between sympatric species of Saguinus and Sciurus.
Am. J. Phys. Anthropol. 65:135--146.
Hall WC, Kaas JH, Killackey H, and Diamond IT (1971) Cortical visual areas in the grey squirrel (Sciurus
carolinensis): A correlation between cortical evoked potential maps and architectonic subdivisions. J.
Neurophysiol. 34:437-452.
Hershkovitz P (1970) Notes on Tertiary platyrrhine monkeys and a description of a new genus from the late Mio-
cene of Columbia. Folia Primatol. 12: 1-37.
Hershkovitz P (1977) Living New World Monkeys (Platyrrhini), Vol. I. Chicago: The University of Chicago Press.
Ide C (1976) The fine structure of the digital corpuscle of the mouse toe pad, with special reference to nerve fi-
bers. Am. J. Anal. 147:329--356.
Johansson RS (1978) Tactile sensibility in the human hand: Receptive field characteristics of mechanoreceptive
units in the glabrous skin area. J. Physiol. (London) 281:101-123.
Johansson RS (1979) Tactile afferent units with small and well demarcated receptive fields in the glabrous skin
area of the human hand. In DR Kenshalo (ed.): Sensory Functions of the Skin of Humans. New York: Ple-
num Press, pp. 129- 145.
Johansson RS, and Vallbo AB (1976) Skin mechanoreceptors in the human hand. An inference of some population
properties. In Y Zotterman (ed.): Sensory Functions of the Skin in Primates. Oxford: Pergamon Press, pp.
171-184.
Johansson RS, and Vallbo AB (1979) Tactile sensibility in the human hand: Relative and absolute densities offour
types of mechanoreceptive units in the glabrous skin area. 1. Physio!. (London) 286:283-300.
Jouffroy FK, Godinot M, and Nakano Y (1991) Biometrical characteristics of primate hands. Hum. Evol.
6:269--306.
Kaas JH, Hall WC, and Diamond IT (1972) Visual cortex of the grey squirrel (Sciurus carolinensis): Architectonic
subdivisions and connections from the visual thalamus. J. Compo Neurol. 145:273-305.
Koprowski JL (1994) Scillrlls carolinensis. Mammal. Species 480: 1-9.
Krause W (1860) Die terminalen Koperchen der einfach sensiblen Nerven. Hanover: Hahn.
Le Gros Clark WE (1959) The Antecedents of Man. Edinburgh: Edinburgh University Press.
Le Gros Clark WE (1963) History of the Primates. An Introduction to the Study of Fossil Man. 4th ed. Chicago:
The University of Chicago Press.
Lemelin P (1996a) Relationships between hand morphology and feeding strategies in small-bodied prosimians.
Am. J. Phys. Anthropol. Suppl. 22: 148 (abstract).
Lemelin P (1996b) The Evolution of Manual Prehensility in Primates: A Comparative Study of Prosimians and
Didelphid Marsupials. Ph.D. Dissertation, State University of New York at Stony Brook.
Martin RD (1986) Primates: A definition. In B Wood, L Martin, and P Andrews (eds.): Major Topics in Primate
and Human Evolution. Cambridge: Cambridge University Press, pp. 1-31.
McClearn D (1992) Locomotion, posture, and feeding of kinkajous, coatis, and raccoons. J. Mamm. 73:245--261.
Mittermeier RA, and van Roosmalen MGM (1981) Preliminary observations on habitat utilization and diet in eight
Surinam monkeys. Folia Primatol. 36:1-39.
Munger BL, and Ide C (1988) The structure and function of cutaneous sensory receptors. Arch. Histol. Cytol.
51:1-34.
Napier JR (1956) The prehensile movements of the human hand. J. Bone Jt. Surg. 38B:902-913.
Napier JR (1961) Prehensility and opposability in the hands of primates. Symp. Zool. Soc. Lond. 5: 115--132.
Napier JR (1993) Hands (Revised ed.). Princeton: Princeton University Press.
Nelson H, Jurmain R, and Kilgore L (1992) Essentials of Physical Anthropology. St. Paul: West Publishing Co.
Niemitz C (1984) Synecological relationships and feeding behavior of the genus Tarsius. In C Niemitz (ed.): Biol-
ogy of Tarsiers. Stuttgart: Gustav Fischer Verlag, pp. 59--76.
144 P. Lemelin and B. W. Grafton

Polyak S (1957) The Vertebrate Visual System. Chicago: The University of Chicago Press.
Quilliam TA, and Ridley A (1971) The human Meissner corpuscle. J. Anat. 109:338-339 (abstract).
Rasmussen DT (1990) Primate origins: Lessons from a neotropical marsupial. Am. J. Primatol. 22:263-277.
Rosenberger AL (1977) Xellothrix and ceboid phylogeny. 1. Hum. Evo!. 6:461-481.
Ross C (1995) Allometric and functional influences on primate orbit orientation and the origins of the Anthro-
poidea. J. Hum. Evol. 29:201-227.
Ross C (1996) Adaptive explanation for the origins of the Anthropoidea (Primates). Am. J. Primatol. 40:205-230.
Rothe H (1971) Some remarks on the spontaneous use of the hand in the common marmoset (Callithrixjacchus).
Proc. 3rd Int. Congr. Primat., ZUrich (1970) 3:136-141. Basel: Karger.
Rylands AB (1989) Sympatric Brazilian callitrichids: The black tufted-ear marmoset, Callithrix kuhli, and the
golden-headed lion tamarin, Leolltopithecus chrysomelas. 1. Hum. Evol. 18:679-695.
Servos P, Goodale MA, and Jakobson LS (1992) The role of binocular vision in prehension: A kinematic analysis.
Vision Res. 32:1513-1521.
Snowdon CT, and Soini P (1988) The tamarins, genus Saguillus. In RA Mittermeier, AB Rylands, A Coimbra-
Filho, and GAB Fonseca (eds.): Ecology and Behavior of Neotropical Primates, Vol. 2. Washington, D.C.:
World Wildlife Fund, pp. 223- 298.
Soini P (1987) Ecology of the saddle-back tamarin Saguillus luscicollis illigeri on the Rio Pacaya, northeastern
Peru. Folia Primatol. 39:11-32.
Sussman RW, and Kinzey WG (1984) The ecological role of the Callitrichidae: A review. Am. J. Phys. Anthropol.
64:419-449.
Terborgh J (1983) Five New World Primates. Princeton: Princeton University Press.
Thorington RW, Jr. (1968) Observations of the tamarin Saguillus midas. Folia Primatol. 9:95-98.
Thorndike EE (1968) A microscopic study of the marmoset claw and nail. Am. 1. Phys. Anthropol. 28:247-262.
Vallbo AB and Johansson RS (1978) Tactile sensory innervation of the glabrous skin of the human hand. In G
Gordon (ed.): Active Touch: The Mechanism of Recognition of Objects by Manipulation. Oxford: Per-
gamon Press, pp. 29-54.
Winkelmann RK (1963) Nerve ending in the skin of primates. In J Buettner-Janusch (ed.): Evolutionary and Ge-
netic Biology of Primates, Vol. I. New York: Academic Press, pp. 229-259.
Winkelmann RK (1964) Nerve endings of the North American opossum (Didelphis virginiana): A comparison
with nerve endings of primates. Am. J. Phys. Anthropol. 22:253-258.
Winkelmann RK (1965) Innervation of the skin: Notes on a comparison of primate and marsupial nerve endings.
In AG Lyne and BF Short (eds.): Biology of the Skin and Hair Growth. New York: American Elsevier, pp.
171-182.
Wood Jones F (1916) Arboreal Man. London: Edward Arnold.
 9

TAIL-ASSISTED HIND LIMB SUSPENSION AS A

TRANSITIONAL BEHAVIOR IN THE
EVOLUTION OF THE PLATYRRHINE
PREHENSILE TAIL

D. Jeffrey Meldrum

Departments of Biological Sciences and Anthropology

Idaho State University
Pocatello, Idaho 83209

1. INTRODUCTION

The ate lines (Ateles, Lagothrix, Brachyteles. and Alouatta) are distinguished among
the New World primates by the presence of a prehensile tail, equipped with a naked volar
pad covered with dermatoglyphic friction skin (Geoffroy Saint-Hilaire, 1829). This adap-
tation plays a significant role in the definition of the feeding and locomotor niche of the
atelines (Rosenberger and Strier, 1989). Atelines exhibit modifications of the sacral and
caudal vertebrae (Ankel, 1972; German, 1982), caudal musculature (Lemelin, 1995) and
cerebral cortical representation of the tail (Falk, 1980). The capuchin monkey (Cebus)
also displays prehensile abilities in its relatively shorter tail, but lacks the volar pad and
other distinctive caudal morphologies present in the atelines, suggesting prehensile tails
evolved in parallel in Cebus and the ate lines (Rosenberger, 1983; Lemelin, 1995).
Several explanations have been offered as proximate causes for the evolution of pre-
hensile tails in the platyrrhine primates. These include: to accommodate increased body size
(e.g., Napier, 1967; Grand, 1972), to exploit a frugivorous diet (e.g., Mittermeier and
Fleagle, 1976; Cant, 1977), to reduce distance and time traveling between feeding patches
(Milton, 1984; Cant, 1986), to enhance the ability to feed in terminal branches (Grand,
1972, 1984; Bergeson, 1996), and to utilize fragile forest structures (e.g., Emmons and Gen-
try, 1983; Christoffer, 1987). As with most evolutionary explanations, it seems most reason-
able that a combination of these factors is likely responsible, since no one is completely
unique to the atelines or Cebus, or for that matter, other prehensile-tailed mammals.
In this paper I offer a hypothesis proposing a mechanism for the evolution of pre-
hensile tails in some New World monkeys. The mechanism is certainly not independent of

145
146 D. J. Meldrum

or exclusive of the explanations enumerated above, but entails a transitional behavior that
may have served as a source for the key innovation of tail prehension, within the context
of the influences of body size, support characteristics, foraging strategies, etc. This hy-
pothesis springs from observations of tail-assisted hind limb suspension in atelines and
other primates not generally considered as possessing a prehensile tail.
Before exploring this hypothesis directly, I will discuss the generally accepted and
somewhat problematic definition of prehensile tails. Next, the distribution of prehensile
abilities of the tail will be briefly reviewed. Then, the associated behavior of hind limb
suspension and foot reversal will be examined and their correlated morphologies dis-
cussed. Finally, the proposed mechanism of tail-assisted hind limb suspension as a transi-
tional behavior in the evolution of the ateline prehensile tail will be considered together
with analogies drawn from non-primate taxa.

2. DEFINITION OF "PREHENSILE"

It is necessary to consider the definition and characterization of a prehensile tail be-

fore exploring this hypothesis further. Emmons and Gentry (1983) have suggested a rather
circumscribed definition that is generally adopted by investigators:

A prehensile tail is one which can support alone the weight of the suspended body; semi-pre-
hensile tails can be wrapped around branches and support a significant part, but not all, of the
body weight (p. 513).

This definition places primary emphasis on the ability of the "prehensile" tail alone
to support body weight. It is implied in this definition that "semi-prehensile" tails are used
in conjunction with other appendages to support the suspended body weight.
A singular emphasis on body weight support, however, may have resulted in many
prehensile qualities of primate tails being overlooked or trivialized. According to Web-
ster's New Universal Unabridged Dictionary, the adjective prehensile conveys the quality
of seizing or grasping, "as the prehensile tail of some monkeys" (McKechnie, 1983). It is
this connotation of grasping ability that was given preeminence in earlier discussions of
New World primate prehensile tails. Consider, for example, Elliot's (1913) description of
Ateles' prehensile tail:

The tail is unsurpassed, if not unequaled, in its flexibility, always in motion, the tip as sensi-
tive as that of the elephant's trunk, grasping with an unshakable firmness anything and every-
thing it may touch, and fulfilling in the highest degree ... the purposes of a fifth hand. By it,
fruits or other desirable objects otherwise unattainable are seized and brought within reach of
the mouth or hands, and it also can hold its possessor suspended in the air... " (pp. 21-22)

This more encompassing characterization takes into account the grasping and ma-
nipulative capabilities of primate tails that have attained the greatest development in spi-
der monkeys and other atelines. It is the dexterity and sureness of grip endowed by the
modifications to the distal end of the ateline tail combined with the increased robusticity
that permits body weight to be supported unassisted. These manipUlative abilities, if not
the ability to support total body weight, however, are also present to a lesser degree in
many other primate tails, some considered "semi-prehensile," as well as many generally
considered non-prehensile. This more general connotation of prehensile will be employed
throughout the remainder of this paper.
Tail-Assisted Hind Limb Suspension as a Transitional Behavior 147

3. DISTRIBUTION OF PREHENSILE TAILS

Prehensile capabilities, in the sense of seizing or grasping, have been noted in the
tails of primates typically considered as having non-prehensile tails. Early examples of
such field observations include tail use in such catarrhine primates as Cercocebus al-
bigena (Allen, 1925; Haddow, 1952; Tappen, 1960). Jones and Sabater Pi (1968) observed
that C. albigena wraps the tip of the tail around branches, in a prehensile fashion, when at
rest. Rose (1974) tabulated reports of the occurrence of prehensile tail use in a number of
cercopithecoid as well as platyrrhine taxa generally considered non-prehensile tailed.
These include species of Old World monkeys such as guenons, the already mentioned
mangabeys, and macaques, and among the New World primates, Aotus, Callicebus,
Saimiri, and the callitrichines (see Rose, 1974, Table 2, for specific references).
Observations of behavior by captive primates further reveals a diversity of prehen-
sile capabilities. For example, Dandelot (1956) described the tail of captive Cercopithecus
lhoesti as "semi-prehensile," noting its crooked tip and frequent twining about supports
(Figure 1). Karrer (1970) also referred to the tail of captive Macaca irus (=.fascicularis) as
semi-prehensile and described its employment in food and object retrieval and social in-
teraction. Erwin (1974) also noted similar but less dexterous manipulations by captive
Macaca mulatta.

Figure I. The tail of L'hoesti's monkey (Cercopithecus Ihoesti) engaged in various prehensile activities (from
Dandelot, 1956).
 148 D. J. Meldrum

I queried primate keepers and field researchers via the Internet Bulletin Board, Pri-
mate-Talk, and received numerous responses relating observations of prehensile tail use by
cercopithecoid monkeys. For example, KB Swartz (pers. comm.) related this observation
of M. fascicularis:

They would stick their tails through the grid in the floor of the cage and use their tails to re-
trieve dropped bits of food. I first saw it when they retrieved Froot-Loops that I was using to
adapt them to my presence. They would curl the top 113 of the tail around the Froot-Loop and
pick it up and bring it into the cage.

Also, reminiscent of Dandelot (1956) are observations by J Moore (pers. comm.):

C. lhoesti at the San Diego Zoo appear to use tail semi-prehensile, as balance or brace, regu-
larly. Tail is carried in crooked position, which remains under anesthesia (but is not skeletal, it
can be bent out).

Finally, from CA Bramblett (pers. comm.):

Infant vervet and Sykes monkeys have relatively prehensile tails. They wrap the infant's tail
around mother's tail when the infant is clinging ventrally to provide additional attachment to
the mother. If you hold a guenon infant in your hand, the infant's tail will wrap with a substan-
tial grip around your arm.

This last observation is reminiscent of a published figure of an infant squirrel mon-

key (Saimiri sciureus) suspended by its tail from its keeper's finger (Rosenblum and Coo-
per, 1969). Furthermore, it highlights the fact that some infant monkeys are capable of
greater prehensile tail use than are the adults of the same species, such as some colobines
and baboons (Rose, 1974).

4. ASSOCIATED POSITIONAL BEHAVIORS AND MORPHOLOGIES

Rose (1974) made particular note of the distinction between tail use during postural
behaviors and those employed during locomotor behaviors. The use of the tail in stereo-
typed "non-prehensile-tailed" primates was generally associated with postural activities.
Prehensile tail use during locomotor activities is largely restricted to the atelines, and es-
pecially the spider monkey, in which the tail is used in concert with the arms during
brachiation, and displays specializations for hyperextension associated with an orthograde
suspensory posture (Lemelin, 1995). The atelines also frequently use the prehensile tail in
postural activities, including hindlimb suspension (e.g., Carpenter and Durham, 1969;
Mendel, 1976; Mittermeier, 1978; Fleagle and Mittermeier, 1980; Grand, 1984; Schon
Ybarra, 1984; Cant, 1986; Fontaine, 1990).
During personal observations of positional behavior of the Malagasy lemurs housed
at the Duke Primate Center, I became immediately impressed by the habit of many of the
lemurs to posture suspended from their hind limbs. Similar observations have been made
in the wild (Meldrum et aI., 1997). Some taxa were particularly adept at this, including the
acrobatic sifak, Propithecus, the black lemur, Eulemur macaco, and the ruffed lemur,
Varecia. This behavior was employed on horizontal supports of various diameters, includ-
ing the small-branch milieu. It was also adopted when descending large trunks head first.
Further discussion of this behavior and its anatomical correlates, as an alternative strategy
Tail-Assisted Hind Limb Suspension as a Transitional Behavior 149

Figure 2. A comparison of the hindlimb-suspension and tail-bracing postures of the bearded saki (Chiropotes sa-
tanas) [redrawn from van Roosmalen et al. (1981») and the red ruffed lemur (Varecia variegata rubra) [redrawn
from a photograph by the author).

for expanding the feeding kinesphere, are presented elsewhere (Meldrum et aI., 1997). In
association with hind limb suspension in Varecia, the tail was frequently draped over the
support during hind limb suspension (Figure 2). It was not simply acting as a counterbal-
ance, but was forcefully braced against the superior surface of the horizontal support, or
braced against the side of a trunk or over adjacent supports when descending a vertical
trunk head first.
Hind limb suspension is a postural behavior employed to varying degrees by a
number of primate and non-primate mammals. The skeletal correlates of this behavior
have been examined in a number of taxa (e.g., Jenkins and McCleam, 1984; Meldrum et
aI., 1997). In a study of the locomotor behavior and anatomy of two platyrrhine primates,
the sympatric sakis, Pithecia and Chiropotes, a number of these skeletal correlates were
observed in Chiropotes (Fleagle and Meldrum, 1988). Specific contrasts were made be-
tween their respective tarsal elements. Features distinguishing Chiropotes from Pithecia
are associated with the enhancement of plantartlexion and supination and are comparable
to those cited by Jenkins and McCleam (1984) for hindfoot reversal and/or hind limb sus-
pension in a number of mammalian taxa. On the basis of the anatomical correlates and an-
ecdotal field observations (van Roosmalen et aI., 1981), Fleagle and Meldrum (1988)
suggested that hind limb suspension may constitute an important component of Chiropotes
 150 D. J. Meldrum

positional behavior. This prediction has since been borne out through field observations by
W Kinzey (pers. comm.) and Walker (1994, 1997). Just as observed in Varecia, a remark-
ably similar use of the tail during hind limb suspension was noted and figured by van
Roosmalen et al. (1981) in Chiropotes (Figure 2). They made particular note, based on
brief field observations by Mittermeier (1977), that hind limb suspensory postures by Chi-
ropotes significantly increased the feeding sphere, in spite of the absence of a prehensile
tail.
Ankel (1962, 1963) and German (1982) have identified a number of distinctive
skeletal features of the ateline prehensile tail including large sacral foramina and neural
canal, an initial caudal segment comprising additional shorter elements, a longest caudal
element situated more distally in the tail, and distal vertebrae that are shortened and flat-
tened. Meldrum and Lemelin (1991) have observed that the shortening of the distal seg-
ment elements may be present to a lesser extent in some taxa that use their tails in a
semi-prehensile fashion during postural behaviors. This is suggested in a contrast between
the caudal vertebrae of Pithecia and Chiropotes. Figure 3 illustrates a plot of the ratio of
proximal width/length of individual caudal vertebrae for a small sample of extant and fos-
sil platyrrhines. It demostrates the relative shortening of the distal elements in the tail of
Chiropotes as compared to Pithecia. Also noteworthy is an inflexion in the curve for Chi-
ropotes (n=2), caused by a relatively shorter single caudal segment roughly two-thirds

1 V
0.9
j
!
j
r
-+- Ateles

---
0.8
Chiropotes

-e- Pithecia
0.7
-.-
l
Cebupithecia
lJ
...J
r'.\ A'
~
0.6
a.. l/ y .AI'
0.5 - \ .... V
"\

iA
,/
'\ ~
I\.. ~
1\I\. ....... ~
~

...
0.4
.... rs.. ~ ~ -....
;-

I ~ ....
1\ l--'
... .....J-
~
..... ....,
--
0.3 ~ '- ~
1\ X'
~ rs- --A. .JIll.
r...
...E)

IY ~
~
0.2
7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31

Caudal Vertebra #
Figure 3. Graph illustrating the relationship between the proximal width and length of individual caudal elements
for selected platyrrhine primates (n=2 for each taxon).
Tail-Assisted Hind Limb Suspension as a Transitional Behavior 151

down the length of the tail. It suggests a correspondence to the flexion point of the tail as
it is braced over the support during hind limb suspension. A similar but less marked inflex-
ion is present in the much more robust tail of Ateles (n=2) at about the same point. Further
investigation is needed to establish the functional correlation of this flexion point and tail
draping behavior in these and other taxa.
The holotype of the fossil pitheciine Cebupithecia preserves much of the tail and
portions of the tarsals. This provides the opportunity to further examine the relationship
between skeletal correlates of hind limb suspension and tail use in an extinct platyrrhine
with a relatively long tail. The tarsals of Cebupithecia lack the traits correlated with hind
limb suspension. Cebupithecia had a less robust tail, and although the distal elements are
incomplete, the overall similarity to those of Pithecia is evident (Meldrum and Lemelin,
1991 ).
I have also noted similar tail use in some cercopithecoids under semi-natural condi-
tions (Meldrum, 1989). Examination of cinefilms made by Rollinson (197S) revealed ex-
amples of tail twining and bracing during hind limb suspension or other below-branch
reaching activities in guenon species, such as Cercopithecus cephus. For example, when
attempting to reach fruits hanging from a small diameter horizontal support, a mustached
monkey suspended from its hind limbs while bracing with its tail (Figure 4a). As the hind
limbs became more extended and the foot more supinated, the prehensile twining of the
tail about the support became more pronounced (Figure 4b).
In a second example, a mustached monkey with body oriented perpendicular to a
horizontal support of moderate diameter, leaned out over the side of the support to investi-
gate something. Its tail was draped over an adjacent support in a manner very similar to
that observed in Varecia and Chiropotes (Figure Sa). As the monkey leaned farther out to
the side of the primary support, displacing its center of gravity, the tail twined more pre-
hensily around the secondary support (Figure Sb) until finally, the very tip of the tail was
curled tightly about the support and the tensed tail was noticeably arched indicating force-
ful flexion (Figure Sc).

5. HYPOTHETICAL TRANSITIONAL BEHAVIOR

Clearly, a common utilization of the primate tail is correlated with positional behav-
iors that place the center of gravity other than directly above the arboreal support. These
tail uses range from a non-contacting passive counterbalance, to a relatively active flexing
brace, to very active modified prehensile grip. Such utilization of the tail is present in·
nearly all long-tailed primates to some degree, and limited prehensile use of the tail is
more widespread than is often recognized. From these observations I propose the follow-
ing hypothesis: that hind limb suspension, assisted by tail-bracing and/or twining, served
as a transitional positional behavior in the evolution of the highly derived platyrrhine pre-
hensile tail (Figure 6).
Indirect corroboration of this hypothesis is to be found in the proposals by Emmons
and Gentry (1983) and Christoffer (1987) that the fragile vegetation and limited Iianas of
the Neotropics, by comparison to the Paleotropics, played a role in selection for and
elaboration of prehensile tails. It is noteworthy that the evolution of prehensile tails has
occurred independently in a number of neotropical mammalian families in addition to the
ate lines, including numerous didelphids, procyonids, mermecophagids, erethizontids, as
well as several reptiles and amphibians. It is unlikely, however, that the tail serves identi-
cle behavioral roles in all of these taxa.
152 D. J. Meldrum

Figure 4. A mustached guenon (Cercopithecus cephus) reaching for fruit suspended below a horizontal support.
See text for further discussion.

Of further significance is that parallels can be drawn to the distribution of prehen-

sile-tailed mammals, including phalangerids, petaurids, burramyids and tarsipedids (Ber-
geson, 1996), and correlated forest types in Australia. Of particular significance to the
hypothesis of tail-assisted hind limb suspension is the presence of a robust divergent hal-
lux among the marsupials of both Australia and South America, associated with hind limb
grasping/suspension and hind foot reversal. The convergent functional morphology of
hindfoot reversal in prehensile-tailed mammals such as the neotropical procyonid, Patos,
and various didelphids, has been described by Jenkins and McCleam (1984). The combi-
nation of grasping halluces and prehensile tails in non-primate taxa would appear to lend
support to the hypothesis advocated here.
Tail-Assisted Hind Limb Suspension as a Transitional Behavior 153

Figure 5. A mustached guenon (Cercopithecus cephus) leaning to the side and below a large horizontal support.
See text for futher discussion.
154 D. J. Meldrum

Figure 6. Representation of strategies for utilization ofthe feeding kinesphere, and the hypothetical role oftail-as-
sisted hindlimb suspension as a transitional behavior in the evolution of the platyrrhine prehensile tail.

6. CONCLUSIONS
In conclusion, it is proposed that: (1) restricted body-weight-support definitions of
prehensile tails may overlook many prehensile capabilities and functions of the general-
ized primate tail; (2) tail bracing and twining, in conjunction with hind limb suspension
selected for greater caudal robusticity and prehensile capabilities in some primate line-
ages, especially ancestral ate lines and Cebus; and (3) prehensile tails were initially a pos-
tural adaptation that, with the advent of the specialized volar pad of the ateline tail,
permitted the tail to acheive a sufficient grip to support body weight unassisted. Only later
did it evolve as an elaborated locomotor adaptation in the spider monkeys.

ACKNOWLEDGMENTS

I would like to thank the organizers of this Conference and acknowledge the numer-
ous discussions with many individuals, which spawned these ideas and encouraged their
formalization, and the anonymous reviewers who helped to focus their presentation.
Tail-Assisted Hind Limb Suspension as a Transitional Behavior 155

REFERENCES
Allen JA (1925) Primates collected by the American Museum Congo expedition. Bull. Am. Mus. Nat. Hist. 47:
283-499.
Ankel F (1962) Vergleichende Untersuchungen ueber die skelettmorphologie des Greifschwanzes suedamerikanis-
cher affen (Platyrrhina). Z. Morph. Oekol. Tiere 52: 131-170.
Ankel F (1963) Zur Morphologie des Greifschwanzes bei suedamerikanishen Affen. Z. Morph. Anthropol.
53:12-18.
Ankel F (1972) Vertebral morphology offossil and extant primates. [n RH Tuttle (ed.): Functional and Evolution-
ary Biology of Primates. Chicago: Aldine Press, pp. 223-240.
Bergeson DJ (1996) Positional behavior and prehensile tail use of Alouatta palliata, Ateles geojJroyi and Cebus
capucinus. Ph.D. Dissertation, Washington University.
Cant JGH (1977) Ecology, locomotion and social organization of spider monkeys (Ateles geoffi·oyi). Ph.D. Disser-
tation, University of California, Davis.
Cant JGH (1986) Locomotion and feeding postures of spider and howling monkeys: Field study and evolutionary
interpretation. Folia Primatol. 46: 1-14.
Carpenter CR, and Durham NM (1969) A preliminary description of suspensory behavior in primates. [n HO
Hofer (ed.): Recent Advances in Primatology. Proc. 2nd IntI. Congr. Primatol. 2: 147-154.
Christoffer C (1987) Body size differences between New World and Old World, arboreal tropical vertebrates:
Cause and consequences. J. Biogeography. 14:165-172.
Dandelot P (1956) Note sur Ie comportment de deux Cercopitheques de I'Hoest en captivite. Mammalia
20:330--331.
Elliot DG (1913) A Review of the Primates. Vol II. New York: American Museum of Natural History.
Emmons LH, and Gentry AH (1983) Tropical forest structure and the distribution of gliding and prehensile-tailed
vertebrates. Am. Nat. 121:513-524.
Falk D (1980) Comparative study of the endocranial casts of New and Old World monkeys. [n RL Ciochon and
AB Chiarelli (eds.): Evolutionary Biology of the New World Monkeys and Continental Drift. New York:
Plenum Press, pp. 275-292.
Fleagle JG, and Meldrum DJ (1988) Locomotor behavior and skeletal morphology of two sympatric pitheciine
monkeys, Pithecia pithecia and Chiropotes satan us. Am. J. Primatol. 16:227-249.
Fleagle JG, and Mittermier RA (1980) Locomotor behavior, body size, and comparative ecology of seven Surinam
monkeys. Am. J. Phys. Anthropol. 52:301-314.
Fontaine R (1990) Positional behavior in Saimiri boliviensis and Ateles geojJroyi. Am. J. Phys. Anthropol.
82:485-508.
Geoffroy Saint-Hilaire I (1829) Remarques sur les caracteres generaux singes americains, et descrition d'un genre
nouveau, sous Ie nom d'Eriode. Mem. Mus. Hist. Nat. (Paris) 17:121-165.
German RZ (1982) The functional morphology of caudal vertebrae in New World monkeys. Am. J. Phys. Anthro-
pol. 58:453--459.
Grand TI (1972) A mechanical interpretation of terminal branch feeding. J. Mammal. 53: 198-201.
Grand TI (1978) Adaptations of tissue and limb segments to facilitate moving and feeding in arboreal folivores. [n
GG Montgomery (ed.): The Ecology of Arboreal Folivores. Washington, DC: Smithsonian Institution
Press, pp. 231-242.
Grand TI (1984) Motion economy within the canopy: Four strategies for mobility. In PS Rodman and JGH Cant
(eds.): Adaptations for Foraging in Nonhuman Primates. New York: Columbia University Press, pp. 54-72.
Haddow AJ (1952) Field and Laboratory studies on an African monkey, Cercopithecus ascanias schmidti Mat-
schie. Proc. Zool. Soc. Lond. 122:297-394.
Jenkins FA Jr, and McLearn D (1984) Mechanisms of hind foot reversal in climbing mammals. J. Morph.
182:197-219.
Jones C, and Sabater Pi J (1968) Comparative ecology of Cercocebus albigena (Gray) and Cercocebus torquatus
(Kerr) in Rio Muni, West Africa. Folia Primatol. 9:99-113.
Lemelin P (1995) Comparative and functional myology of the prehensile tail in New World monkeys. J. Morph.
224:351-368.
McKechnie JL, ed. (1983) Webster's New Universal Unabridged Dictionary, 2nd Edition. New York: Simon and
Schuster.
Meldrum DJ (1989) Terrestrial adaptations in the feet of African cercopithecines. Ph.D. Dissertation, State Univer-
sity of New York at Stony Brook.
Meldrum DJ, and Lemelin P (1991) Axial skeleton ofCebupithecia sarmientoi (Pitheciinae, Platyrrhini) from the
middle Miocene of La Venta, Colombia. Am. J. Primatol. 25:69-89.
156 D. J. Meldrum

Meldrum OJ, Dagosto M, and White J (1997) Hindlimb suspension and foot reversal: An adaptation for increased
feeding kinesphere in primates and other arboreal mammals. Am. J. Phys. Anthropol. 102:85-102.
Mendel F (1976) Postural and locomotor behavior of Alouatta palliata on various substrates. Folia Primatol.
36:1-19.
Mittermeier RA (1977) Distribution, synecology and conservation of Surinam monkeys. Ph.D. Dissertation, Har-
vard University.
Mittermeier RA (1978) Locomotion and posture in Ateles geojJroyi and Ateles paniscus. Folia Primatol.
30:161-193.
Napier JR, and Napier PH (1967) A Handbook of Living Primates. New York: Academic Press.
Norconk MA, and Kinzey WG (1994) Challenge of neotropical frugivory: Travel patterns of spider monkeys and
bearded sakis. Am. 1. Primtol. 34: 171-183.
Rollinson J (1975) Interspecific comparisons of locomotor behavior and prehension in eight species of African for-
est monkey: A functional and evolutionary study. Ph.D. Dissertation, University of London.
Rose MD (1974) Postural adaptations in New and Old World monkeys. In FA Jenkins Jr (ed.): Primate Locomo-
tion. New York: Academic Press, pp. 201-222.
Rosenberger AL (1983) Tale oftails: Parallelism and prehensility. Am. J. Phys. Anthropol. 60: I 03-1 07.
Rosenberger AL, and Strier KB (1989) Adaptive radiation of the ateline primates. 1. Hum. Evol. 18:717-750.
Rosenblum LA, and Cooper RW, eds. (1969) The Squirrel Monkey. New York: Academic Press.
Schon Ybarra MA (1984) Locomotion and postures of red howlers in a deciduous forest-savanna interface. Am. J.
Phys. Anthropol. 63:65-76.
Tappen N (1960) Problems of distribution and adaptation of the African monkeys. Curro Anthropol. 1:91-120.
van Roosmalen MGM, Mittermeier RA, and Milton K (1981) The bearded sakis, genus Chiropotes. In AF Coim-
bra-Filho and RA Mittermeier (eds.): Ecology and Behavior of Neotropical Primates, Vol. 1. Rio de Ja-
neiro: Academia Brasileira de Ciencias, pp. 419-441.
Walker SE (1994) Positional behavior and habitat use in Chiropotes satanas and Pithecia pithecia. In B Thierry,
JR Anderson, JJ Roeder, and N Herrenschmidt (eds.): Current Primatology, Vol. I: Ecology and Evolution.
Proc. XIV IPS Congress. Strasbourg: Universite Louis Pasteur, pp. 195-201.
Walker SE (1996) The evolution of positional behavior in the saki/uakaris (Pithecia, Chiropotes and Cacajao). In
M Norconk, A Rosenberger, and P Garber (eds.): Advances in Primatology: Adaptive Radiations of Neot-
ropical Primates. New York: Plenum, pp. 335-363.
 10

UNIQUE ASPECTS OF QUADRUPEDAL

LOCOMOTION IN NONHUMAN PRIMATES

Susan G. Larson

Department of Anatomical Sciences

School of Medicine
State University of New York at Stony Brook
Stony Brook, New York 11794-8081

1. INTRODUCTION

Quadrupedal walking and running are certainly not the first things that come to mind
when one considers unique aspects of primate locomotion. However, there is a growing
body of information about how the form of quadrupedalism displayed by primates differs
from that of non primate mammals (see Vilensky, 1987, 1989). One of the most distinctive
characteristics of primate quadrupedalism is that they typically utilize a diagonal se-
quence/diagonal couplets walking gait pattern (i.e., foot falls in sequence: left hind, right
fore, right hind, left fore, with diagonal limbs moving as a pair), in contrast to the almost
universally employed lateral sequence walking gait (left hind, left fore, right hind, right
fore) of nonprimate mammals (Howell, 1944; Prost, 1965, 1969; Hildebrand, 1967; Rol-
linson and Martin, 1981; Vilensky, 1989; Vilensky and Larson, 1989). This difference in
gait pattern is not trivial, since a diagonal sequence/diagonal couplet walking gait creates
a strong potential for interference between the ipsilateral hind and forelimbs (Figure 1).
The potential for hind/forelimb interference is exacerbated in primates by their long limbs
(due to their relatively longer limb bones, Alexander et aI., 1979), and by their propensity
to use relatively longer stride lengths than nonprimate quadrupeds (Vilensky, 1980; Alex-
ander and Maloiy, 1984; Reynolds, 1987). As a result, many primate quadrupeds must
regularly "overstride" during walking, that is, touch down with their hind foot ahead of
their ipsilateral hand by passing it either "inside" or "outside" of the forelimb (Hildebrand,
1967; Reynolds, 1985b; Larson and Stern, 1987; see Figure 1). Another distinctive aspect
of primate gait utilization is the infrequent use of a running trot (defined as diagonal limbs
moving synchronously with relative stance duration of each limb less than 50%; see
Hildebrand, 1967). Primates generally progress directly from a walk to a gallop, and even

157
 158 S. G. Larson

Forelimb Outside

Forelimb Inside

Figure 1. Side view of a chimpanzee knuckle-walking. Figures are drawn from videotape records. The chimpan-
zee in the upper row is walking with its right forelimb outside the overstriding right hind limb, while the chimpan-
zee in the lower row is walking with its right forelimb inside the overstriding right hind limb.

when they do run·, they do not trot (Hildebrand, 1967; Vilensky et aI., 1988). Finally, pri-
mates have been shown to rely more on their hind limbs for both support and propulsion
than do nonprimate mammals (Kimura et aI., 1979; Kimura, 1985; Demes et aI., 1992, 1994).
To date, no integrated explanation for this unique set of characteristics of primate
quadrupedal ism has been proposed. Based on the kinetic data indicating that primates sup-
port more weight on their hind limb than nonprimates, several researchers have attributed
the differences in gait patterns between primates and nonprimates to differences in body
weight distribution (Iwamoto and Tomita, 1966; Tomita, 1967; Kimura et aI., 1979,
Kimura, 1985; Rollinson and Martin, 1981). Vilensky (1989) and Vilensky and Larson
(1989), however, argue that the actual location of the center of mass in quadrupedal pri-
mates is not dissimilar to that of other mammals, and point out that animals with markedly
posteriorly placed centers of gravity, such as rabbits (some of which employ walking gaits
(Dagg, 1974)), use a lateral sequence walk.
The present study will review some additional aspects of quadrupedal locomotion in
primates, namely patterns of limb motion and of muscle recruitment, and attempt to link
these various pieces of information into some coherent whole.

2. OBSERVATIONS

2.1. Limb Kinematics

Mammalian quadrupeds can be loosely divided into two groups: cursors (adapted for
rapid walking and running; generally over 3 kg), and non-cursors (not speed adapted,

• Running in this case refers to a symmetrical gait wherein the footfalls of the two fore- and two hind feet are
equally spaced, and each foot is on the ground less than 50% of the total stride interval (Hildebrand, 1967). It
thus differs from galloping which is an asymmetrical gait in which the footfalls of the fore- and hind pairs of
limbs are not evenly spaced.
Unique Aspects of Quadrupedal Locomotion in Nonhuman Primates 159

therefore including all other quadrupedal locomotor modes; usually 3 kg or less) (Jenkins,
1971; Alexander and Maloiy, 1984). Non-cursors tend to walk with very flexed limbs, and
as Jenkins (1971) documents, their limb movements also deviate considerably from
parasagittal planes. Small mammals use large maximum limb excursion angles (the angle
the limb passes through between foot touchdown and lift-off) (McMahon, 1975, 1984)
during walking, and for the forelimb, Jenkins and Weijs (1978) and Fischer (1994) suggest
that scapular rotation and translation account for most of this motion, with relatively mi-
nor contributions by motion at the glenohumeral and elbow joints. As mammals get larger,
they stand and walk with more extended limbs that move in roughly parasagittal planes.
They move their limbs through smaller maximum excursion angles (McMahon, 1975),
and if the differences between opossums (Jenkins and Weijs, 1978), cats (Miller and Van
der Meche, 1975; English, 1978a,b) and dogs (Nomura et aI., 1966; Tokuriki, 1973) can be
taken to reflect a size-related trend, this implies that forelimb motion involves less scapu-
lar rotation. t
Quadrupedal primates cover an extremely wide size range from about 60 g to about
160 kg (Jungers, 1985), bridging the size ranges of cursors and non-cursors. Other than
lorisines, however, which have been studied quite extensively (Dykyj, 1980; Jouffroy et
aI., 1983; Jouffroy and Petter, 1990; Demes et aI., 1990; Ishida et aI., 1990), very few ki-
nematic data exist for very small or very large quadrupedal primates. More data are avail-
able on quadrupedal monkeys, in the size range of approximately 3 to 20 kg, comparable
to smaller cursors such as cats and dogs. Like cats and dogs, primates tend to walk with
extended limbs that move roughly in parasagittal planes, although arboreal primate quad-
rupeds are said to use somewhat more flexed and abducted limbs than terrestrial species
(e.g., Napier, 1967; Grand 1968a,b, 1984; Fleagle, 1977; Morbeck, 1979). Also like cur-
sors, primates display only limited scapular rotation during forelimb movement during
walking (generalizing from the vervet examined cineradiographically by Whitehead and
Larson, 1994). Significant kinematic differences in patterns of limb excursion, however,
exist between primate and nonprimate quadrupeds.
Figure 2 displays a series of schematic forelimbs (humeral plus radial/ulnar seg-
ments) at touchdown and lift-offfor a variety of mammals during a walking gait. Although
it would be desirable to include scapular position in this analysis, data exist for only a few
taxa. What information does exist suggests a similar initial position at touchdown in most
mammals including primates, with the scapular spine being approximately 45-60 degrees
to horizontal (see Figure 6). Greater differences exist in scapular position at lift-off (see
Nomura et aI., 1966; Tokuriki, 1973; Miller and Van der Meche, 1975; English, 1978;
Jenkins and Weijs, 1979; and Whitehead and Larson, 1994). The nonprimate mammalian
quadrupeds in Figure 2 display a forelimb touchdown in which the glenohumeral joint re-
mains retracted, that is, the humerus is behind a vertical line through the shoulder. The pri-
mate quadrupeds, on the other hand, begin support phase with their humerus in a
protracted position, i.e., ahead of vertical. At the end of support phase, the humerus in the
primates portrayed in Figure 2 is less retracted than that of the nonprimate quadrupeds. It
may be that primates end support phase with a less retracted forelimb as a means ofreduc-
ing some of the potential for interference between the forelimb and the ipsilateral hind
limb. If the supporting forelimb is imagined as pivoting over the wrist during a step as in

t Although it has been suggested that an aclaviculate condition in cursorial mammals is related to increased an-
teroposterior excursion of the shoulder (e.g., Eaton, 1944; Howell, 1944), Jenkins (I 974) reports that clavicular
loss mainly serves to confine shoulder motion to a parasagittal plane with no change in anteroposterior shoulder
excursion.
160 S. G. Larson

~ .(":" "---Humerus
~---AadiUS'Ulna
" ...........

Figure 2. Left forelimb postures during a

walking step in a variety of mammals (moving
.,'................. . .
from left to right). All limb segments (humerus
/Opossum and radius-ulna) are drawn the same length in
all animals. The solid lines represent the posi-
tion of the limb at touchdown, and the dashed
,..~ .. "~'''''' lines represent the limb at the end of support
phase at lift-off. The non-cursors in the top two
/ Squirrel rows display highly flexed limb postures,
... Monkey
whereas the larger cursors (dog, cat and pri-
mates) use more extended limb postures. In all
nonprimates illustrated here, the humerus is in a
retracted position at touchdown, but is pro-
.... Long Tail tracted in all of the primates. Data for the non-
: Macaque cursors are from Jenkins (1971); for the dog are
from Tokuriki (1973); for the cat are from
Miller and Van der Meche (1975); for the squir-
rel monkey are from Vilensky et al. (1994), for
the spider monkey are from Schmitt (1994), and
Patas for the remaining primates are from Schmitt
Monkey
(1995).

Figure 2, then the primate forelimb appears to pass through a more symmetrical angular
excursion than in nonprimates, that is, approximately half of the excursion occurs prior to
when the shoulder is over the wrist, and half occurs after. Most of the excursion of the
forelimb in nonprimates occurs beyond the point when the shoulder is directly over the
wrist. Unfortunately, besides the monkey species displayed in Figure 2, kinematic data on
forelimb position during walking are not available for many other nonhuman primates.
Extreme protraction of the forelimb at touchdown during walking is clearly characteristic
oflorisines (Jouffroy et aI., 1983; Jouffroy and Petter, 1990; Demes et aI., 1990; Ishida et
aI., 1990), and personal observations indicate that a protracted humerus at touchdown is
typical of all monkeys and large-bodied apes. Published figures support this conclusion
(e.g., Muybridge, 1957, Plates 142 and 143; Hildebrand, 1967, Figure 5; Grand, 1968a,
Figure 4; Rose, 1979, Figure 4-2; Larson and Stern, 1987, Figure 1; Larson and Stern,
1989b, Figure 2; Meldrum, 1991, Figures 4, 9, and 10). Lemelin (pers. com.) reports that a
protracted forelimb is also observed in other walking quadrupedal prosimians besides
lorisines.
Primate quadrupeds also begin hind limb support phase with the limb highly pro-
tracted. They end it with the hind limb equally retracted resulting in significantly greater
angular excursions at the hip during walking than what is observed in nonprimates
(Reynolds, 1987). The difference is most pronounced in suspensory/climbing adapted spe-
cies. Reynolds (1987) concludes that large angular excursion at the hip coupled with rela-
tively long limb segments (Alexander et aI., 1979) produce the relatively long hind limb
Unique Aspects of Quadrupedal Locomotion in Nonhuman Primates 161

stride lengths during walking that have been reported for quadrupedal primates (Vilensky,
1980; Alexander and Maloiy, 1984; Reynolds, 1987).
It is unclear whether the more protracted forelimb position observed during a walk-
ing step in primate quadrupeds is also associated with increased forelimb angular excur-
sion. The average angular excursion for the primates displayed in Figure 2 (65 degrees) is
similar to the average for the non-cursors (66 degrees), and a little higher than the average
for the cat and dog (57 degrees). McMahon (1975, 1984) has described a gradual inverse
relation between body size and maximum limb excursion (11 M .. 10), which would predict
that the larger bodied primates should display relatively smaller limb excursion angles
than the non-cursors. The relation described by McMahon, however, is for animals travel-
ing at the speed at which the trot/gallop transition occurs, and all of the animals portrayed
in Figure 2 are traveling at much slower speeds. Reports vary on how angular excursion
varies with speed (see Vilensky, 1987), making it impossible at present to determine if pri-
mate forelimbs display the same increase in angular excursion as documented for primate
hind limbs.
Nonetheless, the position of the primate humerus during a walking step can be re-
lated to a long forelimb stride length. Throughout support phase, the glenohumeral joint in
primates is more protracted than in nonprimates, meaning that the joint is more open (ob-
tuse angle between the scapula and humerus) thus increasing the effective length of the
limb. This augments the already elongated forelimbs of primates due to relatively longer
limb bones (Alexander et ai., 1979), making an overall increase in limb length mainly re-
sponsible for the relatively long forelimb strides of primates.

2.2. Muscle Activity Patterns

Based on observations of electromyographic (EMG) activity of homologous limb
muscles during walking in different animals, Goslow and coworkers have suggested that
muscle recruitment patterns in vertebrates are quite conservative (Jenkins and Goslow,
1983; Peters and Goslow, 1983; and Goslow et ai., 1989). Goslow et al. (1989) illustrated
the point by contrasting the patterns of shoulder muscle recruitment in a lizard and an
opossum, two rather dissimilar vertebrates (see Figure 3). Despite differences in limb ori-
entation and shoulder structure, four major muscle groups showed similar patterns of ac-
tivity: the latissimus dorsi and the pectoralis muscles acted primarily during support
phase, the deltoids acted during swing phase, and the supracoracoideus - supraspinatus/in-
fraspinatus homologue were biphasic. This result was especially surprising for the supra-
coracoideus - supraspinatus/infraspinatus homologue since they are configured quite
differently as Figure 3 shows. Goslow and colleagues also observed certain similarities to
patterns of shoulder muscle recruitment of birds (Dial et al., 1987, 1991), which led them
to propose what may be called the neuromuscular conservation hypothesis: Motor patterns
of homologous muscles have been maintained during the evolution of tetrapods, and a
primitive organization of the neural control components has persisted in derived groups
despite differences in morphology (Goslow et al., 1989).
EMG data suggest that the neuromuscular conservation hypothesis may indeed ap-
ply to the primate hind limb. Okada et al. (1978) reported that the recruitment pattern of
the vastus lateralis, biceps femoris, gastrocnemius and tibialis anterior in a Japanese ma-
caque and a spider monkey during walking were similar to data reported for cats and dogs
except for small timing differences, and Kimura et al. (1979) obtained similar results on
the same four muscles in a baboon and chimpanzee. Jungers et al. (1980) undertook a de-
tailed study of quadriceps femoris activity in Lemur fulvus, patas and woolly monkeys,
 162 S. G. Larson

infraspinatus

REPTILE MAMMAL
(Varanus exanthematicus) (Didelphis virginiana)
.. Support .. Swing .. .. Support • Swing ..

--
Latissimus

Pectoralis

-
Deltoids

Supracoracoideus • • • • •
(Infraspinatus)
(Supraspinatus)

Figure 3. Electromyographic activity of homologous muscles in the Savannah Monitor lizard and opossum during
walking. Bars represent the most consistent activity. In both animals, the latissimus dorsi and pectoralis muscles
are active mainly in support phase, the deltoids mainly in swing phase, and the supracoracoideus - infraspi-
natus/supraspinatus homologues are biphasic. Redrawn from Goslow et al. (1989).

and also observed only a small delay in recruitment of the quadriceps complex relative to
nonprimate patterns. Like Okada et al. (1978) and Kimura et al. (1979), Jungers et al.
(1980) suggested that such timing differences could be due to the differences in footfall
patterns between primates and other quadrupeds. Finally, Vangor and Wells (1983) studied
14 muscles of the hip and thigh of two spider monkeys, two woolly monkeys and two
patas monkeys, and found "a surprising degree of similarity in phasic activity patterns" (p.
130) to what had been observed in nonprimates, but noted that the primate data showed a
high level of variability.
When the forelimb is considered, a very different impression of the degree of con-
servation of primitive motor programs in primates emerges. Larson and Stern (1987) stud-
ied the pattern of shoulder muscle recruitment in chimpanzees during knuckle-walking.
The authors began with the premise that chimpanzees are basically suspensory/climbing
adapted primates that have had to make certain compromises in order to continue to walk
quadrupedally since their forelimbs are adapted for mobility rather than stability (e.g.,
widened thorax with a dorsally placed scapula, large, globular humeral head, round radial
head permitting a wide range of supination/pronation, distal ulna removed from articula-
tion with the carpals, etc.). The African apes are already unique in that they walk on their
knuckles, and it would seem likely that there would be other ways in which their mode of
quadrupedal locomotion would be distinct.
The EMG results from that study for the muscles illustrated by Goslow as support-
ing the neuromuscular conservation hypothesis are presented in Figure 4 (since chimpan-
zees overstride, inside and outside forelimbs were analyzed separately). In the chimpanzee
the latissimus dorsi and pectoral muscles acted most consistently at the end of swing,
Unique Aspects of Quadrupedal Locomotion in Nonhuman Primates 163

Forelimb Outside Forelimb Inside

Pectoralis
Major

Latissimus
Dorsi

Middle
Deltoid

2199
Supra-
spinatus

Infra-
spinatus

Support Phase Swing Phase Support Phase Swing Phase

Figure 4. EMG activity matrices for chimpanzee shoulder muscles during knuckle-walking. Matrices for forelimb
outside and forelimb inside steps are displayed separately. Each matrix represents a summary of the frequency and
relative magnitude of activity for several steps. Blackened areas indicate highly consistent EMG activity, occur-
ring two-thirds or more of the time. Hatched areas reflect frequent but less consistent activity, occurring one-third
to two-thirds of the time. Heights of these areas reflect amplitude of activity (determined from spike height) as a
percentage of the maximum amplitude observed during the recording session, which is represented by the height
of the box. The numbers in each box represent the number of individuals/number of steps analyzed for each mus-
cle. The pattern of muscle activity in the chimpanzee is different from that reported for the lizard and opossum by
Goslow et at. (I 989}--see Figure 3.

whereas these muscles were active through most of support phase in the lizard and opos-
sum. Only during forelimb outside steps was the pectoralis major of chimpanzees active in
support phase. Larson and Stem (1987) interpreted this activity as being due to the fact
that the arm is slightly abducted during forelimb outside steps, thus requiring the action of
a forelimb adductor to resist the tendency for the hand to slip out to the side. So at least
during forelimb inside steps, there is a very different pattern of recruitment of these two
humeral retractors in the chimpanzee compared to nonprimate vertebrates.
In Goslow's example of the lizard or opossum, the deltoids were recruited during
swing phase and the supra- and infraspinatus were biphasic. + Larson and Stem (1987) re-
ported that in the chimpanzee, the middle deltoid was most consistently active during sup-
port phase of knuckle-walking, and was only occasionally active during swing phases
when the forelimb was outside (Figure 3). Supraspinatus was biphasic as predicted during
forelimb outside steps, but was frequently only active during support phase in forelimb in-
side steps. The pattern of recruitment for the infraspinatus was similar to that of the su-
praspinatus except that it was less frequently biphasic in forelimb outside steps showing

! Although Goslow et at. (1989) characterized the supracoracoideus - supraspinatus/infraspinatus homologues as

biphasic, they described the propulsive phase activity as more intense.
 164 s. G. Larson

consistent activity only during support phase, and was essentially uniphasic during fore-
limb inside steps.
Larson and Stern (1987) concluded from this study that the pattern of recruitment of
shoulder muscles in the chimpanzee during quadrupedal locomotion, especially the ab-
sence of humeral retractor activity during support phase, was related to the chimpanzee's
mobile forelimb and the need to protect it from excessive locomotor stresses. The unique
pattern of muscle recruitment, therefore, reflected an aspect of their adaptation to suspen-
sory/climbing habits, and was related to the high degree of functional differentiation be-
tween their fore- and hind limbs (Kimura et aI., 1979; Kimura, 1985; Demes et aI., 1992,
1994).
Stern et al. (1977) have reported that two other highly arboreal, suspensory/climbing
adapted primates, spider and woolly monkeys, do not recruit the latissimus dorsi during
stance phase of walking, supporting this interpretation. The pectoralis major of the woolly
monkey was also inactive during support phase, although it was recruited in the spider
monkey. It should be noted, however, that spider monkeys typically walk with abducted
arms (Schmitt, 1994) raising the possibility that the pectoralis is again being used more as
an adductor than as a humeral retractor in this species.
As a contrast to these highly arboreal primate species, Larson and Stern (1989a,b)
undertook a similar set of studies of shoulder muscle recruitment in the vervet monkey, a
primate that is semiterrestrial (Struhsaker, 1967; Dunbar and Dunbar, 1974; Fedigan and
Fedigan, 1988), and has a dorsoventrally deep thorax, laterally placed scapula and shoul-
der joint morphology more like that of a cat or opossum than a chimpanzee. The EMG re-
sults, however, were more like those of the chimpanzee than like nonprimates. This was
true for the humeral retractors, as well as for the deltoids and the supra- and infraspinatus
(summarized in Figure 5). Unpublished EMG data recently collected in this lab on the
patas monkey, arguably the most terrestrially adapted primate, demonstrated the same pat-
tern of muscle recruitment. These results are surprising because the vervet or patas mon-
key forelimb is not considered to be adapted for mobility as is the chimpanzee's. Indeed,
force plate studies have shown that it is only the more suspensory/climbing adapted pri-
mates that show a strong dichotomy between the forelimb and hind limb in participation in
weight support and propulsion (Kimura et aI., 1979; Kimura, 1985; Demes et aI., 1994).
This common pattern of muscle recruitment, therefore, is not simply a similar mechanical
response to similar functional demands. It is the case that nearly all primates that have
been studied, however, do display at least slightly higher peak vertical forces on their hind
limbs than on their forelimbs, distinguishing them from all nonprimates (Demes et aI.,
1994). These shared characteristics of patterns of force distribution and muscle recruit-
ment despite differences in morphology and limb use suggest that primates may display a
neuromuscular conservation of their own. That is, a common motor program distinguishes
the primate clade from other tetrapods. The EMG results also suggest that the most pro-
found changes in motor control of limb motion involve the forelimb.

3. DISCUSSION

The various theories that have been offered to explain the origins of primates all
share some emphasis on the use of clawless grasping extremities in an arboreal (small
branch) setting to reach discontinuous supports and/or collect food. I propose that reach-
ing out with clawless grasping extremities also forms the foundation for the various dis-
tinctive characteristics of primate quadrupedal locomotion described above.
Unique Aspects of Quadrupedal Locomotion in Nonhuman Primates 165

Pectoralis
Major

Latissimus
Dorsi f-----'I-I---...- ...- -__

Figure 5. Representative raw EMG traces for five

shoulder muscles in the vervet monkey during walk- Middle
ing. Each trace represents two seconds of activity. Deltoid
Traces for the pectoralis major and the latissimus
dorsi come from two different recording sessions,
and the traces for the other three muscles come from
a third recording session. Steps of approximately the Supra-
same speed and cycle duration were selected for dis- spinatus
play, but the traces have also been graphically ad-
justed to simplify presentation. The pattern of muscle
recruitment displayed by the vervet monkey is more Infra-
similar to that of the chimpanzee than to that of non- spinatus
primate mammals.

Using clawless grasping extremities to travel or forage in an arboreal habitat, espe-

cially on small branches, requires the ability to reach out precisely to gain a secure grip on
a particular branch or object. This demands versatility in the neural control of limb move-
ment. Indeed, Georgopoulos and Grillner (1989) report that while uncomplicated locomo-
tion on an even surface in cats requires very little supraspinal input, more complex
locomotor tasks such as ladder walking are impossible after transection of the corticospi-
nal tract. Studies of the rate of cortical neuron discharge indicate that the corticospinal in-
put is not related to the control of either equilibrium or propulsion, but is directly involved
in the correct positioning of the limbs, particularly the forelimbs. Georgopoulos and Grill-
ner (1989) also report that the same involvement of the corticospinal system is observed in
forelimb manipulatory movements. They propose that ability to use the forelimb for ma-
nipulation evolved from the ability to accurately position the limb during locomotion, and
that as the supraspinal control necessary to precise locomotor limb positioning developed,
so did the fine control of forelimb movements for manipulation.
Vilensky (1989) and Vilensky and Larson (1989) have proposed that these evolu-
tionary changes in the neurological control of forelimb movements associated with the
ability for precise limb positioning during locomotion and use of the forelimb for manipu-
lation and exploration of the environment might account for the preference among pri-
mates for diagonal sequence gaits. Noting that while diagonal sequence gait is the
dominate mode of quadrupedal walking among primates, spontaneous displays of lateral
sequence gaits are frequently seen throughout the Order. However, although a few nonpri-
mates habitually use diagonal sequence walking gaits, there are essentially no observa-
tions of occasional use of diagonal sequence gaits among nonprimates that habitually use
lateral sequence walking gaits. Vilensky and Larson (1989) conclude that quadrupedal pri-
mates are capable of more versatility in their locomotor mode than are most nonprimate
quadrupeds, which, in turn, suggests less rigidity in the control of limb movements during
 166 s. G. Larson

locomotion. Unlike cats, it has been shown that primates require supraspinal input to pro-
duce stepping motions even on flat surfaces (Eidelberg et ai., 1981).
Vilensky (1989) and Vilensky and Larson (1989) propose that a greater degree of cor-
tical control of limb movement in primates may have resulted in less dependence on spinal
cord based program generators. In a study done on rabbits, Viala and Vidal (1978) demon-
strated that hind limb-driving spinal circuits strongly influenced the rhythm of the forelimb
circuits, and suggested that this ability of the hind limbs to control the forelimbs may be a
factor in the prevalence oflateral sequence gaits in quadrupeds. Vilensky (1989) and Vilen-
sky and Larson (1989) suggest that changes in spinal circuitry associated with increased
cortical control of limb motions in primates may have simply eliminated or overridden any
inherited spinal mechanisms tending to favor lateral sequence gaits. If these changes also al-
lowed the movements of each forelimb to exert some direct influence on the ipsilateral hind
limb, then the result would be a preference for diagonal sequence gaits in primates (begin-
ning with the forelimb, diagonal sequence gaits appear as LfLhRfRh and lateral sequence
gaits LfRhRfLh). Primates appear to emphasize ipsilateral fore-hind control, and nonpri-
mates contralateral fore-hind control (or, ipsilateral hind-fore control).
Precise placement of the forelimb in a discontinuous arboreal habitat in order to
reach and grasp food or supports also requires enhanced mobility at the forelimb joints.
However, all else being equal, increasing the mobility of joints can only be accomplished
at the expense of stability, and cannot develop if the forelimbs are subjected to large dis-
ruptive locomotor forces. Therefore, some means of limiting such forces must be devel-
oped. There are a variety of ways that joint forces engendered during locomotion can be
reduced. One is to alter locomotor performance such as by moving more slowly. Another
is through alteration of muscle recruitment patterns. Muscles of the forelimb that are used
to help propel an animal forward among nonprimates are usually inactive during support
phase in primate quadrupeds. This agrees with force plate data demonstrating that nonpri-
mates display higher peak propulsive forces on their forelimbs, whereas primates show
higher peak propulsive forces on their hind limbs (Kimura et ai., 1979; Kimura, 1985;
Pandy et ai., 1988). Interestingly, all primates that have been studied appear to "spare"
their forelimb from some of the disruptive forces associated with quadrupedal locomotion
in this manner regardless of whether or not the forelimb displays particular adaptations to
increased mobility. This suggests that this reduction in use of forelimb propulsive muscles
is a component of the basic adaptation of reaching out with a clawless grasping hand in an
arboreal setting that arose early in the evolution of primates to become characteristic of
the Order.
Another way in which primate quadrupeds can limit the disruptive locomotor forces
acting on their forelimbs is by supporting more of their body weight on their hind limbs,
as has been demonstrated by force plate studies (Kimura et aI., 1979; Kimura, 1985; De-
mes et ai., 1992, 1994). Although this had been thought to be due to a more posterior posi-
tion of the center of gravity, Reynolds (1985a,b) has argued that this alteration in weight
distribution is due to an active transfer of weight to the hind limbs by the contraction of
hind limb retractors. Reynolds (1985b) has shown that the hind limb retractors will be
most effective at shifting weight posteriorly without generating large propulsive forces if
they act when the hind limb is protracted. He thus relates the highly protracted hind limb
at touchdown in primates to an increase in the effectiveness of the posterior weight shift
mechanism (Reynolds, 1987). Unlike the alterations of muscle recruitment patterns, the
posterior weight shift mechanism seems to be most pronounced (greatest disparity be-
tween fore- and hind limbs in weight distribution, most extreme degree of limb protrac-
tion) in those primates displaying higher degrees offorelimb mobility.
Unique Aspects of Quadrupedal Locomotion in Nonhuman Primates 167

The protracted position of primate hind limbs and the accompanying larger angular
excursions documented by Reynolds (1987), also contribute to increased relative hind
limb stride length in primates. Demes et al. (1990) suggest that relatively long strides in
lorisines may be an arboreal adaptation in that it would be a means of achieving high
walking speeds without increasing stride frequency. High frequency gaits, which entail
steeply increasing, relatively high peak forces, could cause branch swaying that is not
only dangerous but also energy costly. Low frequency gaits would minimize such sway,
which may be especially important in lorisines that rely on quiet, stealthful movements
both to capture prey and avoid predators (Demes et aI., 1990). Long strides with low fre-
quency gaits would also generally increase stability by increasing hand contact time. Al-
though long low frequency strides may have originally evolved as adaptations to cryptic
habits among early primates, branch sway is unlikely to occur at the natural walking fre-
quency of any arboreal primate (Alexander, 1991), suggesting that long low frequency
strides may have been maintained in primates as a general mechanism to help maintain
stability on branches.
Schmitt (1995) has proposed that long stride length in primates may also be in-
volved in another mechanism for reducing disruptive forces on the limbs during locomo-
tion. He suggests that certain components of a long stride, namely, longer step length,
increased contact time, and protraction of the limb when coupled with joint yield increase
the compliance of primate quadrupedal walking gaits. By increasing step length and con-
tact time, peak stresses acting on the limb are reduced by increasing the time over which
the reaction forces act (see McMahon, 1985, McMahon et aI., 1987, and Schmitt, this vol-
ume). The highly protracted fore- and hind limbs of primates change the "angle of attack"
of the limbs, thereby reducing the vertical landing speed, and the vertical stiffness of the
body (McMahon et aI., 1987). If limb protraction results in a greater angular excursion for
a compliant limb, then the body's center of mass follows a flatter trajectory (B1ickhan,
1989; Farley et aI., 1993) thereby also helping to increase stability on a branch and reduce
the tendency to cause branch sway. In addition, limb protraction attenuates the impact ac-
celeration peak of the body, implying that less shock due to foot or hand strike passes up-
ward through- the body (McMahon et aI., 1987).
The fact that primates do not typically display a running trot, which is a high fre-
quency, high stiffness gait, may also be a reflection of their tendency to avoid high peak
stresses on the limbs (Schmitt, 1995). Primates typically move from a walk or running
walk to a gallop, which is a more compliant gait than the trot. As Preuschoft et al. (1996)
have noted, asymmetric gaits such as a gallop involve longer contact times than trotting
and thus limit peak substrate reaction forces. Preuschoft and Gunther (1994) suggest that
another reason why primates avoid a trot is that their long relatively heavy limbs with long
pendulum lengths limit the speed of the recovery stroke making a high frequency gait such
as trotting difficult (see also, Preuschoft et aI., 1996). In addition, and unlike most nonpri-
mates that change gaits at fairly predictable speeds (Heglund et aI., 1974), quadrupedal
monkeys sometimes exhibit a walk at speeds at which they would be expected to gallop,
or gallop at very low speeds when they might be expected to walk or trot. These differ-
ences are quite idiosyncratic, with two individuals ofthe same species displaying different
patterns of gait transitions (Vilensky et aI., 1990). This variability and individuality sug-
gests that the ability to avoid a trot may be an additional by-product of a greater degree of
supraspinal control of locomotion in primates.
Effective use of a clawless grasping hand in an arboreal habitat will also be en-
hanced by elongation of the forelimb. Cartmill (1974) has related relative forelimb elonga-
tion to the ability to hold onto a vertical support. Whereas clawed mammals can climb in
 168 S. G. Larson

trees by interlocking with the substrate, without claws, an animal must rely on friction to
prevent slipping off of a branch or trunk. The ability to hold onto a large vertical support
is dependent on the size of the central angle that the animal's limbs can subtend. Longer
limbs can subtend larger central angles. When climbing up or down a vertical tree trunk,
however, a clawless animal must retain a more secure grip with its forelimbs than with its
hind limbs, since failure of the upper grip will cause it to fall backward away from the
tree, whereas failure of the lower grip will result in the animal falling forward into the tree
(Cartmill, 1974, 1985). In addition, clawless mammals can augment pedal friction by lean-
ing away from the support. Thus primates not only have long limbs, but display relative
forelimb elongation with increasing body size as a component of a basic adaptation to
tree-dwelling habits (Cartmill, 1974; Jungers, 1985).
Other mammals such as many carnivores or ungulates also display limb elongation,
but mainly through lengthening the distal elements of the limb (see Figure 6), including
the fingers (Alexander et aI., 1979). This is generally understood as a means of reducing
the inertial properties of the limb by concentrating mass proximally thereby reducing limb
momentum and the cost of reversing the limb's direction with each stride (Smith and Sav-
age, 1956; Hildebrand, 1988; however, see Steudel, 1994). However, in so doing, these
animals have sacrificed any ability to grasp with their hands or feet. The proximal and in-
termediate limb segments of primates have been lengthened, thus permitting the mainte-
nance of grasping extremities.
The protracted and extended limb postures of primates further increase the effective
length of their limbs (see Figure 6). However, if limb elongation were the only goal, this
conceivably could have been achieved solely through lengthening the bony elements. Al-
though extended limb postures have been recognized as a means of reducing the load arms
for substrate reaction forces as body size increases to limit joint and bone stresses
(Biewener, 1983, 1989, 1990) and reduce energy costs (Kram and Taylor, 1990), primate
quadrupeds display the most extended limb postures at the beginning and end of stance
phase when substrate reaction forces are low. They typically flex their elbow or knee
through midstride, when reaction forces are high, as part of their compliant gait (Schmitt,
1995). This suggests that the unique protracted limb positions displayed at touchdown by
primate quadrupeds are not related to the reduction of substrate reaction force load arms. I
suggest that limb protraction, especially forelimb protraction, is more fundamentally asso-
ciated with reaching out with a grasping extremity to travel and forage in the discontinu-
ous small branch arboreal habitat. Lemelin (1996) has reported on the many parallels in
hand proportions between primates and highly arboreal marsupials, such as Caluromys

Figure 6. Position of forelimb at touchdown in different mammals. Superimposed black lines represent scapular.
humeral and radial-ulnar limb segments. The forefoot touches down ahead of the shoulder in all individuals. How-
ever, in the horse, dog, cat and capybara, this is due mainly to the elongated distal limb elements since the
humerus is positioned behind a vertical line through the shoulder. In the baboon, the anterior position of the hand
results from the protracted position of the humerus and extended elbow. Figures drawn from photographs in Muy-
bridge (1957), and are not to scale.
Unique Aspects of Quadrupedal Locomotion in Nonhuman Primates 169

and Marmosa, which also use grasping extremities to travel in a small branch habitat and
to capture prey. Significantly, these arboreal marsupials also display a protracted forelimb
posture during locomotion (Lemelin, pers. com.).
Finally, the protraction of the hind limb and accompanying increase in hind limb an-
gular excursion observed in primates may have had another consequence. Schmitt and
Larson (1995) suggest that the protracted position of the hind limb is also related to the
occurrence of plantigrady in arboreal primates. They note that a highly protracted hind
limb at touchdown results in much of the body weight being placed behind the foot. This,
in turn, tends to force the heel down after touchdown. Although heel contact is not com-
mon among primates, it is seen among the species displaying the most dramatic shift in
weight from the fore- to the hind limbs such as chimpanzees, orangutans, spider, woolly,
and howling monkeys, and gibbons (the logical end-product of the posterior weight shift
mechanism (Reynolds, 1985a,b, 1987)). Schmitt and Larson (1995) suggest that this impe-
tus for heel contact may be a factor in the eventual evolution of the heel strike that has be-
come characteristic of human bipedalism.
Protracted limb positions, limb elongation, and greater supraspinal control are cen-
tral elements underlying the various ways in which the manner of quadrupedal locomotion
of primates differs from that of nonprimates. It seems likely that these changes came about
through some combination of feedback loops (see Figure 7), where reaching out for small
branches using grasping hands and manually foraging for food and/or capturing prey se-
lected for greater cortical input into the control of forelimb motion, which also enhanced
the manipulatory abilities of the forelimb. Improved reaching ability requires some fore-
limb elongation as well as increased forelimb mobility, which can only be brought about if
the necessity for limb stabilization is relaxed. This entails reducing some of the disruptive
locomotor forces that affect the forelimb. In order to spare the forelimb from some of
these forces, muscle recruitment patterns were changed, mechanisms for shifting weight
from the forelimb to the hind limb, and for reducing the magnitude of peak forces acting

Traveling and foraging

~ i~ small b~anch habit.~t ~ Improved
I ncreased st ab I'I't

;Y ~ /
uSing grasping extremities manipulatory

aTes
I y, ...
~ducti°t m b,.noh sway

I
Lengthened stride, Protracted limb Greater cortical
increased contact ~ postures ~ control of ~ ~reater versatil.itY

1 /
time Limb elongation forelimb motion In locomotor skills

~Compliantgait
Posterior weight shift
~ Development of
suspensory postures
A~ered muscle recrurtment and locomotion?

Sparing offorelimb I( ) Increased forelimb

from high locomotor Increased functional mobility
differentiation between
stresses fore- and hind limb

Figure 7. Interaction of factors involved in the evolution of primate quadrupedalism. Use of grasping extremities
to travel and manually forage in a small branch habitat required changes in habitual limb posture and greater corti-
cal control of limb movements. These, in tum, allowed improvement in manipulatory skills and greater versatility
in locomotor abilities. Such changes, however, could only be brought about through increased joint mobility that
required the development of mechanisms to limit disruptive locomotor stresses. See text for further discussion.
 170 S.G. Larson

on the limbs by increasing limb compliance were developed. These mechanisms exploited
the protracted limb postures and limb elongation that had been adopted for reaching and
climbing and perhaps exaggerated their magnitude. Increasing cortical control of forelimb
movements overrode spinal cord pattern generators, leading to greater versatility in gait
patterns and perhaps to a preference for diagonal sequence gaits.
The reduction of peak forces, especially on the forelimb, brought about by these
various mechanisms, and the morphological changes enhancing limb mobility that they
permitted, have been pivotal in the development of the locomotor versatility-including
various forms of climbing, scrambling, and bridging behaviors-that has come to charac-
terize arboreal primates. These changes also constituted a major step toward the "emanci-
pation of the forelimb" from compressive joint forces that Wood Jones (1926) and others
have emphasized as the central evolutionary factor leading to the development of suspen-
sory postures and locomotion as well as human bipedalism.

ACKNOWLEDGMENTS

I wish to thank Daniel Schmitt for many stimulating discussions about the topics con-
tained in this paper, Jack Stern who collaborated on all of the EMG research, and Marianne
Crisci for animal handling during the EMG experiments. Thanks also to Jack Stern, William
Jungers and three anonymous reviewers for helpful comments on earlier versions of this pa-
per, and to Luci Betti for preparation of some of the text figures. This material is based on
work supported by the National Science Foundation under Grant SBR 9507078.

REFERENCES
Alexander R McN (1991) Elastic mechanisms in primate locomotion. Z. Morph. Anthrop. 78:315-320.
Alexander R McN, Jayes AS, Maloiy GMO, and Wathuta EM (1979) Allometry of the limb bones of mammals
from shrews (Sorex) to elephant (Loxodonta). J. Zoo\., Lond. 189:305-314.
Alexander R McN, and Maloiy GMO (1984) Stride lengths and stride frequencies of primates. J. Zoo\., Lond.
202:577-582.
Biewener AA (1983) Allometry of quadrupedal locomotion: The scaling of duty factor, bone curvature and limb
orientation to body size. J. expo Bio\' 105: 147-171.
Biewener AA (1989) Scaling body support in mammals: Limb posture and muscle mechanics. Science 245:45-48.
Biewener AA (1990) Biomechanics of mammalian terrestrial locomotion. Science 250: I 097-11 03.
Blickhan R (1989) The spring-mass model for running and hopping. J. Biomech. 22:1217-1227.
Cartmill M (1974) Pads and claws in arboreal locomotion. In FA Jenkins Jr. (ed.): Primate Locomotion. New York:
Academic Press, pp. 45-83.
Cartmill M (1985) Climbing. In M Hildebrand, DM Bramble, KF Liem, and DB Wake (eds.): Functional Verte-
brate Morphology. Cambridge: Belknap Press, pp. 73--88.
Dagg AI (1974) Running, Walking and Jumping. London: Wykeham.
Demes B, Jungers WL, and Nieschalk U (1990) Size- and speed-related aspects of quadrupedal walking in slender
and slow lorises. In FK Jouffroy, MH Stack, and C Niemitz (eds.): Gravity, Posture and Locomotion in Pri-
mates. Firenze: II Sedicesimo, pp. 175-197.
Demes B, Larson SG, Stern JT Jr., and Jungers WL (1992) The hindlimb drive of primates - theoretical reconsid-
eration and empirical examination of a widely held concept. Am. J. Phys. Anthropo\. Supp\. 14: 69.
Demes B, Larson SG, Stern JT Jr., Jungers WL, Biknevicius AR, and Schmitt D (1994) The kinetics of "hind limb
drive" reconsidered. J. Hum. Evo\. 26:353-374.
Dial KP, Goslow GE Jr., and Jenkins FA Jr. (1991) The functional anatomy of the shoulder in the European star-
ling (SIIII/US vulgaris). J. Morph. 207:327-344.
Dial KP, Kaplan SR, Goslow GE Jr., and Jenkins FA Jr. (1987) Structure and neural control of the pectoralis in pi-
geons: implications for flight mechanics. Anat. Rec. 218:284-287.
Unique Aspects of Quadrupedal Locomotion in Nonhuman Primates 171

Dunbar RIM, and Dunbar EP (1974) Ecological relations and niche separation between sympatric terrestrial pri-
mates in Ethiopia. Folia Primatol. 21:36--60.
Dykyj D (1980) Locomotion of the slow loris in a designed substrate context. Am. J. Phys. Anthropol.
52:577-586.
Eaton TH Jr. (1944) Modification of the shoulder girdle related to reach and stride in mammals. J. Morph.
75:167-171.
Eidelberg E, Walden JG, and Nguyen LH (1981) Locomotor control in macaque monkeys. Brain 104:647-663.
English AW (I 978a) Functional analysis of the shoulder girdle of cats during locomotion. J. Morph. 156:279-292.
English AW (I 978b ) An electromyographic analysis of forelimb muscles during overground stepping in the cat. J.
expo BioI. 76: 105-122.
Farley CT, Glasheen J, and McMahon TA (1993) Running springs: Speed and animal size. J. expo BioI. 185:71-86.
Fedigan L, and Fedigan LM (1988) Cercopithecus aethiops: A review offield studies. In A Gautier-Hion, F Bour-
liere, J Gautier (eds.): A Primate Radiation: Evolutionary Biology of the African Guenons. Cambridge:
Cambridge Univ. Press. pp. 389-411.
Fischer MS (1994) Crouched posture and high fulcrum, a principle in the locomotion of small mammals: The ex-
ample of the rock hyrax (Procavia capensis) (Mammalia: Hyracoidea). J. Hum. Evol. 26:510-524.
Fleagle JG (1978) Locomotion, posture and habitat use of two sympatric leaf-monkeys in West Malaysia. In DJ
Chivers and J Herbert (eds.): Recent Advances in Primatology, Vol. I, Behavior. New York: Academic
Press, pp. 331-336.
Georgopoulos AP, and Grillner S (1989) Visuomotor coordination in reaching and locomotion. Science
245: 1209-1210.
Goslow GE Jr., Dial KP, and Jenkins FA Jr. (1989) The avian shoulder: An experimental approach. Amer. Zool.
29:287-301.
Grand T (I 968a) Functional anatomy of the lower limb of the howler monkey (Alouatta caraya). Am. J. Phys. An-
thropol. 28:163-181.
Grand T (l968b) Functional anatomy of the upper limb. In: Biology of the Howler Monkey (Alouatta caraya).
Bibl. Primat., No.7, Basel: Karger, pp. 104-125.
Grand T (1984) Motion economy within the canopy: Four strategies for mobility. In P Rodman and JGH Cant
(eds.): Adaptations for Foraging in Nonhuman Primates: Contributions to an Organismal Biology of
Prosimians, Monkeys and Apes. New York: Columbia Univ. Press, pp. 54-72.
Heglund NC, Taylor CR, and McMahon TA (1974) Scaling stride frequency and gait to animal size: Mice to
horses. Science 186:1112-1113.
Hildebrand M (1967). Symmetrical gaits of primates. Am. J. Phys. Anthropol. 26:119-130.
Hildebrand M (1988) Analysis of Vertebrate Structure. 3rd Ed. New York: John Wiley and Sons.
Howell AB (1944) Speed in Animals. Chicago: Univ. Chicago Press.
Ishida H, Jouffroy FK, and Nakano Y (1990) Comparative dynamics of pronograde and upside down horizontal
quadrupedalism in the slow loris (Nycticebus coucang). In FK Jouffroy, MH Stack, and C Niemitz (eds.):
Gravity, Posture and Locomotion in Primates. Firenze: II Sedicesimo, pp. 209-220.
Iwamoto M, and Tomita M (1966) On the movement order offour limbs while walking and the body weight distri-
bution to fore and hind limbs while standing on all fours in monkeys. J. Anthropol. Soc. Nippon
74:228-231.
Jenkins FA Jr. (1971) Limb posture and locomotion in the Virginia opossum (Didelphis marsupialis) and in other
non-cursorial mammals. J. Zool., Lond. 165:303-315.
Jenkins FA Jr. (1974) The movement of the shoulder in c1aviculate and aclaviculate mammals. J. Morph.
144:71-83.
Jenkins FA Jr., and Goslow GE Jr. (1983) The functional anatomy of the shoulder of the savanna monitor lizard
(Varanus exanthematicus). J. Morph. 175:195-216.
Jenkins FA Jr., and Weijs WA (1979) The functional anatomy of the shoulder in the Virginia opossum (Didelphis
virginian a). J. Zool., Lond. 188:379-410.
Jouffroy FK, and Petter A (1990) Gravity-related kinematic changes in lorisine horizontal locomotion in relation
to position of the body. In FK Jouffroy, MH Stack, and C Niemitz (eds.): Gravity, Posture and Locomotion
in Primates. Firenze: II Sedicesimo, pp. 199-208.
Jouffroy FK, Renous S, and Gasc JP (1983) Etude cineradiographique des deplacements du membre anterieur du
Potto de Bosman (Perodicticus potto) au cours de la marche quadrupede sur une branche horizontale. Ann.
SC. Nat. Zool. l3eme ser. 5:75-87.
Jungers WL (1985) Body size and scaling of limb proportions in primates. In WL Jungers (ed.): Size and Scaling
in Primate Biology. New York: Plenum Press, pp. 345-381.
Jungers WL, Jouffroy FK, and Stem JT Jr. (1980) Gross structure and function of the quadriceps femoris in Lemur
fulvus: An analysis based on telemetered electromyography. J. Morph. 164:287-299.
 172 S. G. Larson

Kimura T (1985) Bipedal and quadrupedal walking of primates: Comparative dynamics. In S Kondo (ed.): Primate
Morphophysiology, Locomotor Analyses and Human Bipedalism. Tokyo: Univ. of Tokyo Press, pp.
81-104.
Kimura T, Okada M, and Ishida H (1979) Kinesiological characteristics of primate walking: Its significance in hu-
man walking. In ME Morbeck, H Preuschoft, and N Gomberg (eds.): Environment, Behavior, and Mor-
phology: Dynamic Interactions in Primates. New York: Gustav Fischer, pp. 297-311.
Kram R, and Taylor CR (1990) Energetics of running: A new perspective. Nature 346:265--267.
Larson SG, and Stern JT Jr. (1987) EMG of chimpanzee shoulder muscles during knuckle-walking: Problems of
terrestrial locomotion in a suspensory adapted primate. J. Zool. Lond. 212:629--(,55.
Larson SG, and Stern JT Jr. (1989a) The role of supraspinatus in the quadrupedal locomotion of vervets (Cercopi-
thecus aethiops): Implications for interpretation of humeral morphology. Am. J. Phys. Anthropol.
79:369-377.
Larson SG, and Stern JT Jr. (1989b) The role of propulsive muscles of the shoulder during quadrupedalism in
vervet monkeys (Cercopithecus aethiops): Implications for neural control of locomotion in primates. J.
Motor Behavior 21:457-472.
Lemelin P (1996) The evolution of manual prehensility in primates: A comparative and functional analysis in
prosimian primates and didelphid marsupials. Ph.D. Dissertation, State University of New York at Stony
Brook.
McMahon TA (1975) Using body size to understand the structural design of animals: Quadrupedal locomotion. J.
Appl. Physiol. 39:619--(,27.
McMahon TA (1984) Muscles, Reflexes and Locomotion. Princeton: Princeton Univ. Press.
McMahon TA (1985) The role of compliance in mammalian running gaits. J. expo BioI. 115:263-282.
McMahon TA, Valiant G, and Frederick EC (1987) Groucho running. 1. Appl. Physiol. 62:2326-2337.
Meldrum DJ (1991) Kinematics of the cercopithecine foot on arboreal and terrestrial substrates with implications
for the interpretation of hominid terrestrial adaptations. Am. J. Phys. Anthropol. 84:273-290.
Miller S, and Van der Meche FGA (1975) Movements of the forelimbs of the cat during stepping on a treadmill.
Brain Res. 91:255--269.
Morbeck ME (1979) Forelimb use and positional adaptation in Colobus guereza: Integration of behavioral, eco-
logical and anatomical data. In M Morbeck, H Preuschoft, and N Gomberg (eds.): Environment, Behavior,
and Morphology: Dynamic Interactions in Primates. New York: Gustav Fisher, pp. 95--118.
Muybridge E (1957) Animals in Motion. In LS Brown (ed.): Animals in Motion. New York: Dover. (Originally
published by Chapman and Hall. London, 1899)
Napier JR (1967) Evolutionary aspects of primate locomotion. Am. J. Phys. Anthropo!. 27:333-342.
Nomura S. Sawazake H, and Ibaraki T (1966) Co-operated muscular action in postural adjustment and motion in
dog, from the viewpoint of electromyographic kinesiology and joint mechanics. IV. About muscular activ"
ity in walking and trot. Jap. J. Zootech. Sci. 37:221-229.
Okada M, Kimura T, Ishida H, and Kondo S (1978) Biomechanical aspects of primate quadrupedal ism. In E As-
mussen and K Jrorgensen (eds.): Biomechanics 6A. Baltimore: Univ. Part Press, pp. 119-124.
Pandy MG, Kumar V, Berme N, Waldron KJ (1988) The dynamics of quadrupedal locomotion. J. Biomech. Engr.
110:230--237.
Peters SE, and Goslow GE Jr. (1983) From salamanders to mammals: Continuity in musculoskeletal function dur-
ing locomotion. Brain Behav. Evo!. 22: 191-197.
Preuschoft H, and Gunther MM (1994) Biomechanics and body shape in primates compared to horses. Z. Morph.
Anthropol. 80: 149-165.
Preuschoft H, Witte H, Christian A, and Fischer M (1996) Size influences on primate locomotion and body shape,
with special emphasis on the locomotion of 'small mammals.' Folia Primato!. 66:93-112.
Prost JH (1965) The methodology of gait analysis and gaits of monkeys. Am. 1. Phys. Anthropol. 23: 215--240.
Prost JH (1969) A replication study on monkey gaits. Am. 1. Phys. Anthropol. 30:203-208.
Reynolds TR (I 985a) Mechanics of increased support of weight by the hindlimbs in primates. Am. J. Phys. An-
thropol. 67:335--349.
Reynolds TR (I 985b) Stresses on the limbs of quadrupedal primates. Am. J. Phys. Anthropol. 67:351-362.
Reynolds TR (1987) Stride length and its determinants in humans, early hom in ids, primates, and mammals. Am. J.
Phys. Anthropo!. 72: 101-116.
Rollinson J. and Martin RD (1981) Comparative aspects of primate locomotion. with special reference to arboreal
cercopithecines. In MH Day (ed.): Vertebrate Locomotion. Symposia of the Zoological Society of London.
No. 48. London: Academic Press, pp. 377-427.
Rose MD (1979) Positional behavior of natural populations: Some quantitative results of a field study of Colobus
guereza and Cercopithecus aethiops. In ME Morbeck, H Preuschoft, and N Gomberg (eds.): Environment.
Behavior, and Morphology: Dynamic Interactions in Primates. New York: Gustav Fischer, pp. 74-93.
Unique Aspects of Quadrupedal Locomotion in Nonhuman Primates 173

Schmitt D (1994) Forelimb mechanics as a function of substrate type during quadrupedal ism in two anthropoid
primates. J. Hum. Evol. 26: 441-457.
Schmitt D (1995) A kinematic and kinetic analysis of forelimb use during arboreal and terrestrial quadrupedalism
in Old World monkeys. Ph.D. Dissertation, State University of New York at Stony Brook.
Schmitt D, and Larson SG (1994) Heel contact as a function of substrate type and speed in primates. Am. J. Phys.
Anthropol. 96:39--50.
Smith JM, and Savage RJG (1956) Some locomotor adaptations in mammals. 1. Linn. Soc. (Zool.) 42:603--{)22.
Stern JT Jr., Wells JP, Vangor AK, and Fleagle JG (1977) Electromyography of some muscles of the upper limb in
Ateles and Lagothrix. Yrbk. Phys. Anthropol. 20:98--507.
Steudel K (1994) Locomotor energetics and hominid evolution. Evol. Anthropol. 3:42-48.
Struhsaker TT (1967) Ecology of vervet monkeys (Cercopithecus aethiops) in the Masai-Ambesoli Game Reserve,
Kenya. Ecology 48:891-094.
Tokuriki M (1973) Electromyographic and joint-mechanical studies in quadrupedal locomotion. I. Walk. Jap. J.
Vet. Sci. 35:433-448.
Tomita M (1967) A study on the movement patterns of four limbs in walking. 1. Observation and discussion on the
two types of the movement order of four limbs seen in mammals while walking. J. Anthropol. Soc. Nippon
75:120-146.
Vangor A, and Wells JP (1983) Muscle recruitment and the evolution of bipedality: Evidence from telemetered
electromyography of spider, woolly, and patas monkeys. Ann. Sci. Nat., Zool., Paris, Ser. 135: 125-136.
Viala D, and Vidal C (1978) Evidence for distinct spinal locomotion generators supplying respectively fore- and
hindlimbs in the rabbit. Brain Res. 155:182-186.
Vilensky JA (1980) Trot-gallop transition in a macaque. Am. J. Phys. Anthropol. 53:347-348.
Vilensky JA (1987) Locomotor behavior and control in human and nonhuman primates: comparisons with cats and
dogs. Neurosci. Biobehav. Rev. //:263-274.
Vilensky JA (1989) Primate quadrupedalism: How and why does it differ from that of typical quadrupeds. Brain
Behav. Evol. 34:357-364.
Vilensky JA, Gankiewicz E, and Townsend DW (1988) Effects of size on vervet (Cercopithecusaethiops) gait pa-
rameters: A cross-sectional approach. Am. 1. Phys. Anthropol. 76:463-480.
Vilensky JA, Gankiewicz E, and Townsend DW (1990) Effects of size on vervet (Cercopithecus aethiops) gait pa-
rameters: A longitudinal approach. Am. 1. Phys. Anthropol. 81:429-439.
Vilensky JA, and Larson SG (1989) Primate locomotion: Utilization and control of symmetrical gaits. Ann. Rev.
Anthropol. 18: 17-35.
Vilensky JA, Moore AM, and Libii IN (1994) Squirrel monkey locomotion on an inclined treadmill: Implications
for the evolution of gaits. 1. Hum. Evol. 26:375-386.
Whitehead PF, and Larson SG (1994) Shoulder motion during quadrupedal walking in Cercopithecus aethiops: In-
tegration of cineradiographic and electromyographic data. 1. Hum. Evol. 26:525-544.
Wood Johns F (1926) The Arboreal Man. New York: Hafner.
 11

FORELIMB MECHANICS DURING ARBOREAL

AND TERRESTRIAL QUADRUPEDALISM IN
OLD WORLD MONKEYS

Daniel Schmitt

Department of Biological Anthropology and Anatomy

Duke University School of Medicine
Box 3170 DUMC
Durham, North Carolina 27701

1. INTRODUCTION

For over a century it has been known that primates have highly mobile grasping
forelimbs with the supportive functions shifted more strongly to the hindlimbs, unlike
most mammals where all four limbs share a fairly equal role in weight support. Darwin
(1871) was the first to recognize this distinction between forelimb and hindlimbs and to
articulate its evolutionary significance. Since that point many researchers have developed
theories of primate locomotor evolution that suggest that the amount of compressive
weight support experienced by the forelimb of primates was gradually reduced thus facili-
tating the use of the forelimb in tension and then allowing its complete removal from loco-
motion in humans (Wood Jones, 1926; Le Gros Clark, 1959; Napier and Davis, 1959;
Napier, 1967; Stern, 1976; Ripley, 1979; Reynolds, 1981, 1985a,b; Cant, 1988; Rose,
1991). Fundamental to this scenario is the belief that the change in the role of the primate
forelimb is directly related to adaptations to arboreal quadrupedal ism by primates.
This view of primate evolution is intuitively appealing. At present, however, it is
based on untested suppositions concerning the mechanics of primate quadrupedalism. The
scenario described above presupposes that arboreal locomotion requires a very different
role for the forelimb than does terrestrial quadrupedal ism. This is based on the fact that an
arboreal environment poses mechanical challenges that are very different from a terrestrial
one. Unlike the ground, tree branches are not generally level, they are finite in diameter,
and the surface is discontinuous, often presenting animals with sharp directional changes
and large gaps between branches (Cartmill, 1985). Additionally, food in an arboreal envi-
ronment may be distributed at many levels and is often found on terminal branches. There-
fore, quadrupedal locomotion in an arboreal habitat requires primates to have highly

175
176 D. Schmitt

mobile forelimbs that (1) can be held flexed in order to better maintain balance by lower-
ing the center of gravity, (2) can be used to reach and grab in many planes, and (3) allow
for rapid changes of direction and gap crossing in a discontinuous support environment
(Wood Jones, 1926; Grand 1968a,b; Jenkins, 1974; Cartmill, 1985; Alexander and Ker,
1990; Kram and Taylor, 1990).
Naturalistic, anatomical, and laboratory studies support this analysis. Although de-
tailed kinematics are unavailable from the field, qualitative locomotor data on primates in
natural settings support the belief that primate arboreal quadrupeds walk with flexed fore-
limbs (Grand, 1968a,b; Rose, 1973; Morbeck, 1979; Cant, 1988; Dunbar, 1989). Anatomi-
cal studies suggest that arboreal quadrupedal primates have a greater range of forelimb
joint motion due to decreased osteological stabilization of the forelimb joints, and reduced
robusticity of the forelimb bones in comparison to those of terrestrial primates (e.g.,
Ashton et aI., 1965, 1968; Jolly, 1967; Manaster, 1975, 1979; Feldesman, 1976; Morbeck,
1977,1979; O'Connor and Rarey, 1979; Rodman, 1979; Bown et aI., 1982; Schaffler and
Burr, 1984; Schaffler et ai., 1985; Harrison, 1989; Rose, 1988, 1989; Jungers and Burr,
1992, 1995). Kinetic studies of ground quadrupedaJism in primarily arboreal and primarily
terrestrial primate species has shown that primates have a unique distribution of vertical
ground reaction forces in which the hindlimbs serve as the primary supportive organs
(Kimura et aI., 1979; Kimura, 1985, 1992; Reynolds, 1981, 1985a, b, Demes et aI., 1992).
The reduction of vertical compressive force on the forelimb has been interpreted as a
means of protecting a highly mobile (and therefore inherently less stable) forelimb against
disruptive locomotor stresses (Reynolds, 1981, 1985a,b).
Clearly all the presently available data supports the notion that arboreal quadrupeds
have highly gracile and mobile forelimbs that are regularly held in a flexed posture. There
is, however, a serious paradox in this description of primate arboreal posture. A flexed
limb posture should result in higher joint and bone stresses if substrate reaction forces are
equal in a crouched and extended posture as illustrated in Figure 1 (Biewener, 1982,
1983a,b, 1989, 1990). This is because the greater distance of the substrate reaction resul-
tant to the joints yields a decreased mechanical advantage of the antigravity muscles and
increases the bending moments and muscle force required to maintain this posture

Figure 1. Schematic depiction of the effect of

crouched posture on bone and joint stresses in a
limb, based on models of Biewener (1989, 1990).
In (a) and (b) the substrate reaction resultant
force (arrow) and the limb segment lengths are
identical. The only change from (a) to (b) is in-
creased elbow flexion. This results in clearly
larger moment arms and moments along the limb.
This reduces the "effective mechanical advan-
tage" of antigravity muscles, increases the force
necessary to maintain posture, and increases bone
[I) Extended h) Flexed
and joint stresses.
Forelimb Mechanics during Arboreal and Terrestrial Quadrupedalism in Old World Monkeys 177

(Biewener, 1982, 1983a,b, 1989,1990). As a result of this model we might expect that ar-
boreal primates would have more robust forelimbs with greater stabilization to prevent
distraction of the shoulder or elbow. Because anatomical evidence described above (espe-
cially Manaster, 1975, 1979; Schaffler and Burr, 1984; Schaffler et aI., 1985; Jungers and
Burr, 1992, 1995) show that this is not the case, only two other possible solutions to this
apparent conflict remain: (1) either primate quadrupeds do not use the postures described
above, or (2) a change in the substrate reaction forces accompanies this change in posture
thus reducing the bone and joint stresses.

2. MATERIAL AND METHODS

In an attempt to resolve this conflict I documented the kinematics and kinetics of

forelimb use on terrestrial and "arboreal" supports in a group of Old World monkeys. I se-
lected five species that included primarily arboreal, primarily terrestrial, and "semi-terres-
trial" (Ripley, 1979) species (Table 1; Struhsaker, 1967; Rose, 1973, 1977; Rodman, 1979;
Wheatley, 1982; Chism and Rowell, 1988; Fedigan and Fedigan, 1988; Fleagle, 1988;
Dunbar, 1989; Nakigawa, 1989; Rawlins, 1993). All nine subjects were trained to walk
within a plexiglas enclosure (6 m x 1 m x 1 m) on a wooden runway and on raised hori-
zontal poles of varying diameters (Table 1). The arrangement of recording equipment is il-
lustrated in Figure 2. During locomotion, vertical, fore-aft, and mediolateral components
of the substrate reaction force data were recorded from a force plate mounted in the run-
way for terrestrial locomotion or attached to the poles for "arboreal" locomotion. Simulta-
neously, subjects were videotaped with electronically shuttered video cameras from
lateral, frontal and overhead views. The methods used are described in detail elsewhere
(Schmitt, 1994, 1995) and will only be summarized below.
For collecting data on terrestrial quadrupedalism the force platform was mounted in
the center of a runway flush with its surface with a Y2 inch gap on all sides to avoid vibra-
tions from limbs other than the one making contact. For gathering data on "arboreal" quad-
rupedalism, animals were trained to run along PVC pipes coated with a nonslip surface of

Table 1. Primary habitat and number of walking steps collected for each subject
on the poles and the ground

Pole (cm)
Subject Primary habitat 1.25 2.5 6.25 8.75 11.25 Grnd.
M. fascicularis (f) arboreal' 10 \I 10 10 \3
M.fascicularis (rn) arboreal' 12 9 10 10
M. mulatta (f) serni-terrestriae 10 8 10 II
M. mulatta (f) semi-terrestriae 7 10 \0 \0 \0
C. aethiops (m) semi-terrestriae 9 7 8 \0 \I \0
C. aethiops (m) semi-terrestrial 3 5 6 10 8 9 9
P. anubis (f-juvenile) terrestrial4 \0 8 12 \I \0
P. anubis (m) terrestrial4 6 10 5 10
E.patas (m) terrestrialS 10 10 9 10
'Rodman. 1979; Wheatley. 1978. 1980; Cant. 1988; Fleagle. 1988.
2Rawlins. 1976. 1993; DeRousseau. 1988; Fleagle. 1988; Dunbar. 1989.
3Struhsaker. 1967; Dunbar and Dunbar 1974; Fedigan and Fedigan. 1988; Fleagle. 1988.
4Rose 1977; Fleagle. 1988.
SKingdon. 1971; Chism and Rowell. 1988; Fleagle. 1988; Nakagawa. 1989.
 178 D. Schmitt

Figure 2. Diagram of the experimental enclosure, with a subject traveling on a raised horizontal pole. The force
plate can be seen underneath the runway with a cord connected to the charge amplifiers. Three cameras are placed
in lateral, frontal and overhead positions. Halogen lights illuminate the subject. Infrared emitters/sensors are
placed along the subject's path. In the cabinet, from bottom to top, are the monitor (with split-screen image), the
VCR, 100 HZ filters, oscilloscope (with camera), and special effects generator.

sand and paint. To insure collection of force data from single footfalls without influence of
limbs in contact with other portions of the pole, a 30 cm-long central segment was attached
to the surface of the force plate and separated from two non-instrumented segments by a
2 cm-wide gap on either side (Figure 2). Signals from the force platform were passed
through a low-pass filter with a 100 Hz cut-off and displayed on an oscilloscope as deflec-
tions of three beams representing the vertical, fore-aft, and mediolateral forces (Figure 3).
Images from two of the three video cameras directed at the animal, as well as a
fourth camera that recorded force traces on the oscilloscope, were combined by a special
effects generator so that kinetic and kinematic data were precisely coordinated. For each
animal on each substrate, lateral and frontal images of the animal crossing the plate were
simultaneously displayed as a split field. Vertical, fore-aft, and transverse force curves
were superimposed on the split image using previously established methods (Stern et aI.,
1977; Reynolds, 1981; Larson and Stern, 1989). Subject velocity was determined for each
step either using visible markers on the enclosure or using a series of four infrared sensors
placed at 3/4 meter intervals along the runway that triggered a series of four super-bright
LED lamps visible to the lateral camera.
Using frame-by-frame playback on a VCR, steps in which peak forces were not ob-
scured because of contact by another limb and in which the animal was traveling in a rela-
tively straight path, with no obvious acceleration or deceleration, were selected for
analysis. Angular data from the forelimb and substrate reaction force data were digitized
using a microcomputer, video frame grabber board, and video analysis software.
Forelimb Mechanics during Arboreal and Terrestrial Quadrupedalism in Old World Monkeys 179

_m_t_e_ffi_/~(_L~)______~~~___________
medial (M)

propulsive (P)

braking (8)

vertical (V)

Figure 3. Schematic diagram depicting vertical, braking-propulsive, and mediolateral substrate reaction force
traces. The magnitude of the three components is measured as the deflection of the oscilloscope beam and these
data are used to calculate the magnitude and orientation of the SRR using the following formulae: Sagittal Resul-
tant, (r)=(V2+p2)0.5; Total SRR, (R)=(r+e)O.5; Sagittal Angle= arc cos (Vir); Mediolateral angle = arc cos (r/R).

The substrate reaction resultant (SRR) magnitude and orientation have been found to
be more useful than component forces for analyzing bone and joint morphology (see
Jenkins and Weijs, 1979; Biewener, 1983b, 1989; Full and Tu, 1990; Full et aI., 1991). As
can be seen in Figure 3, these values were calculated by first measuring the height of the
deflection from the baseline for all three substrate reaction components at the vertical
peak, braking peak, propulsive peak, the braking-propulsive transition, and at midsupport
(as defined by Larson and Stern [1989]). Then the magnitude (as a percentage of subject
body weight) and orientation (in degrees deviation from vertical) of the SRR was calcu-
lated in a sagittal and transverse plane using formulae shown in Figure 3 and described in
Schmitt (1994, 1995). Substrate reaction force data for the hindlimb were also collected in
order to calculate the forelimb/hindlimb vertical force ratio on all supports.
Joint angles at the shoulder, elbow, and wrist (Figure 4) were calculated from lateral,
frontal, and overhead images for the same kinetic points described above as well as for
touchdown and lift-off. I identified joint centers initially on sedated subjects and relocated
those points on the videotapes (Schmitt, 1995). To correct for any error caused by out-of-
plane rotation of the forelimb, true intersegmental angles were calculated using an algo-
rithm that calculates 3-D coordinate data for each joint based on the arccosine relation
between the actual length of a segment and its projected length. The formulae for this pro-
cedure (Figure 4) have been described in detail in Schmitt (1994, 1995) and Chan et ai.
(nd).
Additionally, three other variables were measured for this analysis: (l) contact time,
(2) shoulder height, and (3) vertical oscillation of the shoulder. To calculate contact time
the duration of the vertical force was determined from the oscilloscope image. Shoulder
height, used as a measure of "crouching", was calculated trigonometrically using the
known lengths of limb segments and the angles of the joints. By creating right triangles
from these data as shown in Figure 4, three vertical distanceg..-from shoulder to elbow, el-
180 D. Schmitt

a) b)
_D1 -----.

I \ I

:
I
I
,m2
wz: ' I
I

Y
I
I I

I -X2-,t-Ll
r
H2
\
\
j D2
-----:'
------.
./
k
\
\

..1'3 I_~-
D3

Figure 4. Diagram illustrating calculations used in the analysis. (a) Calculation of height (H) and contact distance
(D) using midsupport as an example. A, arm length; B, forearm length; C, palm length; a, arm angle; g, elbow an-
gie; I+h, wrist angle. All other angles and distances are calculated. For example, at midsupport, vertical distance
from shoulder to elbow (H I )=A *sin(a) and total shoulder height=Hl +H2+H3 where H2 and H3 are calculated in a
manner similar to the calculation of HI. Similarly, horizontal distance from shoulder to elbow (0 I )=A *cos(a) and
total distance from point of contact to shoulder=Dl +02+03 where 02 and 03 are calculated in a manner to the
calculation of 01. (b) Calculation of moment arms (i.e., m2) and moments (i.e., SRR magnitude*m2) using mid-
support as an example. Dashed arrow, SRR; k, sagittal angle of the SRR. Horizontal distance of the SRR from ver-
tical (A2)=tan(k)*(H2+H3); horizontal distance of the SRR from the elbow (x2)=(D2+D3)-A2; the moment arm
(m2)=x2*sin(Z). The moment at the elbow then is the total magnitude of the SRR*m2. Moment arms and mo-
ments at the shoulder are calculated in a similar manner.

bow to wrist, and wrist to ground-were calculated. These three distances were added to-
gether to determine the height of the shoulder from the substrate (expressed as a percent-
age of the animal's potential fully-extended height). Finally, from these data on shoulder
height the vertical oscillation of the forelimb was calculated by subtracting the minimum
shoulder height from the maximum shoulder height for each step.
U sing methods similar to those used to calculate shoulder height, the moment arms
and moments of the substrate reaction force resultant at each joint were also calculated
(Figure 4). Following the method of Biewener (l983b), the SRR is placed at the head of
the metacarpals. Right triangles were created in which the vertical height of the joint, the
limb length, and the SRR vector could be used to calculate the distance (in centimeters) of
the SRR vector from the joint and then to calculate a perpendicular distance from the SRR
to the joint at any point during support phase (Figure 4). From these data it was then pos-
sible to calculate the relative moment (expressed as a percentage of body-weight multi-
plied by centimeters for the moment arm or %bwcm) around each joint by mUltiplying the
magnitude of the SRR by the moment arm at each point in support phase.
The goal of this project is to compare kinetic and kinematic values during support
phase of arboreal and terrestrial locomotion. Normally one would compare each value at
specific points (e.g., midsupport) during support phase. Because animals differed in the
timing of kinetic and kinematic points during support phase between the pole and ground,
direct point-by-point comparison was difficult. To correct for this confounding factor, ki-
nematic and kinetic values across substrates were compared by calculating the minimum,
Forelimb Mechanics during Arboreal and Terrestrial Quadrupedalism in Old World Monkeys 181

maximum, and mean values for all kinematic and kinetic variables. This was done in the
following manner. Values for each step were calculated, followed by calculation of aver-
age maximum, minimum, and mean value for each substrate for each subject. The maxi-
mum, minimum, and mean values for each step were also tested for correlations with
speed on each substrate using conservative nonparametric Spearman correlation methods.
Data that were significantly (P<0.05) correlated with speed were compared across sub-
strates within taxa, using a standard analysis of covariance (ANCOVA); the variable of in-
terest was the dependent variable and speed was the independent (effectively held
constant) variable (Sokal and Rohlf, 1981). The result is the calculation of an adjusted
mean and standard error in which the effect of speed has been taken into account. Figures
and tables presented in this paper show only a comparison of terrestrial travel to travel on
the smallest pole on which each subject would walk.

3. RESULTS

Subjects walked at broadly overlapping speeds ranging between 0.6 - 2 mls on the
poles and ground. There were few significant differences in speed across substrates for
any subject. Maximum arm protraction and retraction, maximum elbow flexion, and
shoulder height all showed a significant negative correlation with speed. In contrast SRR
magnitude and vertical peak force showed significant positive correlations with speed.
With the exception of the patas monkey, braking forces showed significant positive corre-
lations with speed, while propulsive forces showed a negative correlation. This latter re-
sult suggests that at higher speed subjects were decelerating. The influence of this pattern
is discussed below.
Height at the shoulder indicates the position of the shoulder relative to the substrate
and gives some estimation of whether a subject is "crouching" and lowering its center of
gravity. As can be seen in Figure 5 all subjects showed a reduced shoulder height while
traveling on an "arboreal" support. The maximum shoulder height was significantly differ-
ent on the pole versus the ground for all subjects except the vervet monkeys (Figure 5). It
is clear that these subjects are crouching in order to lower their center of gravity through-
out at least some part of support phase on arboreal supports.
The reduced shoulder height is brought about partly by differences in the degree of
protraction and retraction of the arm, which can lower the height of the shoulder at the be-
ginning and end of support phase on arboreal supports (Table 2). Only the macaques and
the young female baboon, perhaps because of her age, showed significant increases in pro-
traction on a pole compared to the ground. This pattern reflects the arboreal habits of the
macaques and probably the high mobility in the juvenile female baboon.
In contrast, maximum arm retraction shows a consistent pattern. At liftoff all sub-
jects retracted their arm more on the pole as compared to the ground, although for the ba-
boons and patas this substrate-related difference was nonsignificant (Table 2).
If the forelimb were a single rigid strut, the shoulder would always rise to the same
maximum height during support phase. The elbow joint, however, can moderate the rise of
the shoulder during support phase. The minimum elbow angle is a measure of the elbow
flexion that can reduce the height of the shoulder. Figure 6 shows that all subjects in-
creased elbow flexion on "arboreal" supports relative to the ground, although for the
vervet monkeys this change was nonsignificant. This change in elbow angle from touch-
down to midsupport is probably the major component in maintaining a lower vertical
height throughout the middle portions of support phase.
182 D. Schmitt

M. fascicularis

90

85
..c:
0080
c
0)

.D
.5 75
...o
"0
...... 70
~

65

60 ~-----------------------------------
TO BRAKE MSP VERT BIP PROP LO
Support Phase Points
Maximum shou lder height ground vs. pole, P=O.OOOI

M. mulatta

85

80

65

60 ~--------------------------------------
TD BRAKE MSP VERT BIP PROP LO
Support Pha e Points
Maximum shou lder height ground vs. pole, P=O.0049

Figure 5. Displacement curves for the height of the shoulder as a percentage of forelimb length during support
phase. Dashed lines with circles indicate the adjusted-y mean values on the smallest pole on which subjects would
walk. In some cases (e.g., M. fascicularis) the smallest pole on which one subject walked was smaller than the
smallest pole on which the other subject would walk. In these cases the data were pooled and, therefore, the means
represent the mean of the values for the smallest pole on which each individual subject would walk. Solid lines
with squares indicate adjusted-y mean values during terrestrial locomotion. Error bars indicate one standard error
of the adjusted-y mean value on each substrate. Support phase is divided into touchdown (TO), braking peak
(BRAKE), midsupport (MSP), vertical peak (VERT), braking/propulsive transition (B/P), propulsive peak (PROP)
and liftoff (LO). Significance levels for comparisons of the pole versus the ground using ANCOVA are given at
the base of each graph.
Forelimb Mechanics during Arboreal and Terrestrial Quadrupedalism in Old World Monkeys 183

c. aethiops

90

85
..:
Co
c: 80
I!)

.0
.§ 75
U....
o
...... 70
tI::
65

60 L------------------------------------
TD BRAKE MSP VERT BIP PROP LO
Support Phase Points

Ground and pole not significantly different.

P. anubis

90

85
..:
Co
c: 80 I -.l'" ..,...... 1
"'r - -" . .
I!)

.0
L .....
,. '1- 1'" .... :;
.§ 75
U.... ~ !
o
...... 70

65

60 ~-------------------------------------
TD BRAKE MSP VERT BIP PROP LO
Support Pha e Points
Maximum shoulder height ground vs. pole, P=O.0018

Figure 5. (continued)
184 D. Schmitt

E. patas

90

..<:: 85
Oil
c
<U

.D
.§ 80
~
....
.E
~ 75

70 ~----------------------------------------
TD BRAKE MSP VERT BIP PROP LO
Support Phase Points
Maximum shoulder height ground VS. pole, P=O.0002

Figure 5. (continued)

These kinematic results clearly suggest that either joint moments and stresses are
larger in arboreal quadrupedal primates or there is a change in either magnitude or orienta-
tion of the SRR. Figure 7 illustrates the pattern of SRR for subjects in this study. The
maximum SRR magnitude was significantly lower on poles versus the ground for all sub-
jects except the vervet monkeys (Figure 7). This reduction is driven primarily by a differ-
ence in vertical force on the pole versus the ground. The rhesus macaques experienced
vertical peak forces that were 45% of those seen on the ground. This was the maximum re-
duction among all the species. This was followed by the female baboon (60%), the male
baboon and M.fascicularis (75%), the patas (80%), and finally the vervet monkeys (93%).
In contrast, most subjects showed no significant differences in fore-aft forces on the pole

Table 2. Maximum arm protraction (touchdown) and arm retraction (liftoff) angle in
degrees. Pole, smallest pole on which subjects would walk. Values under Pole and
Ground are the adjusted-y means and one standard error (in parentheses).
P, significance values for comparisons of adjusted-y means using ANCOVA

Arm protraction Ann retraction

Subject Pole Ground P Pole Ground P
M. /ascicu/aris 118(4.7) 106 (3.2) 0.04 32.5 (2.5) 40 (1.7) 0.02
M. mulatta liS (6) 99.5 (4) 0.01 22 (4) 35 (2.5) 0.03
C. aethiops 96 (2.7) 95 (1.9) ns 22 (1.6) 34 (1.1) 0.0001
P. anubis (f) III (4) 103 (3.6) 0.002 43 (1.9) 47 (2.1) ns
P. anubis (m) 99 (2) 111.5 (3) 0.0001 38 (2.3) 44 (3) ns
E. patas liS (3.3) 113.5 (3) ns 44(1.1) 47 (2) ns
ns, not significant
Forelimb Mechanics during Arboreal and Terrestrial Quadrupedalism in Old World Monkeys 185

M. fascicularis

180

170

~ 160
...
0.)

00

~ 150
S!
00
; 140
,
! ... ..- ...i-
-
130

120L----------------------------------------
TO BRAKE MSP VERT TRANS PROP LO
Support Phase Point
Maximum elbow flexion ground vs. pole, P=O.0016

M. mulatta

170

160 !
,-.. 150
'"OJOJ
~ 140
OJ
2-

.
~ 130
00
c

-- •
- /
'" 120
... - J..

110

100~-----------------------
TO BRAKE MSP VERT SIP PROP LO
Support Phase Points
Maximum elbow flexion ground vs. pole, P=O.0004

Figure 6. Displacement curves for the anterior elbow angle during support phase. Same conventions as in Figure
5.
186 D. Schmitt

c. aethiops
170

160

____ 150
</)
<)
<)

~
<)
140
~
<) 130
be
c:
C<l 120 --f---t---
110

100 L-------------------------------------
TO BRAKE MSP VERT BIP PROP LO
Support Phase Events
Ground and pole not significantly different

P. anubis

180

170

~160
...
<)
0/)

2-150
<) ,,
a 140
be
,
130 '--1-
120 ~-------------------------------------
TO BRAKE MSP VERT BIP PROP LO
Support Phase Points
Maximum elbow flexion ground vs. pole, P=O.0297

Figure 6. (continued)
Forelimb Mechanics during Arboreal and Terrestrial Quadrupedalism in Old World Monkeys 187

E. patas

180

170
---'"
<1)

,,
<1)

~ 160 ~
~
<1>
~ 150
<tI
',.. ,
140
'i- -
130 L-------------------------------------
TD BRAKE MSP VERT BIP PROP LO
Support Phase Points
Maximum elbow flexion ground vs. pole P=O.0056

Figure 6. (continued)

versus the ground. This latter result may represent an interaction between a reduction in
fore-aft forces due to lower accelerations of the limbs and an increase in fore-aft forces
due to increased protraction/retraction of the forelimbs on a small pole.
The fact that vertical forces were different across substrates, but that fore-aft forces
remained relatively unchanged has important implications. The sagittal orientation of the
SRR is influenced by the relationship of the magnitude of the vertical force and the mag-
nitude of the fore-aft forces. If the vertical force is reduced but the fore-aft force is not, the
SRR vector is drawn farther away from vertical than when the vertical force is relatively
high. This pattern is enhanced by those subjects where braking forces increased with
speed and thus caused the SRR to be drawn more caudally at higher speeds. In all the
cases, except the vervet monkeys, the ratio of the vertical/fore-aft peak forces was smaller
on the pole than it was on the ground. Therefore, the SRR vector was drawn more
caudally during the first half of support phase and then more cranially (Figure 8). As a re-
sult, as the limb was protracted and then flexed the SRR orientation changed at the same
time and maintained nearly equivalent moment arms on the pole versus the ground.
Table 3 displays the ANCOVA results for moment arms and moments at vertical
peak, the point at which SRR is generally highest in magnitude. In most subjects moment
arms at the shoulder at vertical peak are not significantly different on the ground versus
the pole and often they are lower on the pole than on the ground. In contrast, moment
arms are generally slightly larger at the elbow at vertical peak on a pole compared to the
ground. However, these differences are nonsignificant and do not lead to higher moments
at vertical peak (see below). This pattern holds for most points in support phase as well.
Except for the vervet monkeys, moment arms at the shoulder are consistently (though non-
significantly) lower on the pole versus the ground except at propulsive peak, where they
are always higher. Moment arms at the elbow are generally (though nonsignificantly)
larger on the pole versus the ground.
188 D. Schmitt

M. fascicularis

100

80
.E

---
Q/J
'0:; 60 , ,
~
>.
, ,
"0
40 '" '"
t'"
.D

20

OL----------------------------------------
BRAKE MSP VERT B/P PROP
Support Phase Points

Maximum SRR ground vs. pole, P=O.0083

M. mulatta

110

f-'
-f --1- - - 1, ,
50 ,,
30 ~-------------------------------------
1
BRAKE MSP VERT B/P PROP
Support Phase Points
Maximum SRR ground vs. pole, P=O.0002

Figure 7. Displacement curves for the magnitude of the substrate reaction resultant as a percentage of subject
body weight during support phase. Same conventions as in Figure 5.
Forelimb Mechanics during Arboreal and Terrestrial QuadrupedaJism in Old World Monkeys 189

c. aethiops

130

:;: 110
ell
.;:;
~ 90
>.
"0
0
.0
70
ts>::

SO

30
'BRAKE MSP VERT B/ P PROP
Support Phase Points
Ground and pole not significantly different.

P. anubis

140

120

1:
0)
100
·05
~
>- 80
"0
o
.D
*- 60

40

20 L--------------------------------------
BRAKE VERT MSP TRANS PROP
Support Phase Points

Maximum SRR ground vs. pole, P=O .0071

Figure 7. (continued)
190 D. Schmitt

E. patas

105

95
, ~"
, ,
E0/) 85
·v
75
""
" ... ...
~
>.
"0
0
.D 65 ... ...
* 55

45

35
BRAKE MSP VERT B/P PROP
Support Phase Points

Figure 7. (continued)

Brake Vertical BIP Propulsive

a) Representative "arboreal" step

Figure 8. Schematic diagram of how limb po-

sition and SRR orientation change simultane-
ously on the pole ("arboreal") and the ground
(terrestrial). This composite does not represent
anyone subject and indicates the relative orien-
tation of the arm to the horizontal and the fore-
arm to the arm. (a) The figures indicate the
typical condition on a small pole. (b) The fig-
ures indicate the typical condition on the
ground. In both cases, the points represented
are the braking peak (Brake), vertical peak
(Vertical), braking/propulsive transition (B/P),
and the propulsive peak (Propulsive). Touch-
down and liftoff were omitted from this figure
Brake Vertical BIP Propulsive
b) Representallve terrestrial step because no substrate reaction forces were re-
corded at those points.
Forelimb Mechanics during Arboreal and Terrestrial Quadrupedalism in Old World Monkeys 191

Table 3. Moment anns (cm) and moments (%bwcm, see text) around the shoulder and elbow at
vertical peak. Pole, smallest pole on which subjects would walk. Values under Pole and
Ground are the adjusted-y mean and one standard error (in parentheses)
Shoulder Elbow
Variable Subject Pole Ground P Pole Ground P
Moment arms at M. fascicularis 6.3 (I) 6.44 (I) ns 5.4 (0.9) 5.8 (1.2) ns
M. mulatta 5.2 (1.5) 6.8 (1.3) ns 5.4 (I) 4.8 (0.9) ns
C. aethiops 3.5 (0.7) 3.1(1.4) ns 5.2 (0.5) 4.6 (I) ns
P. anubis (f) 3 (1.5) 5 (I) ns 6.2 (I) 4.6 (I) ns
P. anubis (m) 5.6 (\.I) 5.3 (1.4) ns 7.5 (I) 7.2 (\.6) ns
E. patas 4.8 (\,2) 5.8 (1.6) ns 10.2 (1.4) 10.9 (1.3) ns
Moments around M.fascicularis 377 (127) 650 (192) ns 289 (129) 680(101) 0.04
M. mulatta 437 (134) 645 (117) ns 513 (107) 498 (93) ns
C. aethiops 417 (73) 558 (146) 0.02 680 (93) 395 (185) ns
P. anubis (f) 331 (149) 594 (132) ns 620 (122) 480 (108) ns
P. anubis (m) 408(114) 598 (92) 0.04 792 (122) 890(134) ns
E.patas 285 (92) 430(104) ns 600(111) 907 (124) 0.04
P, significance values for comparisons of adjusted-y means using ANCOVA.
ns, not significant

The combination oflower or equal moment arms and an SRR that is lower in magni-
tude yields bending moments along the forelimb that are generally lower than those on the
pole compared to the ground. Table 3 illustrates this pattern in detail for the moments
around the shoulder and elbow at vertical peak and Figure 9 illustrates this pattern for all
points in support phase. In most cases these are not significant differences and the mo-
ments must be considered effectively equal. In cases where they were statistically signifi-
cant (P<O.05) the moments were lower on the pole versus the ground. As a result of these
relatively equal moments, the muscular force required to maintain a crouched posture on a
pole is either less than or no greater than the force required to maintain an extended pos-
ture on the ground.

4. DISCUSSION AND CONCLUSIONS

The prediction that Old World monkeys lower their center of gravity by changing
limb posture during locomotion on relatively small "arboreal" supports was supported.
Therefore, if force magnitude and orientation had remained equal, the model predicted
higher moment arms and moments along the forelimb. However, the moment arms at most
points during support phase, with the exception of propulsive peak, are not significantly
different on a small pole compared to the ground. Therefore, arboreal quadrupedalism
does not necessarily engender large moments around the forelimb joints and therefore
joint reaction forces are not necessarily higher on arboreal versus terrestrial supports.
It is interesting to note that vervet monkeys represent a consistent exception to the
patterns described here. This may be explained by the fact the vervet monkeys were the
smallest animals in my sample. It is possible that the substrates I designed were not a suit-
able challenge for their normal agile locomotion. It may also be the case that the very
flexed elbow position that vervet monkeys maintain on both the ground and on poles bet-
ter prepares these small mammals to make sudden dashes and jumps to escape predators.
Such a posture would be well suited to a small quadrupedal primate that spends time both
 192 D. Schmitt

800
a)
'-
Q)
:2
is 600
..c
en
-0
c
::J
o
'-
co 400
en
.....
c
Q)
E
o
E 200
c:
c:
C/)

O ~~=b~~dbbE~dd~~4d~~~~bk~ddbb~
M. fascicularis M. mulatta C. aethiops P. anubis (f) P. anubis (m) E. patas

1,000
b)

~ 800
0
.n
Q)

-0
C
::J 600
0
'-
CO
en
.....
c
Q)
400
E
0
E
c:
II
C/) 200

o ~~~~~~~~~~~~~~~~~~~~~~~='

M. fascicularis M. mulatta C. aethiops P. anubis (f) P. anubis (m) E. patas

Figure 9. Adjusted-y mean values for the SRR moments around (a) the shoulder and (b) the elbow on the smallest
pole (light bars) on which subjects would walk versus the ground (dark bars) at five points during support phase.
For each species, from left bars indicate braking peak, vertical peak, midsupport, braking/propulsive transition,
and propulsive peak. Touchdown and liftoff were omitted from this figure because no substrate reaction forces
were recorded at those points. At most points SRR moments on the pole are lower than or equal to those on the
ground. Table 3 contains additional data for moment arms and moments at vertical peak.
Forelimb Mechanics during Arboreal and Terrestrial Quadrupedalism in Old World Monkeys 193

on the ground and in the trees and must leap quickly into the branches when approached
by a predator (Struhsaker, 1967; Fedigan and Fedigan 1988; Fleagle, 1988).
The discovery that primate arboreal quadrupeds can maintain poorly stabilized fore-
limbs and a crouched posture simultaneously helps explain how the role of the forelimb
could progressively change throughout primate evolution. It is also important, however, to
understand the mechanism by which primates avoid the increased stresses that theoreti-
cally accompany a crouched posture. By identifying the mechanism I hope to explain how
this pattern first arose among primates.
One possible way to reduce forces on the forelimb is that described by Reynolds
(1981, 1985a,b). He demonstrated that arboreal primate quadrupeds walking on the
ground actively shift weight posteriorly through activation of hindlimb retractors when the
hindlimb is protracted. It is possible that on arboreal supports this mechanism is used to an
even greater degree, thereby reducing the vertical force on the forelimb further. The data
presented below examines this possibility.
Table 4 shows that the reduction of vertical peak force on the forelimbs on a small
pole versus the ground is greater than the reduction in vertical peak force on the hindlimbs
on a small pole versus the ground. This differential reduction supports the idea that weight
is being shifted posteriorly while animals travel on the small poles. The data in Table 4,
however, also demonstrate that all the animals experience an overall reduction in vertical
peak force on both forelimbs and hindlimbs. The pattern of vertical peak force reduction on
both forelimbs and hindlimbs is not consistent with what one might expect if these animals
were actively shifting weight posteriorly in the manner described by Reynolds (1981,
1985a,b). If the hindlimb lever effects described by Reynolds (1981, 1985a,b) are the only
mechanisms employed to adjust substrate reaction forces, then one would predict a reduc-
tion in peak vertical forelimb forces and an increase in peak vertical hindlimb forces on the
pole versus the ground. Because this is not the case in my sample, an additional mechanism
must be invoked to explain the overall reduction in vertical forces from ground to pole.
Spring-mass models of mammalian locomotion (McMahon and Greene, 1978; Taylor,
1978; McMahon, 1985; McMahon et aI., 1987; McMahon and Cheng, 1990; Blickhan,
1989; Farley et aI., 1993) appear to explain this pattern (Figure 10). In a basic model, the
limb is seen as a massless spring with the body mass concentrated on top of the spring (Fig-
ure lOa). When this simple spring-mass system is hopping forward, the stiffness of the
spring controls the contact and flight time. As the leg-spring becomes more compliant, con-
tact time increases and height of the hop decreases. Therefore, time over which to develop

Table 4. Comparison offorelimb and hindlimb

percentages for vertical peak force on the
smallest pole on which subjects would walk by
vertical peak force on the ground. Values under
forelimb and hindlimb are the
adjusted-y mean and range (in parentheses)

Subject Forelimb Hindlimb

M. fascicularis 74 (69--81) 86 (80-92)
M. mulatta 45 (31-56) 81 (78-88)
C. aethiops 93 (90-100) 97 (95-99)
P. anubis (f) 60 (57-70) 80 (73-87)
P. anubis (m) 74 (71-77) 82 (72-92)
E. patas 80 (77-94) 93 (88--98)
194 D. Schmitt

a)

contact phase aerial phase

(based on Blickhan, 1989)

b) ,
-' "

/
,,
/

t·
/
'.
vertical force

c)

Figure 10. Simple spring-mass models of gait. (a) A massless leg spring, with body mass (m) mounted at one end,
hopping in place. Spring compliance affects contact time. A stiffer spring than the one illustrated would remain in
contact with the substrate for a shorter period of time and have a higher and longer aerial phase. The effect of this
increased contact time is shown in b. (b) The vertical force of a relatively stiff spring (dashed line) has a short du-
ration and a high peak. A more compliant spring (dotted line) yields a vertical force with a longer duration and a
lower peak. (c) A similar spring-mass system in forward locomotion as in human walking. The dashed line indi-
cates a relatively stiffer spring. A compliant spring (dotted line) follows a flatter path during contact and maintains
longer contact time. The angle of attack also reduces the vertical landing velocity relative to the hopping spring il-
lustrated in a.

force increases and the vertical peak force at any point in the contact period decreases (Fig-
ure lOb; Taylor, 1985; Blickhan;1989; Kram and Taylor, 1990). This model can be made
more realistic by making the leg-spring walk forward in the manner of an inverted pendu-
lum whose highest point is at midstance (Figure lOc; Blickhan, 1989; Alexander, 1992). A
compliant spring reduces the rise of the mass by collapsing as the limb approaches mid-
stance and thus provides a "flatter ride" and a longer contact time (McMahon, 1985; Blick-
han, 1989; Alexander, 1992). Additionally, as the limb becomes more protracted the "angle
of attack" of the leg-spring increases and the vertical landing velocity decreases thus reduc-
ing the vertical force developed by the spring (McMahon et aI., 1987; Blickhan, 1989;
McMahon and Cheng, 1990). Empirical data on human walking and running support this
model for human locomotion. When human subjects walk or run on a compliant track or
with extreme hip and knee yield during support they experience a flatter path of the center
of gravity, longer contact times, and lower vertical forces (McMahon and Greene, 1979;
McMahon et aI., 1987; Farley, 1992; Farley et aI., 1993; Schmitt et aI., 1996).
Forelimb Mechanics during Arboreal and Terrestrial Quadrupedalism in Old World Monkeys 195

Table 5. Vertical oscillation of the shoulder (maximum

shoulder height - minimum shoulder height). Pole,
smallest pole on which subjects would walk. Values under
Pole and Ground are the adjusted-y mean and one
standard error (in parentheses). P, significance values for
comparisons ofadjusted-y means using ANCOVA
Subject Pole Ground P
M. fascicularis 5.1 (0.41) 6.9 (1.31) 0.04
M. mulatta 4.3 (0.72) 5.0 (0.98) ns
C. aethiops 4.4 (0.89) 4.6 (0.57) ns
P. anubis 5.3 (0.62) 6.2 (1.1) ns
E. patas 4.4 (0.77) 7.7 (1.48) 0.02
ns. not significant

The compliant gait models described above seem to predict some of the results of
my research. First, the subjects in this study experience a larger change in elbow flexion
between touchdown and mid support on a small pole than they do on the ground (Figure
6). This dramatic flexion during support phase increases the compliance of the forelimb on
the pole. Second, the vertical peak force on "arboreal" supports is reduced compared to
those recorded for terrestrial supports (Table 4). Finally, some of my subjects increased
the angle of attack of the forelimb due to greater arm protraction at touchdown (Table 2).
To confirm that these subjects are using a "compliant" walking gait while on "arbo-
real" compared to terrestrial supports I examined additional indicators of compliant gait
that were not included in my original analysis. Vertical oscillation of the shoulder height, a
measure of the "flatness" of the rise and fall of the subject, was lower on the pole than on
the ground in all subjects. This can be seen in the calculated oscillation displayed in Ta-
ble 5 and in Figure 5 that depicts the path of the shoulder. This "flatter ride" is directly re-
lated to the increased flexion at the elbow. Additionally, for all subjects except the vervet
monkeys hand contact is significantly longer on the small pole compared to the ground
(Table 6). This latter variable is a strong indicator that the gait has become more compliant
(Blickhan, 1989). As a result of increased contact time, the time available to develop force
certainly increases and the peak forces throughout support phase will necessarily decrease.
The support for the hypothesis that the subjects in this study were using a compliant
walking gait is based on data easily accessible from videotape without force traces. This is
useful because that is the source of much additional data on primate locomotion. In addi-
tion, I calculated vertical stiffness following the methods of Farley et al. (1993). The mean
vertical stiffness for the entire group was 5.78 on the ground and 3.69 on a small pole.
These data suggest that the subjects of this study reduce compressive forces on their
forelimb by adopting a "compliant" walking gait. It remains to be seen whether this pattern is
part of a basic primate adaptation. Although this will be the subject of future study, I believe
there is evidence suggesting that compliant walking may be common in all primates when
compared to the walking gaits of nonprimates (Schmitt, 1995). The crouched posture de-
scribed here for Old World monkeys is distinct from the posture adopted by noncursorial
small mammals described by Jenkins (1971). Differences between primate postures in general
and the postures of nonprimate mammals have been noted by other authors (Reynolds, 1981,
1985a,b, 1987; Alexander and Maloiy, 1984; Vilensky, 1989; Vilensky and Larson, 1989; Fis-
cher, 1994; Schmitt, 1995; Larson, this volume) and some of these differences suggest that
primates show a unique pattern of compliant walking gait compared to nonprimate mammals.
 196 D. Schmitt

Table 6. Contact time (in seconds) offorelimb with substrate.

Pole, smallest pole on which subjects would walk. Values
under Pole and Ground are the adjusted-y mean and
one standard error (in parentheses). P, significance values
for comparisons of adjusted-y means using ANCOVA
Subject Pole Ground P
M. fascicularis 0.474 (0.023) 0.334 (0.022) 0.0001
M. mulatta 0.376 (0.03) 0.271 (0.025) 0.03
C. aethiops 0.296 (0.019) 0.262 (0.015) ns
P. anubis 0.356 (0.055) 0.291 (0.049) 0.01
E.patas 0.586 (0.015) 0.534 (0.017) 0.01
ns. not significant

Alexander (1977) has reviewed evidence that a "stiff' walk is characteristic of hu-
mans, dogs, cats, and horses whereas a "compliant" walk is characteristic of small birds.
He further speculated that noncursorial mammals may also use a compliant gait. Current
evidence suggests that the limb kinematics of primates is distinct from the kinematics of
noncursorial small mammals and that primates use a relatively compliant gait (Jenkins,
1971; Rollinson and Martin, 1981; Alexander and Maloiy, 1984; Reynolds, 1987; Fischer,
1994; Larson, 1997). Primates are known to have longer stride lengths and larger angular
excursions than nonprimate mammals (Alexander and Maloiy, 1984; Reynolds, 1987).
Animals with long stride lengths and large angular excursions have relatively low stride
frequencies and longer contact times (Heglund and Taylor, 1988, Kram and Taylor, 1990).
In addition, data on elbow excursion in non primate noncursorial mammals indicate that in
many cases the elbow does not go through a period of flexion during support phase, but
rather remains stable and then extends (Jenkins, 1971; Jenkins and Weijs, 1979; Fischer,
1994). Finally, primate quadrupeds protract their fore- and hindlimbs to a greater degree at
touchdown than do other mammals (Reynolds, 1987; Larson, 1997). Thus, primate quad-
rupeds appear to have longer contact times, greater elbow yield, and an increased angle of
attack of their limbs than do nonprimate mammals.
Another strong indication of a compliant gait in primates can be derived from the
work of Demes and colleagues, who found that Loris tardigradus and Nycticebus coucang
increased speed by protracting their hindlimbs using lateral spinal flexion, thus increasing
stride length. These lorises appear to increase their gait compliance as speed increases by
lowering their center of gravity and by maintaining low stride frequencies and longer
stride lengths. Demes et al. (1990) postulate that these results are related to arboreal life in
which an increase in stride length, angular excursion and a reduction in stride frequency
reduces substrate reaction forces that tend to displace fine branches and disturb insect
prey. Similar patterns of compliant locomotion were found in chameleons that travel and
hunt on fine branches (Peterson, 1984).
If it is, in fact, the case that primates are unusual among mammals in adopting a
compliant walking gait, the question that remains is what the selective agent of such a gait
may have been for early primates. As stated in the introduction, arboreal primates must
maintain a crouched posture in order to balance the relatively wide bodies on relatively
narrow supports. In addition, this crouched posture allows them to change direction rap-
idly. In connection with this latter necessity, and the need to reach and grasp in many
planes, comes the need for gracile, mobile limbs. Compliant walking gait allows both a
crouched posture and mobile forelimbs. In addition, the long contact time of compliant
Forelimb Mechanics during Arboreal and Terrestrial Quadrupedalism in Old World Monkeys 197

walking may increase stability, particularly for animals grasping branches, and, therefore,
further helps maintain balance. There are undoubtedly energetic costs of a compliant pos-
ture (McMahon et aI., 1987). However, the advantage of increased balance and thus the
ability to negotiate arboreal supports with potentially rich food sources may offset the
costs (Rodman, 1979).
If this is the value of this posture, why has it evolved in primates and not other mam-
mals? Compliant gait may reasonably be seen as part of an initial and fundamental primate
adaptation to arboreal locomotion. Early primates had to develop a way of climbing vertical
supports with nails. Whereas non-primate mammals can climb a vertical support by inter-
locking with the substrate and can "crawl" up the support with similar kinematics as in
walking (Cartmill, 1974, 1985), primates need long arms to reach around the substrate
(Fleagle et aI., 1981; Yamazaki and Ishida, 1984; Cartmill, 1985; Hirasaki et aI., 1993; Lar-
son, this volume) and to grasp an overhead support. Primates would thus need long limbs
that they were able to protract extensively (see Larson, this volume). With long arms, the
need to crouch for balance was accentuated. Early primates could crouch in manner that in-
volved protraction of the humerus and significant increase in elbow flexion during support
phase, a pattern not always seen in nonprimate mammals (Larson, this volume).
Primates may have developed a "compliant walk" in order to maintain critical postural
and anatomical adaptations for arboreal life on terminal branches. Compliant walking also re-
leases the forelimb from constraints imposed by the need for a high degree of osteological sta-
bilization. This change in both anatomy and function allowed for the evolution of the use of
forelimbs in tension, a feature that is critical for the evolution of antipronograde postures and
ultimately for the removal of the forelimbs from locomotion (Stern, 1976).

ACKNOWLEDGMENTS

This study was part of my thesis research conducted under the patient guidance of
Dr. Susan G. Larson, to whom I am very grateful. I also wish to thank Dr. Jack T. Stern Jr.
and the other members of my dissertation guidance committee, Drs. William L. Jungers,
Brigitte Demes, Farish Jenkins, and Michael D. Rose for their valuable guidance and their
patience. I wish to also thank Marianne Crisci for her expert animal training and assis-
tance in data collection, Yvette Pirrone for her invaluable help in animal training, data col-
lection and analysis, and Luci Betti-Nash for her help with illustrations. Drs. Pierre
Lemelin, W. Scott McGraw, Christine Wall, and Roshna Wunderlich all have provided
useful advice on this project. I am also grateful for thoughtful reviews by Drs. Jeff
Meldrum, Fred Anapol, and two anonymous reviewers. This research was supported by
NSF BNS 8819621, and 8904576, and SBR 9209004, and a Sigma Xi Grant-in-aid of Re-
search. I am particularly grateful to Drs. Elizabeth Strasser, Alfred Rosenberger, Henry
McHenry, and John Fleagle for organizing both the conference at which this paper was
originally presented and this volume.

REFERENCES
Alexander RM (1977) Mechanics and scaling of terrestrial locomotion. In T. Pedley (ed.): Scale Effects in Animal
Locomotion. London: Academic Press, pp. 93-110.
Alexander RM (1992) A model of bipedal locomotion on compliant legs. Phil. Trans. R. Soc. Lond. B
338:189-198.
 198 D. Schmitt

Alexander RM, and Maloiy GM (1984) Stride lengths and stride frequencies of primates. J. Zoo!., Lond.
202:577-582.
Alexander RM, and Ker RF (1990) Running is priced by the step. Nature 346:220-221.
Ashton EH, Oxnard CE, and Spence TF (1965) Scapular shape and primate classification. Proc. Zoo!. Soc. Lond.
145:125--142.
Ashton EH, Healy M, Oxnard CE, and Spence T (1968) The combination of locomotor features of the primate
shoulder girdle by canonical analysis. J. Zoo!. 147:406-429.
Biewener AA (1982) Bone strength in small mammals and bipedal birds: Do safety factors change with body size?
J. expo Bio!. 98:289--30 I.
Biewener AA (l983a) Allometry of quadrupedal locomotion: The scaling of duty factor, bone curvature and limb
orientation to body size. J. expo BioI. 105: 147-171.
Biewener AA (1983b) Locomotor stresses in the limb bones of two small mammals: The ground squirrel and chip-
munk. J. expo BioI. 103:131-154.
Biewener AA (1989) Scaling body support in mammals: Limb posture and muscle mechanics. Science 245:45-48.
Biewener AA (1990) Biomechanics of mammalian terrestrial locomotion. Science 250: I 097-11 03.
Blickhan R (1989) The spring-mass model for running and hopping. J. Biomech. 22: 1217-1227.
Bown T, Kraus M, Wing S, Fleagle J, Tiffany B, Simons E, and Vondra C (1982) The Fayum primate forest revis-
ited. J. Hum. Evol. 11:624-628.
Cant JGH (1988) Positional behavior of long tailed macaques (Macacafascicularis) in northern Sumatra. Am. J.
Phys. Anthropo!. 76:29--37.
Cartmill M (1974) Pads and claws in arboreal locomotion. In FA Jenkins (ed.): Primate Locomotion. New York:
Academic Press, pp. 45--83.
Cartmill M (1985) Climbing. In DM Bramble, KF Liem and DB Wake (eds.): Functional Vertebrate Morphology.
Cambridge: Belknap Press, pp. 73-88.
Chan LK, Schmitt D, and Cole TM III (nd) A method for calculation of 3-D limb position using a single video-
camera. Submitted Am. J. Phys. Anthropol.
Chism J, and Rowell TE (1988) The natural history of patas monkeys. In A Gautier-Hion, F Bourliere, T Gautier,
and J Kingdon (eds.): A Primate Radiation: Evolutionary Biology of the African Guenons. Cambridge:
Cambridge University Press, pp. 413-437.
Darwin C (1871) The Descent of Man. New York: Appleton.
Demes B, Jungers WL, and Nieschalk U (1990) Size- and speed-related aspects of quadrupedal walking in slender
and slow lorises. In F Jouffroy, M Stack, and C Niemitz (eds.): Gravity, Posture, and Locomotion in Pri-
mates. Florence: Sedicesimo, pp. 175--198.
Demes B, Larson SG, Stern JT Jr., Jungers WL, Biknevicius AR, and Schmitt D (1994) The kinetics of primate
quadrupedalism: "Hindlimb drive" reconsidered. J. Hum. Evol. 26:353-374.
Dunbar D (1989) Locomotor behavior of rhesus macaques (Macaca mulatta) on Cayo Santiago. J. Puerto Rican
Health Sci. 8:79--85.
Farley CT, Glasheen J, and McMahon TA (1993) Running springs: Speed and animal size. J. expo BioI. 185:71-86.
Fedigan L, and Fedigan LM (1988) Cercopithecus aethiops: A review of field studies. In A Gautier-Hion, F Bour-
liere, T Gautier, and J Kingdon (eds.): A Primate Radiation: Evolutionary Biology of the African Guenons.
Cambridge: Cambridge University Press, pp. 413-437.
Feldesman MR (1976) The primate forelimb: A morphometric study of diversity. U. Oregon Anthropol. Papers
l:iii-149.
Fischer MT (1994) Crouched posture and high fulcrum, a principle in the locomotion of small mammals: The ex-
ample of the rock hyrax Procavia capensis (Mammalia: Hyracoidea). J. Hum. Evol. 26:501-524.
Fleagle JG, Stern JT Jr., Jungers WL, Susman RL, Vangor AK, and Wells JP (1981) Climbing: A biomechanical
link with brachiation and with bipedalism. Symp. Zool. Soc. Lond. 48:359--375.
Fleagle JG (1988) Primate Adaptation and Evolution. New York: Academic Press.
Full RJ, and Tu MS (1990) Mechanics of six-legged runners. J. expo BioI. 148: 129--146.
Full RJ, Blickhan R, and Ting LH (1991) Leg design in hexapedal runners. J. expo BioI. 158:369--390.
Grand TI (I 968a) The functional anatomy of the lower limb of the howler monkey (Alouatta caraya). Am. J. Phys.
Anthropol. 28: 161-181.
Grand TI (I 968b) Functional anatomy of the upper limb. Biology of the howler monkey (Alouatta caraya). Bibl.
Primatol. 7:104-125.
Harrison T (1989) New postcranial remains of Victoriapithecus from the middle Miocene of Kenya. J. Hum. Evol.
18:1-54.
Heglund NC, and Taylor CR (1988) Speed, stride frequency, and energy cost per stride: How do they change with
body size and gait? J. expo BioI. 138:301-318.
Forelimb Mechanics during Arboreal and Terrestrial QuadrupedaJism in Old World Monkeys 199

Hirasaki E, Kumakura H, and Matano S (1993) Kinesiological characteristics of vertical climbing in Ateles geor-
froyi and Macacafuscata. Folia Primato!' 61: 148-156.
Jenkins FA Jr. (1971) Limb posture and locomotion in the Virginia opossum and other non-cursorial mammals. 1.
Zoo!., Lond. 165:303-315.
Jenkins FA Jr. (1974) Tree shrew locomotion and the origin of primate arborealism. In FA Jenkins (ed.): Primate
Locomotion. New York: Academic Press, pp. 85-116.
Jenkins FA Jr., and Weijs W (1979) The functional anatomy of the shoulder in the Virginia opossum (Didelphis vi/'-
giniana). 1. Zoo!., Lond. 188:379-410.
Jolly C (1967) The evolution of the baboons. In H Vartog (ed.): The Baboon in Medical Research. Vo!' II. Austin:
University of Texas Press. pp. 323-338.
Jungers WL, and Burr DB (1992) Body size, long bone geometry, and locomotion in cercopithecoid monkeys. Am.
J. Phys. Anthropol., Supp!. 14:96.
Jungers WL, and Burr DB (1995) Body size, long bone geometry and locomotion in quadrupedal monkeys. Z.
Morph. Anthropol. 80:89-97.
Kimura T (1985) Bipedal and quadrupedal walking of primates: Comparative dynamics. In S Kondo, H Ishida, T
Kimura, M Okada, M Yamazaki, and J Prost (eds.): Primate Morphophysiology, Locomotor Analyses and
Human Bipedalism. Tokyo: University of Tokyo Press, pp. 47-58.
Kimura T (1992) Hindlimb dominance during primate high-speed locomotion. Primates 33:465-474.
Kimura T, Okada M, and Ishida H (1979) Kinesiological characteristics of primate walking: Its significance in hu-
man walking. In ME Morbeck, H Preuschoft, and N Gomberg (eds.): Environment, Behavior, Morphology:
Dynamic Interactions in Primates. New York: Gustav Fischer, pp. 297-311
Kram R, and Taylor R (1990) Energetics of running: A new perspective. Nature 346:265-267.
Larson SG (this volume) Unique aspects of quadrupedal locomotion in nonhuman primates. In E Strasser, JG
Fleagle, AL Rosenberger, and HM McHenry (eds.): Primate Locomotion: Recent Advances. New York:
Plenum Press, pp. 157-173.
Larson SG, and Stern JT Jr. (1989) The role of propulsive muscles of the shoulder during quadrupedalism in
vervet monkeys (Cercopithecus aethiops): Implications for neural control of locomotion in primates. 1.
Motor Behavior 21:457-472.
Le Gros Clark WE (1959) The Antecedents of Man. Edinburgh: Edinburgh University Press.
Manaster BJ (1975) Locomotor Adaptation within the Cercopithecus, Cercocebus, and Presby tis genera: A multi-
variate Approach. Ph.D. Dissertation, University of Chicago.
Manaster BJ (1979) Locomotor adaptation within the Cercopithecus genus: A multivariate approach. Am. 1. Phys.
Anthropol. 50:169-182.
McMahon TA (1985) The role of compliance in mammalian running. 1. expo BioI. 115:263-282.
McMahon TA, and Cheng GC (1990) The mechanics of running: How does stiffness couple with speed? J.
Biomech.23:65-78.
McMahon TA, and Greene PB (1979) The influence of track compliance on running. J. Biomech. 12:893-904.
McMahon TA, Valiant G, and Frederick EC (1987) Groucho Running. 1. Appl. Physiol. 62:2326-2337.
Meldrum OJ (1991) Kinematics of the cercopithecine foot on arboreal and terrestrial substrates with implications
for the interpretation of hominid terrestrial adaptations. Am. J. Phys. Anthropol. 84:273-290.
Morbeck ME (1979) Forelimb use and positional adaptation in Colobus guereza: Integration of behavioral, eco-
logical, and anatomical data. In M Morbeck, H Preuschoft, and N Gomberg (eds.): Environment, Behavior,
and Morphology: Dynamic Interactions in Primates. New York: Gustav Fisher, pp. 95-118.
Nakagawa N (1989) Activity budget and diet of patas monkeys in Kala Maloue National Park, Cameroon: A pre-
liminary report. Primates 30:27-34.
Napier JR (1967) Evolutionary aspects of primate locomotion. Am. J. Phys. Anthropol. 27:333-342.
Napier JR, and Davis PR (1959) The forelimb skeleton and associated remains of Proconsul africanus. Fossil
Mammals of Africa # 16.
O'Connor BL, and Rarey K (1979) Normal amplitudes of pronation and supination in several genera of anthropoid
primates. Am. 1. Phys. Anthropol. 51:39-44.
Rawlins RG (1993) Locomotive and manipulative use of the hand in Cayo Santiago macaques (Macaca mulatta).
In H Preuschoft and OJ Chivers (eds.): Hands of Primates. Berlin: Springer-Verlag, pp. 21-30.
Reynolds TR (1981) Mechanics of interlimb weight redistribution in primates. Ph.D. Dissertation, Rutgers Univer-
sity.
Reynolds TR (1985a) Mechanics of increased support of weight by the hindlimbs in primates. Am. J. Phys. An-
thropol. 67:335-349.
Reynolds TR (I 985b) Stresses on the limbs of quadrupedal primates. Am. J. Phys. Anthropol. 67:351-362.
Reynolds TR (1987) Stride length and its determinants in humans, early hominids, primates, and mammals. Am. 1.
Phys. Anthropol. 72:101-115.
200 D. Schmitt

Ripley S (1979). Environmental grain, niche diversification, and positional behavior in Neogene primates: An evo-
lutionary hypothesis. In ME Morbeck, H Preuschoft, and N Gomberg (eds.): Environment. Behavior, Mor-
phology: Dynamic Interactions in Primates. New York: Gustav Fischer, pp. 37-74.
Rodman PS (1979) Skeletal differentiation of Macaca fascicularis and Macaca nemestrina in relation to arboreal
and terrestrial quadrupedalism. Am. J. Phys. Anthropol. 51:51-;62.
Rollinson J, and Martin RD (\981) Comparative aspects of primate locomotion with special reference to arboreal
cercopithecines. Symp. Zool. Soc. Lond. 48:377-427.
Rose MD (1973) Quadrupedalism in primates. Primates 14:337- 358.
Rose MD (1977) Positional behavior of olive baboons (Papio anubis) and its relationship to maintenance and so-
cial activities. Primates 18:59-116.
Rose MD (1988) Another look at the anthropoid elbow. J. Hum. Evol. 17: 193-224.
Rose MD (1989) New postcranial specimens of catarrhines from the middle Miocene Chinji Formation, Pakistan:
Descriptions and a discussion of the proximal humeral functional morphology in anthropoids. J. Hum.
EvoI.18:131-162.
Rose MD (\991) The process ofbipedalization in hominoids. In B Senut and Y Coppens (eds.): Origine(s) de la
Bipedie chez les Hominides, Cah. Paleoanthrop. Paris: Editions du CNRS, pp. 37-49.
Schaftler MB, and Burr DB (\ 984) Primate cortical bone microstructure: Relationship to locomotion. Am J. Phys.
Anthropol. 65:191-197.
Schaftler MB, Burr DB, Jungers WL, and Ruff CB (1985) Structural and mechanical indicators oflimb specializa-
tion in primates. Folia Primatol. 45:61-75.
Schmitt D (1994) Forelimb mechanics as a function of substrate type during quadrupedal ism in two anthropoid
primates. J. Hum. Evol. 26:441-458.
Schmitt D (1995) A Kinematic and Kinetic Analysis of Forelimb Use During Arboreal and Terrestrial Quadru-
pedalism in Old World Monkeys. Ph.D. Dissertation, State University of New York at Stony Brook.
Schmitt D, Stern JT Jr., and Larson SG (\ 996) Compliant gait in humans: Implications for substrate reaction
forces during australopithecine bipedalism. Am. J. Phys. Anthropol., Supplement 22:209.
Sokal R, and Rohlf J (\981) Biometry. The Principles and Practice of Statistics in Biological Research. New York:
Freeman.
Stern JT Jr. (\ 976) Before bipedality. Yrbk. Phys. Anthropol. 20:59-68.
Stern JT Jr., Wells JP, Vangor AK, and Fleagle JG (1977) Electromyography of some muscles of the upper limb in
Ateles and Lagothrix. Yrbk. Phys. Anthropol. 20:498--507.
Struhsaker T (1967) Ecology of vervet monkeys (Cercopithecus aethiops) in the Masai-Amboseli Game Reserve,
Kenya. Ecology 48:891-904.
Taylor CR (\ 978) Why change gaits? Recruitment of muscles and muscle fibers as a function of speed and gait.
Amer. Zool. 18:153-161.
Taylor CR (1985) Force development during sustained locomotion: A determinant of gait, speed and metabolic
power. J. expo BioI. 115:253-262.
Vilensky JA (\ 989) Primate quadrupedal ism: How and why does it differ from that of typical quadrupeds? Brain
Behav. Evol. 34:357-364.
Vilensky JA, and Larson SG (1989) Primate locomotion: Utilization and control of symmetrical gaits. Ann. Rev.
Anthropol. 18: 17-35.
Wheatley B (1982) Energetics and foraging in Macacafascicularis and Pongo pygmaeus and a selective advan-
tage for large body size in the orangutan. Primates 23:348--363.
Wood Jones, F (1926) The Arboreal Man. New York: Hafner.
Yamazaki N, and Ishida H (1984) A biomechanical study of vertical climbing and bipedal walking in gibbons. J.
Hum. Evol. 13:563-571.
 III

DATA ACQUISITION AND ANALYTIC

TECHNIQUES
INTRODUCTION TO PART III

Elizabeth Strasser

Since 1965 there have been some major advances in the methods available for
studying aspects of primate locomotion. The first three papers in this section illustrate
some technological advances in data acquisition while the last three papers cover some
analytical issues.
The paper by John Kappleman (Chapter 12) is an up-to-date review of technologies
for acquiring three-dimensional data and their implications for studying primate
biomechanics and locomotion. He covers software programs, various innovations in hard-
ware, and some analytical techniques. Kappelman provides internet web site addresses
where a sampling of software programs can be accessed. He discusses the strengths and
pitfalls of computerized tomography, magnetic resonance imaging and laser digitizers. Fi-
nally, he describes finite element analysis and the virtual museums of tomorrow. After
reading this paper, one is impatient to try everything out!
The application of one technique that Kappelman discusses, laser scanning, is ele-
gantly demonstrated by Leslie Aiello and colleagues in Chapter 13. These authors illus-
trate one use of laser scanning: for interpreting fossil hominid postcranial material. They
address the affiliation of OH35, a hominid tibia and fibula from Olduvai Gorge, with
OH8, the foot attributed to Homo habilis. They examine the fossil material in the context
of homologous measurements of the talocrural joint surfaces from a comparative sample
of hominoids and conclude that the two Olduvai specimens not only come from different
individuals, but most likely from different species. Aiello et al. 's study is a tantalizing ex-
ample of what three-dimensional data acquisition and analysis holds for the future.
Another example of a technological advance is provided by Brigitte Demes (Chapter
14). Demes introduces the reader to strain gauge technology by covering its principles,
theory and practice as well as the history of its application. She describes the two main ar-
eas of strain gauge use in functional morphology and illustrates them with case studies of
her own. The first area is direct bone strain analysis and Demes' case-study is an in vivo
bone strain analysis of the macaque ulna. In this study, she tests some common "assump-
tions inherent in the functional analysis of bone morphology" about how bones are loaded
during locomotion (Demes, pg. 243). The second area is the use of strain gauges to con-
struct custom force transducers. In Demes second case study she uses such a force

203
204 E. Strasser

transducer to measure the takeoff and landing forces generated by Propithecus and Ha-
palemur on a compliant pole. Strain gauge analysis and other experimental methods prom-
ise a geometric increase in our understanding of the forces and factors involved in primate
locomotion.
Charles Oxnard, who was at the original symposium organized by Warren Kinzey,
and his students have applied multivariate methods to literally every skeletal system in
primates as well as to aspects of primate behavior and ecology. His paper in this volume
(Chapter 15) is, as one reviewer commented, an appropriate successor to his 1965 contri-
bution. Oxnard convincingly demonstrates how morphometric studies on individual ana-
tomical units yield information about function, while morphometric studies on
combinations of anatomical units shed light about evolutionary relationships. Further-
more, Oxnard presents an ingenious algebraic analogy that explains why the information
content from studies of individual anatomical units differs from those on combinations of
units. This paper is an elegant summary of the contributions that Oxnard and his students
have made to primatology over the many years since the original conference.
Laurie Godfrey and collaborators (Chapter 16) write about the application, use, and
misuse of heterochrony. They argue that the notion that ontogenetic scaling implies little
to no functional-behavioral differences between ancestor and descendant is naive. It is
also apparent that isometries contribute little to diagnosing heterochronic processes. Fur-
thermore, that allometric commonality somehow necessarily implies commonality of an-
cestral and descendant behavioral trajectories is exposed as nonsense. To paraphrase one
of the reviewers, Godfrey et al. suggest a remedy to some problems in the field that are a
result of over-simplification, inconsistent jargon and sloppy thinking. This paper contrib-
utes significantly to clarifying the issues surrounding heterochrony in general and in spe-
cific application to the evolution of locomotion.
William Jungers and colleagues (Chapter 17) report on a different type of scaling.
They performed scaling analyses of long bone external dimensions and quantitative esti-
mates of long bone cross-sectional geometry based on bone mineral density scans for two
sister clades: Cercopithecinae and Colobinae. They find that positive allometry in the
measures of long bone strength relative to body mass is insufficient to maintain mechani-
cal equivalence across the range of body sizes seen in cercopithecids. As a consequence,
large-bodied animals have to employ behavioral and postural adjustments to make up for
reduced relative bone strength. The authors also find differences in bone strength between
cercopithecines and colobines, which they relate to differences in the compliance of sub-
strates habitually used by the groups. Finally, Jungers et al. report that direct measure-
ments of cross-sectional geometry are superior to measurements of external dimensions to
estimate cross-sectional geometry for understanding long bone biomechanics. This is per-
haps not an unexpected result, but given the logistical problems involved in obtaining di-
rect measurements of cross-sectional geometry, it is comforting that Jungers et al. also
validate the importance of data and inferences drawn from external diameters.
In summary, the collection of papers in this section are illustrative of advances made
since 1965, as well as exciting new technologies on the horizon for all primatologists.
 12

ADVANCES IN THREE-DIMENSIONAL DATA

ACQUISITION AND ANALYSIS

John Kappelman

Department of Anthropology
The University of Texas at Austin
Austin, Texas 78712-1086

1. INTRODUCTION

Advances in computer technology have greatly expanded the range of topics that
can be investigated within the general field of primate locomotion and the field has wit-
nessed great strides in such areas as telemetered electromyography (Jungers and Stern,
1980). The study of the relationship between form and function in living animals remains
a critical focus of functional morphology because this approach offers the only bridge to
understanding the function of extinct species and the nature of evolutionary transitions.
Technological advances have not, however, changed the fact that the fossil record is al-
most exclusively limited to skeletal remains, and studies of this material are necessarily
more limited and are usually restricted to quantifying the shape and size of different skele-
tal elements and how these variables are related to function.
Most past studies have usually been restricted to quantifying shapes or lengths in
two dimensions, or reducing complex three-dimensional geometries to two dimensions for
the purpose of taking linear measurements (see Ruff, 1989), because the techniques for
gathering extensive three dimensional data did not exist. Many recent advances in data ac-
quisition build on these more traditional and time honored measurement techniques (see
Martin, 1989), and the addition of the computer interface permits the rapid capture of two-
dimensional linear measurements (Spencer and Spencer, 1995) and the calculation of true
areal data. Some of the most exciting advances of the past few years are to be found in the
acquisition of both internal and external three-dimensional shape information, with some
of this work based on a necessarily limited number of landmark data points (see
Richtsmeier, 1989; Corner and Richtsmeier, 1993). In addition, advances in data storage,
transmission technologies, and the internet facilitate the sharing of data in ways that were
never before envisioned.

205
206 J. Kappelman

2. COMPUTERS, STORAGE MEDIA, AND SOFTWARE

One of the most important technological innovations of the past twenty years is the
personal computer. Computing power that was formerly in the domain of large mainframe
computers is now found in the hands of nearly every researcher. Research and develop-
ment during recent years has continued the trend of producing ever faster central process-
ing units (CPUs), now with multiple CPUs, and ever larger hard drive storage devices.
With regard to future forecasts, the only prediction that we can be certain of is that what-
ever we purchase today will be replaced by something faster with more storage capability
tomorrow! Developments in read-and-write storage media have continued in lockstep as
well. Optical erasable disks are presently at nearly 5 gigabytes (GB) in capacity, and the
digital audio tape (DAT) technology is not far behind. The cost of write-once CD ROM
technology is now within the reach of many labs, and the long life span and low per-unit
cost make this medium one of the most cost effective means for backing up, sharing, and
storing data. Advances in CD ROM technology with the introduction of the Digital Ver-
sitile Disk (DVD) promise to increase the storage capability of this medium by over an or-
der of magnitude thus making it possible to archive and share nearly any size data set
(Duncan and Kappelman, 1991).
The ubiquity of personal computers has in turn spawned the development of user-
friendly operating systems and a nearly endless supply of software, some of which has
been written specifically for the functional morphologist, but most of which is easily co-
opted from other disciplines. For example, off-the-shelf software that is designed for ar-
chitectural and mechanical design such as Autodesk's AutoCAD® is easily co-opted for

Figure 1. Video capture and input using Autodesk's AutoCAO"', a computer-aided-design program, permits a vari-
ety of traditional and true areal measurements.
Advances in Three-Dimensional Data Acquisition and Analysis 207

such purposes as measuring phalangeal curvature (Duncan et aI., 1994) or limb joint areas
(Hake, 1992; Figure I), while software packages such as the National Institute of Health's
(NIH) Image®, and video image analysis (Spencer and Spencer, 1995) are more particu-
larly designed for the needs of a biomedical researcher or functional morphologist. Other
software programs such as Alias® and Wavefront® (Silicon Graphics), Soft Image® (Mi-
crosoft), or 3D Studio Max® (Autodesk) can be used to produce animations of joint and
muscle movements in two and three dimensions. Sequences of actual primate locomotion
captured on video can be easily downloaded to the computer using a video capture board
and software such as Adobe's Premiere®, and the limb or trunk movements can be studied
frame by frame with some of the packages listed above. Information about these programs
can be accessed at the Internet web sites listed in Table 1.

3. TECHNIQUES OF DATA ACQUISITION

3.1. Inside and Out: Conventional and High Resolution X-Ray

Computerized Tomography (CT) and MRI
Many of the new technologies for acquiring digital data have either been invented in
the past few years or represent refinements of existing technologies. The primary advan-
tage of X-rays and MRI over some of the other digitizing techniques discussed in this arti-
cle is that these techniques permit the imaging of internal as well external structures.
Conventional X-ray studies have witnessed a long history of use in physical anthropology
(e.g., Coolidge, 1933; Weidenreich, 1940) that continues to today (e.g., Mann, 1975;
Winkler et aI., 1996). Even though the shadowgraph technique of passing parallel X-rays
through the object has been around for many years, refinements in conventional X-ray
continue and include polychromatic techniques that improve resolution by varying both
the distance of the object from the X-ray source and absorption times (Wilkins et aI.,
1996). This new technique produces a very high level of resolution.
The development of X-ray CT began as a medical diagnostic tool in 1971 (see Taube
and Adelstein 1987 for a brief history of development) and witnessed a wide number of
applications in physical anthropology (e.g., Tate and Cann, 1982; Conroy and Vannier,
1984, 1987; Vannier et aI., 1985; Ruff and Leo, 1986). Computerized tomography differs
from conventional radiography in that the X-rays are restricted to a plane and the intensi-
ties of the beam before and after it passes through the specimen are measured. The meas-
urement of a single slice of a specimen is completed by either rotating the object 360 0 on
a turntable, or by rotating the X-ray source itself 360 0 around the object, with this latter
method being critical in medical applications so as to minimize motion sickness in the pa-
tient. (This "axial" rotation of the X-ray source around the patient adds the "A" to "CT' to
form the commonly used term "CAT' scan.) In order to complete a scan, the specimen is
either translated vertically on the turntable or the X-ray source is moved along the long
axis of the specimen. Conventional CT has a resolution on the order of 1-2 mm (or more)
thickness, and this does limit the imaging of structures that are below this size, but some
increase in resolution can be obtained by overlapping closely spaced images. Image recon-
struction is completed by using what are usually company-specific algorithms to calculate
the linear attenuation coefficient !l for each point in the specimen.
Once the image reconstruction is completed, analysis can begin on either the single
slices or on the reconstructed 3-D object. The CT images can be viewed as either a digital
file format on the computer screen or printed out as a series of hard copy films. Most com-
 ....
=>
QO

Table 1. Internet addresses for information on 3-D data acquisition hardware, analysis software, and files

Product type Product name and descri ption Internet address

Analysis Software NIH Image®, image analysis and measurement software http://biocomp.arc.nasa.gov/3dreconstruction/software/nihimage.html
Soft Image"', animation software http://www.softimage.com/softimage/
Alias®lWavefront®, animation software http://www.alias.com
Autodesk's AutoCAD®, and 3D Studio® software http://www.autodesk.com
I-DEAS®, FEM software http://www.sdrc.com
ABAQUS", FEM software http://www.abaqus.com
Sculpt", 3-D editing software http://www.engr.colostate.edu/-dga/sculpt.html
VoxBlast®, volume rendering software http://www.vaytek.com
General listing of 3-D reconstruction software http://biocomp.arc.nasa.gov/3dreconstruction/software/
Hardware High Resolution X-ray CT Lab at UT Austin http://www.ctlab.geo.utexas.edu
Laser Design: laser scanners http://www.laserdesign.com/prodsurv.htm
Digibotics, Inc.: laser scanners http://www.digibotics.com
Cyberware, Inc.: laser scanners http://www.cyberware.com
Thermoelastic stress analysis http://www.StressPhotonics.com/
DTM, Inc.: laser sintered 3-D printouts http://www.dtm-corp.com
Data Sets, Files, and Animations US National Library of Medicines: The Visible Human Project http://www.nlm.nih.gov/research/visible/visible_human.html
Smithsonian Laser Scanner Laboratory http://www.digitaldarwins.sarc.msstate.edulbvl.html
W. W. Howell's modern human craniometric data set ftp://utkux.utk.edu!pub/anthro/HOWTXT.ZIP
Virtual Human Osteology Guide http://J,vww.dla.utexas.edu/depts/anthro/kappelman/osteolog.html
3-D animation of human skull http://www.dla.utexas.edu/depts!anthro/kappelman/skull.mpg
Virtual Laboratories for Introductory Physical Anthropology http://www.dla.utexas.edu/depts/anthro/kappelman/currdev.html
CT generated 3-D data sets http://biocomp.arc.nasa.gov/3dreconstruction/datal
'This is not an exhaustive list of all available hardware, analysis software, or Internet addresses with data sets. It is recommended that the interested user conduct a keyword search on the Internet to locate
additional addresses.
:-
;:0::
"Q
l
'"
a
'"
=
Advances in Three-Dimensional Data Acquisition and Analysis 209

panies that build medical CT scanners use proprietary file formats, and the digital images
can only be viewed and analyzed on their hardware with their software. In almost all cases
it is simply not possible to save and export the images in a file format that can be viewed
and analyzed on a personal computer. This is a serious limitation with proprietary CT
hardware and software and often serves to restrict the researcher to analyzing the data on
the hospital computer, and this can be expensive and often difficult to arrange. This limita-
tion has forced many researchers to output the image reconstructions as hard copy films
and then redigitize the image (e.g., Ruff, 1989).
High resolution X-ray CT (HRXCT) (Figure 2), also known sometimes as industrial
CT, offers several advantages over conventional medical CT. First, HRXCT uses a range
of higher energy sources (typically 125-450 kV) than those available in medical CT,
which makes the instrument capable of penetrating much denser objects including rocks
and very heavily mineralized fossils. (Of course, these higher intensities also prevent this
instrument from being used on living organisms.) Second, the X-ray detectors are modu-
lar, and these can be switched between linear and area detector arrays, which increases the
resolution of the instrument. The combination of modular sources and detectors produces
a CT system that can scan a variety of specimens across a wide range of resolutions. For
example, the current resolution available on a medical CT is 1 mm (1000 f..Lm) while that
on a HRXCT ranges from 10 f..Lm to 100 f..Lm. This increase in resolution means that much
smaller objects or structures such as trabecular bone (Figure 3) or even the developing
premolar in the mandible of Alphadon, a tiny early Mesozoic marsupial (Cifelli et ai.,
1996), can be clearly imaged. One other advantage to HRXCT is that the digital files are
not written in a proprietary file format, but are exportable in a variety of formats that are

Figure 2. High resolution X-ray computerized tomography (HRXCT) offers several advantages over conventional
medical CT, including better resolution and non-proprietary file formats. This instrument is produced by Bio-Im-
aging Research, Inc. and is housed at the University of Texas at Austin (see http://www.ctlab.geo.utexas.edu).
210 J. Kappelman

Figure 3. High resolution X-ray CT image

of a human distal femur showing an arrow-
head that fragmented upon impact. High
resolution X-ray CT provides a dramatic in-
crease over the resolution of conventional
medical CT and images tiny biological
structures such as trabecular bone.

easily accessible on all personal computers. Rowe et a1. (1993) have used HRXCT to pro-
duce a digital atlas of the skull of Thrinaxodon, which combines the original coronal sec-
tions along with digitally resliced sagittal and horizontal (transverse) sections that are
presented on CD ROM in both DOS and Macintosh formats. Together these advantages
combine to make HRXCT very useful scientifically.
Research with conventional medical CT can prove to be quite expensive, and access
to instruments housed in hospitals or clinics usually depends upon patient throughput,
which can sometimes limit its use for pure research and non-medically oriented projects.
HRXCT is available to scientists on a commercial basis, but its high per-hour cost serves
to greatly limit the broad application of this technology across the biological and earth sci-
ences. In an effort to make this latter technology more widely available to the scientific
community, a HRXCT laboratory has recently been established at The University of Texas
at Austin. This laboratory became operational in early 1997 and is available to the re-
search community (Rowe et al., 1997). The HRXCT instrument is designed by Bio-Imag-
ing Research, Inc. (Lincolnshire, IL) and includes 200 kV and 420 kV sources as well as
linear and area detectors that offer a resolution down to 10 /lm. Images can be written in a
variety of file formats (e.g., TIFF, BMP, TARGA, PICT, etc.) and on a range of media
(e.g., optical erasable disks, CD ROM, Zip or Jazz disks, DAT tapes) that can be accessed
by all platforms of personal computers. Information on the High Resolution X-ray CT
Laboratory at the University of Texas at Austin can be accessed at the Internet web site
given in Table 1.
Advances in Three-Dimensional Data Acquisition and Analysis 211

Magnetic Resonance Imaging (MRI or MR), also known as Nuclear Magnetic Reso-
nance (NMR) imaging, is a relatively new technology that has to date seen only a few ap-
plications in studies of the functional morphology of hard tissue. Although NMR can in
some cases resolve bone, its greatest utility in medical imaging has been found in the
study of soft tissues. MacLatchy and Bossert (1996) and MacLatchy (1996, this volume)
used a combination of conventional medical CT and NMR technologies to study the hip
joint of a variety of extant and extinct primates. MacLatchy and Bossert (1996) relied on
NMR for the smallest primates because it provided them with nearly an order of magni-
tude increase in resolution over conventional CT. (The NMR resolution of 0.15 mm is at
the upper range of the resolution available for HRXCT.) Hard tissue is not easily resolved
by NMR, and in their studies NMR was used to image external shape only. Given this
limitation with extant material, it is unlikely that NMR will prove very effective for imag-
ing the internal morphology of densely mineralized fossils. Other much less expensive and
less invasive techniques are available for imaging external morphologies (see below).

3.2. Detailing the Surface: Laser Digitizers

Some of the most recent advances in automated 3-D imaging incorporate a laser
scanner to capture the surface topography of a specimen. These scanners differ from the
CT technologies discussed above because in general only the geometrical coordinates of
the outside surface can be gathered. Laser surface scanners offer advantages over medical
CT scanners when only surface coordinates are needed because these scanners do not in
general require proprietary software, but produce data sets that can be analyzed on any
personal computer platform with a variety of "off the shelf' software. Operating costs are
much lower than CT and NMR because laser scanners use standard electric current, have
very low maintenance costs, and do not require a licensed operator. These advantages
combine to make laser scanners an affordable option for many researchers who are inter-
ested in 3-D morphology, and these instruments can now be found in many physical an-
thropology laboratories around the world (e.g., The University of Texas at Austin,
University of Liverpool, Smithsonian Institution). Laser scanners are not the only non-CT
alternative to capture surface coordinates. There are many stylus operated 3-D surface
data gathering systems on the market, but these are generally fairly slow, non-automated,
and labor intensive, and vary dramatically in their degree of resolution. Excellent reviews
of these and many other digitizing systems are given in Hartwig and Sadler (1993) and
Dean (1996).
Laser scanners can be divided into two basic technologies. Both share the use of la-
ser, but the first of these, the optical digitizer, breaks the beam oflight into a plane that il-
luminates a single contour on the surface of the specimen. This scanner uses two CCD
(charge coupled device) cameras to digitize the plane oflight and capture the coordinates
of this single contour. Two options are next available to complete the scan. The laser
source can be translated along a plane that is parallel to the specimen with the collection
of more contour data, or the specimen or the laser source can be rotated so that radial con-
tours of data are collected. This scanning technology excels with convexly shaped speci-
mens. An additional advantage of these systems is that a very large number of data points
can be collected very rapidly (l04 points per second) with a point spacing that varies from
10' _10 2 f.1m (see Table 1: Cyberware Inc. web site). This high speed data collection makes
it possible to scan living specimens, and several scanners offer options that record the sur-
face color of the specimen as well. Perhaps the greatest disadvantage of these systems is
the distortion that results if the surface normals deviate significantly from either the plane
 212 J. Kappelman

of the laser beam or the radial direction of the scan. These distortions, generally most ex-
treme along the edges of a specimen, often require the collection of data from separate
scans of differently oriented views of the specimen. The final rendering of the digitized
specimen is completed by using sophisticated "zipper" software to combine the multiple
scans into a complete model. This processing of a model is computationally intensive and
can require many hours.
The second laser scanning technology, known as adaptive scanning, retains the laser
but keeps the beam tightly focused as a single point of light and uses a triangulation algo-
rithm to measure the exact x and y coordinates of each point on the surface of the speci-
men at a fixed z coordinate or level (Figure 4). In order to eliminate the distortion
resulting from mismeasured surface normals as noted above, the Digibot II® scanner (see
Table 1, web site for Digibotics, Inc., Austin, TX) uses a turntable to rotate the specimen
so as to bring each surface into a position that is exactly normal to the laser. Additional ro-
tations of the specimen complete the collection of points around one horizontal contour,
and the scan of the specimen is completed by translating the laser vertically though the z
axis and collecting additional contours. This scanning method collects data from convex
as well as concave surfaces, and creative mounting of the specimen on the turntable can
produce scans of structures such as eye orbits (Table 1, Internet web site for Virtual Hu-
man Osteology Guide). Digibotics, Inc. has recently added a new lens to their system that
collimates the laser beam to a much smaller diameter. This option produces a point spac-
ing of 500 f..lm in the x, y, and z directions and further increases in resolution are expected.
Because adaptive scanning technology collects each point separately, there is no need for

Figure 4. Three-dimensional laser scanning can be used to digitize the surface topography of nearly any speci-
men. The scanner shown in this figure is produced by Digibotics, Inc. (see Table I) and permits a point spacing of
0.5 mm in the x, y, and z planes (see Figure 6). A three-dimensional animation of the human skull is available for
downloading at the Virtual Human Osteology Guide web site (see Table I).
Advances in Three-Dimensional Data Acquisition and Analysis 213

sophisticated and computationally time-intensive "zipper" software or additional scans to

completely digitize the object. The x, y, z data can be exported to a variety of software
programs for wiremeshing and shading. One potential disadvantage of this system is that
because each data point is collected separately, it can take several hours to scan an object
at high resolution, and this type of laser scanning is not appropriate for living specimens.
These scanners are, however, fully automated and do not require an operator. In the lab at
The University of Texas, long scans are usually run overnight or on the weekend, and this
protocol has been used to complete scans of every element of the human skeleton to pro-
duce a "virtual" human osteology guide and many other teaching applications (see Ta-
ble 1).
It is commonplace for researchers who study physical objects to worry about
whether or not they have collected enough data to quantify the features that they wish to
describe. One important point to keep in mind when studying surfaces and shapes is that
all of the scanning methods described above collect what are usually many more points
than are required to mathematically describe the surface. It is not at all uncommon for la-
ser scanners to collect on the order of 104 _10 6 points for a human skull. Creating a surface
by triangulating or wiremeshing such a large number of points can easily produce models
that are on the very edge of what today's personal computers can manipulate. Such large
models often require more powerful computers such as Silicon Graphics workstations for
analysis. There is no simple way to determine exactly how dense a coverage of points is
required to "describe" the original surface, but there are now several software packages
such as Sculpt® (see Table 1) that are designed to filter data based upon a number of user-
defined criteria so as to reach an optimum between the number of data points and the sur-
face detail of the specimen. Data point density will only increase with more powerful
digitizers, and the filtering programs will in turn witness increasingly wide use.

3.3. To Have and To Hold: 3-D Printouts

Many of the digitizing technologies discussed above have come to physical anthro-
pology directly from the fields of computer-aided engineering and manufacturing. In fact,
many laser scanners are specifically designed for reverse engineering and computer-aided
manufacturing (CAM). One additional area that holds great potential for physical anthro-
pology involves the automated production of 3-D physical printouts from the computer
models. These 3-D printout technologies fall under three general categories: automated
milling, stereolithography (STL), and laser sintering. In each case the digitized object is
written to a file format that computationally "slices" the object into thin layers. The x, y,
and z coordinates of these layers are then used to produce the 3-D printout. Automated
milling physically sculpts the object from wood or Styrofoam. Stereolithography builds
the 3-D object layer by layer by curing epoxy, but this is often a time consuming process.
Laser sintering, a process invented by DTM, Inc. (see Table I for web site) is the newest
addition to this line up of technologies. This process uses a laser to sinter or fuse nylon or
plastic powder into the first layer of the object, which, in turn, is sintered to the next layer,
and so on, until the 3-D printout is completed (Figure 5). Laser sintering is a relatively
rapid process and new research is focused on using more powerful lasers to sinter ceram-
ics and metals so as to transform sintering into a rapid manufacturing process.
One direct application of the 3-D printout technologies involves digitizing fossils or
extant specimens by either CT or laser scanner and using the output technologies to pro-
duce 3-D printouts of the specimens (Kappelman, 1992; Hjalgrim et aI., 1995). This proc-
ess could prove to be especially important in the case of very delicate fossils that could be
214 J. Kappelman

DTM
The Sinterstation 2000 Syste
eo,Luer

Las« OpticsI
5<tIIIIIIag on

Powder led

parte,..
Po'"

Figure 5. There are now a variety of technologies available for producing 3-D printouts of digitized files. This ap-
proach is especially useful for producing reconstructions of fragmentary fossils. The sinterstation shown in this
figure is built by DTM, Inc. (see Table I) and uses a laser to fuse or sinter plastic powder layer by layer into a 3-D
model (see Figure 6). As each layer is sintered, the piston (center of the station) that holds the part translates verti-
cally downward by a fixed distance and a new layer of powder is refreshed and leveled on the top of the piston.
This new layer is next sintered to the part, and the process is repeated until the part is completed.

damaged by full preparation or even physical handling. Furthermore, given that traditional
silicon or rubber molds produce only a limited number of casts, a 3-D printout of the
original fossil could be used to make additional molds. It is important to note that the pre-
sent high cost of these technologies makes it unlikely that traditional plaster- and plastic-
based molding and casting will be completely replaced by the 3-D printout technologies at
any point in the near future.
The richest potential of 3-D printout technology for physical anthropology is
reached when it is combined with the versatility of computer digitizing and modeling. It is
often the case that a fossil will vary in the degree of completeness of its skeletal elements,
especially from either the right and left sides of its body, or element by element, and in the
past a process of "informed sculpting" was used to model the missing anatomies. One use-
ful example is seen in the case of STS 14, an early australopithecine. This specimen pre-
serves much of its pelvic region, but the right innominate is much more complete than the
left. In order to create a more accurate reconstruction of the complete pelvis, the right ele-
ment was digitized with a laser scanner. After digitizing, the file was exported to Auto-
CAD®, and the "mirror" function converted the right innominate into a "left" innominate.
This "new" innominate was next surface meshed, saved in a STL format, and transferred
to DTM's Sinterstation to produce a 3-D printout of the right innominate mirrored as a
left. This 3-D nylon printout can now be combined with the true right innominate to pro-
duce a reconstruction of the pelvis (Figure 6). The range of these types of applications is
endless. Kalvin et al. (1992) reconstructed a cranium from multiple fragments, and Zollik-
ofer et al. (1995) digitized fragmentary pieces of a Neanderthal infant cranium and com-
bined the fragments into a file for a stereolithography 3-D printout. Johanson (1995)
Advances in Three-Dimensional Data Acquisition and Analysis 215

Figure 6. A reconstruction of the fragmentary pelvis of STS 14. The more complete right innominate of a cast of
STS 14 was digitized with a Digibot II laser scanner (see figure 4). This file was input into AutoCAD®, mirrored
to the opposite side, and printed out in 3-D on DTM's sinterstation in order to produce a complete reconstruction
of the pelvis.

followed a similar approach to model a reconstruction of the cranium of A. afarensis. Such

applications are becoming increasing common and are of great utility not only to the re-
searcher, but also to the student because 3-D printouts allow direct tactile contact with the
model and this additional sensory input is often more dramatic than simply viewing a
computer model.

4. INTO THE FUTURE: DATA ANALYSIS IN THREE AND FOUR

DIMENSIONS

It was noted at the beginning of this article that the study of the relationship between
form and function in living animals remains a critical focus of functional morphology, not
only for understanding functional relationships in extant forms but also because this is the
only approach that exists for understanding extinct species (see Aiello et aI., this volume).
One relatively new area of analysis for investigating and understanding the relationship
between form and function is the finite element method (FEM), a technique again bor-
rowed from engineering that holds very exciting possibilities for physical anthropology. It
is widely accepted that the size and shape of the various elements of the skeleton preserve
a record of the genetic fingerprint as well as the past behaviors of the organism. These be-
haviors include the loading histories that the elements have experienced, and their shape is
believed to be related to the way that the structure of the element is designed to resist or
respond to these loads (Thompson, 1971; Wainwright et aI., 1976; Currey, 1984; Lanyon
and Rubin, 1985; Carter 1987; Carter et aI., 1989). Briefly, the finite element method is
 216 J. Kappelman

used to analyze the way that an object responds to a load by first dividing its form up into
a sometimes very large number of smaller but interconnected elements, each of which is
assigned material properties and boundary conditions. A load is next applied to the model
and the stress and strain that each element is subjected to are analyzed. These results indi-
cate the particular regions of the element where compressive and tensile stresses are con-
centrated as well as the magnitude of the stress (Figure 7). Such studies are especially
critical in engineering design and manufacturing because these analyses can be used to
pinpoint the precise area of a load-bearing part or tool that will fail under certain loading
conditions. Redesign of the part and retesting by FEM is used to produce a product that is
optimally designed to resist the applied load that it will experience. A much more com-
plete treatment ofFEM can be found in Rao (1982), Huiskes and Chao (1983), and Baran
( 1988).
Studies of the way that transmitted loads influence the size and shape of skeletal ele-
ments share a broad number of similarities with earlier studies that used the photoelastic
method of analysis (Oxnard, 1973; Ward and Molnar, 1980). Even though most photoelas-
tic and FEM studies have until recently relied on two dimensional models only, or have in-

Von iii ••• su... OCPa) Ho*> P<Ul

-INFINITY
+0.169
+1.67
+3.22
+4.75
+6.28
+7.81
+9.34
+10. 8
+ 12.4
+13.9
+ 15.4
+17.0
+18.1

[2c. [2~

Postel'ior Lateral Posterior Lateral

Figure 7. The finite element method is becoming a common method for analyzing the functional morphology of
extant and extinct species. The internal and external surface morphologies of these two tibiae were digitized with a
conventional CT, and a model consisting of many brick elements was next constructed from this file. The finite
element models of these tibial shafts are from a modem human and a chimpanzee and were completed with the
software program ABAQUS@ (see Table I for web site). This particular FEM study loads both tibiae in a bipedal
stance, and the contours of lighter and darker colors show areas of high and low compression, respectively.
Advances in Three-Dimensional Data Acquisition and Analysis 217

vestigated the transmission of loads along a restricted number of planes, these studies
have provided important insights into the relationship between shape and function (e.g.,
Oxnard, 1973, 1975, 1984; Richmond and Qin, 1996; Spears and Crompton, 1996). It is,
however, the case that almost all skeletal elements have complex shapes in three dimen-
sions and at some junction FEM studies must take this additional shape information into
account. The new imaging technologies of High Resolution CT and laser scanning dis-
cussed above now permit accurate modeling of skeletal complexity on a scale from
10 1_10 6 11m that encompasses nearly the full range of vertebrate hard tissue structures.
Studies that use FEM to investigate complex three-dimensional objects are becoming in-
creasingly common (Hobatho et aI., 1991; Korioth et aI., 1992; Chen and Povirk, 1996;
Edelman and Reeke, 1996; Ryan et aI., 1996; Chen and MacLatchy, 1997; Duncan et aI.,
1997; Ryan, 1997; Ryan and Kappelman, 1997). Other techniques such as strain gauge
studies (Lanyon and Smith, 1970; Lanyon, 1971; Lanyon et aI., 1975; Finlay, 1982), ther-
moelastic stress studies (see Table 1), or photoelastic method (Ryan et aI., 1996) provide
the means for partial but nonetheless critical independent testing of FEM results.
Perhaps the most exciting and also one of the challenging aspects of applying FEM
to studies of skeletal form and function is found in extending this method across the fourth
dimension. Most FEM analyses are concerned with static loading models that attempt to
approximate the highest loading regimes that are experienced by a structure in order to
test for failure. Experiments have, however, demonstrated that most locomotor and masti-
catory behaviors involve complex loading cycles in which the magnitude and direction of
load vary throughout the cycle (see Hannam and Wood, 1989; Korioth et aI., 1992).
Analogous dynamic changes are also witnessed in tracking loading regimes encountered
through the ontogeny of an individual. Dynamic loading cycles are routinely modeled by
FEM in engineering studies, and an example of two revolutions of a disc brake assembly
can be seen at the Internet web sit ..http://www.abaqus.comlapplications/disk-brake/in-
dex.html." We can anticipate tests of dynamic models in studies of primate locomotion,
mastication, and ontogeny in the near future (Langdon et aI., 1991).
Once FEM has been used to model and study the loading parameters of a skeletal
element from an extant species, the technique can be extended to fossil species. Although
other methods, such as strain gauge analysis or loading rigs, offer experimental results for
extant specimens that are broadly comparable to FEM, it is not possible to extend these
physical methods to the study of fossils because neither the mineralized fossil itself nor
cast reproductions of the fossil has the correct material properties of bone. The 3-D imag-
ing and modeling techniques discussed above can be used to produce accurate 3-D models
of the fossil, which in tum can be used in FEM testing. This combination of methods of-
fers what is now one of the only means for testing the relationship between form and func-
tion of skeletal elements in extinct species. Furthermore, only one additional step is
required to use this combination of methods to approach questions about the nature of
evolutionary transitions. Once the skeletal elements from a number of extant and fossil
specimens have been digitized, these files can be edited to model a hypothetical specimen
that has some mix of the morphologies of the actual specimens. This process is known as
"morphing" in computer animation and is commonly used to transform one cartoon char-
acter into another. For the purpose of functional morphology, "morphing" provides the
means for building models with intermediate properties that can in tum be used to test hy-
potheses about the inferred function of intermediate morphologies. Of course, discoveries
of new fossils provide the ultimate test or confirmation of the hypothetical "construction,"
but until that time, this combination of techniques allows the framing and testing of ex-
plicit hypotheses and offers a direct approach to questions of evolutionary transitions.
 218 J. Kappelman

5. SHARING THE DATA: VIRTUAL MUSEUMS FOR ALL

Advances in computer technology and especially storage media and transmission

links greatly facilitate the sharing of raw and processed data. Even tremendously large ta-
bles of raw data or video clips can now be easily copied to CD ROM or other media for
inexpensive distribution, or transmitted via the Internet. Electronic publishing promises to
even more greatly simplify the distribution of raw data, and many journals that publish
hard copy formats also have Internet web sites where primary data are archived. This lat-
ter option is also available to anyone who has access to a web site. In fact, the very impor-
tant modem human craniometric data set of WW Howells, was published in three volumes
(Howells, 1973, 1989, 1995) has now been archived to the web (Howells, 1996: see Table
1).
It has become common practice in most disciplines for authors to report the primary
data that form the basis for their observations. In fact, many journals now require that the
supporting data sets be made available as a condition of publication (e.g., see Nature vol.
384:598). Clearly, the high costs of traditional hard copy text publishing does force some
page length restrictions on published articles, but it seems reasonable to provide the actual
data so that the precise details of the argument can be followed by the reader. In many cases
these data are taken from specimens that are from rare or endangered species that can no
longer be collected in the field, or are from very fragile fossils, or are located in museums
that are difficult to access. Making measurements of these specimens widely available is a
very valuable contribution in and of itself, and in many cases can significantly improve the
quality and sample for any number of future projects. Once a set of standard measurements
has been made, it also makes good sense to publish the actual measurements so that a global
"mean" can be calculated that takes into consideration inter-observer error. A second argu-
ment can be made for making the raw data available to the scientific community because in
most cases the data were collected with the aid of tax or foundation funds and in some sense
then the raw data do in fact belong to the public at large. In most cases authors freely dis-
tribute the raw data once their paper has been published, but there is sometimes a tendency
for authors to regard their raw data as private property. Journals in physical anthropology do
not appear to have any standard policy with regard to requirements for publishing primary
data, and almost any issue of the leading journals in this field will include some articles that
provide a list of specimens along with actual raw measurements, while others report only
transformed z-scores of species' means. The absence of any clear requirement to publish or
otherwise make available the raw data appears to have, in part, contributed to a subculture
of "private ownership" of the raw data.
Primary digital data sets for'even a single specimen that are generated by 3-D laser
scanning or High Resolution CT can easily be several orders of magnitude larger in size
than nearly any data set that consists of linear, areal, or volumetric values, but these data
are also easily shared via storage media or the Internet. It is perhaps the case that an even
stronger argument can be made for freely sharing these sorts of data after the original in-
vestigator has published her or his observations. Unlike a set of linear measurements,
whose exact landmarks may vary somewhat among researchers, 3-D laser scanning and
High Resolution X"ray CT produce a nearly "virtual" copy of the original specimen, and
any number of measurements can then be taken from the digital file (Kappelman, 1993).
Once a high quality, high resolution 3-D scan has been completed, it is unlikely that sec-
ond, third, or fourth scans using the same technology will provide much additional detail
and, in the case of fragile fossils, there is little rationale for exposing the specimen to the
risks associated with additional handling. In fact, predicted reductions in the future alloca-
Advances in Three-Dimensional Data Acquisition and Analysis 219

tion of tax dollars for research may restrict spending to the point where research visits to
archived collections become the exception rather than the rule. The educational and public
relations benefits that would accrue from making extant and fossil collections available to
the entire community as a whole, from grade schoolers to senior scientists, are difficult to
quantify, but such efforts would certainly strengthen the public image of physical anthro-
pology. At this time there is no "virtual museum" of scanned specimens, and building an
Internet museum will require the coordinated efforts of many museums and curators, espe-
cially with regard to setting standards for scan resolution and file formats (see Hartwig
and Sadler, 1993), and resolving the complex legal issue of who owns the "copyright" of
unique specimens or the data themselves. Nonetheless, it is the case that many important
and useful 3-D data sets are already available on the Internet, and any number of files can
be downloaded from some of the web sites listed in Table 1. It is likely that most journals
will soon take full advantage of these advances and require the publication of the raw data
that can, in tum, provide the critical "cornerstones" for building the virtual museums of
the future.

6. CONCLUSION

Advances in computer design, storage media, digitizing technologies, and analysis

techniques have combined to open new possibilities in studies of functional morphology.
Increases in the capacity and variety of digital storage-media and the widespread availabil-
ity of the Internet offer new possibilities for the unlimited exchange of primary data. Al-
though the relatively new technologies of surface laser scanning and High Resolution
X-ray CT produce data sets that can be used to gather simple linear measurements, the
strength of these technologies lies in their ability to address questions of form in three di-
mensions. For example, the actual surface area of even complexly shaped three-dimen-
sional joints or tooth cusps can be accurately measured with CAD software, thus
eliminating two-dimensional approximations of these features as based on linear measure-
ments. Analytical techniques, such as the finite element method, provide the means for as-
sessing and testing how the size and shape of a skeletal element responds to loading, and
this technique offers great promise for both understanding the functional morphology of
fossil species and testing hypotheses about the nature of intermediate morphologies and
evolutionary transitions. Together these advances provide new pathways for scientists to
follow in fulfilling their responsibilities for making their observations widely and freely
available for testing by others.

ACKNOWLEDGMENTS

I wish to thank Elizabeth Strasser, John Fleagle, Alfred Rosenberger, and Henry
McHenry for inviting me to participate in the conference, "Primate Locomotion - 1995,"
and the Wenner-Gren Foundation for Anthropological Research as well as the Anthropol-
ogy Division of NSF for funding to help defray travel and housing expenses. Special
thanks are due to Henry McHenry and his family for their hospitality during the meeting.
Various colleagues at UT have assisted in laboratory work and discussions, and I would
like to thank Claud Bramblett, William Carlson, Alex Duncan, Reuben Reyes, Tim Rowe,
Tim Ryan, Rob Scott, Ron Stearman, and Greg Weiner. I thank Leslie Aiello for discus-
sions about the digital reconstruction of STS 14. Thanks to Paige Hake Sheehan for the
220 J. Kappelman

use of the image in Figure 2, and Tim Ryan for the use of the image in Figure 7. My work
in computer imaging has received support from NSF (DUE 9354427) and The University
of Texas at Austin, and corporate support from Autodesk, Inc., Digibotics, Inc., DTM,
Inc., Intel, Inc., and Tektronix, Inc. Thanks to Elizabeth Strasser and two anonymous re-
viewers for very helpful comments on an early draft of this paper.

REFERENCES
Aiello LC, Wood BA, Key C, and Wood C (1998) Laser scanning and palaeoanthropology: An example from
Olduvai Gorge, Tanzania. In E Strasser, JG Fleagle, AL Rosenberger, and H McHenry (eds.): Primate Lo-
comotion: Recent Advances, New York: Plenum Press, pp. 223-236.
Baran NM (1988) Finite Element Analysis on Microcomputers. New York: McGraw-Hill.
Carter DR (1987) Mechanical loading history and skeletal biology. J. Biomech. 20: I 095-11 09.
Carter DR, Orr TR, and Fyhrie DP (1989) Relationships between loading history and femoral cancellous bone ar-
chitecture. J. Biomech. 22:231-244.
Chen X, and Povirk G (1996) Assessing errors introduced by modeling the anisotropic human mandible isotropi-
cally with finite element method. Am. J. Phys. Anthropol. Supplement 22:83.
Chen X, and MacLatchy LM (1997) The biomechanical significance of the trabecular bone in hominoid proximal
femora. Am. J. Phys. Anthropol. Supplement 24:90-91.
Cifelli RL, Rowe TB, Luckett WP, Banta J, Reyes R, and Howes RI (1996) Fossil evidence for the origin of the
marsupial pattern of tooth replacement. Nature 379:715-718.
Conroy GC, and Vannier MW (1984) Noninvasive three-dimensional computer imaging of matrix-filled fossil
skulls by high-resolution computer tomography. Science 226:456--458.
Conroy GC, and Vannier MW (1987) Dental development of the Taung skull from computerized tomography. Na-
ture 329:625--{)27.
Coolidge HJ Jr. (1933) Pan paniscus. Pygmy chimpanzee from south of the Congo River. Am. J. Phys. Anthropol.
18: I-59.
Comer BD, and Richtsmeier JT (1993) Cranial growth and growth dimorphism in Ateles geojJroyi. Am. J. Phys.
Anthropol. 92:371-394.
Currey J (1984) The Mechanical Adaptations of Bones. Princeton: Princeton Univ. Press.
Dean D (1996) Three-dimensional data capture and visualization. In LF Marcus, M Corti, A Loy, GJP Naylor, and
DE Slice (eds.): Advances in Morphometrics. New York: Plenum Press, pp. 53-69.
Duncan A, and Kappelman J (1991) Image manipulation, analysis, and mass storage on the microcomputer. Am. J.
Phys. Anthropol. Supplement 12:69--70.
Duncan A, Kappelman J, and Shapiro LJ (1994) Metatarsophalangeal joint function and positional behavior in
Australopithecus afarensis. Am. J. Phys. Anthropol. 93:67-81.
Duncan A, Podnos E, Cleghorn NE, and Kappelman J (1997) Development and analysis of 3-D finite element
models of human femoral diaphyses. Am. J. Phys. Anthropol. Supplement 24: 103-104.
Edelman DB, and Reeke GN Jr. (1996) Finite element analysis of hominoid proximal femora. Am. J. Phys. An-
thropol. Supplement 22: I 0 I.
Finlay 18, Bourne RB, and McLean J (1982) A technique for the in vitro measurement of principal strains in the
human tibia. J. Biomech. 15:723-739.
Hake P (1992) An examination of sexual dimorphism in the dentition and postcranium of Papio anubis. B.A. Sen-
ior Honors Thesis, The University of Texas at Austin.
Hannam AG, and Wood WW (1989) Relationships between the size and spatial morphology of human masseter
and medial pterygoid muscles, the craniofacial skeleton, and jaw biomechanics. Am. J. Phys. Anthropol.
80:429--445.
Hartwig WC, and Sadler LL (I 993) Visualization and physical anthropology. In AJ Almquist and A Manyak (eds.):
Milestones in Human Evolution. Prospect Heights: Waveland Press, Inc, pp. 187-222.
Hjalgrim H, Lynnerup N, Liversage M, and Rosenklint A (J 995) Stereolithography: Potential applications in an-
thropological studies. Am. J. Phys. Anthropol. 97:329--333.
Hobatho MC, Darmana R, Pastor P, Barrau JJ, Laroze S, and Morucci JP (1991) Development of a three-dimen-
sional finite element model of a human tibia using experimental modal analysis. J. Biomech. 24:371-383.
Howells WW (1973) Cranial Variation in Man. A Study by Multivariate Analysis of Patterns of Differences among
Recent Human Populations. Papers of the Peabody Museum of Archeology and Ethnology, 67: 1-259.
Advances in Three-Dimensional Data Acquisition and Analysis 221

Howells WW (1989) Skull Shapes and the Map. Craniometric Analyses in the Dispersion of Modem Homo. Pa-
pers of the Peabody Museum of Archeology and Ethnology 79: 1-189.
Howells WW (1995) Who's Who in Skulls. Ethnic Identification of Crania from Measurements. Papers of the Pe-
abody Museum of Archeology and Ethnology 82: 1-108.
Howells WW (1996) Howell's craniometric data on the Internet. Am. 1. Phys. Anthropol. 101:441-442.
Huiskes R, and Chao EYS (1983) A survey of finite elements analysis in orthopedic biomechanics: The first dec-
ade. J. Biomech. 16:385-409.
Johanson DC (1995) Face-to-face with Lucy's family. National Geographic 189:96-117.
Jungers WL, and Stem JT Jr. (1980) Telemetered electromyography of forelimb muscle chains in gibbons (Hylo-
bates tar). Science 208:617-619.
Kalvin AD, Dean 0, and Hublin JJ (1995) Reconstructing human fossils. IEEE Computer Graphics and Applica-
tions 15:12-15.
Kappelman J (1992) Three-dimensional input and output of solid models for studies in functional morphology.
Am. J. Phys. Anthropol. Supplement 14:97.
Kappelman J (1993) Building an evolutionary database using digital imaging and animation software. Am. J. Phys.
Anthropol. Supplement 16: 122.
Korioth TWP, Romilly DP, and Hannam AG (1992) Three-dimensional finite element stress analysis of the dentate
human mandible. Am. J. Phys. Anthropol. 88:69-96.
Langdon JH, Bruckner J, and Baker HH (1991) Paleokinematics of the australopithecine foot with three-dimen-
sional imaging. Am. 1. Phys. Anthropol. Supplement 12: III.
Lanyon LE (1971) Strain in sheep lumbar vertebrae recorded during life. Acta. Orthop. Scand. 42: 102-112.
Lanyon LE, and Rubin CT (1984) Functional adaptation in skeletal structures. In M Hildebrand, OM Bramble, KF
Liem, and DB Wake (eds.): Functional Vertebrate Morphology. Cambridge: Harvard University Press, pp.
1-25.
Lanyon LE, and Smith RN (1970) Bone strain in the tibia during normal quadrupedal locomotion. Acta. Orthop.
Scand.41:238-248.
Lanyon LE, Hampson WGJ, Goodship AG, and Shah JS (1975) Bone deformation recorded in vivo from strain
gauges attached to the human tibial shaft. Acta. Orthop. Scand. 46:256-268.
MacLatchy LM (1996) Another look at the australopithecine hip. J. Hum. Evol. 31:453-476.
MacLatchy LM (1998) Reconstruction of hip joint function in extant and fossil primates. In E Strasser, JG Fleagle,
AL Rosenberger, and HM McHenry (eds.): Primate Locomotion: Recent Advances. New York: Plenum
Press, pp. 111-130.
MacLatchy LM, and Bossert WH (1996) An analysis of the articular surface distribution of the femoral head and
acetabulum in anthropoids, with implications for hip function in Miocene hominoids. J. Hum. Evol.
31:425-453.
Mann AE (1975) Some Paleodemographic Aspects of the South African Australopithecines. Philadelphia: Univer-
sity of Pennsylvania Press.
Martin RD (1989) New quantitative developments in primatology and anthropology. Folia Primatol. 53: 1-246.
Oxnard CE (1973) Form and Pattern in Human Evolution: Some mathematical, physical, and engineering ap-
proaches. Chicago: University Chicago Press.
Oxnard CE (1975) Uniqueness and Diversity in Human Evolution: Morphometric studies of Australopithecines.
Chicago: University Chicago Press.
Oxnard CE (1984) The Order of Man: A Biomathematical Anatomy of the Primates. New Haven: Yale University
Press.
Rao SS (1982) The Finite Element Method in Engineering. New York: Pergamon Press.
Richmond B, and Qin Y-X (1996) Finite element methods in paleoanthropology: The case of phalangeal curvature.
Am. J. Phys. AnthropoJ. Supplement. 22:197.
Richtsmeier JT (1989) Application of finite-element scaling analysis in primatology. Folia PrimatoJ. 53:50-64.
Rowe T, Carlson WD, and BottorffW (1993) Thrinaxodon: Digital Atlas of the Skull. CD ROM, 623 MB, Austin:
University of Texas Press.
Rowe T, Kappelman J, Carlson WD, Ketcham R, and Denison C (1997) High resolution computed tomography: A
breakthrough technology for earth scientists. Geotimes 42:23--27.
Ruff CB (1989) New approaches to structural evolution oflimb bones in primates. Folia Primatol. 53: 142-159.
Ruff CB, and Leo FP (1986) Use of computed tomography in skeletal structure research. Yrbk. Phys. Anthropol.
29: 181-196.
Ryan T (1997) A structural analysis of the cross-sectional shape of the tibia in Homo and Pan using the finite ele-
ment method. M. A. Thesis, The University of Texas at Austin.
Ryan T, and Kappelman J (1997) A structural analysis of tibial shape using the finite element method. Am. 1. Phys.
Anthropol. Supplement 24:202.
222 J. Kappelman

Ryan TM, Scott RS, Duncan A, Kappelman 1, Shapiro L, Grant S, Lewis K, and Steannan R (1996) Finite element
analysis using a 3-D laser scanner. Am. 1. Phys. Anthropol. Supplement 22:206.
Spears IR, and Crompton RH (1996) The mechanical significance ofthe occlusal geometry of great ape molars in
food breakdown. 1. Hum. Evol. 31:517-535.
Spencer MA, and Spencer GS (1995) Technical note: Video-based three-dimensional morphometrics. Am. 1. Phys.
Anthropol. 96:443-453.
Tate 1R, and Cann CE (1982) High-resolution computed tomography for the comparative study of fossil and extant
bone. Am. 1. Phys. Anthropol. 58:67-73.
Taube RA, and Adelstein S1 (1987) A short history of modern medical imaging. In E Guzzardi (ed.): Physics and
Engineering of Medical Imaging. Dordrecht: Martinus Nijhoff, pp. 9-40.
Thompson DW (1971) On Growth and Fonn. Cambridge: Cambridge University Press.
Vannier MW, Conroy GC, Marsh 1L, Knapp RH (1985) Three-dimensional cranial surface reconstructions using
high-resolution computed tomography. Am. l. Phys. Anthropol. 67:299-311.
Wainwright SA, Biggs WD. Currey lD, and Gosline 1M (1976) Mechanical Design in Organisms. Princeton:
Princeton Univ. Press.
Ward SC, and Molnar S (1980) Experimental stress analysis of topographic diversity in early hominid gnathic
morphology. Am. J. Phys. Anthropol. 53:383-395.
Weidenreich F (1943) The skull of Sinanthropus pekinensis: A comparative study on a primitive hominid skull. Pa-
laeontologia Sinica D 10: 1-485.
Wilkins SW. Gureyev TE, Gao D, Pogany A, and Stevenson AW (1996) Phase-contrast imaging using polychro-
matic hard X-rays. Nature 384:335-338.
Winkler LA, Schwartz 1H, and Swindler DR (1996) Development of the orangutan pennanent dentition: Assessing
patterns and variation in tooth development. Am. l. Phys. Anthropol. 99:205-220.
Zollikofer CPE, Ponce de Leon MS, Martin RD, and Stucki P (1995) Neanderthal computer skulls. Nature
357:283-285.
 13

LASER SCANNING AND

PALEOANTHROPOLOGY
An Example from Olduvai Gorge, Tanzania

Leslie Aiello,l Bernard Wood,2* Cathy Key,l and Chris Wood2

IDepartment of Anthropology
University College London
Gower Street
London WC1E 6BT, England
2Department of Human Anatomy and Cell Biology
The University of Liverpool
P.O. Box 147
Liverpool L69 3BX, England

1. INTRODUCTION

Taxonomic affiliation in Plio-Pleistocene fossil hominids is generally based on the

analysis of cranial and dental material where species-specific characteristics are well rec-
ognised (Wood, 1991, 1992). With the exception of features such as the relatively long
neck and small head in the australopithecine femur, taxon-specific features in the postcra-
nial skeleton are largely unknown (Aiello and Dean, 1990). This situation is undoubtedly
the result of the relative absence of associated skeletons in the Plio-Pleistocene fossil re-
cord and the general paucity of postcranial material. It follows that species-specific fea-
tures in the postcranial skeleton have played only a secondary role in taxonomic studies
and that it is often difficult to assign isolated postcranial fossils to specific species.
New technology is, however, making it possible to analyse isolated skeletal ele-
ments and to determine whether or not unassociated fossils could have come from a single
individual, from individuals in the same species or from individuals belonging to different
species. In particular, the technique of laser scanning permits detailed three-dimensional
data to be gathered from complex joint surfaces. Developments in morphometries now al-
low these complex three-dimensional data to be analysed in terms of both species affili-

* Current Address: Department of Anthropology, George Washington University, 2112 G Street, NW, Washington
DC 20052.

223
224 L. Aiello et aL

ation and functional capability. For example, it is possible to compare two reciprocal com-
plex joint surfaces, determine their degree of compatibility or congruency and on this ba-
sis, assess the probability of their association. If one of the reciprocal joint surfaces comes
from a fossil of known affiliation, it is then possible to assess the probability of the other
belonging to that same species.
The purpose .of this paper is to illustrate the use of laser scanning in this context by
addressing a particular question of hominid postcranial fossil affiliation that has been left
unanswered in the literature since the 1960s. This question concerns the affiliation of OH
35, a fossil hominid tibia and fibula from Olduvai Gorge, Tanzania, which was found in
level 22 of the FLK site in 1960. When these bones were first excavated, Louis and Mary
Leakey believed that they belonged to the same individual as the well known OH5 'Zinjan-
thropus' skull, which had been found in 1959 in the same level of the same site (Leakey,
1971). The idea that OH35 could represent Australopithecus boisei was also entertained by
Day (1976) and more recently by Tobias (1991). Tobias (1991) points out that OH5 would
have been between 15-17 years of age at death based on modern human standards and that
the tibia and fibula are from an individual of at least that age. The reason for this determina-
tion is that the distal epiphyses of the tibia and fibula are fully closed. Furthermore, recent
body mass predictions for OH5 ba'sed on cranial measurements indicate a mass of between
39 and 40 kg (Aiello and Wood, 1994). T~e midshaft girth of the OH35 tibia (61.0 mm) sug-
gests a roughly equivalent ma&s. It is virt~ally the same as the mean tibial midshaft girth for
Pan troglodytes (61.7 mm, sd 5.93 mm, N=37), which has a mean body mass of36.4 kg (fe-
male=31.1 kg, males=41.6 kg) (Harvey et aI., 1986).
When OH35 was originally described by Davis (1964). he concluded that the tibia
and fibula could not be assigned to any taxon with confidence. Later, the close compatibil-
ity between OH35 and the OH8 (Homo habilis) foot was pointed out and the possibility
entertained that OH35 might belong to Homo habilis (Leakey et a\., 1964; Leakey, 1971;
Day, 1978). This interpretation was taken up and enthusiastically supported by Susman
and Stern (1982) who suggested that the OH35 tibia and fibula might even come from the
same Homo habilis individual as the OH8 foot. This is despite three points recognised by
Susman and Stern (1982) that would seem to argue against this interpretation. Firstly, the
site at which the OH8 foot was found (FLK NN, level 3) is some 200 m from the OH5 site
(FLK). Secondly, the stratigraphic levels in which the two fossils were found are separated
by 0.5 m. And thirdly, Susman and Stern (1982) believed that the OH8 foot belonged to a
juvenile individual aged between about 13.7 and 13.9 years at death while the closed dis-
tal epiphyses of the tibia and fibula suggest an older individual.
The taxonomic affiliation of the OH35 tibia and fibula will be tested here by compar-
ing the shape of the distal joint surface of the OH35 tibia to the trochlear joint surface of the
OH8 talus. The main aim is to determine whether or not the shapes of these reciprocal joint
surfaces are compatible enough to support the hypothesis that OH35 could have belonged to
the same individual as OH8 or alternatively to the same species (Homo habilis).

2. MATERIALS AND METHODS

The comparative material used in this analysis consists of the reciprocal articular
surfaces comprising 30 hominoid right talocrural joints, 5 males and 5 females each of
adult modern humans, chimpanzees (Pan troglodytes) and gorillas (Gorilla gorilla). The
modern humans are from the Spital fields Collection housed at the Natural History Mu-
seum, London and are individuals of known sex and age. The chimpanzees and gorillas
are from the Powell Cotten collection, Birchington, Kent. The fossil hominid sample is
Laser Scanning and PaJeoanthropoJogy 225

made up of OH35 and OH8 and the Australopithecus afarensis specimens, AL 288-lar (a
right distal tibia) and AL 288-las (a right talus), from Hadar, Ethiopia. In both cases casts
were used for the analysis. AL 288-1 was included in the analysis as an additional com-
parison because it is an associated joint from a single individual. All analysis was under-
taken at the Department of Human Anatomy, the University of Liverpool.
The reciprocal talocrural joint surfaces were scanned using a Cyberware 3030 High
Resolution Colour Laser Scanner fitted with an MM motion platform (Cyberware Labora-
tory Inc., Monterey, CA), running on a Silicon Graphics Indigo 2 XZ. The digitisation
process was controlled by Cyberware's own "Echo" software and each complete object
scan is written out to an ASCII-delimited file. A scanned object was stored as a dense
cloud of three-dimensional data points to an accuracy of ± 0.1 mm. The technique is fully
described in Wood et al. (submitted). The following is an abbreviated summary.
Four scans were made for each talocrural joint. In the first scan, three pieces of 1 mm
lead shot were attached to the talar trochlear surface as markers to provide an orientation to
the scan. This was found to be necessary because of the difficulty of locating precise refer-
ence points on the unmarked scanned image. The three pieces of lead shot were located on
each talar joint surface as follows: (a) at the most posterior point on the fibular facet; (b) at the
most anterior point on the fibular facet; and, (c) at the point where the plane of the anterome-
dial border of the tibial facet intersects the anteromedial border of the trochlear articular sur-
face of the talus when the joint is in the close-packed, dorsiflexed, position (Figure la). The
second, third and fourth scans were required to capture the relevant portions of the articulated
talocrural joint for each pair of tibiae and tali. In the first of these scans (Scan 2), the ta-
locruraljoint was articulated in maximum dorsiflexion ('close-packed' position), and three 5
mm diameter registration spheres were attached to each of the bones of the joint in such a way
that all six spheres (3 on the talus and 3 on the distal tibia) were visible in the scan. These
spheres were painted matt white to ensure that they were clearly captured by the scanning
process. The registration spheres were mounted on 3 cm wires so that they stand out from the
bones to which they were attached (the wires were painted matt black so that the scanner did
not detect them). Once the articulated joint and its 6 registration spheres were scanned, the ta-
lus and tibia were carefully separated so as not to disturb the position of either set of 3
spheres. The articular surfaces of both the talus and tibia were next scanned separately, along
with their own sets of three registration spheres (Scans 3 and 4). This process thus captured
both the position of the talus and tibia as an articulated unit (Scan 2) and the articular surface
of each bone (Scans 3 and 4) otherwise not visible when in articulation.
It is important to emphasise that laser scanning has been used in these analyses as a
data capture device. It produces a glut of data characterising the reciprocal joint surfaces
and the challenge is to usefully analyse these data (Koenerink and van Doorn, 1992). We
have decided here to base the analysis on landmark data rather than on surfaces. The rea-
son for this is that proper statistical models are not yet available for surface analysis and
comparison. We have defined a manageable number of landmarks that represent the tibial
and talar surfaces in the form of cross-sectional curves.
In order to do this, the four scans previously described were integrated using Sur-
facer 4.0 software (3D Imaging International, Inc., Trumbull, CT.). The image was ori-
ented in relation to the 3 lead shot markers that formed the Z plane. Cross-section profiles
were created by projecting a series of curves (straight lines) onto the two articular surfaces
of the talus and tibia point clouds in predefined planes. Five cross-sections (three antero-
posteriorly and two mediolaterally) were constructed through the articulated joint relative
to the triangular orientation plane of the talus (Figure 1b). Portions of the cross-section
curves (Figure 2a,b) that do not articulate were removed. For the mediolateral sections,
226 L. Aiello et at.

A Posterior

z x
Lateral

Anterior

B
AP 25 . AP 50 AP 75

z x

Height
ML50

ML25

~ --

Width

Figure 1. (A) Right talus in superior view. The shaded triangle, the "orientation plane", is a planar surface at Z=O
and its base (the line bc) is parallel to the X axis. Points a, b, and c represent the positions of the lead shot applied
to the talus before scanning. See text for description. (B) Location of the 5 anteroposterior and mediolateral cross-
section curves through the articulated talocrural joint. The AP50 cross-section curve is cut at the middle of the ori-
entation plane base (i.e., at 0.5*Width). The AP25 (lateral) and AP75 (medial) curves are constructed at 25% and
75% of basal width. The ML50 cross-section curve is located at 0.5*height of the orientation plane and the ML25
curve is at 25% of height.
Laser Scanning and PaJeoanthropoJogy 227

the articular portions of the cross-section curves were extracted by constructing a tangent
across the high points of the medial and lateral crests of the talar trochlear surface, cutting
the talus curve at this point and cutting the tibial curve at the locations normal to this (Fig-
ure 2a). For the anteroposterior sections, articular portions of the cross-section curves
were extracted by constructing a tangent across the low points of the anterior and posterior
crests of the tibial articular surface, cutting the tibial curve at this point and cutting the ta-
lus curve at the locations normal to this (Figure 2b).
The tibial and talar cross-sectional curves were analysed firstly on the basis of raw
data that provide information relevant to both the size and the shape of the cross-section
and secondly on the basis of size-corrected data that provide only information relevant to
the shape differences in the cross-section. In order to extract the raw (size and shape) data
from the cross-sections each curve was sampled at 20 even increments along its length
(Figure 2c,d). This resulted in 20 co-ordinates (here termed 8 co-ordinates) that measure
the vertical distance of the curve from its tangent (baseline) at 20 evenly spaced locations
for each tibial and talar joint surface. The congruence between the reciprocal tibial and
talar cross-sections was determined by (1) squaring each of the individual co-ordinate val-
ues for the tibial and talar joint surfaces, (2) subtracting the corresponding talar co-ordi-
nate from the tibial co-ordinate, and (3) taking the average of the summed differences over
the 20 co-ordinates. It should be noted that 81 and 820 for both the tibial and the talar
joint surfaces have the value of 0 and, therefore, contributed nothing to the analysis.
Size-corrected congruence data were gathered by scaling the cross-sections before the
co-ordinate data were collected. This was carried out by scaling the cross-sectional curves
so that in all cases point 1 was at (0,0) and point 20 was at (1,Q) with the value for X pro-
ceeding from 0 to I in increments of O.OS units. The congruence between the reciprocal
cross-sections was then determined in the same fashion as for the non-size-corrected data.
One possible concern with this method is homogeneity of the cross-sectional curves
between specimens. The curves are homogenous in the sense that they are always at a con-
stant location in relation to the reference triangle in the Z plane. When comparing randomly
associated joints, the AP7S curve projected on the tibia may not precisely match the AP7S
curve as projected on the talus. As a result, the 20 points on the tibial curve will not pre-
cisely correspond to the 20 points on the talar curve. Any differences in location would be
due either to size differences between the randomly associated bones (in the non-size cor-
rected analysis) and/or to shape differences. This mismatch would be reflected in the results
of the analysis and would be highly important in assessing the degree of congruency in the
joint. Lack of homogeneity in this sense would be a fundamental feature of the analysis.
It is also important to point out that in this context each curve was sampled at 20
evenly-spaced points along its cord. Any mismatch between say the Sth point on a particu-
lar tibial curve and the Sth point on the corresponding talar curve that may be due to the
angle of the two cords in relation to each other in the articulated joint is expected to be
negligible. On the basis of radiological measurements in the living joint, Wynarsky and
Greenwald (1983) and Jonsson et al. (1984) have determined that the articular cartilage
shows an even thickness across the joint. Such variation in the angle of the cords in rela-
tion to each other would, therefore, be expected to be minimal and would likely be less
than the resolution of the scanner «O.S mm).

3. RESULTS AND DISCUSSION

Congruency between the reciprocal joint surfaces was determined for four cross-sec-
tional curves characterising the talocrural joint, two anteroposterior curves at SO% (APSO)
228 L. Aiello et aL

and 75% (AP75) (most medial) and two mediolateral curves at 25% (ML25) (most anterior)
and 50% (ML50). For each comparative species and each cross-sectional curve, congruency
values were determined for each associated talocrural joint and for every possible intraspe-
cific combination of tibiae and tali. This gives the range of congruency variation expected
for joints from the same individual as well as that expected for a single species. In all cases
except one, associated talocrural joints from single individuals were more congruent than
randomly associated talocrural pairs (Figures 3, 4). The one exception is the AP75 cross-
section for humans based on size-corrected data (Figure 4d). Here a single associated joint
is less congruent than any of the randomly associated joints. Table I provides the summary
congruency statistics for the comparative human, chimpanzee and gorilla samples.
In. reference to the fossil talocrural joints, the results are similar in all cross-sectional
comparisons based on the raw data (containing both size and shape information) (Figure
3). The OH35/8 talocrural joint consistently is less congruent than the Australopithecus
afarensis (AL 288-1) joint. AL 288-1 shows the degree of congruency at the lower end of
the range of variation expected in an associated talocrural joint (Figure 3). In relation to
the human comparative sample, the OH35/8 joint also shows a degree of congruency that
would be compatible with an associated joint (Figure 3, Table 2). However, in relation to
the chimpanzee and gorilla comparative samples the results suggest that although the
OH35/8 joint is compatible with the intraspecific variation found in these species, it is not
necessarily compatible with the congruency that would be expected in an associated joint.
In relation to the gorilla ML25 cross-section and the chimpanzee and gorilla AP75 cross-
sections, the OH35/8 joint shows a lesser degree of congruency than any of the associated
joints from single individuals (Figure 3a, gorillas; Figure 3d, chimps and gorillas; Table
2). Nevertheless, in all cases the congruency is compatible with the intraspecific variation
expected from randomly associated talocrural joints.
Based on size-corrected data, the results are different (Figure 4, Table 3). The
AL288-1 talocrural joint continues to be more congruent that OH35/8 and fully compat-
ible with the congruency expected for an associated joint. However, in the case of both the
AP50 and the AP75 cross-sections, the congruency shown by the OH35/8 joint is less than
that of any associated joint in any of the three comparative species (Figure 4, Table 3).
The same is true for the ML25 cross-section in relation to the human and gorilla compara-
tive samples, although the congruency is compatible with an associated joint based on the
chimpanzee sample. The ML50 cross-section is compatible with an associated joint based
on the variation in all three of the comparative species.
The fact that the anteroposterior cross-sections give a different result than do the
mediolateral cross-sections may simply reflect the differences in the magnitude of vari-
ation in congruency observed for the different cross-sections in the different comparative

Figure 2. (A) Mediolateral cross-section through a schematic left talocrural joint in maximum dorsiflexion. The
articular portions of the cross-section curves are extracted by constructing a tangent across the high points of the
medial and lateral crests of the talar trochlear surface (points I and 2 respectively). The talar cross-section is cut at
points I and 2. The tibial cross-section curve (between points 3 and 4) is cut at the normals, a and b, to this tan-
gent. (8) Anteroposterior cross-section through the left talocrural joint in maximum dorsiflexion. The articular
portions of the cross-section curves are extracted by constructing a tangent across the low points of the anterior
and posterior crests of the tibial articular surface (points 5 and 6 respectively). The tibial cross-section is cut at
points 5 and 6. The talar cross-section curve (between points 7 and 8) is cut at the normals, c and d, to this tangent.
(C) The extracted talar and tibial mediolateral cross sections. (D) The extracted talar and tibial anteroposterior
cross-sections. The XV co-ordinates are obtained at 20 evenly-spaced increments along the length of each curve
segment (only 3 indicated here for each cross-section).
Laser Scanning and Paleoanthropology 229

A 1
v-x

Talus
tangent

Medial Lateral

B Tibia

Anterior Posterior

Talus
Tibia
tangent

20 20
c I.
Talus Tibia
1 2 3 "
~ ~~1'-O)------------------(0~'O~

(0.25,65)

D (0.5,610) (0.75,615)

(~~(1_'O_) _________ (_0-{O~~.OI

7 Talus 8 5 Tibia 8

20
. 20
.
 230 L. Aiello et aL

a b

.
90.------------------------------, 90 ~------------------------------,

80 AL 288-1 Humans 80 AL 288-1 Humans

70 70
_ 60 _ 60
.. 083518
c: 083518 l5o
8 ,
50
50
:p 40 :p 40
0.. 30 0.. 30

20 20
10 10
OUW~~UL-A~----------------~~
0.25 1.75 3.25 4.75 6.25 7.75 9.25 10.75 12.25 2.5 4.5 6.5 8.5 10.5 12.5 14.5 16.5 18.5 20.5 22.5
Difference Difference
90 ~---------------------------.
90 . - - - - - - - - - - - - - - - - - - - - - - - - - - - - ,
80 AL 288-1 Chimps 80 AL 288-1 Chimps
70 70
_ 60 C 60 .,. 083518
l5

~ li ~(~. _"
50 . 083518
o
:p 40 i
0.. 30 .
'
.
I

20 ,
10 10 & ~. >
o . .----------------------:-:-'.
0 0 .25 1.75 3.25 4.75 6.25 7,75 9.25 10.75 12.25 0.5 2.5 4.5 6.5 8.5 10.512.514.516.518.520.522.5
Difference Difference
90 90
AL 288-1 80 i
80
70
... Gorillas
70 I I AL288-1
Gorillas
_ 60 _ 60 I
c:

Ig: '
8 50
083518
~ 50 I 'f 083518

..
:p 40 Q) 40 ;
0.. 30
~ 30

~~o
20 '
10
0 J
0.25 1.75 3.25 4.75 6.25 7.75 9.25 10.75 12.25 0.5 2.5 4.5 6.5 8.5 10.5 12.5 14.5 16.5 18.5 to.5 22.5
Difference Difference

Figure 3. Congruency histograms based on raw data for humans, chimpanzees and gorillas and the fossil ta-
locrural joints. (a) The mediolateral25 cross-section, (b) the mediolateral 50 cross-section, (c) the anteroposterior
50 cross-section, (Ii) the anteroposterior 75 cross-section. Black bars, every possible combination of randomly as-
sociated tibiae and tali. Shaded bars, associatedtalocrural joints from single individuals.

species. The absolute range of variation (standard deviation) in congruency for the medio-
lateral sections is much smaller than for the anteroposterior sections (Figure 5). Because
of the limited range of variation in these mediolateral sections in all of the comparative
samples, they may not be as good indicators as the anteroposterior sections.
The congruency values of the anteroposterior cross-sections also suggest the possi-
bility that OH35 and OH8 may not belong to individuals of the same species. In relation to
the human AP50 and AP75 cross-sections, the OH35/8 joint lies at the extremes of the in-
traspecific variation. This is also true for the AP50 cross-section in relation to the gorilla
comparative sample (Figure 4, Table 3).
The difference between the results based on the raw (size and shape) data and on the
size-corrected data can most probably be explained by the fact that the OH35 distal tibia and
 Laser Scanning and Paleoanthropology 231

c d

90 . -----------------------------__, 90 .-------------------------------~
80 Humans 80 Humans
70 AL 288-1 70
c: _ 60

•
60 I AL 288-1
c
Q) 50.. OH3518 ~ 50
~ 40 I A lii 40 OH3518

~ ~~o I~_~J'~I~. I ...

~ 30
20
L _________ _ _ _- - ' 10
OWE.u~~~~WL~ __ ~~~ __________ ~

1.3 6 .3 11 .3 16.3 21.3 26.3 31.3 36.3 41.3 46.3 51 .3 56.3 5 9 13 17 21 25 29 33 37 41 45

Difference Difference
90 .-----------------------------__, 90 ~------------------------------__,
80 AL 288-1 Chimps 80 . AL 288-1
70 70
Chimps
60 ' _ 60
~ 50 " OH3518 ~ 50

~ :l.\L .~________
__________
~ () .
Q) 40 .) lii 40 OH35/8
~ 30 ;
20
...J 10
O~~·~~~~--------------------~
1.3 6.3 11 .3 16.3 21 .3 26.3 31.3 36 .3 41.346.3 51 .3 56.3 5 9 13 17 21 25 29 33 37 41 45
Difference Difference
90 I 90 .---------------------------------,
80 AL 288-1 Gorillas 80 Gorillas
70 70
C 60 n" _ 60
c
AL 288-1

IT
~~~
P.1 :H35"
~ 50
~ ~~~. OH35/8
20 . n ,
10 " 10 i
o • .• ••• - - - o . .------"--.-.......- ---------------'
1.3 6.3 11 .3 16.3 21.3 26.3 31.3 36 .3 41 .3 46.3 51.3 56.3 1 5 9 13 11 21 25 29 33 37 41 45
Difference Difference

Figure 3c and d.

the OH8 talus are of a comparable size. In the raw data analysis this similarity in size obscures
the differences in shape that are apparent in the size-corrected analysis. The reasonable fit be-
tween these bones that has been noted in the literature (Leakey, 1971; Susman and Stem,
1982) results from this size similarity rather than from any necessary shape similarity.

4. CONCLUSIONS

This analysis has demonstrated the use of laser scanning in the interpretation of fos-
sil hominid postcranial material. By offering a means of analysing complex joint surfaces
in three dimensions, laser scanning provides a powerful source of metrical data in addition
to the traditional metrics normally taken on postcranial bones. This is particularly impor-
 232 L. Aiello et al.

a b
90 ,------------------------------, 90 ~------------------------------~

80
70
I Humans 80 Humans
70 AL 288-1
C 60 AL 288-1 C 60
~ 50 I ' Ql
~
50 'f
OH35/8
~ :'1 ... OH35/8 Ql 40

:! illtn:.
40
CL 30 1
'f
20 .}

10 :111. fl.
OWL~~~.L---~----~--~------~
~.
1.3 8.8 16.3 23.8 31.3 38.8 46.3 53.8 2.5 17.5 32.5 47.5 62.5 77.5 92.5 107.5
Difference Difference

90 . - - - - - -- - - - - -- - - - - -- -- - - -- -- -- -- . 90,-------------------------------,
Chimps 80 Chimps

i III
70
60 AL 288-1
C

•
:L 288-1 ~ 50 I OH35/8
W 40 'f
CL 30 r~ OH35/8 CL 30

20
10
t ILn f1 ..
11 n
20
10
o ...1 • .i...L . ... . . . . . . ' " - - - -- - ---'
17.5 32.5 47.5 62.5 77.5 92.5 107.5
1.3 8.8 16.3 23.8 31.3 38.8 46.3 53.8
Difference Difference

90 1--------- ----- ------- , 90 . - -- - - - - - - -- - -- - - - .

80 ; Gorillas 80 Gorillas
70 I 70 AL 288-1
c 60 i AL 288-1
C
60
1
Ql 50 : Q) 50 ...
u OH35/8
Q; 40 t ...
~
40
OH35/8 Q)
a.
CL
30 ~ 30

~H I~ II :. ~ ----------:-~
20
10
OUL~LL~LL~~~ ______________ ~

1.3 8.8 16.3 23.8 31 .3 38.8 46.3 53.8 2.5 17.5 32.5 47.5 62.5 n.s 92.5 107.5
Difference Difference

Figure 4. Congruency histograms based on size-corrected data for humans, chimpanzees, gorillas, and the fossil
talocrural joints. (a) The mediolateral 25 cross-section, (b) the mediolateral 50 cross-section, (c) the anteroposte-
rior 50 cross-section, (d) the anteroposterior 75 cross-section. Black bars, every possible combination of randomly
associated tibiae and tali. Shaded bars, associated talocrural joints from single individuals.

tant for the comparative analysis of hominid postcranial fossils where traditional mor-
phometries involving lengths and breadths of long bone shafts and joint surfaces may not
provide adequate resolution.
Although this analysis has used a relatively simple morphometric technique to com-
pare congruency of talocrural joints, it has produced results that are entirely consistent
with expectations for the known comparative samples. In particular, talocrural joints from
single individuals have a closer congruency than randomly associated intraspecific pairs.
Furthermore, the AL 288-1 talocrural joint in all analyses shows the close congruency ex-
pected of an associated joint from a single individual. The OH35/8 joint does not show
 Laser Scanning and Paleoanthropology 233

c d
90 , ---------------------------, 90r-----------------------------~

Humans 60 Humans
~~ AL 288-1 70
C 60 .. C 60
50

~ ~~ ~'
Q)

OH35/8 ~ 40 AL 288-1 OH3518

Q. 30 a.. 30 ,
n '"
1~ Ei.1-1i.:_________----'
20 ' 20 '"
10

5 25 45 65 85 105 125 145 165 185 205

Difference
90 .------------------------------, 90
60 Chimps 60
AL 288-1 Chimps
70 70
_ 60 AL 288-1
c:
~ 50 ... 0H3518
iii 40 OH35/8

.
I
30
'"
Q.

o ~~·AL -_ __ _ _ _.uL-~~
5 25 45 65 85 105 125 145 165 185 205 2,5 12,5 22,532,5 42.5 52.562.572.5 62.592.5102.5
Difference Difference

90 .-----------------------------, 90,-----------------------------,
80 Gorillas 60 Gorillas
70 AL 288-1 70

•
60
C 60 AL 288-/
~c.> 50 ~ 50
Q; 40 iii 40 OH35/8
D.
OH35/8 a.. 30 '"
20 20
10 10 .
". '.".". '.'- o " ..:
0~5ua2d5LA4J5La6~5~6-5-1-0-5~12~5~1~45~1-6~5~1-
65--20~5 2.5 12.5 22.532.5 42.5 52.5 62.572.562.592.5102.5
Difference Difference

Figure 4c and d.

this same consistency. The analyses based on raw data (carrying size and shape informa-
tion) suggest that OH35 and OH8 most probably represent different individuals, particu-
larly in comparison to the congruency variation found in the gorilla and chimpanzee
comparative samples, while the analyses based on size-corrected data suggest that they
may even come from different species as originally thought.
While not conclusive, these results suggest that the observed similarity between OH35
and OH8 (Leakey, 1971; Susman and Stem, 1982) is primarily a size-based similarity rather
than necessarily a shape-based similarity. It leaves open the possibility that the difference in
joint shape between OH35 and OH8 may also indicate a difference in functional capabili-
ties. Further studies involving more sophisticated morphometric analyses (e.g., Bookstein,
1991; Marcus et aI., 1996) and larger comparative samples would be expected to provide
better resolution of the taxonomic affiliation of OH35 and OH8 as well as any possible dif-
ference in functional capabilities that might be indicated by these bones.
234 L. Aiello et al.

Table 1. Congruency statistics of cross-sections for the comparative sample of humans,

chimpanzees, and gorillas

AP50 AP75 ML25 ML50

Mean SO Mean SO Mean SO Mean SO
Shape only: Associatedjoint l
Human 25.70 16.50 36.96 25.43 4.03 3.41 4.00 3.40
Chimpanzee 28.10 22.90 13.16 17.68 7.59 8.66 7.60 8.70
Gorilla 12.50 11.40 15.48 11.40 2.36 2.52 2.40 2.50
Shape only: Random pairs2
Human 27.30 17.20 37.30 21.88 8.74 9.39 8.70 9.40
Chimpanzee 29.80 27.00 27.40 27.86 12.76 11.99 12.80 12.00
Gorilla 18.00 17.80 24.12 24.46 7.62 7.17 7.60 7.20
Size and Shape: Associatedjoint
Human 6.10 3.\0 7.78 5.46 0.71 0.54 0.70 0.50
Chimpanzee 3.10 2.50 1.59 2.20 0.70 0.83 0.70 0.80
Gorilla 2.20 2.20 3.12 2.1\ 0.46 0.45 2.\0 2.30
Size and Shape: Random pairs
Human 8.20 7.90 8.71 6.70 1.31 1.44 1.30 1.40
Chimpanzee 4.90 6.10 4.25 4.42 0.97 0.90 0.97 0.90
Gorilla 3.80 4.20 5.30 6.06 2.11 2.30 0.50 0.50
I Associated joints are those known to come from the same individual. N=1O for each species.
2 Random pairs represent every possible intraspecific combination of tibiae and tali. N=90 for each species.

ACKNOWLEDGMENTS

We are grateful to Dr. Derek Howlett, Curator of the Powell Cotton Museum, Birch-
ington, Kent for the loan of the gorilla and chimpanzee material and to Theya Molleson
and the Natural History Museum, London, for the loan of the Spitalfields sample. We
would also like to thank Elizabeth Strasser for the opportunity to participate in the "Pri-
mate Locomotion-1995" symposium and in that way to commemorate the many contribu-
tions of Warren G. Kinzey to the field of primate locomotion. This paper was considerably

Table 2. Congruency position ofOH35/8 based on raw datal

Cross-Section Human Chimpanzee Gorilla

AP50
Associated joint 40.0 20.0 10.0
Random pairs 45.6 27.6 21.1
AP75
Associated joint 20.0 0.0 0.0
Random pairs 84.0 22.2 14.4
ML25
Associated joint 10.0 10.0 0.0
Random pairs 28.9 22.2 43.3
ML50
Associated joint 30.0 40.0 50.0
Random pairs 28.9 26.7 56.7
I Values in each column are the percentage of individuals in the comparative
sample with a degree of congruency that is as large as or larger than that of
the OH35/8 pair. Definitions and sample sizes as in Table I.
Laser Scanning and Paleoanthropology 235

Table 3. Congruency position ofOH35/8 based on

size-corrected data I
Cross-Section Human Chimpanzee Gorilla
APSO
Associated joint 0.0 0.0 0.0
Random pairs 1.1 10.0 1.1
AP75
Associated joint 0.0 0.0 0.0
Random pairs 0.0 22.2 14.4
ML25
Associated joint 0.0 20.0 0.0
Random pairs 18.9 35.6 11.1
ML50
Associated joint 20.0 60.0 30.0
Random pairs 27.8 56.7 44.4
IDefinitions and sample sizes as in Table 2.

40 ~----------------------------------------------------~

c 30
0
~
I
.:;
Q)
0
-0
ro 20
'-
-0
C
ro
U5
10

o 1-__________-,~--------------------_!Ii~j~~r~cf~~t~.t~!~j~~1~a____________~~----------J
6 4 6 4 6 4
Chimpanzee Gorilla Human

Species

Figure 5. Boxplots illustrating the magnitude of variation (standard deviation) in congruency in both random and
associated pairs of anteroposterior cross-sections (open boxes) and in both random and associated pairs of medio-
lateral cross-sections (shaded boxes) for chimpanzees, gorillas, and humans. Values below each boxplot is the
sample size. Note that although there is variation in the standard deviation between species, for each species the
magnitude of the standard deviation of the mediolateral cross-sections is always smaller than that of the anteropos-
terior cross-sections.
236 L. Aiello et al.

improved through suggestions made by Elizabeth Strasser, John Kappelman and two
anonymous referees. The research reported in this paper was supported by The
Leverhulme Trust grant F.134BB to L.A. and by NERC grant GRlH616174 to B.W.

REFERENCES
Aiello LC, and Dean MC (1990) An Introduction to Human Evolutionary Anatomy. London: Academic Press.
Aiello LC, and Wood BA (1994) Cranial variables as predictors of hominine body mass. Am. 1. Phys. Anthropol.,
95:409-426.
Bookstein FL (1991) Morphometric Tools for Landmark Data. Cambridge: Cambridge University Press.
Davis PR (1964) Hominid fossils from Bed I, Olduvai Gorge, Tanganyika. Nature 201:967-970.
Day MH (1976) Hominid postcranial material from Bed I, Olduvai Gorge. In G Isaac and E McCown (eds.): Hu-
man Origins. California: W.A. Benjamin, pp. 363-374.
Day MH (1978) Functional interpretations of the morphology of postcranial remains of early African hominids. In
CJ Jolly (ed.): Early Hominids of Africa. London: Duckworth, pp. 311-345.
Jonsson K, Fredin HO, Cedgrlund CG, and Bauer M (1984) Width of the normal ankle joint. Acta Radiologica Di-
agnosis 25:147-149.
Harvey PH, Martin RD, and Clutton-Brock TH (1987) Life histories in comparative perspective. In BB Smuts, DL
Cheney, RM Seyfarth, RW Wrangham, and TT Struhsaker (eds.): Primate Societies. Stuttgart: Thieme-Ver-
lag, pp. 419-434.
Koenerink JJ, and van Doom AJ (1992) Surface shape and curvature scales. Image and Vision Computing,
10:557-565.
Leakey MD (1971) Olduvai Gorge: Excavations in Beds I and n, 1960-1963, Vol. 3. Cambridge: Cambridge Uni-
versity Press.
Leakey LSB, Tobias PV, and Napier JR (1964) A new species of the genus Homo from Olduvai Gorge. Nature,
202:7-9.
Marcus LF, Corti M, Loy A, Nayler GJP, and Slice DE (1996) Advances in Morphometrics. New York: Plenum.
Susman RL, and Stem IT Jr. (1982) Functional morphology of Homo habilis. Nature, 217:931-934.
Tobias PV (1991) OIduvai Gorge: The Skulls, Endocasts and Teeth of Homo habilis. Vol. 4. Cambridge: Cam-
bridge University Press.
Wood BA (1991) Koobi Fora Research Project IV: Hominid Cranial Remains from Koobi Fora. Oxford: Claren-
don.
Wood BA (1992) Origin and evolution of the genus Homo. Nature, 355:783-790.
Wood BA, Aiello LC, Wood C, and Key C (submitted) A technique for establishing the identity of "isolated" fos-
sil hominid limb bones. Journal of Anatomy.
Wynarsky GT, and Greenwald AS (1983) Mathematical model of the human ankle joint. Journal of Biomechanics,
16:241-251.
 14

USE OF STRAIN GAUGES IN THE STUDY OF

PRIMATE LOCOMOTOR BIOMECHANICS

Brigitte Demes

Department of Anatomical Sciences

School of Medicine
State University of New York at Stony Brook
Stony Brook, New York 11794-8081

1. INTRODUCTION

The strain gauge technique is a relatively recent addition to the catalogue of experi-
mental methods available for functional analyses, especially locomotor studies. Strain
gauges track the deformation of objects they are attached to, thus allowing the reconstruc-
tion of external forces and loads that cause these deformations. They are restricted to sur-
face use, but extrapolations allow us to reconstruct strain patterns through the object (e.g.,
Gross et ai., 1992; see also example in Figure 8). In the field of biomechanics there are
two major applications: the measurement of bone deformations and the instrumentation of
force measuring devices. The data in both fields can be used in interpreting musculoskele-
tal morphology. Functional interpretations of bony morphology have been historically
based on correlations between shape and activity. The interface is the mechanical environ-
ment into which behaviors translate and in which bone develops, maintains and/or
changes its shape. The mechanical demands of particular locomotor modes are commonly
derived from behavioral observations in combination with biomechanical models. Measur-
ing the external forces acting on limbs with force transducers is a first step in testing the
numerous assumptions inherent in this process. Even with this background information,
however, actual loadings of a bone can only be deduced with a certain degree of plausibil-
ity. In vivo measurement of bone strain is currently the only method of directly determin-
ing the major loading regimes caused by the external forces acting on the bone.
In the following I will briefly describe the technique, review its applications in pri-
mate locomotor biomechanics, and give two examples of recent work of my own in this
field: in vivo bone strains recorded on the macaque ulna (Demes et ai., 1997, 1998), and a
custom-designed "force pole" to measure takeoff and landing forces of leaping prosimians
(Demes et ai., 1995).

237
238 B. Demes

2. MECHANICAL BACKGROUND: STRAIN*

Strain is a technical term used to express deformation. It is defined as the change in

length over original length or percentage change in length, and is, therefore, a dimension-
less unit. Out of convenience units are nevertheless assigned, and for small deformations
these are microstrain (/lc).

M
£ =- =0.000001 =11-1£
L

with M = change in length and L = original length.

Strains perpendicular to any given plane are called normal strains and they are either
compressive (c-) or tensile (c+). Strains within any given plane are called shear strains (y).
Within any given strain field (any point on the surface where strain is measured)
there is a maximum and a minimum normal strain at right angles to each other. These are
called the principal strains. In most biomechancial applications, the maximum principal
strain is tensile (positive sign) and the minimum principal strain is compressive (negative
sign), independent from their relative magnitudes. Biaxial compressive or tensile strain is
less common. The magnitude and alignment of the principal strains are indicative of the
major loading regimes. In a simple, symmetrical beam, the mechanical analog most fre-
quently used in biomechanics, axial forces cause either compressive or tensile strains in
line with the long axis of the beam and with the opposite strain quality represented at right
angles to the long axis. This transverse strain component is usually low and results from
the change in width of the beam being tensed (decrease in width, compressive transverse
strain) or compressed (increase in width, tensile transverse strain). The amount of trans-
verse to longitudinal strain is expressed in Poisson's ratio v, and is a function of material
stiffness:

with c,'" = longitudinal strain and cxx = transverse strain.

Poisson's ratios for bone are approximately 0.3-0.4 (Currey, 1984).
Bending results in a composite strain pattern with longitudinal tension at the convex
side of bending and longitudinal compression at the concave side of bending. The transi-
tion occurs at the neutral plane of bending, which is crossed by the principal (tensile and
compressive) strains at 45 degree angles (Figure I). Shear strains (not shown in Figure I)
also result from bending because bending causes a tendency of the fibers in a beam to
slide against each other. The maximum shear strains are at 45 degree angles to the normal
strains, highest at the neutral plane of bending and zero in the outer fibers. Twisting results
in principal strains at 45 degrees to the long axis of a symmetrical beam. These strains are
highest at the surface and decrease to zero towards the centroidal neutral axis of torsion.
Unlike in bending, these strains are distributed uniformly over the surface of the beam.

• Note that engineers distinguish between "true strain" and "engineering strain". Calculations of engineering strain
do not take into account the change in cross-sectional dimensions of the strained object that influences its total
deformation. Engineering and true strain are practically coincident when the strains are small, as is the case in
most biomechanical applications.
Use of Strain Gauges in the Study of Primate Locomotor Biomechanics 239

neutral axis
of bending +--.;,--------:-------f'-.d--4,...--~'--+____;.L--____,4_--r_--¥~!......;w~

- - - tension
compression

Figure 1. Principal strain trajectories on the surface of a beam loaded in three-point bending. Principal tensile
strains are shown as solid lines, compressive strains as dashed lines; their relative magnitudes are visualized as el-
lipses or circles for several different locations on the surface (after Ramm and Wagner, 1967: 126).

Measuring the principal strains at one surface location may not suffice to resolve the
loading patterns of a beam. For example, principal strains at 45 degrees to the long axis
may be encountered in a twisting regime but also in bending close to the neutral axis (Fig-
ure 1). Multiple gauges are required to distinguish between twisting and bending (Gross et
ai., 1992). Loading regimes are also frequently superimposed, which complicates their in-
terpretation. When measuring strain on bones the pattern is further complicated by the fact
that bones are not simple beams. They are composed of more than one material (heteroge-
neous) and may strain differently in different directions (anisotropic); i.e., they respond
with various amounts of deformation for a given external loading regime at different lo-
calities and in different directions. In addition, bones are not symmetrical.

3. STRAIN GAUGE TECHNIQUE

The most commonly used strain gauges are electric resistance gauges. They work on
the basis of Kelvin's principle, which states that the electric resistance of metals is influ-
enced by their state of strain. The gauges consist of a metal wire or foil that is, in the com-
monly used bonded type, fixed to a flexible base (to keep the stiffness low so that the
gauge does not reinforce the object that it is bonded to). The sensitivity of the gauge is de-
termined by the gauge factor and the length of the wire (Dove and Adams, 1964):

fiR=FRE

with fiR = change in electric resistance, F = gauge factor, R = original resistance, E =

strain.
The gauge factor is a measure of the change in resistance per unit of original resis-
tance that will occur per unit of strain applied. The length of the wire determines the origi-
nal resistance. "Good" gauges, i.e., gauges with a high sensitivity should have a high
gauge factor and a high original resistance. Wires in electric resistance gauges are there-
fore arranged in a grid, which allows them to be long and the gauges to be small at the
same time. Most commonly used strain gauges in biomechanical applications have gauge
 240 B. Demes

5mm

Figure 2. Rtctangular rosette (Kenkyujo FRA-I-II-IL). The three elements are stacked on top of each other and
attached to a larger, round epoxy backing that also holds the soldering connections to the wires from the elements.
The gauge is I mm long, the backing 4.5 mm in diameter.

factors of around 2 and an original resistance of 120 n, with lengths and diameters of only
a few mm (Figure 2).
The circuitry most commonly used with electric resistance gauges is a Wheatstone
bridge (Figure 3). It is a circuit of four resistors, one of which is the strain gauge. A low
DC voltage is applied to two contact points of the circuit, and the output voltage is meas-
ured across the other two. Any change in resistance of the gauge produces a DC offset in

;SAL
R3

Vin - BAL

R2
R(i AL

Figure 3. Wheatstone bridge circuitry with two calibration resistors (CAL) and a balancing resistor (BAL). Ym in-
dicates input voltage supplied across one diagonal, YOU! the output voltage measured across the other diagonal.
Use of Strain Gauges in the Study of Primate Locomotor Biomechanics 241

the signal voltage (because voltage is directly proportional to resistance--Ohm's law).

The nominal resistance of all four resistors is originally the same (120 n), i.e., the output
voltage is zero. However, the resistance of the gauge may change during the bonding pro-
cedure--it may be slightly strained and therefore have a higher or lower resistance than
120 n. The Wheatstone circuitry allows one to "balance" the circuitry after the gauge is in
its final position.
The voltage change for a given amount of strain is determined by shunting two resis-
tors that produce a change in output voltage corresponding to a known amount of fJe (e.g.,
± 1000 fJe) with two arms of the bridge: shunt calibration. When the calibration switch is
thrown they produce a change in resistance and output voltage. The measured output volt-
age change is used to calculate calibration factors. The output voltage is amplified and
sampled either directly into a computer (which requires the transformation of the analog
signal into a digital one; i.e., an AID board) or on tape. The length of the cable from the
gauge to the amplifiers should be kept short because it acts as an antenna and adds noise
to the actual signal.
Single element gauges only register strain in one direction. The more frequently
used rosette gauges combine three single elements arranged at known angles to each other
(45 degrees: rectangular rosettes (Figure 2), 60 degrees: delta rosettes). With the informa-
tion from three elements, the magnitudes and directions of the strains in a biaxial strain
field can be resolved. The change in output voltage for each element is transformed into
strain units using the calibration factors. Trigonometric relations as expressed in the so-
called Mohr's circle of strains (Dally and Reilly, 1991; see also Biewener, 1992) are then
applied to calculate strains with reference to particular axes of interest. Most frequently
reported are the principal strains and their angles and the maximum shear strain. The
strain in the direction of the long axis of a bone is a sometimes more intuitive vehicle to
interpret deformations. The gauge orientation on the surface is prerequisite knowledge for
relating strain directions to the geometry of the object the gauge is bonded to.

4. BONE STRAIN MEASUREMENTS IN PRIMATE

BIOMECHANICS
Bonded metallic resistance gauges were developed in 1940 (Dove and Adams,
1964). As early as 1944, they were first applied in the measurement of bone strain. Gurd-
jian and Lissner (1944) surgically implanted strain gauges onto the cranial vaults of dogs
and recorded deformations. They banged the heads of the anesthetized animals and regis-
tered the oscillations of the cranial vault (as well as intracranial pressure changes) in an at-
tempt to understand head injury mechanisms. The single element gauges were about half
the length of the temporal fossa of the dogs, and data reported were selected traces of
strain on an oscilloscope screen. Modem rosettes are available in much smaller sizes, but
reporting selected strain readings only is still common practice.
Another early in vivo strain study on the tibia of a dog was the first one published in
an anthropological journal (Evans, 1953). The first application of strain gauge technique
in primate biomechanics was in 1975, when Lanyon and coworkers in England implanted
a rosette gauge on the anteromedial aspect of a human tibia. The subject was one of the
authors. The size of the gauge was not reported but the incision to implant it was 10 cm
long. Bone strains were recorded during walking and running. During the support phase of
walking, the largest principal compressive strains were about 400 fJe, the peak principal
tensile strain only slightly lower and at an angle of about 50 degrees to the long axis of the
242 B.Demes

bone. The authors interpreted this pattern as the result of the combined effect of muscle
pull and body weight.
Bone strain papers from Hylander's lab at Duke University began coming out in
1977. They reported strains during mastication from the mandible and facial skeleton in
several species of nonhuman primates (e.g., Hylander 1977,1979,1984; Hylander et aI.,
1991; Hylander and Johnson, 1994) and represent by far the bulk of primate bone strain
data. Having substantially altered and improved interpretations of primate jaw shapes,
these papers demonstrate the immense importance of this approach in understanding bone
adaptations.
In comparison, strain gauge analyses of the primate locomotor system are very lim-
ited in number. Aside from the already mentioned human tibia experiment by Lanyon and
coworkers (1975), there are only a few studies on primate postcrania.
In 1977, Young et aI. implanted a miniature strain transducer into the tibia ofa ma-
caque and registered strains during a variety of activities. This was more of a technical re-
port to introduce and test the transducer, and the actual strain data (or the monkey) did not
receive much attention.
Fleagle et aI. (1981) measured strains on the posterior aspect of the ulna of a spider
monkey with a single element gauge aligned with the long axis of the bone. The gauge
registered tensile strains during the support phase of walking, which they ascribed to ante-
rior bending of the ulna. Strains during brachiation and climbing were similar in pattern
but of much lower magnitude.
In a more detailed study, Swartz et aI. (1989) reported strains on the anterior and
posterior surfaces of each of the long bones of gibbon forelimbs during brachiation. Hylo-
batid forelimb bones are very long and slender, suggesting that they experience less bend-
ing and compression than those of other animals (Kummer, 1970) and, instead, are loaded
predominantly in axial tension (Swartz et aI., 1989). Although this was confirmed in the
experiments for the ulna, where principal strains on the anterior and posterior aspects were
both tensile and in line with the long axis of the bone, the radius and humerus did not
seem to experience axial loading. The radius experienced compression on the ventral mid-
shaft and tension on the dorsal midshaft at midswing, with strain directions again in line
with the long axis of the bone, i.e., a bending regime. Compressive strains were higher
than tensile strains, which suggests axial compression superimposed on bending. Swartz et
aI. concluded that the bending is caused by the bone's curvature, which is seven times
greater than that of the ulna. The gibbon radius is curved in the mediolateral plane, how-
ever, and its curvature is unlikely to be responsible for anteroposterior bending. In the an-
teroposterior plane the gibbon radius is straight. In addition, if bending was the
predominant loading regime, the tensile and compressive peaks on opposite cortices
should have occurred simultaneously-which is not the case in the representative strain
traces presented in their Figure 1. Finally, the principal strains on the humerus were ori-
ented at an angle of approximately 45 degrees to the long axis of the bone, which is typi-
cal for a torsional loading regime. These strain patterns do not support their hypothesis
that compressive muscle forces and tensile external weight force will oppose each other
during brachiation (in contrast to being additive in terrestrial locomotion) and thereby re-
duce the net strain in the skeleton. Peak strain magnitudes for the three bones are all
around 1500 ,.u:.
Very recently, Burr et aI. (1996) duplicated the Lanyon et aI. (1975) experiments on
the human tibia, with the subject performing more rigorous exercises. Maximum strains
for zigzag runs on inclined surfaces reached nearly 2000 ,.u:, strains during level walking
were slightly higher than those recorded by Lanyon et aI.
Use of Strain Gauges in the Study of Primate Locomotor Biomechanics 243

5. CASE STUDY # 1: IN VIVO BONE STRAIN ANALYSIS OF THE

MACAQUE ULNA

Bone strain measurements at SUNY Stony Brook's primate locomotion laboratory

were started recently to test certain assumptions inherent in the functional analysis of bone
morphology. When interpreting the cross-sectional geometry oflong bones, for example, lo-
comotor modes are implicitly or explicitly equated with major loading regimes: limbs that
are moved predominantly in sagittal planes during locomotion will experience sagittal bend-
ing, or hind-limb bones in leaping animals will experience high loads (e.g., Demes and
Jungers, 1993). For primates, specifically, a great range of loading patterns is generally as-
sumed because they use their limbs, especially their forelimbs, in a more versatile fashion
than nonprimate mammals. However plausible these assumptions may be, they can only be
verified by directly measuring the deformations of bones during various activities.
The functional anatomy of the ulna and its articulations with the wrist and the
humerus are difficult to interpret. The amount of weight snaring and force transmission
between the two forearm bones is unknown. It is equally unclear whether and to what de-
gree the two bones act as a unit in resisting the loads acting on the forearm; i.e., whether,
in a bending regime, the bending moments would be distributed over the combined cross
sections in a way that one bone experiences predominantly compression, the other one
tension. The ulna and the elbow joint have, however, received considerable attention by
functional morphologists. They display significant variation among primates that corre-
lates with locomotor modes and these elements are frequently represented in the fossil re-
cord (e.g., Fleagle et aI., 1975; Rose, 1988; Harrison, 1989; Richmond et aI., 1998).
Functional interpretations of the medial-and lateral flanges that characterize the cercopi-
thecid elbow have been offered by all of the abQve authors, but, as Rose points out, "a de-
tailed functional interpretation of these features is difficult without good data on the
direction in which loading forces act-at the elbow at different phases of the gait cycle"
(1988: 214).
Experiments were performed on an adult female rhesus macaque. The ulna offered
itself because of its comparatively easy surgical accessibility and muscle-free sections.
Similar experiments are currently under way for the tibia, the most accessible hind-limb
bone. The experiment was performed in collaboration with the Musculo-Skeletal Research
Lab (Director: Clinton Rubin) and follows their protocol (e.g., Gross et aI., 1992). Prior to
surgery, the animal was trained to walk while controlled by a pole attached to a neck col-
lar. This allowed us to run the gauge wires from the animal to the animal trainer and from
there to a computer. Three rosette straiIl gauges (Kenkyujo, 2 mm diameter) were surgi-
cally implanted on the ulna, through a single incision of 5 cm length, approximately 1/3 of
the way up the shaft from the distal end in an area that required no muscle disruption (Fig-
ure 4). The bone surface was exposed by removing a small area of periosteum at each
gauge site. The bone surface was then c!egreased and dried with isopropyl alcohol, and the
gauges glued on with cyanoacrylate. The gauges were located on the (antero)medial cor-
tex (more medial than anterior), the (antero)lateral cortex (more lateral than anterior) and
the postero(lateral) cortex (more posterior than lateral). Wires from the gauges were
passed through small resin flanges that were screwed onto the bone approximately 2-3 cm
proximal to the gauge sites to provide strain relief. The leads were then passed subcutane-

t Stress concentrations around cut holes decrease rapidly, and at a distance from the edge of the hole equal to the
radius of the hole, stresses are usually equal to the applied stresses (Timoshenko, \958).
244 B. Demes

Figure 4. Anteroposterior x-ray of the ma-

caque forearm with gauges attached. Wires
are running subcutaneously to the shoulder.
Bone screws hold plastic flanges (not vis-
ible) for strain relief of the wires coming
from the gauges. The arrow indicates the
level of the gauges.

ously to the shoulder region and surfaced through a small incision between the shoulder
blades where the animal could not reach them. Here they were soldered to a connector.
Connectors and wires were protected by a vest worn by the monkey.
The animal engaged in apparently normal locomotion immediately after surgery and
data were collected then and again two days later. On the day following surgery the animal
was sore and did not locomote normally. The animal was anaesthetized again after the sec-
ond period of data collection, CT scans and X-rays were taken to verify the gauge posi-
tions (Figure 4), and the gauges were then removed. The animal recovered without
complications.
Rosettes were conditioned with a 3 V excitation voltage and amplified with the
Vishay Measurement Group 2110. Data were subsequently sent through a low-pass filter
with a cut-off frequency of 100 Hz, ND converted, and stored in an IBM-compatible com-
puter. Sampling frequency was 100 Hz and sampling periods were 10 seconds long. The
animal was videotaped during the experiments so that strain patterns and activity patterns
could be matched (Figure 5). For synchronization, one of the strain channels was superim-
posed over the live animal video (see Stem et aI., 1977 for more details on the method).
A wide range of locomotor activities was recorded, including walking, galloping and
climbing. For sequences that were visually identified on the videotapes as representative
of a particular locomotor mode, principal strains were calculated using standard formulae
Use of Strain Gauges in the Study of Primate Locomotor Biomechanics 245

Figure 5. Experimental set-up. The wire runs from the animal's shoulder to a pole attached to its neck collar and
held by the experimenter. From there it runs to the computer, being held by a second person to avoid tangling. The
superimposed trace is the strain from one gauge element and was used to synchronize strain data and video im-
ages. Figure taken from video.

(Dally and Riley, 1991) in a custom-written subroutine of the Macintosh-based analysis

program Igor (Wavemetrics). In addition, strain components in the direction of the long
axis were calculated (longitudinal strains). Strain data for walking steps only will be dis-
cussed here.
Figure 6 presents longitudinal strains for a sequence of steps, and Figure 7 shows the
tensile and compressive principal strains and the angle of the tensile (maximum) principal
strain with the long axis of the bone for the three gauge sites for those steps. Peak maxi-
mum and minimum as well as longitudinal strains during the stance phase of all steps ana-
lyzed are presented in Table 1.
Peak strain magnitudes and peak strain angles during the support phase are very
consistent (Table I, Figure 7). Strain magnitudes during the swing phase are low. Strain
angles throughout support phase do not change much whereas they are highly variable for
the swing phase. t The (antero )lateral cortex experiences longitudinal tension, the (an-
tero )medial cortex experiences longitudinal compression (Table I). The postero(lateral)
cortex experiences comparatively low tensile strains. This distribution is compatible with
mediolateral bending. The postero(lateral) gauge is probably located close to the axis of
bending, which is not strictly anteroposterior but slightly oblique, from posterolateral to

t Note, however, that angular changes are less dramatic than suggested by the traces in Figure 7, as negative
(clockwise) angles correspond to positive (counterclockwise) angles minus 180 degrees.
246 B. Demes

(antero)lateral
400

200

postero(lateral)
c:
~:1
-e
...
°i
U)

u
0- 0
(antero)medial
E
-200

-400

-600

-800 Figure 6. Longitudinal y-strains, i.e., strains in the

direction of the long axis of the bone, for the three
gauge sites. The (antero )Iateral and postero(lateral)
-1000 cortices experience tensile strains whereas the (an-
2.0 2.5 3.0 3.5 4.0 tero )medial cortex is in compression during the sup-
port phase of walking. Strains for four consecutive
time (s) steps are shown.

anteromedial. This is confirmed by the calculation of the normal strain distribution for the
entire cross section using combined beam and finite element model analysis (courtesy D.
Polonet; see Rybicki et aI., 1977, and Gross et aI., 1992 for the method). The shades of
gray in Figure 8 are indicative of various amounts of tensile and compressive strains over
the cross section, and the solid line corresponds to the neutral axis of bending. The devia-
tions of the principal strains from the long axis of the bone (Table 1, Figure 7) indicate
that the ulna is not loaded in pure bending during the support phase of walking, but, in-
stead, also experiences torsion. The angles suggest superimposition of a negative (proxi-
mal clockwise) torque.
Mediolateral bending as a major component in the loading regime of the ulna is a
somewhat provocative and unexpected result. The majority of in vivo strain studies of long
bones in other species have identified anteroposterior bending as the predominant loading
regime during the support phase of locomotion: sheep radius (Lanyon and Baggott, 1976),
radius and tibia of horse and dog (Rubin and Lanyon, 1982), horse radius (Biewener et aI.,
1983), goat tibia and radius (Biewener and Taylor, 1986), and horse tibia (Biewener et aI.,
1988). Gross et ai. (1992), on the other hand, report tension not only for the anterior, but
also for a small part of the lateral cortex of the horse metacarpal, and Biewener and Ber-
tram (1993) measured compression on the cranial (anterior) and medial cortices of the
Use of Strain Gauges in the Study of Primate Locomotor Biomechanics 247

c
'!... 400
en
e
CJ
200
0
'E
Cantero)-Iateral
Q) 0
C, -50
c
«I -100

2.0 2.5 3.0 3.5 4.0

Time (s)
c
'!... 400

een
200
-293
CJ 0
'E -200
postero-{Iateral)
Q) 0
C, -50
c
«I -100

2.0 2.5 3.0 3.5 4.0

Time (s)

400
200
c
.;
......en 0

eu -200
-400
'E -600
-800 358 {antero)-medial
-1000

50
Q)
c, 0
c -50
«I
-100
2.0 2.5 3.0 3.5 4.0
Time (s)

Figure 7. Principal strains and angles of the maximum principal strain with the long axis of the ulna for the three
gauge sites and four consecutive steps. The arrows indicate the peak strain for one of these steps and their orienta-
tion relative to the long axis of the bone.
 248 B. Demes

Table 1. Peak ulnar strains (in microstrain) for the stance phase of walking
(Antero )medial (Antero )Iateral Postero( lateral)
N II 17 17
Max. principal strain 361 ± 128 495 ± 67 217±44
Min. principal strain -993 ± 150 -182±28 -273 ± 31
Angle * -107 ±2 12 ±2 29 ± 3
Longitudinal strain -873 ± 138 466 ± 60 99± 29
• Angle of the peak maximum principal strain with the longitudinal axis ofthe bone; positive values
indicate counterclockwise direction.

chicken tibiotarsus, a pattern similar to our findings. Finally, Figure 7a in Yoshikawa et aI.
(1994) indicates mediolateral bending in the tibia of dogs, however, for unnatural, bipedal
locomotion. For the macaque, mediolateral bending with the medial cortex in compression
is probably caused by the weight force vector that passes medially to the limb. The source
of the twisting regime is less obvious. Twisting could result from forearm rotation during
the support phase as well as from muscle forces. Although bending and axial loading re-
gimes are usually emphasized in the other analyses cited above, superimposed torsion is
also indicated by the deviation of the principal strain angles from the long axes of the
bones. These deviations are especially large in the chick tibiotarsus (Biewener and Ber-
tram, 1993).
Peak strain magnitudes in the range of 100 to 1000 Ill> (Table If are low in compari-
son to strains recorded during locomotion for other animals, which most commonly range
in amplitude from 2000-3000 Ill> (Rubin and Lanyon, 1982, 1984). Indeed, this latter level
is considered to be beneficial for bone tissue and to be maintained through kinematic ad-
justments (speed, joint angles) over wide ranges of body sizes and locomotor modes ("dy-
namic strain similarity", Rubin and Lanyon, 1984; Rubin et aI., 1994). Peak strain levels
during locomotion below the 2000-3000 Ill:: bracket are reported, however, making the

anterior
40%

lateral medial
80%
Figure 8. Distribution of normal strains in the cross
section of the macaque ulna at the level of the gauge
sites. The four cross sections represent the strain pat-
terns at 20%, 40%, 60%, and 80% of support phase.
The line indicates the neutral axis of bending, and the
shades of gray the increasing tensile (positive) and
compressive (negative) strains. The numbers are the
posterior maximum strains on the convex and concave side of
bending.

o Strain magnitudes were not substantially higher during more vigorous activities
Use of Strain Gauges in the Study of Primate Locomotor Biomechanics 249

low strain levels in the macaque ulna less unusual (sheep radius: Lanyon and Baggott,
1976; sheep tibia: Lanyon and Smith, 1970; human tibia: Lanyon et aI., 1975; dog tibia:
Bouvier and Hylander, 1984; human tibia: Burr et aI., 1996). Two factors may contribute
to the low strain magnitudes in the macaque ulna: (1) The radius, whose robusticity in ma-
caques is similar to that of the ulna, may be instrumental in resisting the external forces
and moments at the forearm. This is currently being explored in in vitro experiments in
our lab where ulnar bone strain is measured with the interosseous membrane intact and,
subsequently, severed. (2) The predominance of the hind limb in weight bearing and pro-
pulsion may reduce forelimb loads. Kinetic data demonstrate that forces and impulses at
the macaque hind limb are higher than those at the forelimb, though this difference is not
pronounced (Demes et aI., 1994). Tibial bone strain data are expected to clarify this ques-
tion.
Our results (Demes et ai., 1997, 1998) warrant further confirmation before the strain
environment of the macaque ulna during locomotion can be ultimately characterized. With
regard to the functional morphology of the bone and its articulations, the strain data indi-
cate that the radius and ulna do not act as a composite structure to resist bending but, in-
stead, the ulna undergoes a bending regime on its own. As it is "locked" into the trochlea
at the elbow joint it is likely that mediolateral bending results in mediolateral forces at this
joint, which has been suggested by Schmitt's (1994) data on the orientation of the sub-
strate reaction force. The strain data thus confirm earlier notions that the reinforcements of
the cercopithecid elbow on the medial and lateral sides as described by Fleagle et al.
(1975), Rose (1988), Harrison (1989) and Richmond et al. (1998) are indeed structural ad-
aptations to counteract frontal plane forces at this joint.

6. FORCE TRANSDUCERS AND STRAIN GAUGE TECHNOLOGY

Force transducers in locomotor studies are used to quantify the forces exchanged at
the contact between the moving body and the substrate. These are the forces exerted by
the animal and the forces experienced by the animal, the two of them being equal in mag-
nitude and opposite in direction. Force transducers most frequently used in human sports
biomechanics and orthopedics as well as in animal locomotor research are piezoelectric
force platforms (e.g., Kistler plates). They are based on the property of certain materials,
like quartz, to generate an electric charge when experiencing strain. Although the respon-
siveness, sensitivity and precision of such platforms are high their considerable price and
standard design do not recommend them for all applications. Less expensive force plat-
forms can be built using strain gauge technology (e.g., Heglund, 1981; Biewener and Full,
1992; Nigg and Herzog, 1994).
A few studies have been performed using nonstandard force transducers to quantify
substrate reaction forces. Their great advantage is the potential of mimicking natural sub-
strates used by animals in the wild. This is especially relevant for studying arboreal loco-
motion. An easy way of building horizontal, branch-like force transducers is to mount bars
onto force plates. This technique has been used by Ishida et al. (1990) and Nieschalk
(1991) to quantify substrate reaction forces of slow and slender lorises, as well as by
Schmitt (1994, this volume) to study the forces acting on the forelimbs of various monkey
species.
Force transducers using strain gauge technology allow an even greater variability in
design. In 1984, Yamazaki and Ishida used a pole-type force detector to analyze substrate
reaction forces during vertical climbing in gibbons. The same force pole was used to com-
 250 B. Demes

pare the kinetics of vertical climbing in macaques and spider monkeys (Hirasaki et ai.,
1992, 1993). Bonser and Rayner (1996) constructed a force-transducing perch to record
takeoff and landing forces of small birds. A force-transducing handle has been used by
Chang et ai. (1997) to measure superstrate forces of brachiating gibbons and chimpanzees.

7. CASE STUDY # 2: A STRAIN-GAUGED FORCE TRANSDUCER

TO MEASURE LEAPING FORCES
Our force pole for measuring takeoff and landing forces of large-bodied vertical
clingers and leapers (Demes et ai., 1995) differs in various aspects of design and instru-
mentation from previous devices. Field observations in Madagascar had revealed that
large-bodied indriid leapers frequently use compliant supports that sway visibly under the
animals' takeoff and landing forces. Breaking a dogma in force transducer design ("good"
force transducers should be rigid; i.e., have a high frequency response), we constructed a
compliant force pole. It consists of an aluminum pipe instrumented with strain gauges and
a PVC pipe that served as the takeoff and landing site (Figure 9). The pole is anchored in
an aluminum box. Whereas the PVC pipe, mostly because of its large dimensions (diame-
ter ::: 0.11 m), deforms minimally under the animal's impact, the aluminum pipe is bent
considerably and initiates swaying movements of the system. These deformations are reg-
istered by the gauges. The largest deflection is proportional to the peak force during take-
off or landing.

Tak.offl
landing area
2.1 m

1 Gauge
t
Instrumented
section
0.4m
~ .
... _-_ ............
t
Sase
0.3m 0.03m
d = 0.007 m
!
Figure 9. Sifaka leaping from force pole. The takeoff and landing (upper) part of the pole was drawn from two su-
perimposed vidoeimages: the animal is shown prior to acceleration for takeoff and at toeoff, when the pole is
maximally deflected. The dimensions of the pole and the attachment sites of the gauges as well as peak forces are
indicated in the right part of the drawing.
Use of Strain Gauges in the Study of Primate Locomotor Biomechanics 251

The limitations of the described design lie in the fact that the registered deforma-
tions are the combined result of the animal's effort and the compliancy of the pole. When
using rigid measuring devices, registered forces are a direct indicator of the forces gener-
ated by the animal, and force-time-curves allow one to identify, for example, periods of
high accelerations and low accelerations and relate them to joint positions and muscle ac-
tivities. Because of the low natural frequency of the system, particularly rapid changes in
forces will be underestimated.
Resolving for the substrate force and its components was also done in a novel way.
Three gauges (and a fourth as a safeguard) attached around the circumference of the alu-
minum pipe permit a first order approximation of the external forces. Normal strain distri-
bution at the instrumented cross section was calculated using the same algorithm as for the
above described in vivo bone strain experiment and the neutral axis of bending was deter-
mined. To resolve for the forces at the contact point the pole was modeled as a cantilever
beam. The location of the neutral axis of bending and the height of the contact on the pole
(from synchronous video) allowed to resolve for the axial forces and moments necessary
to generate a measured strain distribution (Rybicki et ai., 1977; Gross et ai., 1992). The
takeoff and landing angles of the animals at the moment of toeoff or touchdown were esti-
mated on the video, and the resultant forces calculated from the trigonometric relationship
between axial forces and angles.
Takeoff and landing forces were collected with the compliant force pole for two spe-
cies of vertical clingers and leapers: Hapalemur griseus and Propithecus verreauxi. The
pole was installed in an indoor enclosure at the Duke University Primate Center whose
concrete floor allowed for solid anchorage of its base. The recording equipment was the
same as used for in vivo bone strain data collection (see above). Only peak forces were
evaluated, and they are presented here as multiples of body weight (see Demes et ai., 1995
for a more detailed report).
Not unexpectedly, the peak takeoff and landing forces are higher for the smaller Ha-
palemur grise us (average weight for two animals: 1.0 kg) than Propithecus verreauxi (av-
erage weight for two animals: 4.0 kg; Figure 9). As smaller-bodied animals have
absolutely shorter acceleration distances and relatively more muscle force available, they
accelerate over short periods of time but with high pushoff forces. Larger-bodied animals,
on the other hand, can make use of their absolutely longer limbs, but are restricted by a
more limited force-per-unit-mass ratio; they therefore accelerate over longer periods of
time and with lower pushoff forces (Demes and GOnther, 1989).
Peak takeoff forces were higher than landing forces. On rigid (force platform) sur-
faces the landing impact is usually associated with higher peak forces, and the decelera-
tion period is shorter than the acceleration period (e.g., Preuschoft, 1985). The elastic
properties of the pole require excess force at takeoff, as part of it does not translate into
leaping distance but is "used" to deform the pole. The yield of the pole at landing, on the
other hand, absorbs some of the kinetic energy, increases the duration of the braking pe-
riod, and, consequently, reduces the magnitude of the landing force.
Finally, the timing of the takeoff relative to the movements of the pole are of interest
for the question of whether the animals take advantage of the elastic properties of the pole.
The simultaneous videorecordings clearly indicate that the animals invariably take off be-
fore the elastic recoil of the pole. This is also the case for indriid leapers in their natural
habitats when taking off from compliant trunks (Demes et ai., 1995). It therefore appears
that the vertical clingers and leapers are not able to make use of elastic substrates as power
amplifiers. As they perform work on the substrates that is not returned to them, leaping
from elastic supports must be less efficient than leaping from rigid supports.
252 B. Demes

8. CONCLUSIONS

Strain gauges in primate locomotor studies are used in two ways. The strain gauge
technique is currently the only method that permits tracking the deformations of bone in
vivo and relating them to animal activities. If attached to parts of the environment an ani-
mal is moving on, strain gauges can be also be used to determine the external forces acting
at the contacts. The examples reviewed and provided in this chapter demonstrate the po-
tential contributions of this technique to the field offunctional morphology.
In vivo bone strain measurements provide direct evidence of bone deformation and
are an invaluable tool for testing hypotheses on how bones are loaded during locomotion.
The hypothesized anteroposterior bending for the macaque ulna was not confirmed by the
strains measured in vivo on this bone, which, instead, indicated that mediolateral bending
is the predominant loading regime. Major bone and joint reinforcements at the elbow are
therefore to be expected in this plane.
Knowledge of the magnitude and direction of substrate reaction forces is useful in
identifying the forces and, if combined with kinematic information, moments generated by
and acting on an animal's musculoskeletal system. This will add significantly to our un-
derstanding of muscle distribution as well as joint and bone adaptations. In addition, it
may allow evaluation of environmental variables and their influence on locomotion. In the
example given here, it was demonstrated that large-bodied vertical clingers and leapers on
compliant supports increase their work load and generate and endure highest forces during
the takeoff, rather than the landing.

ACKNOWLEDGMENTS

The in vivo bone strain experiments are a combined effort of members of the Dept.
of Anatomical Sciences and the Musculoskeletal Research Lab of the Dept. of Orthopae-
dics at SUNY, Stony Brook. Contributions of Clinton Rubin, Jack Stern, Ted Gross and
Susan Larson are greatly acknowledged. Michael Hausman and David Reim assisted in
the surgery, and Terry Button made X-ray and CT scans possible. David Pollonet calcu-
lated the normal strain distribution in the bone cross sections and Luci Betti prepared the
artwork. Comments by Susan Larson, Brian Richmond, and three reviewers improved the
manuscript. This research was supported by NSF grants DBS-920961 and SBR-9507078.

REFERENCES
Biewener AA (1992) In vivo measurement of bone strain and tendon force. In AA Biewener (ed.): Biomechanics-
Structures and Systems. Oxford: Oxford University Press, pp. 123-147.
Biewener AA, Thomason J, Goodship A, and Lanyon LE (1983) Bone stress in the horse forelimb during locomo-
tion at different gaits: A comparison of two experimental methods. 1. Biomech. 16:565-576.
Biewener AA, and Taylor CR (1986) Bone strain: A determinant of gait and speed? J. expo BioI. 123:383--400.
Biewener AA, Thomason JJ, and Lanyon LE (1988) Mechanics oflocomotion and jumping in the horse (Equus):
In vivo stress in the tibia and metatarsus. J. Zool., Lond. 214:547-565.
Biewener AA, and Full RJ (1992) Force platform and kinematic analysis. In AA Biewener (ed.): Biomechanics-
Structures and Systems. Oxford: Oxford University Press, pp. 45-73.
Biewener AA, and Bertram JEA (1993) Skeletal strain patterns in relation to exercise training during growth. 1.
expo BioI. 185:51~9.
Bonser RHC, and Rayner JMV (1996) Measuring leg thrust forces in the common starling. 1. expo BioI.
199:435-439.
Use of Strain Gauges in the Study of Primate Locomotor Biomechanics 253

Bouvier M, and Hylander WL (1984) In vivo bone strain on the dog tibia during locomotion. Acta Anat.
118:187-192.
Burr DB, Milgrom C, Fyhrie D, Forwood M, Nysaka M, Finestone A, Hoshaw S, Saiag E, and Simkin A (1996) In
vivo measurement of human tibial bone strain during vigorous activity. Bone 18:405--410.
Chang YH, Bertram JEA, and Ruina (1997) A dynamic force and moment analysis system for brachiation. J. expo
BioI. 200:3013-3020.
Currey J (1984) The Mechanical Adaptations of Bones. Princeton: Princeton University Press.
Dally JW, and Riley WF (1991) Experimental Stress Analysis, 3rd ed. New York: McGraw-Hill.
Demes B, and Giinther MM (1989) Biomechanics and allometric scaling in primate locomotion and morphology.
Folia Primatol. 53:125--141.
Demes B, and Jungers WL (1993) Long bone cross-sectional dimensions, locomotor adaptations and body size in
prosimian primates. 1. Hum. Evol. 25:57-74
Demes B, Larson SG, Stern IT Jr., Jungers WL, Biknevicius AR, and Schmitt D (1994) The kinetics of primate
quadrupedal ism: "hindlimb drive" reconsidered. J. Hum. Evol. 26:353--374.
Demes B, Jungers WL, Gross TS, and Fleagle JG (1995) Kinetics of leaping primates: Influence of substrate ori-
entation and compliance. Am. J. Phys. Anthropol. 96:419-429.
Demes B, Stern IT, Rubin CT, Larson SG, and Hausman MR (1997) Bone strain in the macaque ulna during loco-
motion. Am. 1. Phys. Anthropol. Suppl. 24: 101.
Demes B, Stern IT Jr., Hausman MR, Larson SG, McLeod KJ, and Rubin CT (1998) Patterns of strain in the ma-
caque ulna during functional activity. Am. J. Phys. Anthropol. 106:87-100.
Dove RC, and Adams PH (1964) Experimental Stress Analysis and Motion Measurement. Columbus: C.E. Merrill
Pub. Co.
Evans FG (1953) Methods to study the biomechanical significance of bone form. Am. J. Phys. Anthropol.
11 :413-435.
Fleagle JG, Simons EL, and Conroy GC (1975) Ape limb bone from the Oligocene of Egypt. Science
189: 135--137.
Fleagle JG, Stern IT Jr., Jungers WL, Susman RL, Vangor AK, and Wells JP (1981) Climbing: A biomechanical
link with brachiation and with bipedalism. Symp. Zool. Soc. Lond. 48:359--375.
Gross TS, McLeod KJ, and Rubin CT (1992) Characterizing bone strain distributions in vivo using three triple ro-
sette strain gauges. J. Biomech. 25:1081-1087.
Gurdjian ES, and Lissner HR (1944) Mechanism of head injury as studied by the cathode ray oscilloscope. Pre-
liminary report. J. Neurosurgery 1:393-399.
Harrison T (1989) New postcranial remains of Victoriapithecus from the middle Miocene of Kenya. J. Hum. Evol.
18:3-54.
Heglund NC (1981) A simple design for a force-plate to measure ground reaction forces. J. expo BioI. 93:333-338.
Hirasaki E, Matano S, Nakano,Y, and Ishida H (1992) Vertical climbing in Ateles geojJroyi and Macacafuscata
and its comparative neurological background. In S Matano, R Tuttle, H Ishida, and M Goodman (eds.):
Topics in Primatology, Vol. 3. Tokyo: University of Tokyo Press, pp. 167-176.
Hirasaki E, Kumakura H, and Matano S (1993) Kinesiological characteristics of vertical climbing in Ateles geoj-
froyi and Macaca fuscata. Folia Primatol. 61: 148--156.
Hylander WL (1977) In vivo bone strain in the mandible of Galago crassicaudatus. Am. J. Phys. Anthropol.
46:309--326.
Hylander WL (1979) The functional significance of primate mandibular form. J. Morph. 160:223-240.
Hylander WL (1984) Stress and strain in the mandibular symphysis of primates: A test of competing hypotheses.
Am. J. Phys. Anthropol. 64:1-46.
Hylander WL, Picq PG, and Johnson KR (1991) Function of the supraorbital region of primates. Archs. Oral Bio.
36:273-281.
Hylander WL, and Johnson KR (1994) Jaw muscle function and wishboning of the mandible during mastication in
macaques and baboons. Am. 1. Phys. Anthropol. 94:523-547.
Ishida H, Jouffroy FK, and Nakano Y (1990) Comparative dynamics of pronograde and upside down horizontal
quadrupedal ism in the slow loris (Nycticebus coucang). In FK Jouffroy, MH Stack, and C Niemitz (eds.):
Gravity, Posture and Locomotion in Primates. Firenze: II Sedicesimo, pp. 209--220.
Kummer B (1970) Die Beanspruchung des Armskeletts beim Hangeln. Anthrop. Anz. 32:74--82.
Lanyon LE, and Smith RN (1970) Bone strain in the tibia during normal quadrupedal locomotion. Acta orthop.
Scandinav.41:238--248.
Lanyon LE, and Baggott DG (1976) Mechanical function as an influence on the structure and form of bone. J.
Bone Jt. Surg. 588:436-443.
Lanyon LE, Hampson WGJ, Goodship AE, and Shah JS (1975) Bone deformation recorded in vivo from strain
gauges attached to the human tibial shaft. Acta orthop. Scand. 46:256--268.
254 B. Demes

Nieschalk U (1991) Fortbewegung und Funktionsmorphologie von Loris tardigradus und anderen kleinen quadru-
peden Halbaffen in Anpassung an unterschiedliche Habitate. Ph.D. thesis, Ruhr-Universitlit Bochum.
Nigg BM, and Herzog W (1994) Biomechanics of the Musculo-skeletal System. Chichester: John Wiley and Sons.
Preuschoft H (1985) On the quality and magnitude of mechanical stresses in the locomotor system during rapid
movements. Z. Morph. Anthrop. 75:245-262.
Ramm H, and Wagner W (1967) Praktische Baustatik, Teil3, 5th ed. Stuttgart: B.G. Teubner.
Richmond BG, Fleagle JG, Kappelman J, and Swisher CC III (1998) First hominoid from the Miocene of Ethiopia
and the evolution of the catarrhine elbow. Am. J. Phys. Anthropol. 105:257-277.
Rose MD (1988) Another look at the anthropoid elbow. J. Hum. Evol. 17: 193-224
Rubin CT, and Lanyon LE (1982) Limb mechanics as a function of speed and gait. J. expo BioI. /01: 187-211.
Rubin CT, and Lanyon LE (1984) Dynamic strain similarity in vertebrates: An alternative to allometric limb bone
scaling. J. Theor. BioI. 107:321-327.
Rubin C, Gross T, Donahue H, Guilak F, and McLeod K (1994) Physical and environmental influences on bone
formation. In: CT Brighton, GE Friedlaender, and J M Lane (eds.): Bone Formation and Repair. Am. Acad.
Orthopaed. Surgeons. pp. 61-78.
Rybicki EF, Mills EJ, Turner AS, and Simonen FA (1977) In vivo and analytical studies of forces and moments in
equine long bones. J. Biomech. 10:701-795.
Schmitt D (1994) Forelimb mechanics as a function of substrate type during quadrupedalism in two anthropoid
primates. 1. Hum. Evol. 26:441--457.
Stem JT Jr., Wells JP, Vangor AK, and Fleagle JG (1977) Electromyography of some muscles of the upper limb in
Ateles and Lagothrix. Yrbk. Phys. Anthropol. 20:498--507.
Swartz SM, Bertram JEA, and Biewener AA (1989) Telemetered in vivo strain analysis of locomotor mechanics of
brachiating gibbons. Nature 342:270-272.
Timoshenko S (1958) Strength of Materials. Part II. 3rd edition. New York: D. van Norstand Co.
Yamasaki N, and Ishida H (1984) A biomechanical study of vertical climbing and bipedal walking in gibbons. J.
Hum. Evol. 13:563-571.
Yoshikawa T, Satoshi M, Santiesteban AJ, Sun TC, Hafstad E, Chen J, and Burr DB (1994) The effects of muscle
fatigue on bone strain. J. expo BioI. 188:217-233.
Young DR, Howard WH, and Orne D (1977) In-vivo bone strain telemetry in monkeys (M. nemestrina). J.
Biomech. Eng. 99:104-109.
 15

THE INFORMATION CONTENT OF

MORPHOMETRIC DATA IN PRIMATES
Function, Development, and Evolution

Charles E. Oxnard

Centre for Human Biology

Department of Anatomy and Human Biology
The University of Western Australia
WA, 6009, Australia

1. INTRODUCTION

In the 1965 Wenner Gren Symposium on Primate Locomotion organized by the late
Warren Kinzey, one of the contributions was an exposition of the morphometrics of the
primate shoulder (Oxnard, 1967). That study clearly demonstrated that the morphometrics
of the shoulder arranged the primates on the basis of the way that the shoulder was used in
the different species. Since then, morphometric studies of other individual anatomical re-
gions have provided similar results: that species are arranged in line with functional us-
ages of the particular anatomical regions concerned (Oxnard, 198311984).
However, investigations in which data sets from different anatomical regions are
combined (e.g., combinations of data from several functional units of the postcranium and
cranium, even from different components of the niche including behavior, ecology and
diet) seem to provide for each combination, a different type of information about species
groupings. That is: the groupings of the species that they produce are similar to those that
result from studies of molecular evolution. They thus mirror phylogeny rather than func-
tion. As the data are the same for both individual and combined analyses, it is necessary to
ask why such completely different results are obtained.
One answer may come from asking a theoretical question: what are we actually do-
ing when we combine the information content of data taken from individual anatomical
units into larger and larger data sets involving many units, and, in the end, a large portion
of the whole animal? A second answer may stem from asking the practical question: how,
in arranging the species in the combined studies, are the data variables clustered? We al-
ready know that the clusterings of the variables from individual units provide information

255
256 C. E. Oxnard

ofbiomechanical import that helps explain the reason for the functionally adaptive separa-
tions of the species. Is there any information in the clustering of the variables from com-
bined analyses that speaks to the reason for the evolutionary groupings of the species?

2. INFORMATION CONTENT OF LOCOMOTOR DATA:

FUNCTION VERSUS PHYLOGENY

2.1. Function
Though the morphometric studies of the shoulder in the 1965 Symposium on Pri-
mate Locomotion had seemed to imply primarily locomotor usage (Oxnard, 1967) it was
already starting to be recognised (e.g., Ripley, 1967; Oxnard 1967) that posture, feeding,
playing, escaping and many other behaviors also utilize those same movements of the
shoulder and must also be implicated in shoulder form (later reviewed more fully in
Oxnard, 198311984).
Since then, morphometric investigations of many other individual anatomical units
have been made (both by Oxnard and colleagues, and by other investigators, e.g.,
McHenry and Corrucini, 1975; Feldesman, 1976, 1979; Corrucini and Ciochon, 1978;
Manaster, 1975,-r979; and many others since, .also reviewed in Oxnard, 198311984). Such
studies now cover every individual region of the body, those of Oxnard and colleagues
specifically including the shoulder, elbow, wrist, hand, upper limb as a whole, vertebral
column, hip, knee, foot, and lower limb as a whole. In every case, the results have mir-
rored those of the original symposium contribution on the shoulder; that is, they have pro-
vided arrangements of the species that seem to relate most obviously to the functions of
the relevant anatomical part within overall animal lifestyles.
Thus, for most upper limb units, the non-human primates are arranged in a band
shaped spectrum (as shown not only in Oxnard, 1967, 1968, and 1973, but also in Feldes-
man, 1976, and Corrucini and Ciochon, 1978 as demonstrated in Oxnard, 198311984).
This spectrum (see figure 8.2 of Oxnard, 1983/84) extends from species that use the upper
limbs in highly terrestrial milieux under compression, to those that use the upper limbs in
highly arboreal niches largely under tension. Humans, not using the upper limbs at all in
locomotion, lie outside the band shaped spectrum, uniquely separate from all other pri-
mates (e.g., Oxnard 1975), separate, even, from all other mammals (Oxnard, 1968).
For various lower .limb regions, the non-human primates are arranged in a star-
shaped spectrum (see also figure 8.2 of Oxnard, 1983/84). In the center of the star lie spe-
cies from many different taxonomic groups that have in common that they are generalized
arboreal quadrupeds. In the different rays of the star are highly specialized species, usually
from different taxonomic groups, that are convergent in major aspects of their locomotor
hindlimb function. Again, humans, with their unique lower limb dominated locomotion,
bipedality, are widely separate from all the non-human primates (e.g., Oxnard, 1975,
198311984; Oxnard and Hoyland-Wilkes, 1993; Kidd, 1995; Kidd et ai., 1996; Kidd and
Oxnard, 1997).
This separateness of humans from non-human primates in the form of the various in-
dividual anatomical regions contrasts totally with the unequivocal similarity of humans
and African apes in the various combined studies that employ the same data as the indi-
vidual studies. This is a result that concurs with the various molecular findings.
The functional results from the individual anatomical regions are of especial interest
for fossil assessment because they are capable of indicating, irrespective of taxonomic as-
The Information Content of Morphometric Data in Primates 257

signment, functional adaptation in individual fossil parts. For example, in a series of stud-
ies of different anatomical regions from 1966 to 1973 (summarized in 1975) Oxnard
showed that Australopithecus africanus could not have been simply a biped in the human
manner, the usual assessment of that species in those days. He showed it must have had a
unique combination of locomotor abilities including both bipedality of a type different
from that in modern humans together with arboreal propensities of one kind or another.
This has been further evidenced in recent studies of A. africanus (e.g., Oxnard and Hoy-
land-Wilkes, 1993; Kidd et aI., 1996, 1997). Yet it has taken many analyses by many other
investigators (e.g., Prost, 1980; Tardieu, 1981; Senut, 1981; Feldesman, 1982; Jungers and
Stern, 1983; Stern and Susman, 1983a, b, Susman et aI., 1983; Schmidt, 1984; Zonnen-
veldt and Wind, 1985 and many others since that time) who all studied the more recently
discovered A. afarensis. to gradually insert this idea into the literature. Even so, it has not
been until the last few years that evidence in support of the idea of the combined terres-
trial and arboreal activities of A. africanus (put forward as long ago as Oxnard, 1975) has
been presented by some of those most strongly against the original idea. Thus, a very new
study of previously undescribed australopithecine foot remains points out that A. africanus
"while bipedal, was equipped to include arboreal, climbing activities in its locomotor rep-
ertoire ... it was likely not an obligate terrestrial biped, but rather a facultative biped and
climber" (Clarke and Tobias, 1995:524).

2.2. Phylogeny
In contrast to the situation in individual anatomical units (such as the shoulder, hand,
hip or foot) morphometric analyses of combinations of anatomical units of the postcra-
nium have not particularly resulted in functional understanding. Study of combinations of
anatomical regions (either through [a] combinations of large numbers of specific regions:
shoulder, elbow, wrist, hand, hip, knee, ankle, foot (Oxnard, 1983a), or [b] combinations
of overall proportions of the entire body: upper limb, lower limb, abdomen, thorax, and
head (Oxnard, 1983b), or even most recently [c] combinations of both overall bodily pro-
portions and individual parts of the vertebral column (Milne et aI., 1996 and work in pro-
gress)) seem all to arrange species in line with evolutionary relationships rather than
functional convergences. Let us summarize these findings.
As might be expected of a line of work carried out over more than a thirty year pe-
riod, the theoretically best combined study (i.e., an analysis of a very large number of
variables taken on all units of the body in the same specimens representing the full suite of
primate species) is not possible. Over the years, the specimens available for study never
included quite the same samples, never quite the same specimens, never quite the same
measurements, and so on. Second best sets of combined investigations have been carried
out, however, and because these have been done in three quite different ways yet with the
same results, considerable confidence can be placed in them.
A first set of studies employed Andrew's high dimensional displays for combining
the data from the individual analyses on detailed anatomical units (such as those of the
shoulder, the elbow, etc., as detailed above, Oxnard, 1983/1984). Though such analyses do
take account of correlations within individual studies, they do not take account of correla-
tions between them and therefore the "between individual studies" information is missing.
A second set of studies analyzed only a reduced number of variables representing
aggregated parts of animals (e.g., measures of overall bodily form, Oxnard, 1983/1984)
for all specimens and species. In this case, though the analyses take full account of corre-
lations both within and between the individual regions, the data suffer from the deficit
258 C. E.Oxnard

that, being only measures of overall bodily regions they do not contain the more detailed
information inherent in the measures of the smaller anatomical units.
A third set of studies combines these two types of investigation by looking not only
at detailed data from restricted specific anatomical parts (e.g., measures of the vertebrae)
but also at their correlations with the cruder but none-the-less very important measures of
overall bodily proportions (e.g., Milne et al., 1996, and work in progress). These last stud-
ies thus take into account both the information contained in the correlations within and be-
tween regions and the more detailed information inherent within and between measures of
smaller anatomical units (the various individual vertebrae).
The net result of these various combined studies is almost always groupings of spe-
cies that relate to evolutionary relationship rather than functional adaptation. Thus, each
major group of the primates: lemuriformes, lorisiformes, tarsiiformes, New World mon-
keys, Old World monkeys, and hominoids are clearly demarcated (top frame of Figure 1
showing the result for the combination of overall body proportions). In a part of that com-
bined study that includes only hominoids, humans are grouped with African apes, and this
combined group is clearly separated from Asian apes (figure 8.19 of Oxnard, 1983/4).
Such results thus concur with those of molecular evolution. Though based on the same
data as the investigations of individual anatomical units, these combined results are in to-
tal contrast with the individual studies wherein what is most clearly defined are functional
parallels and convergences, and where phylogenetic groups of the primates are almost to-
tally overlapping (e.g., a study of the shoulder: bottom frame of Figure 1).
The situation for all postcranial regions is summarized in Table 1. Whenever analy-
ses are carried out at the individual bone-joint-muscle unit or individual anatomical region
level, the information about species separations that is most clearly recognizable is about
FUNCTION. There must be evolutionary information within the data from these units, but
it is rarely obvious. When, however, analyses are carried out on the same data, but com-
bined to include as much of the whole postcranium as possible, thus combining many indi-
vidual functional units, then the most obvious information about species separations
reflects EVOLUTION. There definitely is functional information within these combined
data-we know that this must be so because they contain the data of the individual analy-
ses-but the functional information is hidden when the data are combined.

3. INFORMATION CONTENT OF NON-LOCOMOTOR DATA:

FUNCTION VERSUS PHYLOGENY

3.1. Craniodental Data

A similar picture arises from our consideration of studies of the cranium (including
the teeth) in a range of primate adults. Thus, from a series of investigations of measure-
ments of the dentition and skull (Oxnard et al., 1985; Oxnard, 1987; Hayes et al., 1990,
1995, 1996; Hayes, 1994; Pan, 1998; Pan et al., in press) univariate, bivariate and multi-
variate examinations of individual cranial units (e.g., mandibular teeth alone, mandible
alone) generally reveal groupings of species that are associated with groupings of func-
tional variables. These variables include features such as grinding areas and cutting
lengths of teeth and which are related, as might be expected, to primary dietary compo-
nents or masticatory function such as degree of omnivory or herbivory. This has also been
shown by a host of studies by other investigators over the years.
The Information Content of Morphometric Data in Primates 259

'--_ _.....1 1 I
STREPSIRHI ES NEW WORLD OLD WORLD HOMI OIDS
MO KEYS MO KEYS

PROSIMlANS

OLD WORLD MONKEYS

NEW WORLD MONKEYS

HOMINOIDS

Figure 1. Top, an example of the separations of the species in combined studies (in this case the combination of
overall body proportions as displayed in a minimum spanning tree of generalized distances - total length of tree
more than 40 standard deviation units). The separations seem most to reflect phylogeny. The major groups ofpri-
mates are indicated. Even the tarsier (+), long grouped as a prosimian and apparently similar in functional abilities
to bush-babies (x), is well separated from what are now known as strepsirhines and is clustered closer to anthro-
poids (as it is by molecular data). Bottom, an example of the separations of species in individual studies (in this
case an analysis of shoulder dimensions as displayed in the first canonical axis, units in standard deviations). The
separations are related mostly to the function of the shoulder. Cartoons display arm-hanging propensities ranging
from none (left) to extreme (right). The various major taxonomic groups overlap totally. Both studies employ ca-
nonical variates analyses and generalized distances.
260 c. E.Oxnard

Table 1. Summary of information content of morphometric investigations of postcranium

Anatomical units Functional relationship Evolutionary relationship
Obvious Not obvious Obvious Not obvious
Individual Shoulder x x
Arm and forearm x x
Hand x x
Complete upper limb x x
Hip x x
Knee x x
Foot x x
Complete lower limb x x
Combined Shoulder, arm, forearm, hip, thigh, leg, foot x x
Upper limb, lower limb, thorax, abdomen, head x x

The primary reason for our craniodental studies, however, was to examine data from
the individual functional units in combined morphometric analyses to see whether the spe-
cies were grouped in ways associated with phylogenetic relationships and sexual dimor-
phism. These combined results show that, indeed, the species are arranged in ways that
align with their evolutionary relationships. They show furthermore, that sexual dimor-
phism is not a single phenomenon, essentially similar in all species except for degree and
mainly related to overall size. Sexual dimorphism is, rather, different in each species, dif-
ferent in the various anatomical regions, differently aligned in relation to overall size, and
thus far more complex than generally thought. The broad implication is that there are
many different sexual dimorphisms and that these must have evolved separately since the
origin of the various common ancestors of the groups concerned.

3.2. Data Pertaining to the Niche

Even studies of the primate niche concur with this idea that different kinds of infor-
mation are inherent in the data, and that it takes different types of analyses to disentangle
them. Thus, investigation of individual species defined through data pertaining to the
niche gives information about ecological groupings of the species (the niche equivalent of
function). Those same data, when applied to species combined into their higher taxonomic
groups (e.g., into families) provide information pertaining to evolutionary relationships
(Oxnard et aI., 1990).
The aspects of the niche that were studied in these investigations include quantita-
tive measures of locomotion, environment and diet. The multivariate statistical methods
that are used are identical to those used in the parallel morphometric investigations of
these same species (but of course, in these analyses, are termed nichemetrics, Crompton et
aI., 1987; Oxnard et aI., 1990).
The results of the study of individual species show that the various niche variables
(relating to locomotor activities, to the environment within which the activities are carried
out, and to the dietary items that are garnered within the environment and obtained by the
activities) arrange the species into groups that make locomotor, environmental and dietary
sense. For example, several different groups of species are defined in relation to associa-
tions between (a) various locomotor activities (such as richochetal leaping, scurrying, and
slow climbing), (b) various environmental loci (such as small branch undergrowth, main
branch highways, and canopy), and (c) various dietary items (such as leaves, fruits, and
The Information Content of Morphometric Data in Primates 261

animal products). Though not functional in the sense of functional morphology, these
groups are extremely "functional" in the sense of ecological adaptation.
The results of the nichemetric studies are, moreover, closely concordant with the re-
sults of individual morphometric studies on the anatomies of the same series of species
(compare figures 6.1 through 6.6 in Oxnard et ai., 1990). This is what might have been ex-
pected; after all, it is these morphologies that exist in these niches.
The aforementioned lines of argument, however, have forced us to look further into
the nichemetric studies to see if additional information might lie hidden in the associa-
tions. First, studies were undertaken to see if the clusterings of the variables in the study
of individual species also made ecological or "niche" sense in a way similar to that in
which clusterings of morphological variables made functional or biomechanical sense in
the individual morphometric studies. This is easily seen to be so. The original publication
(figure 5.12 of Oxnard et ai., 1990) indicates that the variable clusters are not random or
undecipherable. Thus, the locomotor variables and environmental variables are placed in
each half of a circular part of the analytical space. In contrast, lying on the periphery of
the circle, are each of the dietary variables. Each of these is, individually, about as far dis-
tant from each of the others as it could possibly be.
These variable relationships have now been examined more closely to see if they
make ecological sense. Within the overall picture described above, smaller variable neigh-
borhoods exist. Figures 2 and 3 show two examples. In each case a localized neighbor-
hood of the variables makes sense in relation to the ecological adaptations outlined by the
clustering of the species. Thus one neighborhood of variables (labeled at the upper part of
the plot, Figure 2) includes fruit eating, living in the canopy, using many horizontal sup-
ports, and climbing. This association of variables relates to the species grouping of most
bushbabies but not the two extreme leaping bushbabies that are associated more closely
with the extreme leaping tarsiers. Another neighborhood of variables (Figure 3) includes
leaf-eating, living on large supports, and extreme leaping (labeled on the left hand side of
the plot, Figure 3). This association of variables relates to the species grouping of the vari-
ous extreme leaping indriids together with Lepilemur and Hapalemur. Other variable
neighborhoods make similar adaptive sense. It is evident, then, that the information con-
tent of these data not only identify niche groups of species (as in Oxnard et ai., 1990) but
also niche groups of variables.
This leads to the question: is there any way of examining these niche metric data that
might give information relating to evolution? Thus, though one of the niche groups con-
tains only lorisines (all, therefore, in the same evolutionary group) several of the other
niche groups contain species that are not in the same evolutionary group: for example, the
grouping of the extreme leaping tarsiers along with two of the more extreme leaping bush-
babies, and the grouping of indriids together with Lepilemur and Hapalemur as a second
form of extreme leaping. This implies that the groups are due to adaptive convergences
(niche groups) and not to evolutionary relationship.
The way in which the data cluster species into families was therefore examined. The
result, given the information content about niche groups in the data, achieves some very
interesting separations (Figure 4). Thus, in spite of the extreme differences in a niche
sense between lorisines (mostly slow climbing insectivorous creatures) and galagines
(mostly fast running and leaping frugivorous forms), this analysis achieves a closely co-
herent grouping of both lorisines and galagines (Lorisidae) that is widely separated from
all the other species. Similarly, in spite of the marked adaptive similarities between all tar-
siers and some of the bushbabies, this analysis achieves complete and utter separation of
all tarsiers (Tarsiidae) from all bushbabies that are unequivocally placed together with
262 C. E.Oxnard

PRIN3 FRUIT

2.79

-0.23

-3.24
\
\
\
-6.26 \
\
5.55
\
v'"
'"
","'\
\
\
\
\
\
\ ",'" '"
",'" \
PRIN2 '" \
\

'" '"
",'"

\
\

PRIN 1

Figure 2. Principal components analysis of the groups of variables (circles, locomotor variables; squares, environ-
mental variables; diamonds, dietary variables) responsible for the niche group separations of the species in the
nichemetric studies of prosimians. A local neighborhood of specific variables is identified.

lorisines in the Lorisidae. Yet other separations of major evolutionary interest are readily
evident. One of special interest is the separation of the single species Daubentonia as the
family Daubentoniidae, from all other lemurs, especially indriids with which it has often
been associated in an evolutionary sense (e.g., Tattersall and Schwarz, 1975). This melds
very well with the previously determined unique morphometric separation of Daubentonia
from all other strepsirhines (Oxnard, 1983/1984). In other words, though adaptive infor-
mation is clearly present in these data, so too is information of evolutionary import. Ask-
ing the right question of the data is necessary to allow it to appear.
It might have been sufficient to have assumed that the reason for the various func-
tional (locomotor, masticatory, ecological) arrangements of species in the individual stud-
ies is because the evolutionary information content within individual elements is swamped
by the functionally adaptive parts of that element's form or make-up. Likewise, it might
have been simply assessed that the reason for evolutionary information being apparently
inherent in combinations of m!J,ny elements might be because, whereas functionally adap-
tive information would not summate because function is different in each region, evolu-
tionary information, being the same in each element, of course, would summate and
swamp the different separate functional convergences (Oxnard, 1991, 1992). Only in a
The Information Content of Morphometric Data in Primates 263

PRIN 3

2.79

-0.23

-3.24

LEAPING -

\1 \
\
",'"
'"
",'" \
PRIN2
",'"
'" \
\
",'" \
\

PRIN 1

Figure 3. The same analysis as Figure 2 with a second local neighborhood of specific variables identified.

few situations where functional convergence is extreme (e.g., the similarities between
bushbabies and tarsiers being dependent upon highly convergent leaping) might this not
always occur (e.g., in the study of overall body proportions) and even here, the functional
convergence can, in fact, be separated out (Oxnard, 1978). In an attempt to further expli-
cate these findings, however, we can ask two further questions; first: how, in terms of
theoretical thinking, might the information content of variables be arranged in combined
as compared with individual studies; second, how is the information content of the vari-
ables in the combined investigations actually arranged?

4. THEORETICAL INFORMATION CONTENT OF THE DATA

SETS

The question above, relating to theoretical studies of information content, can be

discussed rather simply in the following manner. Let us assume that any individual mor-
phometric analysis of an individual functional unit contains information that is partly
about functional adaptation and partly about evolutionary relatedness. This means we can
think of the analysis as revealing mainly functional information (f), say 5 parts, but never-
264 C. E. Oxnard

4.81 CHEIROGALEIDAE

-0.28

-5.38

,,
,
........
,,
....\,...
....
.... ....
LORISIDA
12.40
2 ,
,,
,DAUBENTONIA

, 1

Figure 4. The arrangements of the higher taxonomic groups (families) in a canonical variates analysis (standard
deviation units in each axis) of the nichemetric data.

theless also containing (if not so obviously) a lesser degree of evolutionary information
(e), say 3 parts. Of course, these phenomena are really continuous; I am describing them
as discrete bits to simplify matters.
The information in such an analysis might then be written as

= f1 + f2 + f3 + f4 + f5 + e 1 + e2 + e3
= a total of 8 units of information

where the "f's are overt function and the "e"s are covert evolution. This implies that "f'
(biological function) appears as 5/8 of the total information and "e" (evolution) only 3/8
(this lesser amount may be partly why it is less easily recognised). I have used the bold
format to designate the more easily recognizable portions.
Yet the above is too simplistic. It is likely that some of the evolutionary information
will also be "similar to" (i.e., correlated with) some of the functional information. After
all, functional adaptation is a-major part of evolutionary diversity. This is the equivalent of
saying that there will inevitably be at least some interactions between function and evolu-
tion. Let us assume that 2 of the "f's and 2 of the "e"s interact. Then the information in
the analysis might be rewritten as:
The Information Content of Morphometric Data in Primates 265

= fI + f2 + f3 + f4(=e) + fS(=e) + el(=t) + e2(=t) + e3

= again, a total of 8 units of information.

In this case, however, the easily identifiable .biological function, f, appears to be a

greater proportion of the total information, 7/8, exactly because it is readily identifiable.
In contrast, the evolutionary information (e), although appearing in 5 bits out of 8 in our
example, has only I bit (e3) in which it is clearly different from f. The interactive portions
(f4[=e], f5[=e], el[=f] and e2[=f]) will all be seen as fbecause of the interaction. Thus, in
an individual functional unit where there are interactions between function and evolution,
it is easy to see why the smaller part---evolution--may be obscured, and why the larger
portion--function--may appear extremely large indeed.
If, now, we had a series of such analyses taken on different functional units (a, b, c,
d, and e) the following exposition shows how, though each individual unit might greatly
emphasize function, a combined analysis of all units together might sum to something dif-
ferent.
Thus, each individual unit analysis might give an equation like that above, so that
the entire suite of analyses might look like the following:

Unit a = fIa + f2a + f3a + f4a(=ea) + fSa(=ea) + ela(=fa) + e2a(=fa) + e3a

Unit b = fIb + f2b + f3b + f4b(=eb) + fSb(=eb) + elb(=fb) + e2b(=fb) + e3b
Unit c = fIc + f2c + f3c + f4c(=ec) + fSc(=ec) + elc(=fc) + e2c(=fc) + e3c
Unit d = fId + f2d + f3d + f4d(=ed) + fSd(=ed) + eld(=fd) + e2d(=fd) + e3d
and
Unit e = fIe + f2e + f3e + f4e(=ee) + fSe(=ee) + ele(=fe) + e2e(=fe) + e3e.

For each of these units, f appears to be 7/8 of the information even though f and e are ac-
tually split 5/8 and 3/8 respectively.
When, however, we add the data for each individual unit into a single combined-analy-
sis of all units, the totals could look very different. First, the various "f"s cannot be expected
to sum beyond a single individual analysis because the functions in each unit are different.
Second, in contrast, the various "e"s (including those "e"s that are related to "f's) in the indi-
vidual analyses can be expected to sum in the combined analysis because the information
about evolution should be the same for each unit (they are all parts of the same animal). Ac-
cordingly then, the total "e"s are 25 (5 for each equation and "e" can now be written bold) but
no single "f' is any greater than 3 (e.g., 3 "fa"s, 3"fb"s, 3"fc"s and so on). Thus "e" (evolu-
tion) now shines out strongly at 25/40; no single "f' (function) shines out more strongly than
3/40, even though different functions total 25/40 (Le., there is just as much functional infor-
mation present in the combined analysis as in the total of individual analyses).
This line of thinking could, alone, be the reason why function is clearly evident in
individual analyses, but evolution in combined analyses. There are also practical reasons,
however, why the change from function to evolution might occur as individual analyses
are combined. These latter reasons are not mutually exclusive with the above theoretical
line of argument.

5. PRACTICAL INFORMATION CONTENT OF THE DATA SETS

Practical studies of information content depend not only on how, but on why the
variables in the aggregated studies are clustered. We already know that the variables in the
individual studies, in producing the functional arrangements of the species, do so through
266 c. E.Oxnard

variable clusters that make functional sense and that can be shown to be related to
biomechanics. What information (if any) is provided by the way the variables are clus-
tered in the combined studies so that they produce the evolutionary relationships of the
species? Are these variables, in producing the evolutionary arrangements of the species,
clustered in any way that makes evolutionary sense and, if so, are the clusters related to
underlying mechanisms in evolution?
This is a much more pertinent question now than it was in 1965 because much more
is now known about the underlying genetic and developmental mechanisms that are re-
sponsible for adult form. This is certainly the case for the postcranium: i.e., for the limbs
and trunk, experimental grafting studies in bird embryos (e.g., Wolpert et ai., 1975;
Wolpert and Hombruch, 1992; Richardson et ai., 1990), murine homeobox gene studies in
limbs (e.g., Dolle et ai., 1989) and homeobox genes for trunk axis in frogs (e.g., Ruizi and
Melton, 1990). It is also true for the cranium and jaws depending upon studies of neural
crest cell populations giving rise, for instance, to lower jaw morphogenetic units (e.g.,
Hanken and Hall 1993; Atchley, 1993; Jacobson, 1993). It could be true for studies of the
niche, though as far as I am aware no underlying developmental investigations (these
would be very difficult) have so far been carried out in this last arena.
A confluence of function, genetics, development and evolution is certainly the un-
derlying biological determinant of animal structural and functional diversity. Developmen-
tal biologists, in recent years, have been able to look from the gene, through a cascade of
developmental processes and a series of timing events, towards the understanding of the
structure of the adult in exemplar (experimental) species (such as fruit flies, chicks and
quail, rats and mice). It should equally be the case that, in the opposite vein, evolutionary
biologists might be able to look from comparative studies of adult diversity, through the
clusterings of variables, to the underlying genetic and developmental processes that pro-
duced them. In other words, the genetic-developmental studies and quantitative whole-or-
ganism studies should be able, each in their own right, to provide predictive hypotheses
for testing by the other. In this way, Primate Locomotion: Recent Advances, may have a
rather unexpected descendant from the Primate Locomotion Symposium in 1965.

5.1. Clustering of Variables in Combined Studies

5.1.1. Postcranium. The analyses of combined variables for the body overall, which
clearly give evolutionary arrangements of the species (e.g., figures 8.39 and 8.40, Oxnard,
1983/84 and top frame of Figure 1), provide completely different clusters of variables to
those mechanically obvious clusters of variables evident in the individual analyses of
body regions (e.g., figures 7.1 and 7.2, Oxnard 1983/84 and bottom frame of Figure 1).
Yet the new clusters of variables in the combined studies are not randomly arranged or un-
interpretable. One cluster comprises relative lengths of all major segments of limbs and a
second cluster includes relative lengths of all elements of all rays of the hands and feet.
The first can also be described as a cluster of all measures of proximodistal elements of
the limbs, the second as a cluster of all measures of craniocaudal elements of the only
skeletal parts in this study (the cheiridia) that clearly display craniocaudal arrays. Further
investigation of this combined study indicates a special variable cluster comprising all
measures of the fourth manual digit (but not the fourth pedal digit), a feature related to the
difference between strepsirhines and anthropoids (Table 2).
The first two clusters of variables mentioned above bear remarkable similarity to
two of the developmental processes responsible for limb form. Thus, Wolpert et ai. (1975)
clearly demonstrated: first, a proximodistal gradient in limb segment formation and sec-
The Information Content of Morphometric Data in Primates 267

Table 2. Relationships between clusters of limb variables in diverse species of primates and
developmental processes in experimental non-primates
Variable
clusters Anatomical feature Morphological descriptor Developmental process
I Lengths of major segments of limbs Measures of proximodistal elements Proximodistal gradient
2 Lengths of all elements of hand and Craniocaudal measures of cheiridia Craniocaudal gradient
foot
3 (not studied) (Prediction 1)1 Dorsoventral gradient
4 Lengths of all parts of manual digit 4 Strepsirhine feature cf. anthropoids (Prediction 2)2
Ilf morphometric studies had included dorsoventral variables, would they have formed a cluster equivalent to the dorsoventral
gradient in development?
21f developmental studies had included strepsirhines, would there have been a special developmental process relevant to manual
digit four?

ond, a craniocaudal differentiation in digit fonnation (Figure 5 and 6). Though the experi-
ments were carried out on the bird wing, it is not in doubt that these principles apply to all
tetrapod limbs. Further developmental studies of gradients in limb formation (e.g.,
Richardson et aI., 1990) and studies of homeobox genes responsible for the cascade of
processes involved in limb formation (Dolle et aI., 1989) suggest the existence of a third,
dorsoventral, spirally arranged developmental mechanism (Figure 6).
Two questions thus arise. First, if the overall studies of primate morphological diver-
sity had included measures of a dorsoventral spiral arrangement, would they have fonned
a single cluster and if so, would this have been because of the underlying developmental
mechanism? Unfortunately we were not percipient enough to think ofthat at the time. We
will have to go back to obtain those data. Second, if the developmental biologists were
able to carry out their studies on primates, would they find some developmental factor re-
lated specially to the fourth manual digit. This, too, has not been done and would be very
difficult to do in the same manner as the prior experimental studies on birds. But, investi-

19 24

Figure S. A summary of the grafting experiments that demonstrate the exist-

ence ofa proximodistal developmental process in the bird wing (e.g., Wolpert
and colleagues, 1975). Top, the elements of the adult bird wing labeled with
two of the embryological stages responsible for them (stage 19 for the
humerus, radius and ulna; stage 24 for the cheiridia). Middle, the effect, on fi-
nal wing development, of grafting a stage 19 wing tip onto a limb bud with
stage 24 removed. Stage 19 is thus represented twice in a proximodistal se-
quence and as a result the humerus, radius and ulna are formed twice over in
a proximodistal relationship. Finger development is unaffected. Bottom, the
effect, on final wing development, of grafting a stage 24 wing tip onto a bud
with stage 19 tip removed. Stage 19 and 24 of the recipient are thus not repre-
sented and only the fingers are formed in the subsequent adult from the stage
24 of the donor.
268 C. E. Oxnard

II
"",'
,
~ ..............,
~---III
~ IV

'-
IV
, III
~-:..-""
~=:::-
==--_......
II
~~~-...
III
Figure 6. A summary of the grafting experiments and
IV
genetic determinations that demonstrate the existence
of craniocaudal and dorsoventral developmental proc-
esses in the bird wing (Richardson et aI., 1990; Dolle et
Hex 4.4
aI., 1989). Top, normal limb development in the chick
embryo with caudal portion of limb bud responsible for
Hex 4.6 initiating the craniocaudal arrangements of the 3 digits
Wing bud (II-IV) in the adult. Middle, grafting a portion of the
caudal limb bud into a cranial position duplicates the
Hex 4 .7 digits in the adult in reverse order thus implying a cra-
niocaudal process. Bottom, indicates the dorsoventrally
rotated structure of limb materials controlled by the
various homeobox genes (HOX 4.4-4.7) of the limb
buds.

gations of data about the differential development of the fourth digits of hands and feet in
strepsirhine and anthropoid fetuses might allow equivalent information to be obtained.
The existence of the concordances between the developmental experiments (proxi-
modistal gradient and craniocaudal differentiation) and the whole-organism observations
(proximodistal measures as separate from craniocaudal measures) is unlikely to be coinci-
dence. It is highly likely that developmental factors determined from experiments on em-
bryos in two or three experimental species would, if truly generalizable, have effects in
the adult that would also be revealed in comparisons of the ultimate morphology of adults
of large numbers of related species. It is also fascinating that each type of study (process
in embryonic development--dorsoventral organization, and quantitative diversity of adult
form-special arrangements of manual digit four) should be able to provide predictions
for the other (Table 2).

5. 1.2. Cranium Including Teeth. A similar question has been asked of the combined
cranial and odontometric data in the various investigations. That is: in producing sexual
and species separations that seem to be related to evolution rather than function (Oxnard,
1987) how are the variables clustered? Let us look further into that analysis.
The results seem to indicate that the variables are not clustered randomly or in non-
interpretable ways. In fact, two of the variable clusters comprise the combined length di-
mensions of (a) both mandibular incisors and (b) all mandibular molars (Oxnard, 1987).
Though these are variable clusters of lengths of teeth, they reflect the lengths of the rele-
vant portions of the mandibular alveolar bone: (a) its incisor part, (b) its molar part. In
other studies involving measures of the bones themselves (Pan, 1998) two equivalent clus-
ters of variables have been identified: (a) one of measures at the anterior end of the jaw
(the incisor part), the other of measures posteriorly (the molar part).
The Information Content of Morphometric Data in Primates 269

These two pairs of variable clusters are remarkably concordant with the findings of
developmental and molecular biologists in studies of the morphogenesis of the mandible
in experimental animals (e.g., Atchley, 1993). Those investigations imply that the rat jaw
arises from a series of osteogenic cell populations derived from the first neural crest seg-
ment (Figure 7 and top frame of Figure 8). In addition to incisor alveolar and molar alveo-
lar cell populations, these studies also identify populations of cells for the ramus,
Meckel's cartilage, and the coronoid and angular [muscular] and condyloid [joint] ele-
ments.
Even further back in development a set of homeobox genes controls the cascade of
processes from these cell populations, through the developmental units, to the eventual
adult morphological components (Figure 7). Though this work has been done primarily in
the rat, it is highly likely that a similar set of processes and stages is involved in all mam-
malian mandibles. Is it possible that the first two morphometric clusters of variables (Ta-
ble 3) are discernable exactly because they are adult measures reflecting the fundamental
underlying developmental alveolar components?
There are two further clusters of variables in the primate studies: (a) one of these is
the lengths of all canines and (b) a second is all dimensions of the premolars (Table 3). Of
course, rats do not have canines and premolars. Developmental components related to
these areas could, therefore, not be recognised through experimental studies in the rat. If,

First segment neural crest ---------i.~ Chondrogenic cell group

Meckel's cartilage

Osteogenic cell group

Mandibular ramus

Osteogenic cell group Osteogenic cell group

Incisor alveolus Molar alveolus

Homeobox genes Morphometric clusters

Developmental units

Cell populations

Figure 7. Top, the various cell groups arising from the first segment of the neural crest and giving rise eventually
to individual components of the mandible. Bottom, the cascade of processes controlled by homeobox genes re-
sponsible for each of the anatomical units.
270 C. E.Oxnard

Molar Alveolar Bone

Condylar
process

Incisor Alveolar Bone Ramus Angular Process

Canine Alveolar Bone

Pre-molar Alveolar Bone

Incisor Alveolar Bone Ramus Angu lar Process

Figure 8. Top, the components of the mandible of the rat that are determined through various genetic mechanisms
and developmental processes (after Atchley, 1993). Bottom, the presumed components that might be identified if
similar experiments were carried out on a primate or mammal with all tooth types present.

however, the developmental studies had been carried out on a species with a full comple-
ment of types of teeth, would these additional mandibular elements (cell populations,
mandibular units, adult mandibular components) have been identified? This might be rep-
resented as in the bottom frame of Figure 8, which suggests what might be the molecular

Table 3. Relationships between clusters of dental variables in morphometric studies of

primate species and of developmental processes in experimental animals (rats)

Variable
clusters Anatomical feature Morphological descriptor Developmental process
I Lengths of incisors Incisor portion of jaw Incisor alveolar bone
2 Lengths of canine Canine portion of jaw (Prediction I)'
3 Lengths of premolars Premolar portion of jaw (Prediction 2)2
4 Lengths of molars Molar portion of jaw Molar alveolar bone
'If developmental studies could be done on mammalian species with no missing teeth, would individual cell
populations, developmental units, and mandibular components still exist, or would these populations, units
and components be perceived as a spectrum?
2If developmental studies covered an experimental primate, would separate and additional premolar cell popu-
lations, developmental units, and mandibular components exist?
The Information Content of Morphometric Data in Primates 271

and development units in, for example, a primate or other mammal with a larger comple-
ment of tooth types. This prediction from studies of whole-organism morphology is thus
offered to experimental developmental biologists.

5.1.3. Niche. Although there is no developmental information yet available for the
niche, it seemed worth looking at the nichemetric studies to see how the variables were
clustered in that part of the study that seemed to provide information about the relatedness
of species. This is the analysis (Figure 4) that separates the major groups of prosimians in
a way that seems related to evolution: e.g., the clustering together of lorisines and galagi-
nes (as lorisids), the separation of tarsiers in their own group (tarsi ids), and the extreme
separation of Daubentonia from all other species (into its own family).
This particular analysis of the niche data provides the arrangements of the variables
shown in Table 4. The major information that these variable clusters seem to contain re-
lates to six evolutionary groupings and separations of taxa as described in the third col-
umn of the table.
Though these separations involve, of course, individual variables that speak to func-
tional (ecological) adaptation, the particular ways in which the variables are combined
speak to phylogenetic separations. Thus, these results confirm long established phyloge-
netic relationships, such as the links between bushbabies and the various lorises; they pro-
vide information relating to other relationships that have been judged equivocal in the past
although generally settled now, e.g., the question of the separation of tarsiids from strep-
sirhines. If this is true, then other parts of the data may be evidence that can be used to test
hypotheses that are not yet settled; e.g., though some workers believe that aye-ayes are
merely very close relatives of indriids (e.g., Tattersall and Schwarz, 1975), these results
for the Daubentoniidae are further data supporting a view that see the aye-ayes as being so
different from indriids (e.g., Oxnard 1983/84) that it is most likely that they have been
long separate from them.
In this case, links to developmental phenomena are not evident as they were in the
morphometric studies. That may be, however, because, at least at the present time, we do
not know anything about possible "developmental processes" for the behavioral-environ-
mental-dietary complex. That these clusters of variables differ from the functional "eco-
logical" clusters is not, however, in doubt. That they are related to phylogeny of species
groups rather than ecology of individual species seems clear.

Table 4. Relationships between clusters of variables and evolutionary groups of species

Canonical axis Variable clusters Variables involved Evolutionary relationships
Axis 3 loadings in 3 variables falling, crouching, richochetal 1. separates tarsiids from all
leaps, animal diet other groups.
Axis 2 high positive loadings in 3 falling leaps, scurrying, fruit 2. unifies cheirogaleids
variables eating
high negative loadings in undergrowth, vertical 3. separates cheirogaleids
3 other variables supports, animal diet from lorisids
Axis I high positive loadings in 4 slow quadrupedalism, large 4. separates lemurids from all
variables supports, fruit, leaves other groups; 5. clusters
galagines and lorisines as
lorisids
high negative loadings in scurrying, undergrowth, small 6. separates galagines from
4 other variables supports, animal diet tarsiids
272 C. E. Oxnard

Table 5. Summary of information content of all studies

Region Information content of Separates Clusters
Postcranium Individual units Species, in relation to function Variables, such as those related
(locomotion) to biomechanics
All regions combined Species, in relation to evolution Variables, such as genetic and
developmental processes
Cranium and teeth Individual units Species, in relation to function Variables, such as those with
(masticatory) masticatory significance
All regions combined Sexes and species, in relation to Variables, such as genetic and
evolution developmental processes
Niche Individual units Species, in relation to the niche Variables, reflecting ecology
All units combined Families, in relation to Variables, by removing
evolution convergences

6. SUMMARY
As a result of these investigations, therefore, a summary plan can now be drawn up
about the information content of the various analyses (Table 5).
In terms of regional form and function, Table 5 documents that separations of spe-
cies relate mostly to the functional milieux within which individual anatomical regions
operate. Additional proof that this is so is that the clustering of variables seems to relate to
the anatomical adaptations pertinent to those functional milieux.
In contrast, in terms of the whole organism (the same sets of variables but com-
bined) Table 5 demonstrates that the separations of species are most closely linked to what
is known about evolution (and provides information useful for evolutionary controversies
such as the positions of tarsiers and aye-ayes). Additional proof that this is so is that the
clusterings of the variables are most closely allied with the genetic and developmental
processes that underpin whole organism structure and its diversity.
These investigations also speak especially to the situation of humans among the pri-
mates. In terms of form in individual regions, humans are generally quite different from
other primates (e.g., in the upper and lower limbs, figures 7.1 and 7.2 in Oxnard,
1983/84). This seems to be because of regional anatomical adaptations to the totally new
functional and behavioral milieux that humans have come to inhabit.
But in terms of overall form, the functional uniqueness of humans comes to be ap-
propriately buried in close overall similarities with the African apes (e.g., figure 8.19 of
Oxnard, 1983/84). This picture reflects those common molecular, genetic and develop-
mental phenomena that apply as much to humans as they do to African apes. Perhaps this
is as close as we can get, with whole-organism investigations, in resolving the paradox of
the human position. The studies require us to accept and meld two different views of hu-
mans at one and the same time.

ACKNOWLEDGMENTS
I am grateful to many colleagues and students who, over the years, have participated
in these investigations. In addition, I am indebted to my graduate students for permission
to describe the findings in their doctoral dissertations. These individuals are all cited in the
text.
The Information Content of Morphometric Data in Primates 273

The studies could not have been carried out without the help of a number of muse-
ums on three continents and the different curators of the collections that were used (Pow-
ell Cotton Museum, Birchington, UK, British Museum of Natural History, London, UK,
Field Museum, Chicago, USA, Los Angeles County Museum, Los Angeles, USA, Western
Australian Museum, Perth, Australia).
I am especially indebted to Professor F. P. Lisowski and Dr. Len Freedman for com-
ments upon this and related scripts, and for broader discussions of primate morphology
and evolution. I also thank three anonymous reviewers and Elizabeth Strasser for their
helpful comments on the manuscript.
Most of all I wish to document that these ideas have arisen through a series of dis-
cussions with various morphological, genetic and developmental scientists: especially Le-
wis Wolpert, Brian Hall and Paul O'Higgins. The ideas first arose in the preparation of
The Order of Man (198311984). They were especially stimulated by an Australian Acad-
emy of Science Discussion Meeting held in Sydney, 1992, in commemoration of the late
Professor N. W. G. Macintosh. They were further explored in preparation of the Key Note
Lecture for the Primate Locomotion, 1995, Symposium in commemoration of the late Pro-
fessor Warren Kinzey. They are currently the topic of discussions and collaborations with
colleagues in the Department of Anatomy and Developmental Biology, University Col-
lege, London.
The overall research program is supported by several grants from the Australian Re-
search Council and by the Centre for Human Biology, UWA. The final stages were been
greatly aided by my appointment in 1995 as Australian Academy of Science Visiting
Scholar to the Academia Sinica, Kunming Institute of Zoology, PRC, and in 1996 as Visit-
ing Professor in the Department of Anatomy, University of Hong Kong. Professor Y. C.
Wong, Head of Anatomy in Hong Kong, is especially thanked for his support during that
extended period.

REFERENCES
Atchley WR (1993) Genetic and developmental aspects of variability in the mammalian mandible. In J Hanken
and 8J Hall (eds.): The Skull, Volume I: Development. Chicago: The University of Chicago Press, pp
207-247.
Clark RJ, and Tobias PV (1995) Sterkfontein Member 2 foot bones of the oldest South African hominid. Science
269:521-524.
Corrucini RS, and Ciochon RL (\ 978) Morphometric affinities of the human shoulder. Am. J. Phys. Anthropol.
45:19-38.
Crompton RH, Lieberman SS, and Oxnard CE (1987) Morphometrics and nichemetrics in prosimian locomotion.
An approach to measuring locomotion, habitat and diet. Am. 1. Phys. Anthropol. 73: 149-177.
Dolle P, Izpisua-Belmonte J-C, Falkenstein H, Rennucci A, and Doubou1e D (1989) Coordinate expression of the
murine Hox-5 complex homeobox-containing genes during limb pattern formation. Nature 342:767-769.
Feldesman M (1976) The primate forelimb: A morphometric study of locomotor diversity. University of Oregon
Anthropological Papers 10:1-154.
Feldesman M (1979) Further morphometric studies of the ulna from the Omo Basin, Ethiopia. Am. J. Phys. An-
thropol. 51 :409-416.
Feldesman M (1982) Morphometric analysis of the distal humerus of some Cenozoic catarrhines: The late diver-
gence hypothesis revisited. Am. J. Phys. Anthropol. 59:73-76.
Hanken J, and Hall BJ, eds. (1993) The Skull, Vol. I, Development. Chicago: University of Chicago Press.
Hayes V (1994) Sexual dimorphism in the dentition of African Old World monkeys. Ph.D. Dissertation, University
of Western Australia.
Hayes V, Freedman L, and Oxnard CE (\ 990) Taxonomy of savannah baboons: An odontomorphometric approach.
Am. J. Primatol. 22: 171-190.
274 C. E.Oxnard

Hayes V, Freedman L, and Oxnard CE (1995) The differential expression of dental sexual dimorphism in subspe-
cies of Colo bus guereza.lnt. 1. Primatol. 17:971-996.
Hayes V, Freedman L, and Oxnard CE (1996) Dental sexual dimorphism and morphology in African colobus mon-
keys.lnt. J. Primatol. 17:725-757.
Jacobson AG (1993) Somitomeres: Mesodermal segments in the head and trunk. In J Hanken and BJ Hall (eds.):
The Skull, Volume I: Development. Chicago: The University of Chicago Press, pp 42-76.
Jungers WL, and Stem JT Jr. (1983) Telemetered electromyography of forelimb muscle chains in gibbons (Hyfo-
bates far). Science 206:617--619.
Kidd RS (1995) An investigation into the patterns of morphological separation in the proximal tarsus of selected
human groups, apes and fossils: A morphometric analysis. Ph.D. Dissertation, University of Western Aus-
tralia.
Kidd RS, O'Higgins P, and Oxnard CE (1996) The OH8 foot: A reappraisal of the functional morphology of the
hindfoot using a multivariate analysis. J. Hum. Evol. 31:269-291.
Kidd RS, O'Higgins P, and Oxnard CE (1997) Patterns of morphological discrimination in the human talus: A con-
sideration of the case for negative function. Perspect. Human BioI. 3:57--69.
Lieberman SS, Gelvin BR, and Oxnard CE (1985) Dental sexual dimorphisms in some extant hominoids and
rarnapithecines from China: A quantitative approach. Am. J. Primatol. 9:305-326.
McHenry HM, and Corrucini RL (1975) Multivariate analysis of early hominoid pelvic bones. Am. J. Phys. An-
thropol. 46:263-270.
Manaster BJM (1975) Locomotor adaptations within the Cercopithecus, Cercocebus and Presbytis genera: A mul-
tivariate approach. Ph.D. Dissertation, The University of Chicago.
Manaster BIM (1979) Locomotor adaptations within the Cercopithecus genus: A multivariate approach. Am. 1.
Phys. Anthropol. 50: 169-182.
Milne N, O'Higgins P, and Oxnard CE (1996) Metameric variation in the vertebral column of hominoids. Am. 1.
Phys. Anthropol., Suppl. 22: 170.
Oxnard CE (1967) The functional morphology of the primate shoulder as revealed by comparative anatomical,
osteometric and discriminant function techniques. Am. 1. Phys. Anthropol. 26:219-240.
Oxnard CE (1968) The architecture of the shoulder in some mammals. 1. Morph. 126:249-290.
Oxnard CE (1973) Form and Pattern in Human Evolution: Some mathematical, physical and engineering ap-
proaches. Chicago: The University of Chicago Press.
Oxnard CE (1975) Uniqueness and Diversity in Human Evolution: Morphometric studies of australopithecines.
Chicago: The University of Chicago Press.
Oxnard CE (1978) The problem of convergence and the place of Tarsius in primate phylogeny. In D1 Chivers and
KA Joysey (eds.): Recent Advances in Primatology, Vol. 3, Evolution. London: Academic Press, pp
239-247.
Oxnard CE (I 983a) Anatomical, biomolecular and morphometric views of the living primates. In RJ Harrison and
V Navaratnam (eds.): Progress in Anatomy. Cambridge: Cambridge University Press, pp. 113-142.
Oxnard CE (l983b) Sexual dimorphisms in the overall proportions of primates. Am. 1. Primatol. 4:1-22.
Oxnard CE (1983/84) The Order of Man: A biomathematical anatomy of the primates. Hong Kong: Hong Kong
University Press (1983), New Haven: Yale University Press (1984).
Oxnard CE (1987) Fossils, Teeth and Sex: New perspectives on human evolution. Hong Kong: Hong Kong Uni-
versity Press, Seattle: Washington University Press.
Oxnard CE (1991) Anatomies and lifestyles, morphometrics and niche metrics: Tools for studying primate evolu-
tion. 1. Hum. Evol. 6:97-115.
Oxnard CE (1992) Developmental processes and evolutionary diversity: Some factors underlying form in pri-
mates. Archaeol. in Oceania, 27:95-104.
Oxnard CE, Crompton RH, and Lieberman SS (I 990) Animal Lifestyles and Anatomies: The case of the prosimian
primates. Seattle: Washington University Press.
Oxnard CE, and Hoyland-Wilkes C (1993) Hominid bipedalism: The pelvic evidence. Persp. Hum. BioI. No.
4:13-34.
Oxnard CE, Lieberman SS, and Gelvin B (1985) Sexual dimorphism in dental dimensions of higher primates. Am.
1. Primatol. 8:127-152.
Pan R-L (1998) A morphometric approach to the skull of macaques: Implications for Macaca arctoides and M. thi-
betana. Ph.D. Dissertation, University of Western Australia, Perth.
Pan R-L, Jablonski NG, and Oxnard CE (in press) Morphometric analysis of Macaca arctoides and M. thibetana
in relation to other macaque species. Primates.
Prost J (1980) The origin of bipedalism. Am. 1. Phys. Anthropol. 52: 175-190.
Richardson MK, Hornbruch A, and Wolpert L (1990) Mechanisms of pigment pattern formation in the quail em-
bryo. Development 109:81-89.
The Information Content of Morphometric Data in Primates 275

Ripley S (1967) The leaping of langurs: A problem in the study of locomotor adaptation. Am. J. Phys. Anthropol.
26:149-170.
Ruizi lA, and Melton D (1990) Axial patterning and the establishment of polarity in the frog embryo. Trends
Genet. 6:57-67.
Schmidt P (1984) Eine Rekonstruktion des Skellettes von AL 288-1 und deren Konsequenzen. Folia Primatol.
40:283--306.
Senut B (1981) Humeral outlines in some hominoid primates and in Plio-Pleistocene hominids. Am. J. Phys. An-
thropol. 56:275-284.
Stern JT Jr., and Sussman RL (I 983a) The locomotor anatomy of Australopithecus afarensis. Am. J. Phys. Anthro-
pol. 60:279-318.
Stern JT Jr., and Sussman RL (1983b) Functions of peroneus longus and brevis during locomotion in apes and hu-
mans. Am. 1. Phys. Anthropol. 60:256-257.
Susman RL, Stern JT Jr., and Rose MD (1983) Morphology of KNM-ER 3228 and OH 28 innominates from East
Africa. Am. J. Phys. Anthropol. 60:259-260.
Tardieu C (1981) Morpho-functional analysis of the articular surface of the knee joint in primates. In AB Chiarelli
and RS Corrucini (eds.): Primate Evolutionary Biology. Berlin: Springer Verlag, pp. 68-80.
Tattersall I, and Schwartz JH (1975) Relationships among the Malagasy lemurs. The craniodental evidence. In WP
Luckett and FS Szalay (eds.): Phylogeny of the Primates: A Multidisciplinary Approach. New York: Ple-
num Press, pp. 299-312.
Wolpert L, and Hornbruch A (1992) Double anterior chick limb buds and models for cartilage rudiment specifica-
tion. Development 109:961-966.
Wolpert L, Lewis J, and Summerbell D (1975) Morphogenesis of the vertebrate limb. Ciba Foundation Symp.
29:95-129.
Zonnenfeldt FW, and Wind J (1985) A new method for high resolution computed tomography of hominid fossils.
Proc. Taung Diamond Jubilee Symposium, Johannesburg, pp. 427-436.
 16

HETEROCHRONIC APPROACHES TO THE

STUDY OF LOCOMOTION

Laurie R. Godfrey, I Stephen J. King, I and Michael R. Sutherland2

IDepartment of Anthropology
Machmer Hall, Box 34805
2Statistical Consulting Center
Lederle Tower, Box 34535
University of Massachusetts
Amherst, Massachusetts 01003

1. INTRODUCTION

Despite the advances that have been made in documenting ontogenetic changes in
primate skeletal morphology, relatively little is known about the ontogeny of locomotor
behavior. Even less is known about differences among behavioral developmental trajecto-
ries of closely related species. The study of evolutionary changes in developmental trajec-
tories is the purview of heterochrony. Heterochrony is generally defined as the study of
perturbations and displacements in existing ontogenetic pathways caused by changes in
developmental timing and rates. As such, it has been conceived by some researchers (e.g.,
Gould, 1977; Raff et aI., 1990; Zelditch and Fink, 1996; Rice, 1997) as narrowly encom-
passing only a subset of possible evolutionary shifts in ontogeny, and by others (e.g.,
McKinney and McNamara, 1991) as all-inclusive. The goal of heterochrony is to under-
stand the proximate causes of differences among adult phenotypes of closely related spe-
cies. This entails understanding how, through development, morphology and behavior
emerge, and how developmental trajectories themselves evolve. Applied to the study oflo-
comotion, it means understanding the evolutionary basis for variation, if any, in the onto-
geny of positional behavior, and its relation to the ontogeny of form.
There is today a standard vocabulary of heterochronic "processes," derived largely
from the work of Gould (1977) with modifications by others (e.g., Alberch et aI., 1979;
Shea, 1981, 1983, 1986; McKinney, 1986, 1988; McKinney and McNamara, 1991). Gould
(1977) sought to explain how simple developmental perturbations might lead to ''juvenili-
zation" ("paedomorphosis") or "overdevelopment" (Gould's "recapitulation," later called
"peramorphosis") of descendant adults. Descendant adults are juvenilized (or "paedomor-

277
 278 L. R. Godfrey et al.

phic") if they closely resemble ancestral juveniles. They are overdeveloped (or "peramor-
phic") if they compare favorably to what might be expected of an ancestor that developed
beyond its normal endpoint-an ancestral "hyperadult."
Strongly influenced by Huxley's (1932) Problems of Relative Growth, Gould also
linked the study of heterochrony with the study of allometry, or size-correlated shape
change. To Gould, a fundamental task of heterochrony was the diagnosis of the preserva-
tion or disruption, in descendants, of ancestral ontogenetic linkages between size (metric
dimensions) and shape (dimensionless proportions). Gould called ontogenetic change in
size "growth," and ontogenetic change in shape "development." He built a three-parame-
ter clock model showing possible evolutionary outcomes of alterations in 1) rates of
growth (or change in size), 2) rates of development (or change in proportions), and/or 3)
age at reproductive maturation (or some other maturational stage). Preservation, in the de-
scendant, of ancestral size/shape linkages would result in ancestor/descendant allometric
commonality. Changes in ontogenetic size/shape linkages ("dissociation" of size and
shape) could also be diagnosed using the standard tools of bivariate allometry (see Gould,
1977: 239, Fig. 30; see Figure 1). Ancestral growth allometries could be used, following
Huxley (1932), as a criterion of subtraction from which shifts in descendant ontogenetic
pathways could be assessed (Gould, 1966, 1975a,b, 1977).
Over the past two decades, primatologists have increasingly used heterochrony to
understand locomotor form and function. Much of the literature on heterochrony and pri-
mate locomotion takes Gould's ontogenetic criterion of subtraction, and his treatment of
the relationship between heterochrony and allometry, as its point of departure (Shea, 1981,
1983, 1986, 1988, 1992; Gomez, 1992; Inouye, 1992; Ravosa, 1992; Ravosa et aI., 1993;
Taylor, 1995, 1997). The strategy is to compare species' differences in growth allometries.

a b
\..\{\~o
! __
\50(!\8",
- ••••••
-
.'" - - •••• ~ Pr
- ....
........
........ '

.... ....
.~
-~
r- r-
3
Cl
.3 - Descendant
- - Ancestor

(ylx) Pd = (ylx) DESADULT (ylx) Pr =(ylx) DES ADULT

Log Size Log Size

Figure 1. Gould's test of dissociation of size and shape: Ascertain where the line of descendant adult isometry
(dashed) intersects the ancestor's growth allometry. (a) Under paedomorphosis via neoteny, the line of descendant
adult isometry intersects the juvenile portion of the ancestral growth allometry (Pd) and the descendant adult is a
paedomorph. (b) Under peramorphosis via acceleration, the line of descendant adult isometry intersects the hy-
peradult or extrapolated (dotted) portion of the ancestor's growth allometry (Pr) and the descendant adult is a pera-
morpho Note that the results of this test depend not solely on whether the descendant's growth allometry is steeper
or shallower than that of the ancestor, but also on whether the ancestral and descendant growth allometries are
shallower (as in case a, where both are negative) or steeper (as in case b, where both are positive) than the line of
descendant adult isometry.
Heterochronic Approaches to the Study of Locomotion 279

This is usually accomplished by examining the linear approximations of the ontogenetic

relationships between pairs of variables plotted on logarithmic scales. Growth allometries
are identified as "ontogenetically scaled" or "allometric ally dissociated" based on the sta-
tistical significance of differences between their allometric coefficients (or slopes) and in-
tercepts. Inferences are sometimes drawn regarding heterochronic processes (e.g., rate or
time hypo- or hypermorphosis, pre- or postdisplacement, neoteny, acceleration, etc.) and
the functional significance of the evolutionary perturbations. Ontogenetic scaling is some-
times taken to signal evolutionary conservatism (in that shape changes may be "size-re-
quired" and not the specific objects of selective pressure) whereas allometric dissociation
is taken to signal possible evolutionary innovation (Inouye, 1992; Shea, 1992). In addi-
tion, behavioral inferences are sometimes drawn, based on a postulated correspondence
between ontogenetic behavioral and morphological trajectories, or between paedomorpho-
sis (or peramorphosis) in form and juvenile (or hyperadult) behaviors (Gould, 1977;
Doran, 1992). There is a strong conviction that one can learn a great deal about locomotor
form and function by employing an ontogenetic criterion of subtraction (see Ravosa et aI.,
1993), and that, in general, " ... studies attempting to relate form to function by utilizing
the comparative approach among closely-related adults of differing body size would be on
much firmer ground if ontogenetic allometric data were incorporated in order to examine
scaling patterns and utilize the appropriate criterion of subtraction for such comparisons"
(Shea, 1992: 299).
This chapter focuses on heterochronic analyses of primate locomotion. It addresses
the following questions: Do growth allometries reveal heterochronic process? To what ex-
tent does ontogenetic scaling imply functional conservatism, and allometric dissociation
imply the converse-i.e., functional novelty? Does allometric commonality imply com-
monality of behavioral ontogenies? Does paedomorphosis or peramorphosis ofform imply
juvenilization or adultification of positional behavior? Finally, how might heterochrony
contribute to our understanding of the evolution of positional behavior in primates?

2. DO GROWTH ALLOMETRIES REVEAL HETEROCHRONIC

PROCESS?
What is the relationship between heterochrony and allometry? Neither Gould (1977)
nor Alberch et al. (1979) treated this subject thoroughly, and recent treatments have been
inconsistent and sometimes contradictory (see Shea, 1983, 1985, 1988, 1989; McKinney
and McNamara, 1991; Godfrey and Sutherland, 1995b, 1996). It is actually a simple mat-
ter to model the allometric expectations of evolutionary perturbations in any of Gould's
three parameters of heterochronic change, as long as 1) the descendant follows exactly its
ancestor's path of shape change (Gould's "development"), and 2) the traits whose allomet-
ric relationships are being examined are simple power functions of one another. These
conditions are rarely met. Because, even under these ideal conditions, the allometric ex-
pectations of heterochronic perturbations have been and continue to be misunderstood,
they warrant brief clarification here (see Table 1).

2.1. Rate Hypo- and Rate Hypermorphosis

As noted by Shea (1983,1988), as long as relationships between Y and X are simple

power functions and are thus linear on logarithmic scales, several heterochronic processes
 ....
~

Table 1. Allometry and heterochrony under restricted conditions I

What happens to growth allometries in the descendant, which in the
ancestor are:
Ancestral Descendant size Descendant age
Ontogenetic perturbation Positive Negative Isometric size/shape linkages? at maturation at maturation Product
Rate hypomorphosis No change No change No change Preserved Smaller Same Paedomorph
Rate hypermorphosis No change No change No change Preserved Bigger Same Peramorph
Time hypomorphosis No change No change No change Preserved Smaller Younger Paedomorph
Time hypermorphosis No change No change No change Preserved Bigger Older Peramorph
Pre-displacement No change No change No change Preserved Bigger Same Peramorph
Post-displacement No change No change No change Preserved Smaller Same Paedomorph
Neoteny Slopes decrease Slopes increase No change Disrupted Same Same Paedomorph
Acceleration Slopes increase Slopes decrease No change Disrupted Same Same Peramorph
Proportioned giantism Slopes decrease Slopes increase No change Disrupted Bigger Same Neither
Proportioned dwarfism Slopes increase Slopes decrease No change Disrupted Smaller Same Neither
I All growth allometries are linear on logarithmic scales. All ancestral shape paths are conserved.

r
?'
C1
8-
=l'
~
~
,...
'"
Heterochronic Approaches to the Study of Locomotion 281

a b
Rate Hypermorphosis Time Hypermorphosis
Age 6
Age 5
,.._0 Age 5 /0
.......-
>- >-
Cl Cl
o o
-' -'
Age 1

Log X Log X

Figure 2. Two examples of hypennorphosis. (a) Under rate hypennorphosis, the ancestor and descendant follow
common growth allometries at different rates. (b) Under time hypennorphosis, the descendant follows ancestral
growth allometries at the same rate, but for a longer period of time. After Ravosa et al. (1993).

predict coincidence of ancestral and descendant linear regression parameters, or "ontoge-

netic scaling."
These include "rate hypo-" and "rate hypermorphosis," first defined by Shea in
1983. Under rate hypo- or hypermorphosis, ancestors and descendants follow common
growth allometries, but do so at different rates (Figure 2a). In other words, ancestors and
descendants exhibit similar shapes at similar sizes but different ages. The descendant
grows and develops more slowly (rate hypomorphosis) or more quickly (rate hypermor-
phosis) than its ancestor. Ancestral size/shape linkages are preserved.

2.2. Progenesis and Hypermorphosis (= Time Hypo- and Time

Hypermorphosis)
Gould's "progenesis" and "hypermorphosis" (renamed by Shea, 1983, "time hypo-"
and "time hypermorphosis," respectively), also yield coincident ancestral and descendant
allometric slopes and intercepts on logarithmic scales (as long as the relationships be-
tween Y and X are simple power functions; see Figure 2b). Ancestors and descendants fol-
low common allometries (i.e., they exhibit similar shapes [values ofY/X] at similar sizes
[values of X]), and ancestral rates of trait growth remain unchanged, but the duration of
growth decreases (time hypomorphosis) or increases (time hypermorphosis).

2.3. Predisplacement and Postdisplacement

Alberch et al. (1979) defined predisplacement and postdisplacement as positive vs.
negative perturbations in the timing of the onset of the development (or change in shape)
of some body part. Represented on development (i.e., shape, or YIX) vs. age scales, pre-
displacement and postdisplacement yield descendant trajectories that parallel those of the
ancestor, as depicted by Alberch et al. (1979:309, Fig. 20).
Plotted on logarithmic scales (log Y vs. log X), predisplaced or postdisplaced de-
scendant trajectories thus defined will coincide with those of the ancestor. Predisplace-
ment and postdisplacement of the development of some body part entails no change in the
relationships among Y and X components of that part--only a change in the relationship
of those components to other body parts. The allometric prediction of such a change,
282 L. R. Godfrey et al.

whenever ancestral and descendant relationships between log Y and log X are compared,
is the coincidence of allometric slopes and intercepts, or "ontogenetic scaling." Ancestors
and descendants will exhibit similar shapes (Y/X) at similar sizes (values of X), but at dif-
ferent ages. The age difference results from a change in developmental onset age and not
from a change in developmental rate:

2.4. Neoteny and Acceleration

Neoteny and acceleration are two of many ontogenetic perturbations that result in
the "allometric dissociation," or non-coincidence, of at least some allometric slopes and
intercepts. Gould (1977) defined "neoteny" as paedomorphosis acquired via a retardation
in the rate of development (or shape change) with no concomitant change in overall size
or age at maturation. Thus defined, neoteny requires a weakening, from ancestor to de-
scendant, of growth allometries (Shea, 1989; Godfrey and Sutherland, 1996). This means
that all allometries will converge toward isometry (the slopes of ancestral positive al-
lometries will decrease, thereby becoming more weakly positive; the slopes of ancestral
negative allometries will increase, thereby becoming more weakly negative; and ancestral
isometries will not change). Neoteny contrasts with "acceleration," or peramorphosis ac-
quired via an increase in the rate of development (or shape change) with no concomitant
change in overall size or age at maturation. Acceleration requires a strengthening of
growth allometries (the slopes of ancestral positive allometries increase, thereby becoming
more strongly positive; the slopes of ancestral negative allometries decrease, thereby be-
coming more strongly negative; and the slopes of ancestral isometries remain unpertur-
bed). These allometric predictions have been poorly understood (e.g., McKinney and
McNamara, 1991; Klingenberg and Spence, 1993; Vrba et aI., 1994; Rice 1997; see God-
frey and Sutherland, 1995b, 1996 for critiques), despite their having been explicitly de-
scribed by Shea (1989, see pp. 73 and 96)t and others (Godfrey and Sutherland, 1996, see
pp.33-34).
Whenever ancestral slopes are at or near isometry, little or no change in allometric
slopes and intercepts is the expectation of heterochronic processes such as neoteny and ac-
celeration. For example, under neoteny, a further weakening (as expected) of already weak
ancestral growth allometries may not effect a statistically significant change in slope. An-
cestors and descendants may well appear "ontogenetically scaled" for the relationships un-
der consideration. Thus, whenever either the ancestor or descendant shows isometry,

• Many authors claim that predisplacement and postdisplacement predict simple upward or downward transposi-
tions of descendant growth trajectories on log Y vs. log X scales (e.g., McKinney and McNamara, 1991). This is
not merely incorrect for predisplacement and postdisplacement as originally defined by Alberch et al. (1979), but
aiso for other commonly accepted meanings of these terms (e.g., as perturbations in the onset of the growth of
one trait, Y, relative to another, X). The allometric prediction of a perturbation in the onset of the growth of one
trait relative to another is neither "ontogenetic scaling" nor simple upward or downward transposition of ances-
tral growth allometries. Godfrey and Sutherland (I995a) show that a shift in the onset timing of the growth ofY
(relative to X) will generally result in convergence of ancestral and descendant growth trajectories on logarith-
mic scales, because the trajectories cannot share the same origin on original measurement scales. In contrast,
simple transpositions of growth allometries on logarithmic scales describe changes in the scalar rate component
of simple power relationships between Y and X.
t Shea (1989:91) defined neoteny as "allometric dissociation and retardation of shape change." Under neoteny,
"trajectories of positive allometry require slope decreases and/or downward transpositions to yield paedomor-
phosis, while trajectories of negative allometry require slope increases and/or upward transpositions" (Shea,
1989:96)
Heterochronic Approaches to the Study of Locomotion 283

"ontogenetic scaling" is non-diagnostic of the maintenance of ancestral size/shape link-

ages.
Other ontogenetic perturbations predict changes in growth allometries similar to
those expected under neoteny and acceleration. In fact, any case of retardation of develop-
ment (change in Y/X) relative to growth (change in X) yields weakened growth al-
lometries, whereas any case of acceleration of development relative to growth yields
strengthened growth allometries. Thus, for example, proportioned giantism involves a re-
tardation of shape change relative to growth, and it resembles neoteny in its allometric ex-
pectations, whereas proportioned dwarfism involves an acceleration of shape change
relative to growth, and it resembles acceleration (see Godfrey and Sutherland, 1996).

2.5. Complex Allometry, Neomorphosis, and Mosaic Perturbations

Disjunction of ancestral and descendant growth allometries can also result from a
variety of ontogenetic perturbations when Y and X are not related to one another as simple
power functions and are thus not linear on logarithmic scales (see Shea, 1986; Shea et aI.,
1990; Sutherland and Godfrey, in prep.; Table 2). When ancestral allometries are "com-
plex" (i.e., non-linear on logarithmic scales), heterochronic processes such as time or rate
hypo- or hypermorphosis, as well as predisplacement or postdisplacement as originally de-
fined by Alberch et ai. (1979), can yield apparent "allometric dissociation" (Figure 3a; see
Shea et aI., 1990). The descendant follows the same trait-trait trajectory as its ancestor, but
to a different terminus. Ancestral size/shape linkages are preserved, but disjunct slopes or

Table 2. Changes in descendant growth allometries under specified conditions

What happens to descendant What happens to

growth allometries (vis-a-vis ancestral size/shape
Perturbation those of the ancestor) linkages Product
Truncation of ancestral ? Preserved Paedomorph
complex allometries'
Extrapolation of ancestral ? Preserved Peramorph
complex allometries'
Deviation from ancestral ? Disrupted Neomorph
shape path
Dissociation of growth When traits within affected Local linkages Affected growth fields
fields, with local fields are plotted against size preserved; global will be paedomorphic;
retardation of size and surrogates outside of the linkages disrupted descendant as a whole
shape affected field, descendant will be neither
growth allometries will show paedomorphic nor
a decrease in slope and/or peramorphic.
downward transposition via
a decrease in the V-intercept.
Dissociation of growth When traits within affected Local linkages Affected growth fields
fields, with local fields are plotted against size preserved; global will be peramorphic;
acceleration of size and surrogates outside of the linkages disrupted descendant as a whole
shape affected field, descendant will be neither
growth allometries will show paedomorphic nor
an increase in slope and/or peramorphic.
upward transposition via an
increase in the V-intercept.
'We assume that the data are treated as linear on double-logarithmic scales, whether or not they show some curvature that may
indicate the existence of complex (nonlinear) allometry.
 284 L. R. Godfrey et aL

>-C) >-C)
.3 o
....J

Log X Log X

Figure 3. Two examples of extrapolation of ancestral complex allometry. (a) The descendant (dashed curve) fol-
lows its ancestor's (solid curve) complex allometric trajectory to a new terminus. (b) The descendant (dashed
curve) extends each phase of its ancestor's allometric pattern (solid curve) sequentially. The parallel dashed lines
represent hypothetical lines of isometry. At A, the ancestor and descendant are the same size and shape. They are
not the same size and shape midway through the steepest portion of their allometric trajectories (8 for the ances-
tor, and C for the descendant). The shape pathways have diverged. Modified from Shea et al. (1990).

intercepts can occur because neither trajectory is correctly described by a linear regres-
sion. In cases of nonlinear allometry, standard tests of the statistical significance of differ-
ences between ancestral and descendant linear regression parameters on logarithmic scales
will not necessarily diagnose allometric commonality as "ontogenetic scaling." Because of
this problem, Shea et al. (1990) suggest visual examination of the coincidence of ancestral
and descendant allometric patterns.
An interesting question raised by Shea et al. (1990) is what happens when changes
in complex allometric slopes are age- rather than size-dependent (Figure 3b). Suppose, for
example, a descendant extends each phase of its ancestor's allometric pattern sequentially,
as shown in Figure 3b--i.e., maintaining the overall "shape" of the ancestor's ontogenetic
trajectory on logarithmically transformed trait-trait scales. This sort of evolutionary
change has been called "sequential hypermorphosis" (McNamara, 1983; McKinney and
McNamara, 1991; Rice, 1997). The resulting dissociation of ancestral and descendant al-
lometric slopes and intercepts is not an artifact of fitting linear regressions to nonlinear
data. Ancestral linkages between size and shape are actually disrupted. There is, in addi-
tion, a change, from ancestor to descendant, in the sequence of proportions. The descen-
dant is no longer on its ancestor's shape path (Figure 4). This is one of many ways in
which neomorphosis (or entirely novel shapes) can be achieved (Table 2). Unless the de-
scendant follows its ancestor's shape pathway, concepts such as paedomorphosis and
peramorphosis have little meaning.
Upward or downward transpositions of growth allometries (or changes in their Y-in-
tercepts) are likely to result from mosaic evolution, or changes in the rates of growth and
development of parts, when perturbed traits are plotted against unaffected traits (see Ta-
ble 2). This is true whether or not ancestral size/shape linkages are retained within af-
fected parts. Other perturbations can produce upward or downward transposition of
growth allometries. For example, the nonparallel growth allometries produced by shifts in
the onset of the development of some trait plotted against some unaffected trait may nev-
ertheless appear to be parallel if the shift in onset timing occurs well before the age span
covered by the recorded data (see Godfrey and Sutherland, 1995a).
Heterochronic Approaches to the Study of Locomotion 285

0,0,1

°
Figure 4. A shape space can be constructed by converting trait measurements to proportions (such that they al-
ways sum to I), and then plotting these proportions (from to I) on orthogonal axes. Shown here is a 3-dimen-
sional shape space representing traits X (from the origin, 0,0,0 to 1,0,0), Y (from the origin to 0, I ,0), and Z (from
the origin to 0,0, I). In 3-dimensional shape space. all shape paths are constrained to lie on a single plane. Shape
paths of three species, labeled "a" (ancestor), "b" and "c" (two hypothetical descendants), are shown. Arrows indi-
cate directions of ontogenetic change. Descendant b follows its ancestor's shape path, whereas descendant c does
not. Note that the shape paths do not reveal the sizes of species at any shape or their rates of development (age at
any shape). Bivariate comparisons of growth allometries of species a and b, as well as species a and c. may reveal
a combination of allometrically dissociated and ontogenetically scaled relationships. Only species b, however, is a
true paedomorph for the growth field represented by these three traits.

It is clear from the above discussion that inferences regarding heterochronic process
are difficult to draw from an examination of changes in slopes and intercepts of growth al-
lometries. The coincidence of allometric slopes and intercepts may imply: l) rate hypo- or
hypermorphosis (i.e., a negative or positive perturbation in the rates of growth and devel-
opment); 2) time hypo- or hypermorphosis (i.e., a negative or positive perturbation in the
age of maturation, or cessation of growth and development); 3) predisplacement or post-
displacement of onset timing (i.e., a negative or positive perturbation in the timing of the
initiation, or onset, of growth and development of some part); or 4) under ancestral or de-
scendant growth isometry, some process (such as neoteny) that results in the disruption of
ancestral size/shape linkages.
"Allometric dissociation" is similarly a catch-all category comprising many kinds of
heterochronic perturbations. The non-coincidence of allometric slopes and/or intercepts
may imply: l) neoteny, acceleration or any other simple disruption of ancestral linkages
between size and shape, but with preservation of ancestral shape pathways; 2) inappropri-
ate use of linear regression to describe truncated or extrapolated ancestral complex (or
nonlinear) allometries (preservation of ancestral size/shape linkages not withstanding); 3)
neomorphosis via a deviation of the descendant from the ancestor's shape path; 4) a mo-
saic change in the rate of growth and development of a part but not the whole, or in the
onset timing of a part, but not the whole; or 5) some combination of the above.
Given the above considerations, we can specify conditions under which coincidence
of allometric slopes and intercepts must be considered trivial (at least in the sense that it is
non-diagnostic of the preservation of ancestral size/shape linkages). "Ontogenetic scaling"
must be considered trivial whenever either of the two species being compared shows an
286 L. R. Godfrey et al.

Table 3. Assessment of preservation of ancestral size/shape linkages, with and without consideration
of relationships that are non-diagnostic, or "trivial"
Relationships Relationships that
reported to be are non-trivially
ontogenetically ontogenetically
Source Database Taxa Compared scaled Conclusion drawn scaled
Inouye, 1992 Manual rays Pan and Gorilla 150f22 Pervasive 30f22
ontogenetic
scaling
Shea, 1992 Limbs and trunk Cercopithecus 90f9 Pervasive 20f9
talapoin and C. ontogenetic
cephus scaling
Ravosa et aI., Limbs and trunk Propithecus 17 of 19 Pervasive 3 of 19
1993 diadema ontogenetic
edwardsi and P. scaling
tattersalli
Taylor, 1995 Scapula Male and female 60f6 Pervasive lof6
Gorilla gorilla ontogenetic
beringei scaling

isometric relationship between Y and X. This may well affect assessments of the preva-
lence of size/shape conservation in the primate postcranial skeleton (see Table 3). Simi-
larly, allometric dissociation must be considered trivial (or non-diagnostic of the
disruption of ancestral size/shape linkages) whenever the bivariate relationship being ex-
amined is not a simple power function. If ancestral and descendant growth allometries are
both significantly positive or negative, and if both are linear on logarithmic scales, then
coincidence of allometric slopes and intercepts is not trivial. The strengths and shapes of
allometric trajectories must be considered.
The upshot is that cataloguing differences between ancestral and descendant al-
lometric slopes and intercepts does not reveal heterochronic process in a straightforward
manner. "Ontogenetic scaling" (defined as the coincidence of ancestral and descendant
slopes and intercepts of linear ontogenetic regressions plotted on logarithmic scales) can
be produced by many different heterochronic processes, including some that disrupt an-
cestral size/shape linkages. "Allometric dissociation" (defined as the non-coincidence of
ancestral and descendant slopes or intercepts of linear regressions plotted on logarithmic
scales) is, similarly, an expectation of many different heterochronic processes, including
some that preserve ancestral size/shape linkages. Multiple heterochronic processes yield
the same allometric clues. If heterochronic process is to be inferred, and if growth al-
lometries are to be used as diagnostic tools, then the full set of allometric implications of
different heterochronic processes must be understood. These distinctions must be em-
braced if the functional importance of neomorphosis, as opposed to upwardly or down-
wardly scaled paedomorphosis or peramorphosis, is to be addressed.

3. FUNCTIONAL CONSERVATISM AND NOVELTY

Despite the nebulous connection between "ontogenetic scaling," "allometric disso-

ciation," and heterochronic processes, the distinction between ontogenetic scaling and al-
lometric dissociation may be theoretically important and useful for heterochronic analysis.
Heterochronic Approaches to the Study of Locomotion 287

Huxley (1932) studied ontogenetic series of termites, ants, and other organisms; he found
that many organs grow according to a "simple heterogony formula" (that is, Y increases as
a power function of size, or X). Thus, for example, differences in the shape of soldiers and
workers are achieved via feeding discrimination; they are consequential effects of hetero-
gonic growth to different sizes. Huxley suggested that shape differences produced by ex-
trapolation or truncation of the same simple power functions should be adaptively neutral
and of no taxonomic importance, whereas shape differences produced by different growth
allometries should be functionally and taxonomically important. This notion had a strong
influence on the development of Gould's (1966, 1975a,b, 1977) thinking about ontogeny
and phylogeny, as is reflected in Pilbeam and Gould's (1974: 892) reference to ontoge-
netic scaling as the generation of "the same animals at different sizes."
This concept has also had a profound influence on the literature on heterochrony and
primate locomotion. Shea (1988: 242) states, "The primates present many cases where on-
togenetic allometries are dissociated, presumably in response to selection for specific
novel proportions (as opposed to simple size change or growth duration)." Inouye (1992)
uses the distinction between ontogenetic scaling and allometric dissociation as a vehicle to
identify those shape differences in the manual rays of chimpanzees and gorillas that do not
require adaptive explanation vs. those that do. Traits (such as metacarpal head depth) that
are ontogenetically scaled vis-a-vis several size surrogates have "equal expression" in
chimpanzees and gorillas at common sizes (Inouye, 1992: 127), and their differences
among adult chimpanzees and gorillas need no special functional interpretation. In con-
trast, locomotor functional explanations (increased terrestriality) are sought for shape dif-
ferences (such as the relative shortening of gorilla metacarpals and phalanges) that result
from allometric dissociation. Ravosa et al. (1993) use the distinction between ontogenetic
scaling and allometric dissociation in a similar manner: Proportional differences that arise
through differential extension of common growth allometries in diademed and Verreaux'
sifakas are interpreted as by-products of selection for differences in size (in different eco-
logical regimes), whereas traits that show allometric dissociation are treated as requiring
independent functional explanation (possibly related to differences in locomotion). Fi-
nally, on the basis of "predominant" ontogenetic scaling of the scapulae of male and fe-
male mountain gorillas, Taylor (1995: 442) concludes that sexual differences in gorilla
locomotor behavior do not depend on "unique morphological adaptations to different eco-
logical niches."
If systemic size perturbations produce descendants that are ontogenetically scaled,
then ontogenetic scaling can be viewed as a kind of null hypothesis for the expected de-
gree of shape change that might occur without selection for change in shape. In other
words, it might reveal the shape changes that can be expected if selection operates not to
change shape per se, but only to increase or decrease overall size. Thus, ontogenetic scal-
ing may be a powerful analytical tool, providing an interesting and viable alternative to
the common premise that interspecific shape differences are functionally meaningful and
selected. It may also provide a "criterion of subtraction" from which deviations from (ap-
parently) passive (or, at least, ontogenetically size-correlated) shape changes can be quan-
tified. Allometric truncation or extension via ontogenetic scaling might be used to predict
proportional differences among species that differ in given amounts of size. Essentially,
ontogenetic scaling might be taken as 1) the appropriate test of the hypothesis that inter-
specific differences in adult proportions are merely allometric consequences of overall
size changes, and 2) "the proper 'criterion of subtraction' with which to assess deviations
from expected allometric baselines" (Shea, 1992: 284). Ontogenetic scaling might signal
evolutionary conservatism and allometric dissociation evolutionary innovation.
288 L. R. Godfrey et al.

Lande (1985) shows, however, that ontogenetic scaling is neither an obvious nor
generally appropriate null hypothesis for passive, size-correlated, evolutionary shape
change (also see Lande, 1979). Arguing from a quantitative genetics perspective, he ob-
serves that selection on body size alone will effect shape change in a direction determined
by the complex genetic variance and covariance structure of the ancestor, and not by its
phenotypic variance and covariance structure. For particular sets of variables, there may
be a strong correlation between genetic and phenotypic variance/covariance matrices, but
such a relationship cannot be assumed. Thus, according to Lande (1985: 26), the null hy-
pothesis that selection on body size alone will extrapolate ontogenetic allometries "is in-
correct in theory and may be seriously misleading in practice." There is also no guarantee
that pre-adult genetic variances and covariances will be strongly correlated with adult
variances and covariances; thus, if selection operates to alter adult morphologies, pre-
adult variances and covariances may have little bearing on evolutionary change in shape
(Lande, 1982). Cheverud (1982) gives further reasons to doubt that, under selection for
change in size, evolution will extrapolate ontogenetic allometries. If the slopes and lengths
of ontogenetic allometries are genetically correlated, then ontogenetic scaling is an un-
likely outcome. Finally, as Huxley recognized, many traits do not scale ontogenetically as
simple power functions of size. When they do not, there can be no theoretical reason to
expect size perturbations to maintain ancestral linear allometries on logarithmic scales.
Ultimately, the degree to which systemic size perturbations preserve ancestral al-
lometries must be assessed empirically. Growth allometries of transgenic and normal mice
have been studied specifically to address this question (Shea et aI., 1987, 1990; Shea,
1988); transgenic mice differ from normal mice in their systemic levels of growth hor-
mone (GH) and insulin-like growth factor I (lGF-I). The results, however, are equivocal.
Of 11 postcranial skeletal comparisons (Shea et aI., 1990:29), six show ontogenetic scal-
ing on the basis of coincidence (on logarithmic scales) of their linear regression parame-
ters and a P value for rejecting ontogenetic scaling set at <0.01. Ifwe set P at <0.05, then
only 3 of the II selected postcranial comparisons exhibit ontogenetic scaling. For cranial
comparisons (Shea et ai., 1990:29), 15 of 19 comparisons show ontogenetic scaling at P
<0.01. Six of these 15 cases of ontogenetic scaling are trivial, and a 7th is not ontogeneti-
cally scaled when P is set at <0.05. Thus only 8 or 9 of 19 selected cranial comparisons
show non-trivial ontogenetic scaling. Whereas some cases of "allometric dissociation"
may involve extrapolation of nonlinear allometries, this cannot be assessed from the infor-
mation offered the reader.
Empirical support for functional neutrality under ontogenetic scaling is also mixed.
Differences in postcranial shape among some species of guenons appear to be a function
of passive shape change under selection for change in size (Shea, 1992). Talapoin mon-
keys are smaller, but leap more frequently, than moustached monkeys, yet they have
higher intermembral indices, higher humerofemoral indices, and higher crural indices. Ac-
cording to Shea (1992: 299), these counterintuitive inter- and intralimb proportions could
not be anticipated outside the context of the pervasive ontogenetic scaling of talapoin and
moustached monkeys: " ... the particular body proportions of the talapoin monkey owe
more to its evolutionary size decrease via allometric truncation, than they do to the results
of concerted selective efforts to alter individual skeletal elements in relation to specific lo-
comotor kinematics and frequencies."
Other studies of postnatal scaling patterns in primates do not support the functional
neutrality of ontogenetic scaling. Thus, for example, Ravosa et al. (1993) observe that the
positive allometry of the intermembral indices of ontogenetically-scaled sifaka species fits
Cartmill's (1974) biomechanical model of vertical climbing. Critics of Gould's use ofbi-
Heterochronic Approaches to the Study of Locomotion 289

variate allometry as a "criterion of subtraction" (for example, Clutton-Brock and Harvey,

1979; Smith, 1980; Jungers, 1984) have offered numerous examples of cross-species dif-
ferences in form and function that correlate strongly with size. Most investigators today,
including proponents of the use of .ontogenetic scaling as a criterion of subtraction (e.g.,
Shea, 1988, 1992; Gomez, 1992), would agree that extrapolation of ancestral growth al-
lometries does not necessarily imply functional conservatism. Function must be assessed
independently of ontogenetic patterns of growth and development.
There is, of course, no necessary connection between the degree of interspecific
shape differentiation and the mechanism through which those differences are achieved.
Both ontogenetic scaling and allometric dissociation can effect any amount of shape dif-
ference between ancestor and descendant. Under ontogenetic scaling, differences in adult
proportions depend on the strength of the allometric coefficients and on the degree of size
differentiation between ancestral and descendant adults. Whenever traits scale isometri-
cally, there will be no correlation between size (X) and shape (Y/X), regardless of the
strength of the correlation between traits Y and X. Only if the scaling coefficients them-
selves have no adaptive significance will shape differences generated through ontogenetic
scaling be functionally neutral. It may well be that allometric dissociation (rather than on-
togenetic scaling) is a prediction of functional conservatism. For example, allometric dis-
sociation may be a means to maintain similar shapes at different sizes. If shape is
important (regardless of size), then allometric dissociation may signal behavioral conser-
vatism.
Not all studies of ontogeny and allometry in primate locomotion focus on the dis-
tinction between ontogenetic scaling and allometric dissociation. Another approach uses
ontogenetic allometries to ascertain the direction and degree to which shape changes post-
natally (Jungers and Fleagle, 1980; Jungers and Hartman, 1988; Jungers and Cole, 1992;
Falsetti and Cole, 1992). For each species, the critical questions are: 1) the degree of de-
parture from postnatal isometry; and, 2) the adaptive significance of species-specific pat-
terns of nonisometric scaling. Shape differences that are present at birth are contrasted
with those that arise postnatally. Both are evaluated adaptively. The degree to which post-
natal scaling patterns preserve or exaggerate shape differences that are present at birth is
also assessed. Ontogenetic allometric extrapolation is viewed as but one of a number of
means to achieve adult shape differences, and is not imbued with functional neutrality.
A major finding of this research is that patterns of postnatal nonisometric scaling are
generally adaptive, and may well preserve or exaggerate interspecific shape differences
that are already manifested at birth. In other words, prenatal shape divergence and postna-
tal patterns of nonisometric scaling can carry the same functional signals. Thus, Jungers
and Cole (1992) showed that the larger-bodied siamangs (Hylobates syndactylus) have
relatively longer forearms and forelimbs at birth than do smaller gibbons (H. lar), and
that, in both species, brachial and humerofemoral indices scale with postnatal positive al-
lometry. These differences serve to enlarge the feeding sphere and modify the frictional
constraints on climbing in the larger-bodied animals. Siamangs, however, are not over-
grown (or peramorphic) gibbons. For many postcranial characteristics, "bivariate scaling
patterns were predominantly allometric and were similar in both species, but postnatal
starting points were usually quite different" (Jungers and Cole, 1992: 98). Jungers and
Hartman (1988) assess the functional and phylogenetic significance of differences in post-
natal nonisometric scaling patterns and prenatal shape divergence among hominoids in
general. Both function and phylogeny playa role in establishing shape differences among
species, but the phylogenetic signal tends to be weak. The human pattern of postnatal
growth allometry is very different from those of other hominoids; it exaggerates differ-
290 L. R. Godfrey et al.

ences already manifested at birth, and is clearly related to bipedalism. In general, postnatal
allometries effect proportional changes that are compatible with biomechanical expecta-
tions; thus, for example, humerofemoral indices increase with increasing size among
large-bodied climbers, but not among bipeds.

4. ALLOMETRIC COMMONALITY, BEHAVIORAL ONTOGENY,

AND THE EVOLUTION OF BEHAVIOR

Does allometric commonality imply commonality of behavioral ontogenies? Does

paedomorphosis or peramorphosis of form imply juvenilization or adultification of posi-
tional behavior? These two questions are, of course, not the same. Paedomorphosis or
peramorphosis can be generated without allometric commonality-i.e., through the weak-
ening or strengthening of ancestral growth allometries. Ontogenetically scaled paedo-
morphs and peramorphs will resemble their ancestor's juvenile or hyperadult stage in both
size and shape, whereas paedomorphs or peramorphs produced under allometric dissocia-
tion will resemble their ancestor's juvenile or hyperadult stage in shape but not size.
The idea that paedomorphosis or peramorphosis inform has direct implications for
behavior (regardless of whether or not ancestral size/shape linkages are preserved) has
deep roots outside the primate literature (see, for example, Olson, 1973; Gould, 1977;
Feduccia, 1980; Coppinger and Coppinger, 1982; Coppinger et aI., 1987; Lawton and
Lawton, 1986; Hafner and Hafner, 1988; MacDonald and Smith, 1994; Livezey, 1995).
This idea has not gained a strong foothold in primatology; primatologists have tended to
show more interest in the mechanical and functional dissimilarities of similarly-shaped or-
ganisms that differ in size (Demes and Gunther, 1989; Strasser, 1992). It is widely appre-
ciated that, to maintain mechanical equivalence at different sizes, linear proportions must
change. The meaning ofpaedomorphosis and peramorphosis is also somewhat ambiguous
under size/shape dissociation, since such "paedomorphs" and "peramorphs" can never be
paedomorphic or peramorphic for all characteristics. This is because tissues, cells, and cell
structures have size constraints that are independent of body mass; they do not scale up or
down proportionally as organisms increase or decrease in overall mass. Developing a set
of behavioral expectations for such organisms is not straightforward.
In contrast, there is no a priori reason why paedomorphs or peramorphs that have
evolved via rate or time hypo- or hypermorphosis cannot be paedomorphic or peramorphic
for all characteristics. Allometric commonality may thus provide a null hypothesis for be-
havioral or functional differences among closely related species that differ in size, pro-
vided that there is some behavioral or functional significance to the ontogenetic changes
in ancestral morphology. Thus, for example, if a descendant is juvenilized in size and
shape, one might posit that the behavioral differences between descendant and ancestral
adults will resemble, in nature and degree, those that distinguish ancestral juveniles from
ancestral adults (Shea, 1986).
Evidence collected to date, however, reveals a nebulous correspondence between al-
lometric commonality and the commonality of behavioral trajectories-at least for pri-
mates. For example, consider the locomotor behavioral ontogenies of pygmy and common
chimpanzees. Doran (1992) demonstrated that Pan paniscus and Pan troglodytes follow
similar locomotor ontogenies. With increasing age, there is a concomitant decrease in sus-
pensory behavior and an increase in quadrupedalism. Doran also showed that the two spe-
cies do not follow this developmental trajectory to the same terminus. Instead, there is a
close match between the locomotor behavioral profiles of adult P paniscus and "Stage 3
Heterochronic Approaches to the Study of Locomotion 291

infant" P troglodytes. In P troglodytes, Stage 3 infants are older than 2 years of age, but
they do not yet travel independently of their mothers. This stage is characterized by rela-
tively greater amounts of suspensory behavior and quadrupedalism and less quadruma-
nous climbing and scrambling than is common in adult P troglodytes. Adult P paniscus
exhibit more suspensory and quadrupedal behaviors than do adult common chimpanzees;
they more closely resemble much younger common chimpanzees. Doran (1992) suggests
that this behavioral difference results from juvenilization via ontogenetic scaling.
Doran's (1992) argument is appealing, but breaks down in some of its details. The
argument has two facets: The first is the proposition (derived from the work of Shea,
1986) that, in many aspects of its postcranial anatomy, P paniscus is paedomorphic vis-a-
vis P troglodytes due to a truncation of common scapular allometries. The second is the
proposition that the behavioral resemblance of adult P paniscus to Stage 3 infant P trog-
lodytes is a by-product of ontogenetic scaling.
We repeated Shea's (1986) test of the significance of differences among the slopes
and intercepts of six scapular growth allometries in P paniscus and P troglodytes (Ta-
bles 4 and 5). Five of the six show ontogenetic scaling. This result differs slightly from
that of Shea (1986) for the same set of six comparisons; Shea found ontogenetic scaling in
all six relationships. However, two of these five (II and VI) show isometry in P troglo-
dytes. There are no significant differences between the adult ratios for these two relation-
ships in P paniscus and P troglodytes-that is, adults of the two species exhibit the same
"shape." The other three relationships show significant negative allometry in P troglo-
dytes (e.g., Figure Sa). Given the truncated allometries and the smaller adult scapulae of P
paniscus, one might expect the latter species to be paedomorphic for these three relation-
ships. Only one, however, reveals a significant difference between adult ratios in the pre-
dicted direction (see Table 5, ratio III). This is the relationship between the morphological
length of the scapula and infraspinous breadth. It is this relationship that argues most
strongly for paedomorphosis in P paniscus.
But even this last example is problematic (Figure 5b). As Shea (1986) pointed out,
the relationship between the morphological length of the scapula and infraspinous breadth
in P paniscus and P troglodytes is not linear on logarithmic scales. The problem here is
that, due to this curvilinearity, shape changes rather little until close to the end of the com-
mon trajectory, where the curve flattens. It is the greater extension of the flattened part of
the curve in P troglodytes that is responsible for the significant difference between the
adult ratios of morphological length to infraspinous breadth in P paniscus and P troglo-
dytes: Whereas it is true that the values of this ratio in adult P paniscus are typical for
young P troglodytes (including Infant Stage 3), they are also typical for much older juve-
nile and subadult, and even many adult, P troglodytes (Figure 5c).t Furthermore, the size
of adult P paniscus scapulae is hardly similar to that of Stage 3 infant P troglodytes. Thus,
P paniscus adult scapulae are not most similar in size and shape to those "ancestral" indi-
viduals with similar behavioral profiles (Figure 6).

• Shea (1986) reports significant differences between adult ratios for three relationships in Pan troglodytes and P.
paniscus - I and IV (in addition to III). We found no significant difference for Ratio I (the relationship between
morphological length and total breadth); in any case, Ratio I suffers the same problem that we describe for Ratio
III. Ratio IV is the relationship between infraspinous breadth and supraspinous breadth. For the two species of
Pan, we found the slopes of the regressions of log infraspinous breadth to be (barely) significantly different;
therefore, we do not treat the observed difference between adult means for Ratio IV as a product of ontogenetic
scaling.
t Shea (1986:486) makes this same point when he states that "adult pygmy chimpanzees have scapulae of the same
size and shape as those [of] subadult and small adult common chimpanzees" (emphasis ours).
 ~
Table 4. Comparison of six scapular growth allometries in Pan paniscus and Pan troglodytes I

Regression of Parameters Pan paniscus Pan troglodytes Significance 2

I. Log morphological length (y) on log total breadth (x) N 25 75
k (SE) 0.920 (0.047) 0.936 (0.044) 0.855 (0.025) 0.879 (0.026) NS
Allometry? J Isometry Negative
y-intercept (SE) 0.018 (0.095) -0.016 (0.091) 0.\34 (0.051) 0.087 (0.052) NS
II. Log morphological length (y) on log supraspinous
breadth (x) N 27 85
k (SE) 0.905 (0.052) 0.934 (0.048) 0.971 (0.026) 1.00 (0.027) NS
Allometry? Isometry Isometry
y-intercept (SE) 0.285 (0.092) 0.234 (0.086) 0.179 (0.046) 0.128 (0.047) NS
III. Log morphological length (y) on log infraspinous
breadth (x) N 29 88
k (SE) 0.803 (0.072) 0.856 (0.076) 0.748 (0.033) 0.793 (0.034) NS
Allometry? Negative Isometry Negative
y-intercept (SE) 0.503 (0.125) 0.411 (0.134) 0.568 (0.057) 0.488 (0.060) NS
IV. Log infraspinous breadth (y) on log supraspinous
breadth (x) N 30 95
k(SE) 0.884 (0.1 06) 0.929 (0.110) 1.122 (0.055) 1.267 (0.061) Yes
Allometry? Isometry Positive
y-intercept (SE) 0.159 (0.190) 0.171 (0.194) -0.215 (0.096) -0.469 (0.107) NS (Yes)
V. Log morphological length (y) on log spine length (x) N 26 71
k (SE) 0.929 (0.03) 0.930 (0.028) 0.927 (0.014) 0.933 (0.014) NS
Allometry? Negative Negative
y-intercept (SE) 0.034 (0.06) 0.032 (0.058) 0.040 (0.027) 0.028 (0.027) NS
VI. Log total breadth (y) on log spine length (x) N 29 90
k (SE) 0.906 (0.049) 0.946 (0.047) 0.977 (0.025) 1.006 (0.026) NS
Allometry? Isometry Isometry
y-intercept (SE) 0.222 (0.100) 0.142 (0.096) 0.089 (0.050) 0.033 (0.051) NS
r
~
IResults of least squares analysis (Model I) in normal font. Results of maximum likelihood analysis (Model II) in bold font. c:'l
2Significance of difference in slope (k) or in y-intercept between Pall palliscus and Pall troglodyte.•. Yes, P<0.05; NS, not significant, 1'>0.05.
c
JJudged to be negative, isometric or positive, depending on significance of difference of slope (k) from 1.00. Negative allometry, k significantly <1.00 (P<O.05); Isometry, k insignificantly different from
=-:;0
to
'<
1.00 (1'>0.05); Positive allometry, k significantly> 1.00 (P<0.05). When not otherwise indicated, Models I and II yield identical allometric conclusions.
!a
co
:-.
 ::t:
...
;-
..,
:;
..,=-
Q
=
;;.
>
'C
'C
ci
~
...=-
'"
Table 5. Test of hypothesis that Pan paniscus is paedomorphic (vis-a-vis Pan troglodytes) due to truncation of common allometries S
....
Ontogenetically Theoretical expectation Observed (comparison of adult
...=-
rJJ
Ratio l Observed allometries scaled? (comparison of adult ratios) ratios), Significance Hypothesis supported? e-
Q.
'<
Isometric (P. p.) or Negative (P.t.) Yes P. paniscus > P. troglodytes P. paniscus =P. troglodytes NS No. Pan paniscus adults are not So
paedomorphic. t""
:;
Q
II Consistently isometric Yes P. paniscus = P. troglodytes P. paniscus = P. troglodytes NS Yes, but trivial. No difference in adult shape.
III Consistently negative 2 Yes P. paniscus > P. troglodytes P. paniscus > P. troglodytes P<O.05 Yes. Pan paniscus adults are paedomorphic a
s.
s·
through ontogenetic scaling.
IV Isometric (P. p.) or Positive (P.t.) No No. There is no ontogenetic scaling. =
V Consistently negative Yes P. paniscus > P. troglodytes P. paniscus = P. troglodytes NS No. Pall paniscus adults are not
paedomorphic.
VI Consistently isometric Yes P. paniscus = P. troglodytes P. paniscus = P. troglodytes NS Yes, but trivial. No difference in adult shape.
IRatio I (morphological length X lOO/total breadth); Ratio II (morphological length X lOO/supraspinous breadth); Ratio III (morphological length X IOO/infraspinous breadth); Ratio IV (infraspinous
breadth X lOO/supraspinous breadth); Ratio V (morphological length X lOO/spine length); Ratio VI (total breadth X lOO/spine length).
2These results are based on least squares analysis, following Shea (1986). On the basis of maximum likelihood analysis, this observation changes to Isometric (in P. paniscus) and Negative (in P. troglo-
dytes). All other results are identical for least squares (Modell) and maximum likelihood (Modelll) analyses.

....
~
294 L. R. Godfrey et al.

An even more compelling example of discordance between allometric and behav-

ioral ontogenies is that of talapoin and moustached monkeys. Shea (1992) found these
guenons to be predominantly ontogenetically scaled in their postcranial anatomy (albeit
largely via growth isometries; see Table 3). Adult talapoins are closest, in postcranial size
and shape, to infant moustached monkeys. Whereas talapoin morphology may have
evolved via allometric truncation from a guenon similar to a moustached monkey, adult
talapoin locomotor behavior certainly did not evolve via truncation of the locomotor be-
havioral ontogeny of such an ancestor.
In summary, the above examples offer little support for the notion that "paedomor-
phosis via ontogenetic scaling" preserves the behaviors of "ancestors" of corresponding
size and shape. Because behavior is potentially influenced by so many variables, we must
devise rigorous tests for our heterochronic hypotheses. Furthermore, many reported exam-
ples of "pervasive" or "predominant" ontogenetic scaling are based largely on traits that
are isometric or nearly so; the little change in proportions that these relationships imply
contrasts markedly with the dramatic ontogenetic changes in positional behavior that so
often occur. Growth isometries are decidedly unhelpful in elucidating ontogenetic changes
in behavior.

5. ONTOGENETIC SCALING, DEVELOPMENTAL

CONSERVATISM, AND BEHAVIORAL COMMONALITY: AN
ILLUSTRATIVE EXAMPLE

Thus far we have examined patterns of postnatal scaling among closely related spe-
cies. In this section we extend the analysis to distantly related species. If the distinction
between ontogenetic scaling and allometric dissociation is meaningful, then one might
reasonably expect distantly-related species that differ in morphology and behavior
throughout their life cycles to exhibit predominant allometric dissociation. Ontogenetic
scaling in behaviorally distinct and phylogenetically distant species might be expected to
reflect developmental conservatism.
We examined Shea's (1986) set of six scapular growth allometries in Macaca fas-
cicularis and Hylobates lar (Tables 6 and 7; Figure 7). We anticipated that these species
would exhibit significantly greater dissociation of growth allometries than the two species
of Pan. For each pair of regressions, we tested the significance of differences between re-
gression parameters (Y-intercepts and slopes). Following Shea (1986), ontogenetic scaling
was diagnosed whenever neither showed a significant difference. Mean adult ratios were
calculated and compared. Adults of both species were identified by full adult dental devel-
opment and fusion of the basioccipital suture. Table 7 reports the significance of the dif-

Figure 5. (a) The relationship between log morphological length and log infraspinous breadth in two species of
Pan. A comparison of growth allometries reveals "ontogenetic scaling" with paedomorphosis of Pan paniscus.
Original measurements are in mm. (b) The same relationship showing the inherent curvilinearity (and flattening in
the adult portions of the two species' ranges) on logarithmic scales. (c) The same example on original measure-
ment scales (in mm.). Ellipses show the ranges of adults of both species, and of older infant Pan troglodytes (in-
cluding Infant Stage 3, aged by dental developmental stage). Note that, on original measurement scales, lines of
shape identity (or isometry) are obtained by connecting individual points to the origin (0, 0). The ellipses indicate
that adult Pan paniscus are the same shape as infant and small adult Pan troglodytes. They are not the same size
and shape as older infant Pan troglodytes.
Heterochronic Approaches to the Study of Locomotion 295

Pan paniscus
2.2 a

.c 2 •
"E>
c:
Q)
-l <>
(ij
1.8 Pan troglodytes
()
.6>
o
(5 1.6
.c
...
o
a. o Pan troglodytes
~ 1.4 • Pan paniscus
0>
.3 1.2~------ __________------____________ ----~

1.2 1.4 1.6 1.8 2 2.2

Log Infraspinous Breadth

2.2
b
.c
"E>
c: 2
~
(ij
() 1.8
.6>
o
-a...
(5
1.6
o <>
~
0> 1.4
.3
1.2 ~------ ________------_----____- - - - _
1.2 1.4 1.6 1.8 2 2.2
Log Infraspinous Breadth

Infraspinous Breadth
 :'"

r
?='
C'J
8-
:::j'
Figure 6. (a) Ontogenetic series of scapulae of Pan troglodytes. Top. adult male (MCZ 48686). Middle. juvenile of approximately 8 years (MCZ 17685). Bottom. infants of approxi-
mately 3-4 years (MCZ 42129 and 34101). (b) Adult male Pan paniscus (MCZ 38020. top. and MCZ 38018. bottom). ~
~
,....
'"
 ::
;-
'"
~
:r
Table 6. Comparison of six scapular growth allometries in Macaca fascicularis and Hylobates lar l Q
.
=
;;'
Regression of Parameters Macaca fascicularis Hylobates lar Significance 2
>
'C
I. Log morphological length (y) on log total breadth (x) N 49 54 'C
k (SE) 0.836 (0.029) 0.832 (0.028) 0.796 (0.028) 0.816 (0.028) NS DO
,.,Cl
Allometry? ' Negative Negative :r
y-intercept (SE) 0.380 (0.044) 0.360 (0.045)
!l
0.410 (0.047) 0.377 (0.047) NS
S
II. Log morphological length (y) on log supraspinous breadth (x) N 49 55 ;.
k (SE) 0.721 (0.052) 0.652 (0.033) 0.730 (0.034) 0.760 (0.035) NS '"
V1
Allometry? Negative Negative C'
y-intercept (SE) 1.018 (0.048) 1.034 (0.032) 0.625 (0.053) 0.578 (0.054) Yes ~
III. Log morphological length (y) on log infraspinous breadth (x) N 49 54 So
k (SE) 0.826 (0.032) 0.856 (0.031) 0.863 (0.056) 0.949 (0.061) NS ~
g
Allometry? Negative Negative Isometry a
y-intercept (SE) 0.461 (0.047) 0.440 (0.045) 0.625 (0.074) 0.512 (0.080) NS s.
IV. Log infraspinous breadth (y) on log supraspinous breadth (x) N 49 54 §'
k (SE) 0.825 (0.067) 0.764 (0.039) 0.699 (0.054) 0.776 (0.059) NS
Allometry? Negative Negative
y-intercept (SE) 0.717 (0.061) 0.691 (0.038) 0.227 (0.085) 0.107 (0.092) Yes
V. Log morphological length (y) on log spine length (x) N 49 55
k (SE) 1.021 (0.010) 1.023 (0.010) 0.991 (0.013) 0.995 (0.012) NS
Allometry? Positive (barely) Isometry
y-intercept (SE) -0.104 (0.017) -0.108 (0.017) -0.085 (0.023) -0.093 (0.023) NS
VI. Log total breadth (y) on log spine length (x) N 49 54
k (SE) 1.162 (0.038) 1.230 (0.040) 1.175 (0.040) 1.216 (0.041) NS
Allometry? Positive Positive
y-intercept (SE) -0.477 (0.066) -0.563 (0.070) -0.492 (0.074) -0.570 (0.076) NS
I Resultsof least squares analysis (Modell) in normal font. Results of maximum likelihood analysis (Model II) in bold font.
2Significance of difference in slope (k) or in y-intercept between Macacafascicularis and Hylobates lar. Yes, P<0.05; NS. not significant, P>0.05.
'Judged to be negative, isometric or positive, depending on significance of difference of slope (k) from 1.00. Negative allometry, k significantly < 1.00 (P<0.05); Isometry, k insignificantly different from
1.00 (P>0.05); Positive allometry, k significantly> 1.00 (P<0.05). When not otherwise indicated, Models I and II yield identical allometric conclusions.

~
.....
298 L. R. Godfrey et al.

Table 7. Test of hypothesis that Hylobates lar is peramorphic (vis-ii-vis Macacafascicularis)

due to an extension of common allometries
Theoretical
expectation Observed
Observed Ontogenetically (comparison of (comparison of adult Hypothesis
Ratio' allometries scaled? adult ratios) ratios), significance supported?
Consistently Yes H.lar<M. H. lar < M. Yes. Hylobates lar
negative fascicularis fascicularis P<O.05 adults are
peramorphic.
II Consistently No No. There is no
negative ontogenetic scaling.
III Consistently Yes H.lar < M. H. lar > M. No. Hylobates lar
negative 2 fascicularis fascicularis P<O.05 adults are
paedomorphic.
IV Consistently No No. There is no
negative ontogenetic scaling.
V Isometric (H. I.) Yes H.lar > M. H. lar < M. No. Hylobates lar
or Positive (M. fascicularis fascicularis P<O.05 adults are
f) paedomorphic.
VI Consistently Yes H.lar > M. H. lar > M. Yes. Hylobates lar
positive fascicularis fascicularis P<O.05 adults are
peramorphic.
'Ratio I (Morphological length X 100/total breadth); Ratio II (Morphological length X 100/supraspinous breadth); Ratio III
(Morphological length X 100/infraspinous breadth); Ratio IV (Infraspinous breadth X 100/supraspinous breadth); Ratio V
(Morphological length X 100/spine length); Ratio VI (Total breadth X 100/spine length).
2 These results are based on least squares analysis, following Shea (1986). On the basis of maximum likelihood analysis. this ob-
servation changes to Negative (in M. fascicularis) and Isometric (in H. lar). All other results are identical for least squares
(Model I) and maximum likelihood (Model II) analyses.

ferences between adult ratios, and, for those instances for which significant differences
exist, the directionality of the difference.
Four of the six bivariate comparisons of M. fascicularis and H. lar show "ontoge-
netic scaling" (e.g., Figure 8; Table 7). Given the striking differences in locomotor onto-
geny and adult morphology and behavior between these two species, this result seems
counterintuitive. Surely, nobody would claim that the morphology of the scapula of the
slightly larger-bodied H. lar (see Smith and Jungers, 1997) is similar to that of a hy-
peradult, or overdeveloped, M.fascicularis. Nor would anyone seek a behavioral corollary
for the signal of predominant ontogenetic scaling. No extrapolation of the locomotor be-
havioral ontogeny of an "ancestor" such as M. fascicularis would produce a species re-
sembling a "descendant" such as H. lar.
Does the predominant ontogenetic scaling of the scapulae of M. fascicularis and H.
lar imply that scapular development in Catarrhini is highly conservative? Is it possible
that the four ontogenetically scaled relationships reflect conservatism of certain aspects of
scapular development, whereas the two that show allometric dissociation carry all of the
information of functional import?
Upon scrutiny of the four relationships that show "ontogenetic scaling," it becomes
clear that the answer is "no." If scapular development were largely conservative in the
above-described manner, then we would expect adult shape differences to be passive con-
sequences of allometric extrapolation. The larger-bodied species (in this case, H. lar)
should be peramorphic for all shape differences that arise via ontogenetic scaling. Pera-
morphosis requires the larger-bodied species to manifest the smaller adult ratios whenever
Heterochronic Approaches to the Study of Locomotion 299

Figure 7. Top, infant (Mel 35643, male) and adult (Mel 35613, male) Macaca fascicularis. Bottom, infant
(Mel 35456, male) and adult (Mel 41415. male) Hylobates lar.

the common allometries are negative, and the larger adult ratios whenever the common al-
lometries are positive, In only two of these four examples do the observed adult ratios dif-
fer in the "expected" direction (Table 7). Thus, the four examples, taken together, fail to
support a hypothesis of developmental conservatism.
The two examples that do support peramorphosis via ontogenetic scaling in the larger-
bodied H. lar are flip sides of the same coin. These are relationships I (between log morpho-
logical length and log total breadth) and VI (between log total breadth and log spine length),
Because morphological length and spine length are highly redundant (see relationship V;
Figure 9), the negative allometry of relationship I (where log total breadth is taken as the X
variable) and positive allometry of relationship VI (where log total breadth is taken as the Y
variable) are also redundant. Because total scapular breadth is a reflection of both infraspi-
nous breadth and supraspinous breadth, true ontogenetic conservatism should imply conser-
vatism in the relationships between each of the latter two variables and morphological (or
spine) length. In M. fascicularis and H. lar, these expectations are not met. Indeed the two
examples of "allometric dissociation" (relationships II and IV) involve supraspinous and in-
fraspinous breadth (as related to each other or to morphological length),
 300 L. R. Godfrey et aL

2.2
Pp
Pt

:: 2
0)
C
j
~ 1.8
"a'
o
'0
e- 1.6
~

o
:IE • Macaca fascicularis
0)
o o Hylobates lar
..J 1.4 • Pan paniscus
<> Pan troglodytes

1.2~------~--------~--------~-------+--------~------~~-----
1 1.2 1.4 1.6 1.8 2 2.2

Log Total Breadth

Figure 8. The relationship between log morphological length and log total scapular breadth in Macacafascicu-
laris and Hylobates lar, as well as Pan paniscus and Pan troglodytes. Both species pairs show "ontogenetic scal-
ing" and all four species exhibit weak negative allometry for this relationship. The HylobateslMacaca example
supports the "expectation" that differential extension of common growth allometries will reveal differences in
adult proportions. The Pan comparison does not.

...en
~ 2

c
CD
..J
'ii 1.8
.E
en
0
'0
..
~
a.
0
1.6
•
o
Macaca fasc;cular;s
Hylobates lar
:IE
en • Pan pan;scus
0
..J
1.4
• <> Pan troglodytes

1.2
•
1.3 1.5 1.7 1.9 2.1
Log Spine Length
Figure 9. The relationship between log morphological length and log spine length shows ontogenetic scaling in
Macaca fascicularis and Hylobates lar. However, these variables are strongly redundant. The relationship between
them is virtually identical in other primate species, including Pan troglodytes and Pan paniscus (which also show
ontogenetic scaling for this relationship).
Heterochronic Approaches to the Study of Locomotion 301

Macaca fascicularis and Hylobates lar do differ from one another in their scapular
ontogenies more than the two species of Pan. However, a tabulation of the degree to
which the same six scapular bivariate relationships show ontogenetic scaling does not
make this exceedingly obvious. Although H. lar is larger in body mass (on average) than
M. !ascicularis, all scapular dimensions are not larger. Shape differences, even for suites
of traits that show "ontogenetic scaling," do not fall along a simple spectrum from paedo-
morphosis to peramorphosis. For regressions that show "allometric dissociation," Macaca
and Hylobates do not exhibit the same shapes at different sizes; rather, they exhibit differ-
ent shapes throughout their ontogenies (Figure 10). Within a multivariate context (Table 8,
Sutherland and Godfrey, in prep.), it is easy to document the degree of separation of shape
pathways and the relative contributions of different trait complexes to that separation. In

Macaca fascicularis
1.7 a

-
L:
"0
til
1.6
•
•
• •••
#• • • •+ Hylobates lar

-
l!? 1.5 ••• ••
a:l
• • 0
CJ)
:::J 1.4 ~\
0
c:
'0.. 1.3
• 8

-
CJ)
til
L.
1.2
E
Cl
1.1
••
0
....J 00

1 °e
0.4 0.6 0.8 1 1.2 1.4 1.6 1.8
Log Supraspinous Breadth

50 b Macaca fascicularis

L:
-ctil 40
Hylobates lar
Q)
L.
a:l 30 o
CJ)
:::J
0
c:
'0.. 20

-
CJ)
til
L.

E 10

0
0 10 20 30 40 50 60
Supraspinous Breadth

Figure 10. (a) The relationship between log infraspinous breadth and log supraspinous breadth in Macacafascicu-
laris and Hylobates lar. (b) The same relationship on original measurement scales. Shown here are lines of
isometry for individuals situated at either end of each species' range. Note that there is little ontogenetic shape
change (for this relationship) in either species (i.e., adults of both species exhibit proportions that occur in very
young individuals). Furthermore, there is no shape commonality (for this relationship) at any point in the species'
ontogenies. Under allometric dissociation, shapes may be similar at different sizes, or shapes may differ through-
out ontogeny. This example conforms to the latter situation.
302 L. R. Godfrey et al.

Table 8. Euclidean distances between stage centroids in 5-dimensional shape space'

Distance separating stage centroids of Macaca Distance separating centroids of Distance separating
fascicularis and Hylobates lar infants and adults Pan species' centroids
Infants Juveniles Subadults Adults Macaca Hylobates Adults
0.117 0.128 0.\32 0.138 0.040 0.032 0.D28
I The five axes are: the morphological length of the scapula, the morphological breadth of the scapula, infraspinous breadth, su-
praspinous breadth, and spine length-each is scaled as a proportion of the whole.

other words, paths in shape space can be used to measure neomorphosis and to isolate the
ways in which developmental pathways differ throughout ontogeny. In this case, it is not
scapular breadth, but the combination of infraspinous and (especially) supraspinous
breadth that distinguishes the two species (cf. Figure lla,b).
Table 8 documents the separation (or Euclidean distances in shape space) between
developmental stage centroids for Macaca and Hylobates. We constructed a 5-dimensional
shape space from the five scapular traits that were used to assess ontogenetic scaling vs.
allometric dissociation in M.fascicularis, H. lar, and the two species of Pan. The separa-
tion of stage centroids for Macaca and Hylobates is nearly as great at infancy as at adult-
hood, and is more than four times the distance between adult centroids of the two species
of Pan. The total distances that Macaca and Hylobates travel through shape space, from
infancy to adulthood, are small. Each traverses about a quarter of the total distance sepa-
rating the two species' adult centroids. Macaca and Hylobates occupy their own "niches"
in shape space, and maintain those niches throughout life. Prenatal development is critical
to establishing species-specific shape differences. If there is common directionality to
postnatal shape change, it is a slight decrease in scapular length relative to scapular
breadth (Figure 11 b}-a phenomenon that appears to relate to the mechanical efficiency of
forelimb protraction and retraction in individuals of different body size.

6. HETEROCHRONY IN PRIMATOLOGY: WHERE NOW?

There are numerous ways in which ontogenetic research can contribute to our under-
standing of primate postcranial skeletons and locomotor behavior. One approach has
largely dominated the literature on heterochrony in primate locomotion. It values ontoge-
netic studies for their potential to reveal that an observed interspecific shape difference is
correlated with (and possibly caused by) factors that generate evolutionary changes in
size. Certainly, passive (size-correlated) shape change represents a viable and interesting
alternative to the premise that interspecific shape differences are functionally meaningful
and "deliberately" selected. Many primatologists use "ontogenetic scaling" (or the coinci-
dence of ontogenetic linear regression parameters on logarithmic scales) both to test the

Figure 11. Scapular ontogeny in Macacafascicularis and Hylobates lar as glimpsed in two 3-dimensional con-
structions of shape space. Plotted here are the relative proportions for sets of three variables, averaged over four
dental developmental stages (infants,juveniles, subadults, and adults). Mean values for these proportions are given
for each stage, in the order X, Y, and Z, top to bottom. Arrows show the direction of ontogenetic change in shape
(from infants to adults). (a) The three variables shown here are "allometrically dissociated" on bivariate plots. (b)
The three variables shown here are "ontogenetically scaled" on bivariate plots. Even traits that are "ontogeneti-
cally scaled" when examined as pairs on bivariate plots may not follow the same trajectories when examined as
multivariate sets in shape space.
Heterochronic Approaches to the Study of Locomotion 303

Hylobates

:50
.31 I

~ ..
~ ,42 Macaca
aJ .41 ,32 :'i3
ci5 .32 .26 .31 ,45'
1!1 ,27, ,30 ~6
r-Jg
~
1,25
~ '-- ...
'Q-
~ f45\' i Hylobates
.44 ,39 1.46
~
....- ,40 , 16, .37
, 16 ,.17
304 L. R. Godfrey et al.

prevalence of size-correlated shape change among closely related species, and to construct
a "criterion of subtraction" from which deviations from size-correlated shape changes can
be quantified. It is also sometimes used to construct a null hypothesis for size-correlated
changes in function.
The efficacy of this approach depends on 1) trait-trait ontogenetic relationships be-
ing linear on logarithmic scales, and 2) passive (size-correlated) evolutionary shape
change extrapolating or truncating ancestral ontogenetic allometries. Given the caveats
that we have discussed in this paper, we would suggest that "ontogenetic scaling" be used
as the null hypothesis for passive (size-correlated) evolutionary changes in shape or in be-
havior only with extreme caution. We offer the following recommendations:
1. It should be recognized that selection for change in size alone can result in dis-
sociated growth allometries. Indeed, extrapolation or truncation oflinear or non-
linear ontogenetic allometries is a prediction of selection for change in size (and
not shape) only under a limited set of conditions. One such condition (i.e.,
strong correlation between ontogenetic phenotypic and genetic variance/covari-
ance matrices) can be tested by examining the heritability of traits.
2. The linearity of trait-trait relationships on logarithmic scales should be assessed.
When trait-trait relationships are not linear on logarithmic scales, linear regres-
sion on logarithmic scales should not be used to test the hypothesis that shape
differences are correlated with (and possibly caused by) differences in size.
Multivariate tools that do not depend on a priori assumptions regarding the na-
ture of trait-trait relationships can be employed to test the preservation or dis-
ruption of ancestral size/shape linkages (Sutherland and Godfrey, in prep.). They
can be used to assess the commonality of patterns of change in allometric
slopes, including the sort of heterochronic process that has been called, by some,
sequential hypo- or hypermorphosis.
3. Growth isometries should be documented and treated with reservation. Com-
monality of growth isometries cannot demonstrate that shape changes passively
with size. To the contrary, isometric growth demonstrates no relationship be-
tween size and shape. Furthermore, under ancestral growth isometry, "ontoge-
netic scaling" is the prediction of some heterochronic processes that disrupt as
well as those that preserve ancestral size/shape linkages (and is therefore diag-
nostically trivial).
4. Adult ratios must be compared to verify that shape does indeed differ in the di-
rection or directions predicted by extrapolation or truncation of ancestral onto-
genetic allometries.
5. Function must be assessed independently of the ontogenetic trajectories through
which changes in skeletal form are achieved.
Ontogenetic studies can contribute to our understanding of locomotor skeletal anat-
omy and locomotor behavior in other ways. Only through ontogenetic research can we ap-
preciate the degree to which interspecific shape differences are manifested at birth. Only
through ontogenetic research can we discover the interplay between size and age as triggers
of change in form, and the impacts of variation in absolute rates of growth and development
on adult morphology and behavior. Ontogenetic research can reveal developmental disso-
ciation of parts. Dissociation of parts is not the same as allometric dissociation. Allometric
dissociation can be generated by processes that preserve ancestral developmental pathways
(but not size/shape linkages). Under nonlinear allometry, allometric dissociation can even
be produced by processes that preserve ancestral size/shape linkages. By distinguishing dis-
Heterochronic Approaches to the Study of Locomotion 305

sociation of parts (or developing modularity) from allometric dissociation, investigators

may be able to identify sites of locally targeted selection. Distinguishing allometric disso-
ciation from ontogenetic scaling does not address this problem.
In this paper, we have reviewed the relationship between heterochrony and al-
lometry. We have shown that tests of the coincidence of allometric slopes and intercepts
are only partially successful in revealing heterochronic processes. They have uncertain
utility in revealing size-correlated (and possibly size-required) evolutionary shape change.
They do not diagnose developmental or functional conservatism. Allometric commonality
does not reveal commonality of behavioral ontogenies. Testing any heterochronic proces-
sual model means understanding the model's full set of predictions and how those predic-
tions differ from those of other models. The utility of diagnostic tools depends on their
ability to identify features that distinguish alternative models. Progress in understanding
those evolutionary processes that underlie variation in skeletal form and function means
being willing to modify our processual models, and the tools we use to test them, when in
fact those models or those diagnostic tools do not work.

ACKNOWLEDGMENTS

Research for this paper was supported in part by NSF grant GER 9450175 to LRG and
by the Statistical Consulting Center of the University of Massachusetts. We thank Elizabeth
Strasser for inviting us to submit this paper, and for her patience and encouragement along
the way; we thank William Jungers, Brian Shea, Richard Smith, and an anonymous re-
viewer for their insightful comments on our first draft. We are grateful to Darren Godfrey
and Emily Heinisch for their skillful preparation of the figures. Data analyzed here were
collected by Stephen J. King or drawn from the literature. We thank Maria Rutzmoser and
the curatorial staff of the Museum of Comparative Zoology at Harvard University for access
to scapulae of Macaca jascicularis, Hylobates lar, Pan troglodytes, and Pan paniscus, and
for permission to photograph individuals belonging to this collection. We also thank AW
Crompton for the use of photographic copy equipment in his laboratory.

REFERENCES
Alberch P, Gould SJ, Oster GF, and Wake DB (1979) Size and shape in ontogeny and phylogeny. Paleobiology
5:296-317.
Cartmill M (1974) Pads and claws in arboreal locomotion. In FA Jenkins (ed.): Primate Locomotion. New York:
Academic Press, pp. 45-S3.
Cheverud J (1982) Relationships among ontogenetic, static, and evolutionary allometry. Am. J. Phys. Anthropo\.
59:139-149.
Clutton-Brock TH, and Harvey PH (1979) Comparison and adaptation. Proc. R. Soc. Lond. B. 205:547-565.
Coppinger L, and Coppinger R (1982) Livestock-quarding dogs that wear sheep's clothing. Smithsonian 13:65-73.
Coppinger R, Glendinning J, Matthay C, Smith C, Sutherland M, and Torop E (1987) Degree of behavioral
neoteny differentiating canid polymorphs. Ethology 75:89-108.
Demes B, and Gunther MM (1989) Biomechanics and allometric scaling in primate locomotion and morphology.
Folia Primato\. 53:125-141.
Doran OM (1992) The ontogeny of chimpanzee and pygmy chimpanzee locomotor behavior: A case study of pae-
domorphism and its behavioral correlates. J. Hum. Evo\. 23:139-157.
Falsetti AB, and Cole TM III (1992) Relative growth of the postcranial skeleton in callitrichines. J. Hum. Evo\.
23:79-92.
Feduccia A (1980) The Age of Birds. Cambridge, Ma.: Harvard University Press.
306 L. R. Godfrey et aL

Godfrey LR, and Sutherland MR (I 995a) Flawed inference: Why size-based tests ofheterochronic process do not
work. J. Theor. BioI. 172:43--61.
Godfrey LR, and Sutherland MR (1995b) What's growth got to do with it? Process and product in the evolution of
ontogeny. J. Hum. Evol. 29:405-431.
Godfrey LR, and Sutherland MR (1996) Paradox of peramorphic paedomorphosis: Heterochrony and human evo-
lution. Am. J. Phys. Anthropol. 99: 17-42.
Gomez AM (1992) Primitive and derived patterns of relative growth among species of Lorisidae. J. Hum. Evol.
23:219-233.
Gould SJ (1966) Allometry and size in ontogeny and phylogeny. BioI. Rev. 41: 587-640.
Gould SJ (1975a) Allometry in primates, with emphasis on scaling and the evolution of the brain. In FS Szalay
(ed.): Approaches to Primate Paleobiology. Karger: Basel. Contr. Primatol. 5:244-292.
Gould SJ (1975b) On the scaling of tooth size in mammals. Am. Zool. 15:351-362.
Gould SJ (1977) Ontogeny and Phylogeny. Cambridge, Ma.: Harvard University Press.
Hafuer JC, and Hafner MS (1988) Heterochrony in rodents. In ML McKinney (ed.): Heterochrony in Evolution: A
Multidisciplinary Approach. New York: Plenum Press, pp. 217-235.
Huxley JS (1932) Problems of Relative Growth. London: Methuen & Co.
Inouye SE (1992) Ontogeny and allometry of African ape manual rays. J. Hum. Evol. 23: I 07-138.
Jungers WL (1984) Aspects of size and scaling in primate biology with special reference to the locomotor skele-
ton. Yrbk. Phys. Anthropol. 27:73-1)7.
Jungers WL, and Cole MS (1992) Relative growth and shape of the locomotor skeleton in lesser apes. J. Hum.
Evol. 23:93--105.
Jungers WL, and Fleagle JG (1980) Postnatal growth allometry of the extremities in Cebus albifrons and Cebus
apel/a: A longitudinal and comparative study. Am. J. Phys. Anthropol. 53:471-478.
Jungers WL, and Hartman SE (1988) Relative growth of the locomotor skeleton in orang-utans and other large-
bodied hominoids. In JH Schwartz (ed.): Orang-utan Biology. Oxford: Oxford University Press, pp.
347-359.
Klingenberg CP, and Spence JR (1993) Heterochrony and allometry: Lessons from the water strider genus Lim-
noporus. Evol. 47: 1834-1853.
Lande R (1979) Quantitative genetic analysis of multivariate evolution applied to brain:body size allometry. Evol.
33:402-416.
Lande R (1982) A quantitative genetic theory of life history evolution. Ecology 63:607-615.
Lande R (1985) Genetic and evolutionary aspects of allometry. In WL Jungers (ed.): Size and Scaling in Primate
Biology. New York: Plenum Press, pp. 21-32.
Lawton MF, and Lawton RO (1986) Heterochrony, deferred breeding, and avian sociality. In RF Johnston (ed.):
Current Ornithology, Volume 3. New York: Plenum Press, pp. 187-221.
Livezey BC (1995) Heterochrony and avian flightlessness. In KJ McNamara (ed.): Evolutionary Change and Het-
erochrony. New York: John Wiley & Sons, pp. 169-193.
McDonald MA, and Smith MH (1994) Behavioral and morphological correlates of heterochrony in Hispaniolan
palm-tanagers. The Condor 96:433-446.
McKinney ML (1986) Ecological causation of heterochrony: A test and implications for evolutionary theory. Pa-
leobiology 12:282-289.
McKinney ML (1988) Classifying heterochrony: Allometry, size, and time. In ML McKinney (ed.): Heterochrony
in Evolution, A Multidisciplinary Approach. New York: Plenum Press, pp. 17-34.
McKinney ML, and McNamara KJ (1991) Heterochrony: The Evolution of Ontogeny. New York: Plenum Press.
McNamara KJ (1983) Progenesis in trilobites. In DEG Briggs and PO Lane (eds.): Trilobites and Other Early
Arthropoda: Papers in Honour of Professor H. B. Whittington, FRS. Special Papers in Paleontology 30.
London: Paleontological Association, pp. 59-68.
Olson SL (1973) Evolution of the rails of the South Atlantic islands (Aves: Rallidae). Smithson. Contr. Zool.
152:1-53.
Pilbeam DR, and Gould SJ (1974) Size and scaling in human evolution. Science 186: 892-901.
Raff RA, Parr BA, Parks AL, and Wray GA (1990) Heterochrony and other mechanisms of radical evolutionary
change in early development. In MH Nitecki (ed.): Evolutionary Innovations. Chicago: University of Chi-
cago, pp. 71-98.
Ravosa MJ (1992) Allometry and heterochrony in extant and extinct Malagasy primates. J. Hum. Evol.
23:197-217.
Ravosa MJ, Meyers OM, and Glander KE (1993) Relative growth of the limbs and trunk in sifakas: Hetero-
chronic, ecological and functional considerations. Am. J. Phys. Anthropol. 92:499-520.
Rice SH (1997) The analysis of ontogenetic trajectories: When a change in size or shape is not heterochrony. Proc.
Natl. Acad. Sci. USA 94:907-912.
Heterochronic Approaches to the Study of Locomotion 307

Shea BT (1981) Relative growth of the limbs and trunk in the African apes. Am. 1. Phys. Anthropol. 56: 179-201.
Shea BT (1983) Allometry and heterochrony in the African apes. Am. 1. Phys. Anthropol. 62:275-289.
Shea BT (1985) Bivariate and multivariate growth allometry: Statistical and biological considerations. J. Zool.,
Lond. 206:267-290.
Shea BT (1986) Scapula form and locomotion in chimpanzee evolution. Am. J. Phys. Anthropol. 70:475-488.
Shea BT (1988) Heterochrony in primates. [n ML McKinney (ed.): Heterochrony in Evolution. New York: Plenum
Press, pp. 237-266.
Shea BT (1989) Heterochrony in human evolution: The case for neoteny reconsidered. Yrbk. Phys. Anthropol.
32:69-101.
Shea BT (1992) Ontogenetic scaling of skeletal proportions in the talapoin monkey. 1. Hum. Evol. 23:283--307.
Shea BT, Hammer RE, and Brinster RL (1987) Growth allometry of the organs in giant transgenic mice. Endocrin.
121: 1924-1930.
Shea BT, Hammer RE, Brinster RL, and Ravosa MJ (1990) Relative growth of the skull and postcranium in giant
transgenic mice. Genet. Res. Camb. 56:21-34.
Smith R1 (1980) Rethinking allometry. J. Theor. BioI. 87:97-111.
Smith RJ, and Jungers WL (1997) Body mass in comparative primatology. J. Hum. Evol. 32:523-559.
Strasser E (1992) Hindlimb proportions, allometry, and biomechanics in old world monkeys (Primates, Cercopi-
thecidae). Am. 1. Phys. Anthropol. 87:187-213.
Sutherland MR, and Godfrey LR (in prep.) A multivariate matrix model for heterochronic analysis.
Taylor AB (1995) Effects of ontogeny and sexual dimorphism on scapula morphology in the mountain gorilla (Go-
rilla gorilla beringei). Am. J. Phys. Anthropol. 98:431-445.
Taylor AB (1997) Relative growth, ontogeny, and sexual dimorphism in Gorilla (Gorilla gorilla gorilla and G. g.
beringei): Evolutionary and ecological considerations. Am. 1. Primatol. 43: 1-31.
Vrba ES, Vainys JR, Gatesy JE, De Salle R, and Wei K-Y (1994) Analysis ofpaedomorphosis using allometric
characters: The example of the Reduncini antelopes. Syst. BioI. 43:92-116.
Zelditch ML, and Fink WL (1996) Heterochrony and heterotopy: Stability and innovation in the evolution of form.
Paleobiology 22:241-254.
 17

BODY SIZE AND SCALING OF LONG BONE

GEOMETRY, BONE STRENGTH, AND
POSITIONAL BEHAVIOR IN CERCOPITHECOID
PRIMATES

William L. Jungers,l David B. Burr,2 and Maria S. Cole3

IDepartment of Anatomical Sciences

School of Medicine
SUNY at Stony Brook
Stony Brook, New York 11794-8081
2Departments of Anatomy and Orthopedic Surgery
Biomechanics and Biomaterials Research Center
Indiana University Medical Center at Indianapolis
Indianapolis, Indiana 46202-5120
3Department of Anatomy
University of Health Sciences
College of Osteopathic Medicine
Kansas City, Missouri 64106-1453

1. INTRODUCTION

Allometry in the strictest biometrical sense--size-correlated differences in shape -

explains nothing. It is also not a biological "principle" (Smith, 1980; Jungers, 1984; Jung-
ers et aI., 1995; contra Gould, 1975; contra Martin, 1993). Rather, allometry is merely a
quantitative description or signal that mayor may not serve to test an explicit hypothesis.
Without explicit hypotheses of how and why things should change as a function of body
size (i.e., similarity criteria), allometry cannot be diagnosed except with respect to the
statistical, dimensional null hypothesis of "isometry" or geometric similarity. In special
circumstances, isometry can itself be a hypothetical criterion of biological similarity (Al-
exander et aI., 1979; Biewener, 1990; Prothero, 1992). If such criteria cannot be specified
and justified a priori, it also follows that even when allometry is discovered, it cannot be
assumed that the observed size-correlated differences are evidence of size-required
changes sufficient to insure "functional equivalence" (Smith, 1980). Empirical lines used

309
310 W. L. Jungers et al.

to describe allometric patterns of interspecific scaling can rarely, if ever, be rationalized

into meaningful, adaptive "criteria of subtraction" for the subsequent analysis of residuals
(Smith, 1984; Jungers et aI., 1995). The scaling of mammalian long-bone dimensions
makes these points clearly and unequivocally: although long bone robusticity is expected
to increase with body size according to most biomechanical theories, positively allometric
distortions in the shape of the long bones of larger vertebrates do not produce functional
equivalence in any mechanical or behavioral sense. To the contrary, further behavioral and
structural modifications are still required to maintain adequate safety factors at larger
body sizes (Biewener, 1982, 1990; Rubin and Lanyon, 1984; Selker and Carter, 1989; Ber-
tram and Biewener, 1990; Demes and Jungers, 1993; Jungers and Burr, 1994).
One of the very few explicit similarity criteria in organismal biology was devel-
oped by McMahon (1973, 1975) and applied widely to long bone scaling, "elastic simi-
larity", whereby diameters were predicted to increase disproportionately relative to
lengths in order to prevent Euler buckling; i.e., strong positive allometry of diameters
was the specified expectation such that length scales to diameter to the two-thirds power.
This criterion was qualified (Economos, 1983), modified (Hokkanen, 1986), generalized
to other loading regimes (Prange, 1977; McMahon, 1984; Alexander, 1988), and fre-
quently rejected for a wide array of mammals, including primates (e.g., Alexander et aI.,
1979; Aiello, 1981; Jungers, 1984; Casinos et aI., 1986; Ruff, 1987; Bertram and
Biewener, 1990; Demes et aI., 1991; Biknevicius, 1993; Demes and Jungers, 1993). It is
now recognized that no universal allometric similarity criterion exists for vertebrate long
bones (Rubin and Lanyon, 1984). Instead, the scaling criteria themselves appear to
change as a function of body size (Biewener, 1989, 1990, 1991). Biewener's valuable
synthesis of theory and data predicts that for terrestrial mammals in the size range from
100 g to 300 kg (i.e., encompassing almost the entire range of extant primates), only
slightly positive allometry of limb bone diameters is likely to occur in concert with
changes in limb posture (more extended limbs in larger species) and muscular mechani-
cal advantages (better mechanical advantages in larger animals). In terms of the dynam-
ics of positional behavior, this multifactorial response probably implies somewhat
reduced agility and compromised "athletic ability" in larger species (Alexander, 1989).
We propose to test aspects of this general hypothesis here in two different ways for cer-
co pithecoid monkeys: (1) via analysis of external long bone diameters for a large data
set of wild-collected individuals, and (2) by similar analysis of a subset of these speci-
mens using a cross-sectional methodology based on photon absorptiometry and linked
closely to bone strength (Martin and Burr, 1984; Schaffler et aI., 1985; Martin, 1991;
Jungers and Burr, 1994). Note that Biewener's comprehensive model, unlike that of
McMahon and others, provides no exact magic number of how diameters and cross-sec-
tional dimensions need to scale allometric ally for long bones to cope with gravity-related
strains and stresses. Indeed, precise "equivalence" remains an elusive and quixotic con-
cept in scaling, even in the realm of bone biomechanics.
As Strasser (1992) has observed, cercopithecoids are an excellent "natural experi-
ment" for the study of size and postcranial scaling. There is a similar Bauplan for the en-
tire group (Schultz, 1970), and the two major clades - cercopithecines and colobines-
are clearly monophyletic (Strasser and Delson, 1987). Adult body mass of cercopithecines
ranges from just over 1 kg (talapoin monkeys) to over 30 kg (mandrill males); colobines
range from under 5 kg (olive colobus) to over 20 kg (proboscis monkey males) (Smith and
Jungers, 1997). At the same time, there are noteworthy clade-specific differences in loco-
motor behaviors (e.g., more leaping in colobines) and substrate preferences (more terre-
striality in cercopithecines) that should have mechanical and structural consequences
Body Size and Scaling of Long Bone Geometry in Cercopithecoid Primates 311

(reviewed in Strasser, 1992). We also place our findings about cercopithecoid scaling into
a broader anthropoid context and discuss several eco-behavioral implications of our re-
sults.

2. SCALING OF HUMERAL AND FEMORAL EXTERNAL

DIAMETERS AT MID SHAFT

2.1. Measurements and Skeletal Sample

It is relatively easy to capture data on the external dimensions of long bones. Here
we use sliding calipers to measure inter-articular lengths of the humerus and femur as well
as anteroposterior (a-p) and mediolateral (m-l) diameters at midshaft. Our skeletal sample
is listed in Table I along with sex-specific averages for body mass. Our colobine sample
includes 78 adult individuals of known body mass (data taken from field notes, museum
records and specimen labels); this includes data on 15 colobine species that range in body
mass here from 4.57 kg (female Procolobus verus) to 21.4 kg (male Nasalis larvatus). The
cercopithecine sample is comprised of 131 individuals, all of which have associated body
masses except for Miopithecus talapoin; we used the sex-specific averages reported by
Gautier-Hion (1975) for talapoins. The body mass range for cercopithecines used here is
1.12 kg for female talapoins to 29 kg (chacma baboon males) and a single giant mandrill
male at 45 kg. Trimming the size distribution of cercopithecines to exclude the talapoins
and the mandrill had no significant effect on any of our results or inferences.

2.2. Statistical Methods

We have estimated the relevant scaling parameters--slope and intercept--{)f lengths
and diameters on body mass in log-log space (base e) using the Model II reduced major
axis (RMA) line-fitting method (Ricker, 1984; Rayner, 1985; McArdle, 1988). Ordinary
least squares regression is clearly inappropriate because it assumes that there is no error
term associated with the X (mass) variable. Sex-specific means of all variables were cal-
culated for each species; these are then used as our raw data. Despite sexual dimorphism
in body size in most species and unbalanced sample sizes, preliminary Model II compari-
sons of slopes and elevations (see below) disclosed no significant differences between
sex-specific samples (all P values were> 0.5). We have adjusted degrees of freedom
downward, however, due to significant phylogenetic inertia by using the variance compo-
nents approach developed by Smith (1994), as implemented by a maximum likelihood
procedure in the mainframe version of SAS (1985). Our reduced, "effective" sample size
used for calculating confidence limits and statistical testing is therefore 22 rather than
either 209 (individuals) or 57 (sex-specific averages). Colobines and cercopithecines were
also examined separately and compared via Clarke's (1980) t-test for RMA slopes and
Tsutakawa and Hewett's (1977) "quick test" for elevations. Rejection of statistical null hy-
potheses requires a probability value less than 5 percent (i.e., P<O.05). Ninety-five percent
confidence intervals for the RMA slopes are computed following the recommendations of
Jolicoeur and Mosimann (1968). If the 95% interval of a slope does not include the value
for geometric similarity, the slope is said to describe significant allometry. Note that sig-
nificant positive allometry of diameters (Biewener's expectation) need not imply "strong"
positive allometry of the magnitude predicted by elastic or static stress similarity criteria
(McMahon, 1975).
 Table 1. Taxa, samples and average (sex-specific) body masses for the two analyses (external diameters and bone mineral analyzer geometry) ~
N

External dimensions "Mineral" geometry External dimensions "Mineral" geometry

Cercopithecine species N Mass (kg) N Mass (kg) Colobine species N Mass (kg) N Mass (kg)
Allenopithecus nigroviridis- m 2 5.5 Colobus angolensis- m 2 9.66
Cercopithecus aethiops- m 4 5.62 5.9 C. angolensis- f 9.1
C. aethiops- f 6 3.57 3.6 C. guereza- m 7 10.4 5 10.5
C. ascanius- m 4 5.45 C. guereza- f 10 8.04 5 8.84
C. ascallius- f 2.48 C. polykomos- f 2 7.91
C. cephus- m 3 3 Piliocolobus badius- m 4 8.91
C. cephus- f 3 2.67 P. badius- f 1 7.15
C. mitis- m 7 7.98 Procolobus verus- m 5 4.77
C. mitis- f 6 3.89 P. verus- f 2 4.57 3.64
C. neglectus- m 2 6.9 Presbytis entellus- m 23.6
C. lIeglectus- f 3 4.25 P. frontata- m 2 5.57
C. diana- m 5.4 P. hosei- f 1 5.57
C. petaurista- m 6 P. rubicunda- m 3 5.68
Miopithecus ralapoin- m 5 1.38 P. rubicunda- f 2 6.14
M. talapoin- f 6 1.12 P. melalophos- m 2 6.69 I 6.86
Erythrocebus patas- f 4.9 P .melalophos- f 3 6.88 I 6.88
Lophocebus albigena- m 4 8.89 8.5 Trachypithecus cristata- m 7 7.09 8 6.78
L. albigena- f 2 6.77 6.9 T. cristata- f 6 5.95 7 5.88
Macaca arctoides- m 2 10.1 T. obscura- m 2 7.58
M. arctoides- f 6.02 I. obscura- f 2 6.6
M. fascicularis- m 10 5.14 8 4.91 T. phayrei- m 7.05
M. fascicularis- f 8 3.18 8 3.18 T. phayrei- f 7.05
M. nemestrina-m 2 9.25 2 9.25 Pygathrix nemaeus- m 2 10.9 2 10.9
M. nemestrina- f 5 5.78 7 5.95 Nasalis larvatus- m 4 21.4 4 21.4
M. mulatta- m 2 9.46 N. larvatus- f 5 10.6 5 10.6
Mandrillus sphinx- m I 45
Papio h. hamadryas- m 4 20.5
~
P. h. cynocephalus- m 17 24.4 12 23.6
P. h. cYllocephalus- f 14 13.4 10 12.3
r...
c
P. h. ursin us- m 2 29 =
P. h. ursillus- f 2 16 ~.,...
Theropithecus gelada- f 13.8 ~
I All body masses are sample-specific. wild-collected values except for MiopitheclIs ta/apoill. Body mass data for M. ta/apoill are taken form Gautier-Hion (1975). It
Body Size and Scaling of Long Bone Geometry in Cercopithecoid Primates 313

2.3. Results
As a group, cercopithecoids exhibit slight, but not significant, positive allometry of
humeral and femoral lengths (Table 2, Figure I) and all external diameters except for the
humeral a-p dimension (Table 2, Figure 2). The a-p humeral diameter scales with signifi-
cant positive allometry (RMA slope = 0.405), but the five other variables have RMA
slopes for which the 95% confidence intervals include 0.333. Most of this pattern of scal-
ing is preserved when cercopithecines and colobines are analyzed separately; humeral a-p
diameter remains the only significant departure from isometry in both clades. Femoral
length slopes are now less than 0.333 in both, but not significantly so. None of the slopes
are significantly different between colobines and cercopithecines, but there are two sig-
nificant differences in elevation. At any given body mass, cercopithecines tend to have
predictably greater m-l humeral diameters, and colobines possess predictably longer
femora (also see Strasser, 1992).

3. SCALING OF HUMERAL AND FEMORAL CROSS-SECTIONAL

GEOMETRY (BONE MINERAL)

3.1. Measurements and (Sub )Sample

Logistical constraints (e.g., availability and portability of bone mineral analyzers;
restrictions on the transport of radioactive materials) prevented us from extending our
analysis from external diameters to cross-sectional geometry for our entire sample, but it
is worth noting that important research in this area has often been based on relatively
small samples (e.g., McMahon, 1975; Alexander et aI., 1979; Biewener, 1982; Selker and
Carter, 1989). Studies such as this one focusing on specimens of known body mass are ex-
tremely rare. The subset of species and specimens that were used for this analysis is also
listed in Table 1. Sex-specific mean masses are again reported by species, and it can be
seen that the ranges (total, cercopithecine, and colobine) are comparable to the full cer-
copithecoid sample used above. The total range is from 3.6 kg (female vervet) to 23.6 kg
(male Darajani baboons); the colobines range from 3.64 kg (olive colobus female) to 21.4
kg (proboscis monkey males). A total of90 adult individuals were included in this part of
the study, 51 cercopithecines and 39 colobines.
We have measured the cross-sectional, midshaft geometry of the humeri and femora
using a noninvasive method based on single photon-absorptiometry (Martin and Burr,
1984). A Norland-Cameron Bone Mineral Analyzer is depicted schematically at the top of
Figure 3 with an elliptical bone section. Each long bone was wrapped in a water-filled, tis-
sue equivalency bag and scanned in the a-p and m-l anatomical planes; an average of three
scans in each direction for each individual was used in all subsequent calculations to re-
duce the possibility of measurement error (Yezerinac et aI., 1992). A hypothetical absorp-
tion curve is seen at the bottom of Figure 3; this curve is integrated to calculate bone
mineral content (BMC), cortical area and second moment of area (SMA). Figure 4 is a
sample a-p scan of a female Darajani baboon that reveals the peaks associated with the
cortices. This method adjusts all cross-sectional geometrical measurements for differential
mineralization and microscopic voids in cortical bone (Schaffier et aI., 1985). It has also
been demonstrated that estimates of long bone strength computed this way are highly cor-
related with actual bending strength (to failure) in primate long bones (r = 0.94; Martin,
1991).
 .....
""
-

Table 2. Scaling of external dimensions of the humerus and femur in cercopithecoid primates I

Cercopithecoidea Cercopithecinae Colobinae Comparisons2

Variable N Slope In intercept r Slope In intercept Slope In intercept Ho: = slope H o, = elevation
Humeral A-P Diameter 22 0.405* 1.4641 0.94 0.401 * 1.4891 0.96 0.444* 1.3609 0.91 accept accept
Humeral M-L Diameter 22 0.354 1.6217 0.94 0.351 1.6672 0.97 0.381 1.5141 0.94 accept Reject cerc > colo
Femoral A-PDiameter 22 0.359 1.6675 0.97 0.353 1.6749 0.97 0.385 1.6191 0.94 accept Accept
Femoral M-L Diameter 22 0.347 1.7133 0.97 0.338 1.7148 0.98 0.365 1.6931 0.94 accept Accept
Humeral Length 22 0.349 4.2947 0.97 0.345 4.3084 0.98 0.380 4.2243 0.92 accept Accept
Femoral Length 22 0.338 4.5096 0.93 0.323 4.4962 0.96 0.306 4.6283 0.87 accept Reject colo> cerc
I Reduced major axis line-fitting of In-In data.
2SIope comparisons based on Clarke's (1980) t-test; elevation comparisons based on "quick test" (Tsutakawa and Hewett, 1977).
'significant positive allometry.
N, "effective sample size" based on variance components (Smith, 1994); total N is 57 sex-specific averages. r, Pearson's correlation coefficient.

~
...r
=
=
~
;;!
sa
~
Body Size and Scaling of Long Bone Geometry in Cercopithecoid Primates 315

6.0

e
E 5.5 -
.c
OQ
c::
0>
...l
';
,of
t
<-
II>
E
~ LoY - 0,349'LnX +4.29468
:: 45
c:: r=0,974
...l

I a 6 Cercopilbecines
4,0 o Co}oblnel
0 2 3 4
Ln Body Mass (kg)
6,0

E
.§ 5.5 -
...co
.c
c::
Q.)
..J
5.0 -
~
0
E
Q)
LL. 4S rz; LnY =033S"LoX +4.5095S
c: r =0_929
...l
b 6 Cercopit becinel
o Colobioe,
2 4
Ln Body Mass (kg)

Figure 1. Bivariate scaling of (a) humerus length and (b) femoral length in cercopithecoids. Cercopithecines are
triangles; colobines are circles. The reduced major axis (RMA) slope of the In-In relationship is provided along
with the equation for the line and the parametric correlation coefficient. Neither relationship departs significantly
from isometry. The separate slopes for cercopithecines and colobines are not significantly different, but colobines
possess predictably longer femora.

The relevant measurements derived from the bone mineral analyzer include a-p and
m-l bone widths (BW), cortical areas (CA), and second or cross-sectional moments of in-
ertia (SMA) in both a-p and m-l planes. Section modulus (Z) was calculated in each plane
as [SMN(O.5 x BW)]; Z is proportional to bending strength. Torsional rigidity is esti-
mated by the polar moment of area (J), computed for nearly circular sections as the sum of
a-p and m-l SMAs. Theoretical indices of compressive and bending strength can also be
calculated (Alexander, 1983):

Compressive Strength = CNBody Mass [cm2/kg]

Bending Strength =Z/(O.5 x Length x Body Mass) [cm2/kg]

Mass was used here as a surrogate for force because the two differ only by the gravita-
tional constant. Theoretical bending strength has also been called an index of athletic abil-
316 w. L. Jungers et al.

E 3.5 -,--- - - -
E
...vv.... 3.0 ~
C
.~
0
0.. 2.5
I
<
'"
....
0
c
v
LnY. =0.3S9' LnX + 1.66748
U. r=0.969
~
a l:, Ccrcopitbccioc,
...:I 1.5 a Colobioes
0 2 3 4
Ln Body Mass (kg)

] 3 .5 I
...
....vV 3.0 t-

E
0'"
...:I
:e
I 25 ~
'"v... 2.0
.,E LnY=0.3S3'LnX+I .62173
r=0.943
::z:: b l:, Ccrcopitbecioc.
c
...:I 1.5 o Colobioes
0 123 4
Ln Body Mass (kg)

Figure 2. Bivariate scaling of external diameters in cercopithecoids: (a) the RMA slope for a-p midshaft femoral
diameter is slightly positively allometric (i.e., > 0.333) but not statistically significantly so; (b) the RMA slope for
m-I humeral midshaft diameter is also not significantly different from isometry, but cercopithecines have predict-
ably larger diameters than colobines.

ity or maneuverability by Alexander (1989). Dimensionality alone indicates that both indi-
ces should decrease with increasing body size unless there is very strong positive al-
lometry of the cross-sectional variables in the numerators (Jungers and Burr, 1994).
Although these indices are static formulations of strength, they can be made more "dy-
namic" by acknowledging that forces within and acting upon animals may scale not in
proportion to body mass, but to body mass raised to some fractional power; if Alexander's
(1980, 1985a; also see Biewener, 1982) findings can be generalized, loads acting on bones
may be approximately proportional to massO. 67 • This is probably accomplished in part
through the size-related postural and behavioral modifications inherent in Biewener's
model of terrestrial biomechanics (e.g., higher duty factors and relatively lower peak sub-
strate forces in larger animals). Accordingly, a second set of indices of theoretical strength
were calculated with massO. 67 in addition to those using massl.o.
Body Size and Scaling of Long Bone Geometry in Cercopithecoid Primates 317

DETECTOR

BMC =LI(x)L:.x
SMA =Lx2I(x)L:.x

Figure 3. Schematic illustration of the bone mineral ana-

lyzer and an elliptical bone section is seen above. The hypo-
thetical absorption curve figured below is integrated to Baseline
provide bone mineral content (BMC). cortical areas and sec- Intensity
ond moments of area (SMA) adjusted for mineral content. Photon
"I" refers here to photon intensity; "NA" is the neutral axis Intensity
or centroid. Less intensity (the downward deflection of the .t.x II-x-I
curve) implies more bone mineral (see Martin and BUIT '" - NA or Centroid
[1984] for a full explanation of the method). Scan Distance, x

3.2. Statistical Methods

Slopes and intercepts of log-log bivariate relationships were again calculated using
RMA line-fitting methods. Cercopithecoids are first analyzed as a group followed by sepa-
rate analyses for colobines and cercopithecines. Due to very uneven sample sizes for dif-
ferent species and sexes, and for maximal comparability to the analysis of external
diameters, sex-specific species averages were again used throughout this part of the study.
In order to control partially for phylogenetic inertia and to reduce inflated degrees of free-
dom, effective sample sizes were estimated by the number of species rather than the
number of sex-specific groups (i.e., 12 rather than 20; see Smith and Jungers, 1997; also
see Felsenstein, 1985). Colobines and cercopithecines are compared again by Clarke's t-
test for RMA slopes and by the "quick test" for elevations.
Spearman rank order correlations (rho) were computed between body size and the
indices of strength to test for "allometry" (sensu Mosimann and James, 1979); a signifi-
cant negative correlation, for example, would indicate that strength decreases predictably
with increasing body size (i.e., "negative allometry"). Locally weighted smoothing or
LOWESS regressions (Cleveland, 1981; HardIe, 1990) were fit to each of these relation-

SAMPLE SCAN'A-P HUMERUS AT MIDSHAFT

O.83E+OI

o
~
a:
z
o
i=
a..
a:
o
Ul
(J)
«
Figure 4. Sample bone mineral scan of a
female Darajani baboon humerus. The ab-
sorption peaks indicate the locations of the O.OOE+OO
anterior and posterior cortices. O.OOE+OO BONE WIDTH (CM) O.l4E+OI
318 w. L. Jungers et al.

ships ("tension" or window width was set at 0.7, a relatively stiff factor). This nonpara-
metric approach for detecting trends within scatter plots assumes nothing about mono-
tonicity or probable shape of the underlying distribution. A 2-by-2 matrix was created by
tabulating the number of colobines and cercopithecines above and below each LOWESS
curve, and the significance of clade differences was assessed by the G-statistic for inde-
pendence in a row-by-column contingency table.

3.3. Results
Table 3 summarizes the information about scaling of cross-sectional geometric vari-
ables and bone lengths in our subset of Old World monkeys. Although slope values for hu-
meral and femoral lengths in cercopithecoids as a group are again all greater than the null
isometric expectation of 0.333, they are not significantly greater. The colobine results for
these two variables are very similar to the larger sample considered above, but the cer-
copithecine slopes are higher within the subset, and the femoral length slope is greater
than the humeral (although neither are significantly different from 0.333). Slopes for hu-
meral cortical area are greater than the isometric value of 0.67 regardless of grouping
(lowest in colobines), but this positive allometry is not statistically significant (Figure 5a).
Femoral cortical area (Figure 5b), however, scales in a significantly positively allometric
fashion for cercopithecoids and for cercopithecines (but not for colobines considered
alone despite the 0.738 slope value). Humeral J (Figure 6a) is much like humeral CA
(positive allometry with slopes greater than the isometric expectation of 1.333, but with
confidence limits that include this isometric value), and femoral J (Figure 6b) scaling re-
calls that seen for femoral CA (significant positive allometry in cercopithecoids and cer-
copithecines, but not in colobines). Humeral SMAs and Zs are slightly positively
allometric in both a-p and m-l planes for cercopithecoids as a group (i.e., slope values ex-
ceed isometric values, but not significantly so). Colobines exhibit significant positive al-
lometry for humeral a-p SMA, and cercopithecines have humeral Zs that are significantly
allometric in both anatomical planes. Femoral SMAs and Zs are all significantly allomet-
ric for cercopithecoids and cercopithecines, but not for colobines.
Despite these differences between cercopithecines and colobines, with the former
exhibiting more frequent significant departures from isometry in the direction expected by
Biewener's model, none of the slopes are significantly different from each other (i.e., the
null hypothesis of slope equality is accepted for all 14 variables in Table 3; P>0.05). Nev-
ertheless, cercopithecines have predictably greater slope values for all 12 geometric vari-
ables. There are also several significant differences in elevations between the two clades.
Across the range of body sizes, cercopithecines have predictably larger humeral cortical
areas, humeral polar moments of area, humeral second moments of area, and humeral sec-
tion moduli; colobines have predictably longer femora at any given body size.
Despite the pervasive trend for geometrical properties to exceed isometric values,
and often significantly so, this positive allometry alone is clearly not sufficient to main-
tain comparable levels of theoretical strength as body size increases. If force acting on
bones is indeed proportional to body mass, then there is a predictable decrement in all in-
dices of bone strength at larger body sizes (Table 4). All six rank order correlations are
significantly negative. Figures 7 and 8 illustrate this finding for both compressive and
bending indices of strength. However, if forces acting on bones are really proportional to
massO. 67 , then a rather different picture emerges. None of the indices now have a signifi-
cant negative relationship. Rather, the three humeral indices show no significant relation-
ship with size at all, and femoral strengths actually increase with increasing body size.
 i
~
~ .
.,
=-f(l=
e!.
Ej"
JIQ

Table 3. Scaling of cross-sectional geometry (mineral) of the humerus and femur in cercopithecoid primates I So
~
JIQ
=
Cercopithecoidea Cercopithecinae Colobinae Comparisons
Variable N Slope In intercept Slope In intercept Slope In intercept H,,: = slope
f
H,,: = elevation ;
Humeral Cortical Area 12 0.732 -3.1707 0.93 0.767 -3.1322 0.98 0.683 -3.1724 0.96 accept Reject cerc >colo C"l
g
Femoral Cortical Area 12 0.776* -2.9767 0.98 0.800* -3.0221 0.99 0.738 -2.8963 0.97 accept Accept EI
Humeral J 12 1.491 -6.5439 0.94 1.532 -6.4020 0.98 1.438 -6.6531 0.99 accept Reject cerc >colo
12 1.536* -6.1505 0.99 1.590* -6.2218 0.99 1.472 -6.0491
Femoral J 0.99 accept Accept
!;.
Humeral A-P SMA 12 1.551 -7.3865 0.94 1.566 -7.1797 0.98 1.536* -7.5933 0.99 accept Reject cerc >colo n
Humeral M-L SMA 12 1.456 -7.1460 0.94 1.524 -7.0707 0.98 1.362 -7.1554 0.98 accept Reject cerc >colo
Femoral A-P SMA 12 1.547* -6.8953 0.98 1.618* -6.9846 0.99 1.461 -6.7658 0.99 accept Accept
j
Femoral M-L SMA 12 1.531 * -6.8072 0.99 1.567* -6.8591 0.99 1.487 -6.7318 0.99 accept Accept ~
~
Humeral A-P Z 12 1.161 -5.8731 0.94 1.191* -5.7610 0.98 1.127 -5.9694 0.99 accept Reject cerc >colo Q
Q:
Humeral M-L Z 12 1.110 -5.7340 0.94 1.174* -5.7011 0.98 1.018 -5.6969 0.98 accept Reject cerc >colo "CI
Femoral A-P Z 12 1.175* -5.5254 0.99 1.238* -5.6191 0.99 1.090 -5.3749 0.99 accept Accept :::l.
Femoral M-L Z 12 1.166* -5.4670 0.99 1.210* -5.5373 0.99 1.105 -5.3550 0.99 accept Accept .,EI
Humeral Length 12 0.372 1.9711 0.95 0.365 1.9905 0.98 0.386 1.9335 0.91 accept Accept ~
Femoral Length 12 0.363 2.1850 0.94 0.377 2.1309 0.97 0.303 2.3412 0.93 accept Reject colo >cerc
1Reduced major axis line-fitting ofln-In data. (Same conventions as in Table 2.)
N, "effective sample size" based on number of species rather than number of sex-specific averages (20).

Co>
:;;
320 w. L. Jungers et al.

o r---------,----------r--------~r_------__.

11 Cercopithecines
• Colobines

<
U
===-2.0
...:l
LnY= 0.732 * LnX-3.17065
r = 0.929
-3.0 '--________---'-__________.1......_ _ _ _ _ _ _ _- - ' -_ _ _ _ _ _---'------'
o 1.0 2.0 3.0 4.0
Ln Body Mass (kg)
o r---------r---------~--------~------~

11 Cercopithecines
• Colobines

----
M
5 -1.0
~
'-'

<
U
=-2.0
~

...:l

LnY= 0.776 * LnX-2.97666

r= 0.985
-3.0 '--_ _ _ _--'-_ _ _ _ _-'-_ _ _ _ _-'-_ _ _--'----'
o 1.0 2.0 3.0 4.0
Ln Body Mass (kg)
Figure S. Bivariate scaling of (a) humeral cortical area and (b) femoral cortical area in the subset of cercopithe-
coid monkeys. Cercopithecines are triangles; colobines are circles. The RMA slope for the humerus is slightly al-
lometric (>0.67), but not significantly so; however, cercopithecines have predictably more cortical area than
colobines. The RMA slope for the femur is significantly positively allometric, and there are no predictable c1ade-
specific differences in cortical area (but recall that colobines have longer femora).

Clearly, how loads acting on bones scale is of critical importance, and if animals can
somehow reduce peak forces at larger body sizes, then decreases in strength can be
avoided with only the modest structural allometries documented here.
The scaling trends seen above, especially the series of significant elevational differ-
ences between the two clades, result in some consistent differences in theoretical strengths
between cercopithecines and colobines. More specifically, G-tests of the distribution of
taxa above and below LOWESS curves disclose significant differences between clades for
all indices of strength except femoral compressive strength. In all other comparisons, cer-
copithecines possess predictably stronger bones than their colobine counterparts. These
trends are easily discernible in Figures 7 and 8. This difference holds no matter which esti-
mate of force is used.
Body Size and Scaling of Long Bone Geometry in Cercopithecoid Primates 321

-1.0

--
.-------"""T"-----,...-----~----~

.".
S -2.0

--~

_~ -3.0

E~
S
:::I -4.0

===-5.0
...:l LnY= 1.491 * LnX-6.54387
r = 0.943
-6.0 L--_ _ _ _..L.._ _ _ _...l..._ _ _ _----1_ _ _----1----J a
o 1~ U ~ ~

Ln Body Mass (kg)

-1.0 .-------"""T"-----~----"""T"-~--~

Cercopithecines
/j.
• Colobines
-2.0

--o
~

~ -3.0

EQ

~ -4.0
~
=
...:l -5.0 LnY= 1.536* LnX-6.15048
r= 0.992
b
-6.0 0
1.0 2.0 3.0 4.0
Ln Body Mass (kg)
Figure 6. Bivariate scaling of (a) humeral polar moment of area and (b) femoral polar moment of area in Old
World monkeys. The RMA slope for the humerus is not significantly different from isometry, but cercopithecines
possess predictably larger polar moments than colobines. The femoral slope is significantly positively allometric
(> \.33), but there are no clade-specific differences in elevations.

4. DISCUSSION

4.1. General Trends and Evaluation of Model Expectations

Although there are differences in the details of how external diameters and cross-
sectional geometry scale within cercopithecoid long bones (see below), the results of both
types of analyses are consistent with and corroborate the structural component of
Biewener's multifactorial model for "terrestrial" locomotion and long bone allometry. Pri-
mates, including cercopithecoids, are different from most other mammals in various as-
pects of their anatomy and quadrupedalism: they have longer and more robust limb bones
(Alexander, 1985b; Kimura, 1991; Polk et aI., 1997), their long bones tend to be less
322 W. L. Jungers et al.

Table 4. Spearman rank order correlations (rho) between body mass and theoretical bone
strength in cercopithecoid primates I

Variable Force oc Mass Sig. Force oc Mass 0.67 Sig.

Humeral compressive strength -0.71 P<O.OOI -0.03 ns, P>0.9
Femoral compressive strength -0.82 P<O.OOI 0.52 P<O.OI
Humeral a-p bending strength -0.49 P<0.03 0.11 ns, P>0.6
Humeral m-I bending strength -0.69 P<O.OOI -0.01 ns, P>0.9
Femoral a-p bending strength -0.63 P<0.005 0.52 P<0.05
Femoral m-I bending strength -0.62 P<0.005 0.62 P<0.005
I Mass was used as "force" here without multiplying by the gravitational constant; load-sharing by the upper
and lower limb was assumed to be roughly equal under both force regimes.
Sig., significance; ns, non-significant.

0.04 r----.-----.----.-----.----~

A Cercopithecines
• Colobines

rJ:J 0.03 A
rJ:J
<
~
-.. A
<
U
A
0.02
•
rho = -0.708
•
p < 0.001
0.01
a
0 5 10 15 20 25
BODY MASS (kg)

0.045r----,------.------.-----,--------,

0.040
• A Cercopithecines
• Colobines
r.r.J
r.r.J
•
~ 0.035

~
0.030
• •
0.025
rho = -0.815
p < 0.001
0.020'--_ _ _-L-_ _ _--'-_ _ _--'-_ _ _--'-_ _---'c...:=...J
o 5 10 15 20 25
BODY MASS (kg)

Figure 7. Theoretical compressive strength of the (a) midshaft humerus and (b) midshaft femur as a function of
body mass in cercopithecoids. Strength decreases in both elements as size increases; the LOWESS curve follows
this decline. Cercopithecines tend to have stronger humeri at any given body size than do colobines. Force acting
on the bones is assumed to be proportional to body mass, but changes in posture and locomotor behavior probably
also change allometrically to preserve adequate safety factors.
Body Size and Scaling of Long Bone Geometry in Cercopithecoid Primates 323

-
0.09
~
=>
A A Cercopithecines
~ 0.08 • Colobmes
==
E-<

'"
Z
~
0.07
...;!
III
d 0.06 A
*
(I:l
(I:l
•
-:a
-< 0.05
6
=-. 0.04
rho = -0.495
• •
N
0.03
p < 0.005
• a
0 5 10 15 20 25
BODY MASS (kg)

-
0.08
~
=>
~
A Cercopithecines
• Colobines
~ 0.07 A
'"
z~
...;!
A
A
A
III
d 0.06 -
*(I:l
en
•
-
~
=-.
~
N
0.05
rho = -0.632
• •
p<0.OO5
0.04
b
0 5 10 15 20 25
BODY MASS (kg)
Figure 8. Theoretical a-p bending strength of the (a) humeral midshaft and (b) femoral midshaft in Old World
monkeys. Strength decreases as a function of body size; the LOWESS curves track this decline in bending
strength. Behavioral changes presumably compensate for this apparent lack of functional equivalence. Note that
cercopithecines have predictably stronger bones than colobines, and this may reflect clade-specific differences in
the compliance of their habitual substrates (i.e., stiffer terrestrial substrates are experienced by many cercopi-
thecines).

curved (Swartz, 1990), they tend to take longer strides with lower stride frequencies (Al-
exander and Maloiy, 1984), they support more of their body weight on their hindlimbs
(Kimura et aI., 1979; Reynolds, 1985; Demes et aI., 1994), and their footfall sequences are
unusual for mammals (Hildebrand, 1967; Shapiro et aI., 1997). Many of the cercopithe-
coids are also better characterized as arboreal quadrupeds rather than as terrestrial quadru-
peds (Rowe, 1996), and the mechanical properties of an arboreal substrate differ
considerably from those experienced on the ground (Alexander, 1991; Cant, 1994;
Schmitt, 1994). Nevertheless, the pervasive finding here for cercopithecoids was slightly
positive allometric scaling of supporting dimensions of the long bones, only some of
which were statistically significant departures from isometry. This closely matches
 324 W. L. Jungers et al.

Biewener's general conclusions for quadrupedal species within the size range of Old
World monkeys (Biewener, 1990).
The uniformly negative allometry of theoretical bone strength (with forces assumed
to be proportional to body mass or weight) in cercopithecoids indicates that this slight de-
gree of allometric reinforcement is insufficient to maintain mechanical equivalence at
larger body sizes. In other words, long bone allometry alone is an inadequate solution to
size-related increases in loads and stresses in this group; hence, it would be illogical and
simply wrong to view these allometries as evidence of "functional equivalence". Recall
that Biewener's model predicts other types of size-required changes to maintain reason-
able safety factors, and these include changes in limb posture and muscular mechanical
advantages. We currently possess little more than anecdotal evidence for primates, but
there is reason to believe that larger primate quadrupeds do move with more extended
limb joints (see Figure 9, adapted from Fleagle, 1988); this could rotate the substrate reac-
tion force acting on each limb more into line with the limb's long axis and thereby reduce
its flexing moment arm. Bending moments would be reduced as a consequence, and ex-
tended limbs could then support weight-related loads with less muscular effort, especially
if bony lever arms of the extensor musculature are also longer and/or more favorably ori-
ented in larger species (e.g., re-oriented and relatively longer olecranon processes for the
triceps brachii muscle mass; see Jolly, 1972; Rose, 1993). Although much more data are
needed, some of these suspicions have been confirmed recently for cercopithecine fore-
limbs by Schmitt (1995), and cercopithecoids therefore appear to present the full package
of expectations developed by Biewener and others for truly terrestrial mammals. If peak
forces per unit body mass are somehow reduced at larger cercopithecoid body sizes, then
the more "dynamic" version of the strength indices (the third column in Table 4) might be
more realistic; as such, they would predict that bony stresses and strains do not increase
much, if at all, with increasing body size due to a combination of structural and behavioral
modifications (Biewener, 1982, 1993; Rubin, 1984; Rubin and Lanyon, 1984).
Limb postures may differ between small and large monkeys, but it is not clear if
overall "athletic ability", agility and/or dynamic maneuverability have also been sacrificed
to any great degree by larger cercopithecoids. For example, in McGraw's (1996) study of
Tai Forest (Ivory Coast) cercopithecoids, body size alone did not predict well the observed
frequencies of climbing versus leaping behaviors; e.g., the most frequent climber was one
of the smallest monkeys and leaping was correlated more with phylogeny than size (colo-

a _____lllL_ " " ' -_ _ __

Figure 9. Cartoon contrast between a smaller, more flexed monkey on the left versus a larger, more extended
monkey on the right. These are some the predicted size-related differences in behavior that are believed to com-
pensate for only slightly positive allometry of supporting dimensions. Adapted from Fleagle (1988).
Body Size and Scaling of Long Bone Geometry in Cercopithecoid Primates 325

o.os
D

...
.=
t>I)
s:: 0.04
...~
rJ'l

.::
G)

'"'"
G)
Figure 10. Theoretical compressive ~ 0.03
strength of the midshaft humerus in Neot- eo
ropical and Old World monkeys (open U
squares) and African apes (filled squares).
...'"
::I
The addition of smaller monkeys such as
the squirrel monkey (extreme left open
6 0.02
::I
squares) and the larger apes has little im- ::t:
pact on the size-related trends docu- • gorilla
mented for cercopithecoids alone. The
rapid decline in theoretical strength levels 0.01
off by approximately 20 kg; the LOWESS o 10 20 30 40 so 60 70
curve follows this decay graphically.
Body Mass (kg)

bines were the most frequent leapers at Tai). Comparable data on the largest cercopithe-
coids would be very interesting, but are currently lacking.
The negatively allometric pattern of theoretical strengths documented for cercopithe-
coids remains intact if New World monkeys, including small squirrel monkeys, are added to
the sample (see Jungers and Burr, 1994, for details of the platyrrhine sample). The largest
quadrupedal anthropoids are African apes, and even if chimpanzees and gorillas are added
to the overall picture with platyrrhines and cercopithecoids (Figure 10), this negatively al-
lometric pattern for strength still persists, although it does appear to level off quickly after
approximately 20 kg. Comparable size-related decrements in the theoretical bone strength
of prosimian primates have also been documented (Demes and Jungers, 1993), and
Biewener's model probably obtains for all primates whether they are quadrupedal or not. In
fact, if Asian apes (gibbons, siamang and orang-utans) are considered together as a size-
graded group of suspensory primates (Figure II), their theoretical strengths also decrease
and can be superimposed onto the quadrupedal trends. It is also the case that orang-utan lo-
comotion is considerably less dynamic than that practiced by lesser apes, so functional
safety factors may differ little among them despite dramatic differences in body size.

4.2. External Dimensions versus Cross-Sectional Geometry

Only the humeral a-p diameter was significantly positively allometric in the analysis
of external dimensions, whereas significant departures from isometry were more common
for the cross-sectional data collected with the bone mineral analyzer. There are several
possible, and not mutually exclusive, reasons for this difference. Although information
drawn from external diameters can be and often is correlated with geometrical data ex-
tracted from cross-sections (e.g., Jungers and Minns, 1979; Ruff, 1989), a more complete
picture of bone distribution is provided by approaches that take into account the internal
boundaries of cortical bone (Biewener, 1982; Ruff, 1989; Ruff and Runestad, 1992). In
human evolution, for example (Ruff et aI., 1993; Grine et aI., 1995), differences in the
amount of cortical bone and corresponding diaphyseal strengths are profoundly affected
326 w. L. Jungers et al.

0.035

..=
co1'1 0.030 °
0 lesser apes
...
0) 0
0
cil ~
.::
0)
0.025
rIl

...c:>.
rIl
0)

e0 Ql
0.020
Co)
°female orangs
<ii...
e::s
0)

0.015
II:
Figure 11. Theoretical compressive
male orangs strength of the midshaft humerus in
0.010 L -_ _--'--_ _...J...._ _---'-_ _ _'----=o----'
lesser apes and orang-utans. The size-re-
lated decrease in strength seen in these
o 20 40 60 80 100
suspensory primates mirrors that seen in
Body Mass (kg) more pronograde quadrupeds.

by significant variation in medullary diameters-information simply unavailable to the

most precise calipers. Theoretical and empirical measures of bone strength are clearly
more closely related to sectional moduli and cortical areas than they are to external diame-
ters alone. As such, the allometric trends in cross-sectional geometry are probably tracking
the most relevant biomechanical signals (i.e., cross-sectional variables could be allometric
even if external dimensions are consistently closer to isometry). It should also be noted
that the method used here that weights cortical bone distribution by mineral content (Mar-
tin and Burr, 1984; Gilbert et ai., 1989; Ruff, 1989) could also contribute to observed dif-
ferences from external diameter scaling. Finally, the cross-sectional analysis presented
here is based only on a subset of the sample used for external diameters, and some of the
observed differences could simply reflect these different sampling strategies (e.g., the ab-
solute ranges of body mass are slightly different). Data and inferences drawn from exter-
nal diameters are certainly valuable, and many of the general conclusions offered here
based on midshaft diameters are similar to those derived from more detailed cross-sec-
tional geometry. Given a choice, however, we would still opt for the cross-sectional data
whenever a more complete picture of long bone biomechanics is desired.

4.3. Clade-Specific Differences and Substrate Influences

For both cross-sectional variables and external diameters, no significant differences
were found between colobines and cercopithecines in slope values. Oddly, however, cer-
copithecines had greater slopes for all geometric variables, but colobines had greater
slopes for all external diameters. Significant differences were found for several elevations,
with cercopithecines having more bone in cross-section than colobines of comparable
body size for all humeral variables. The finding that colobines had predictably longer
femora than cercopithecines is perhaps not surprising given the greater incidence of leap-
ing in colobines (Fleagle, 1988; Strasser, 1992; McGraw, 1996). With the exception of
their femoral compressive strength, cercopithecines had predictably stronger long bones
than colobines. One might argue that the longer femora of colobines contributes to their
Body Size and Scaling of Long Bone Geometry in Cercopithecoid Primates 327

lower bending strengths for this element (since the load arm is proportional to bone
length), but this argument would fail to account for the same pattern in humeral compres-
sive and bending strengths. Why, then, are cercopithecine long bones structurally rein-
forced in comparison to those of colobines?
Arboreal and terrestrial substrates have quite different mechanical properties. In par-
ticular, branches and trees are usually more compliant than the ground, and this difference
should impact on the mechanics of locomotion (Alexander, 1991; Cant, 1994). In fact,
Schmitt (1994, 1995, this volume) has documented reductions in peak forces acting on
primates moving on branches compared to their locomotion on the ground; substrate reac-
tion force directions are also predictably different in the mediolateral plane. Cercopi-
thecines as a group are more terrestrial than colo bines, and it seems reasonable to connect
differences in habitual substrate stiffness to differences in long bone geometry and
strength. In other words, despite their propensity for leaping from and to arboreal sup-
ports, quadrupedal colobines may experience relatively lower peak stresses because of
greater substrate compliance, and hence require less bone in cross-section to resist bend-
ing moments and axial loads. The greater commitment to terrestriality seen in cercopi-
thecines requires them to move more frequently on stiffer substrates; higher peak forces
may be one of the prices of this substrate preference, and stronger long bones may be a
necessary corollary of this habitat difference (also see Burr et aI., 1989).
We now know much about the scaling of bone geometry (although its developmental
basis remains highly theoretical [van der Meulen and Carter, 1995]), but the study of how
substrate mechanics are connected to primate locomotor biomechanics and morphology is
still very much in its infancy (Alexander, 1991; Crompton et aI., 1993; Cant, 1994;
Schmitt, 1995; but also see McMahon [1984] on compliant running tracks, and Demes et
al. [1995] on compliant supports in leaping prosimians). We also need a great deal more
and better focused information about the allometry of primate locomotion itself, especially
quantitative data on size-correlated differences in kinematics and kinetics.

ACKNOWLEDGMENTS

We thank the editors for inviting us to contribute to this volume and for the opportu-
nity to celebrate the memory of Warren Kinzey. We gratefully acknowledge the assistance
of the curators of valuable skeletal material for giving us access to specimens in their charge
(at the Smithsonian Institution, the American Museum of Natural History, the Museum of
Comparative Zoology, the Field Museum of Natural History, the University of Texas, the
Tervuren Museum, the Powell-Cotton Museum, the British Museum of Natural History, and
the University of Gottingen). R. Susman and S. McGraw kindly allowed us to measure
specimens in their private collections. We also thank Luci Betti-Nash for her skillful art-
work. Discussions with Brigitte Demes about bone biomechanics and with Eric Delson
about cercopithecoid body size were very helpful. We also offer our sincere thanks to Beth
Strasser, Steve Churchill, and two of the three anonymous reviewers for their careful read-
ing of our paper and their valuable suggestions. This research was supported by NSF Grants
BNS 8606781 and SBR 9507078, the Leakey Foundation, and the Boise Fund.

REFERENCES
Aiello LC (1981) The allometry of primate body proportions. Symp. Zool. Soc. Lond. 48:331-358.
328 W. L. Jungers et al.

Alexander R McN (1980) Forces in animal joints. Engineering in Medicine 9:93-97.

Alexander R McN (1983) On the massive legs of a Moa (Pachyornis elephantopus, Dinornithes). J. Zoo!., Lond.
20/:363-376.
Alexander R McN (1985a) The maximum forces exerted by animals. J. expo BioI. //5:231-238.
Alexander R McN (1985b) Body size and limb design in primates and other mammals. In WL Jungers (ed.): Size
and Scaling in Primate Biology. New York: Plenum Press, pp. 337-343.
Alexander R McN (1988) Elastic Mechanisms in Animal Movement. Cambridge: Cambridge University Press.
Alexander R McN (1989) Mechanics offossil vertebrates. J. Geol. Soc. Lond. 146:41-52.
Alexander R McN (1991) Elastic mechanisms in primate locomotion. Z. Morph. Anthropol. 78:315-320.
Alexander R McN, Jayes AS, Maloiy GMO, and Wathuta EM (1979) Allometry of the limb bones of mammals
from shrews (Sorex) to elephant (Loxodonta). J. Zool., Lond. 189:305-314.
Alexander R McN, and Maloiy GMO (1984) Stride lengths and stride frequencies of primates. J. Zool., Lond.
202:577-582.
Bertram JEA, and Biewener AA (1990) Differential scaling of the long bones in the terrestrial Carnivora and other
mammals. J. Morph. 204:157-169.
Biewener AA (1982) Bone strength in small mammals and bipedal birds: Do safety factors change with body size?
J. expo BioI. 98:289-301.
Biewener AA (1989) Mammalian terrestrial locomotion and size. BioSci. 39:77fr-.783.
Biewener AA (1990) Biomechanics of mammalian terrestrial locomotion. Science 250: I 097-11 03.
Biewener AA (1991) Musculoskeletal design in relation to body size. J. Biomech. 24 (suppl. I): 19-29.
Biewener AA (1993) Safety factors in bone strength. Calcif. Tissue Int. 53 (suppl.):S68-S74.
Biknevicius AR (1993) Biomechanical scaling of limb bones and differential limb use in caviomorph rodents. J.
Mammal. 74:95-\07.
Burr DB, RuffCB, and Johnson C (1989) Structural adaptations of the femur and humerus to arboreal and terres-
trial environments in three species of macaque. Am. J. Phys. Anthropol. 79:357-368.
Cant JGH (1994) Positional behavior of arboreal primates and habitat compliance. In B Thierry (ed.): Current Pri-
matology, Vol. I. Ecology and Evolution. Strasbourg: Universite Louis Pasteur, pp. 187-193.
Casinos A, Bou J, Castiella MJ, and Viladiu C (1986) On the allometry of long bones in dogs. J Morph.
190:73-79.
Clarke MRB (1980) The reduced major axis of a bivariate sample. Biometrika 67:441-446.
Cleveland WS (1981) LOWESS: A program for smoothing scatterplots by robust locally weighted regression.
Amer. Statistician 35:54.
Crompton RH, Sellers WI, and Gunther MM (1993) Energetic efficiency and ecology as selective forces in the sal-
tatory adaptation of pro simian primates. Proc. Roy. Soc. Lond.(B) 254:41-45.
Demes B, and Jungers WL (1993) Long bone cross-sectional dimensions, locomotor adaptations and body size in
prosimian primates. J Hum. Evol. 25:57-74.
Demes B, Jungers WL, Gross TS, and Fleagle JG (1995) Kinetics of leaping primates: Influence of substrate ori-
entation and compliance. Am. J. Phys. Anthropol. 96:419-430.
Demes B, Larson SG, Stern JT Jr., Jungers WL, Biknevicius AR, and Schmitt D (1994) The kinetics of primate
quadrupedalism: "hindlimb drive" reconsidered. J. Hum. Evol. 26:353-374.
Demes B, Jungers WL, and Selpien K (1991) Body size, locomotion, and long bone cross-sectional geometry in
indriid primates. Am. J. Phys. Anthropol. 86:537-547.
Economos AC (1983) Elastic and/or geometric similarity in mammalian design? J. Theor. BioI. 103: 167-172.
Felsenstein J (J 985) Phylogenies and the comparative method. Am. Natur. 125: 1-15.
Fleagle JG (1988) Primate Adaptation and Evolution. New York: Academic Press.
Gautier-Hion A (1975) Dimorphisme sexuel et organisation sociale chez les cercopithecines forestiers Africains.
Mammalia 39:365-374.
Gilbert JA, Skrzynski MC, and Lester GE (1989) Cross-sectional moment of inertia of the distal radius from ab-
sorptiometric data. J. Biomech. 22:751-754.
Gould SJ (1975) On the scaling of tooth size in mammals. Am. Zool. 15:351-362.
Grine FE, Jungers WL, Tobias PV, and Pearson OM (1995) Fossil Homo femur from Berg Aukas, Northern Na-
mibia. Am. J. Phys. Anthropol. 97:151-185.
Hardie W (1990) Applied Nonparametric Regression. Cambridge: Cambridge University Press.
Hildebrand M (1967) Symmetrical gaits of primates. Am. J. Phys. Anthropol. 26: 119-130.
Hokkanen JEI (1986) Notes concerning elastic similarity. J. Theor. BioI. 120:499-501.
Jolicoeur P, and Mosimann JE (1968) Intervalles de confiance pour la pente de I'axe majeur d'une distribution
norrnale bidimensionnelle. Biom.-Praxim. 9: 121-140.
Jolly CJ (1972) The classification and natural history of Theropithecus (Simopithecus) (Andrews, 1916), baboons
of the African Plio-Pleistocene. Bull. Brit. Mus. (Nat. Hist.) Geol. Ser. 22: 1-123.
Body Size and Scaling of Long Bone Geometry in Cercopithecoid Primates 329

Jungers WL (1984) Aspects of size and scaling in primate biology with special reference to the locomotor skele-
ton. Yrbk. Phys. Anthropol. 27:73-97.
Jungers WL, and Burr DB (1994) Body size, long bone geometry and locomotion in quadrupedal monkeys. Z.
Morph. Anthropol. 80:89-97.
Jungers WL, Falsetti AB, and Wall CE (1995) Shape, relative size, and size-adjustments in morphometries. Yrbk.
Phys. Anthropol. 38:137-161.
Jungers WL and Minns RJ (1979) Computed tomography and biomechanical analysis of fossil long bones. Am. J.
Phys. Anthropol. 50:285-290.
Kimura T (1991) Long and robust limb bones of primates. In A Ehara (ed.): Primatology Today. Amsterdam: El-
sevier Science Publishers, pp. 495-498.
Kimura T, Okada M, and Ishida H (1979) Kinesiological characteristics of primate walking: its significance in hu-
man walking. In M Morbeck, H Preuschoft, and N Gomberg (eds.): Environment, Behavior and Morphol-
ogy: Dynamic Interactions in Primates. New York: Fischer, pp. 297-311.
Martin RB (\991) Determinants of the mechanical properties of bones. 1. Biomech. 24 (suppl. 1):79--88.
Martin RB, and Burr DB (\ 984) Non-invasive measurement of long bone cross-sectional moments of inertia by
photon absorptiometry. 1. Biomech. 17:195-201.
Martin RD (\ 993) Allometric aspects of skull morphology in Theropithecus. In NG Jablonski (ed.): Theropithecus.
The Rise and Fall of a Primate Genus. Cambridge: Cambridge University Press, pp. 273-298.
McArdle BH (1988) The structural relationship: regression in biology. Can. J. Zool. 66:2329-2339.
McGraw WS (\ 996) The positional behavior and habitat use of six sympatric monkeys in the Tai Forest, Ivory
Coast. Ph.D Dissertation, State University of New York at Stony Brook.
McMahon TA (\ 973) Size and shape in biology. Science 179: 1201-1204.
McMahon TA (\975) Using body size to understand the structural design of animals: quadrupedal locomotion. 1.
Appl. Physiol. 39:619-627.
McMahon TA (1984) Muscles, Reflexes, and Locomotion. Princeton: Princeton University Press.
Mosimann JE, and James FC (1979) New statistical methods for allometry with applications to Florida red-winged
blackbirds. Evolution 23:444-459.
Polk JD, Demes B, Jungers WL, Heinrich RE, Biknevicius AR, and Runestad JA (I 997) Cross-sectional properties
of primate and nonprimate limb bones. Am. J. Phys. Anthropol., suppl. 24:188.
Prange HD (\977) The scaling and mechanics of arthropod exoskeletons. In TJ Pedley (ed.): Scale Effects in Ani-
mal Locomotion. London: Academic Press, pp. 169-181.
Prothero J (1992) Scaling of bodily proportions in adult terrestrial mammals. Am. J. Physiol. 262:R492-R503.
Rayner JMV (1985) Linear relationships in biomechanics: the statistics of scaling functions. J. Zool., Lond.
206:415-439.
Reynolds TR (1985) Stress on the limbs of quadrupedal primates. Am. 1. Phys. Anthropol. 67:351-362.
Ricker WE (\ 984) Computation and uses of central trend lines. Can. J. Zool. 62: 1897-1905.
Rose MD (\993) Functional anatomy of the elbow and forearm in primates. In DL Gebo (ed.): PostcranialAdapta-
tion in Nonhuman Primates. DeKalb: Northern I\linois University Press, pp. 70-95.
Rowe N (1996) The Pictorial Guide to the Living Primates. East Hampton: Pogonias Press.
Rubin CT (\ 984) Skeletal strain and the functional significance of bone architecture. Calcif. Tissue Int. 36:S 11-
S18.
Rubin CT, and Lanyon LE (1984) Dynamic strain similarity in vertebrates: an alternative to allometric limb bone
scaling. J. Theor. BioI. 107:321-327.
RuffCB (1987) Structural allometry of the femur and tibia in Hominoidea and Macaca. Folia Primatol. 48:9-49.
RuffCB (1989) New approaches to structural evolution of limb bones in primates. Folia Primatol. 53:142-159.
Ruff CB and Runestad JA (1992) Primate limb bone structural adaptations. Ann. Rev. Anthropol. 21 :407-433.
RuffCB, Trinkaus E, Walker A, and Larsen CP (1993) Postcranial robusticity in Homo. I: Temporal trends and
mechanical interpretation. Am. J. Phys. Anthropol. 91:21-53.
SAS (\985) SAS Users's Guide: Statistics, Version 5. Cary: SAS Institute Inc.
Schaffier MB, Burr DB, Jungers WL, and RuffCB (1985) Structural and mechanical indicators of limb specializa-
tion in primates. Folia Primatol. 45:61-75.
Schmitt D (1994) Forelimb mechanics as a function of substrate type during quadrupedalism in two anthropoid
primates. 1. Hum. Evol. 26:441-458.
Schmitt D (1995) A kinematic and kinetic analysis of forelimb use during arboreal and terrestrial quadrupedal ism
in Old World monkeys. Ph.D Dissertation, State University of New York at Stony Brook.
Schmitt D (this volume) Forelimb mechanics during arboreal and terrestrial quadrupedalism in primates. In E
Strasser, JG Fleagle, AL Rosenberger, and HM McHenry (eds.) Primate Locomotion: Recent Advances.
New York: Plenum Press, pp. 175-200.
330 w. L. JungersetaL

Schultz AH (1970) The comparative uniformity of the Cercopithecoidea. In J Napier and P Napier (eds.): Old
World Monkeys. New York: Academic Press, pp. 39-51.
Selker F, and Carter DR (1989) Scaling of long bone fracture strength with animal mass. J. Biomech.
22:1175-1183.
Shapiro LJ, Anapol FC, and Jungers WL (1997) Interlimb coordination, gait, and neural control of quadrupedal ism
in chimpanzees. Am. J. Phys. Anthropo\. 102:177-186.
Smith RJ (1980) Rethinking allometry. J. Theor. Bio. 87:97-111.
Smith RJ (1984) Determination of relative size: the "criterion of subtraction" problem in allometry. J. Theor. Bio.
108:131-142.
Smith RJ (1994) Degrees of freedom in interspecific allometry: an adjustment for the effects of phylogenetic con-
straint. Am. J. Phys. Anthropo\. 93:95-107.
Smith RJ, and Jungers WL (1997) Body mass in comparative primatology. J. Hum. Evol. 32:523--559.
Strasser E (1992) Hindlimb proportions, allometry, and biomechanics in Old World monkeys (Primates, Cercopi-
thecidae). Am. J. Phys. Anthropo\. 87:187-213.
Strasser E and Delson E (1987) Cladistic analysis of cercopithecid relationships. J. Hum. Evo\. 16:81--99.
Swartz SM (1990) Curvature of the forelimb bones of anthropoid primates: overall patterns and specializations in
suspensory species. Am. J. Phys. Anthropo\. 83:477-498.
Tsutakawa RK, and Hewett JE (1977) Quick test for comparing two populations with bivariate data. Biometrics
33:215-219.
Van der Meulen MCH, and Carter DR (1995) Developmental mechanics determine long bone allometry. J. Theor.
BioI. 172:323--327.
Yezerinac SM, Lougheed SC, and Handford P (1992) Measurement error and morphometric studies: statistical
power and observer experience. Syst. Bio. 41:471-482.
 IV

FOSSILS AND RECONSTRUCTING THE

ORIGINS AND EVOLUTION OF TAXA
INTRODUCTION TO PART IV

Henry M. McHenry

There has been a spectacular enrichment of the fossil record of monkeys, apes and
people in the last two decades. There has been a concomitant improvement of methodol-
ogy to interpret this record. The final six chapters reflect this progress.
As Carol Ward points out in Chapter 18, the two best known early Miocene hominoid
genera, Proconsul and Afropithecus, are very different in their craniodental morphology. In a
previous study, Ward and co-authors (Begun et ai., 1997) found that the most parsimonious
cladogram defined by 240 characters separates these genera into two widely separate clades.
The face and anterior dentition of Afropithecus appears to be specialized for sclerocarp forag-
ing similar to that seen in Cacajao and other members of the Pithecinae of South America as
described by Kinzey (1992). In this respect Afropithecus resembles the middle Miocene ge-
nus, Kenyapithecus (McCrossin et ai., Chapter 19). The craniodental morphology of Procon-
sul, on the other hand, is adapted to a more generalized frugivorous diet. But the locomotor
skeletons of the two genera are remarkably similar. Both have some peculiarities seen in mod-
em Hominoidea such as a well-defined zona conoidea of the distal humerus associated with
use of the forelimb in a wide variety of positions of pronation and supination. Both share a
suite of traits associated with arboreal quadrupedal ism without the forelimb-dominated
climbing and suspensory features characteristic of modem hominoids. "The disparity in cra-
niofacial and dental form between Afropithecus and Proconsul" Ward (pg. 350) points out,
"suggests that early hominoid locomotor adaptations did not limit or constrain ecological di-
versification, and may even have facilitated the initial hominoid radiation."
By Middle Miocene times (16-14 m.y.a.) terrestrial adaptations appear in both Cer-
copithecoidea (Victoriapithecus) and Hominoidea (Kenyapithecus) as Monte McCrossin
and his collaborators present in Chapter 19. This is a remarkable new insight made possi-
ble especially by the authors and their crews' heroic efforts on Maboko Island of Kenya.
Both fossil genera have features of the shoulder, elbow, and foot that are associated with
the more terrestrially adapted modem cercopithecoids. The paleoenvironment is wood-
land, not open-country savanna. The morphology of the face and anterior dentition of Ken-
yapithecus shows the key traits associated with sclerocarp foraging as seen in
Afropithecus. Chiropotes. Pithecia and Cacajao. Perhaps, the authors speculate, terrestri-
ality in Hominoidea arose as an adaptation to this type of foraging.

333
334 H. M. McHenry

In what context did hominid bipedalism arise? As McCrossin et al. point out, the terres-
trial adaptation of Kenyapithecus is one clue: it is likely that terrestrial quadrupedal ism pre-
ceded bipedalism. Alternative views still hold strong appeal such as that articulated by Tuttle
(1974) and others that the immediate ancestors of the first bipeds were arboreal, small-bodied
apes whose terrestrial traverses were bipedal similar to those seen in modem species of Hylo-
bates. But the next two chapters (20 and 21) lend support to the view that the immediate an-
cestors of hominid bipeds were terrestrial quadrupeds not arboreal hylobatians.
Kevin Hunt (Chapter 20) approaches bipedal origins from his experience with ob-
serving free-ranging chimps in East Africa. With 701 hours of observations behind what
he says, the fact that 80% of chimp bipedalism involves feeding carries authority. He
points out that the morphology of A. afarensis and other early hominids corresponds to
what one would expect from an animal that, like the modem chimps, stands bipedally to
feed on plant material either on the ground (mostly) or in trees. Robert Foley and Sarah
Elton (Chapter 21) follow by explaining the energetics of hominoid locomotion with mod-
els predicting the costs of different forms of locomotion in various contexts. The results
show that it is likely that bipedalism evolved from an ancestor who spent a majority of its
daylight hours on the ground and only a smaller part of its activity in the trees.
In Chapter 22 Russell Tuttle and collaborators reflect on the key evidence for early
hominid bipedalism and particularly the footprints of Laetoli and the foot morphology of
the Hadar australopithecines. By their extensive analyses the footprints are indistinguish-
able from modem people. The Hadar fossils, however, retain many primitive features in-
cluding long and curved toes. From this it is reasonable to propose that there is more than
one kind of bipedal hominoid in the Pliocene. It is also likely, they argue, that squatting
played a significant role in the early stages of bipedalism.
The best known early hominid biped, Australopithecus afarensis, is assessed by its
original describers as having" ... adaptation to full and complete bipedality" (Johanson et
aI., 1982:386) and a forelimb " ... not primarily involved in locomotor behavior" (Johanson
et aI., 1982:385). But as Hunt, Tuttle et aI., and many others point out, there are many pe-
culiarities of its postcranial skeleton that need to be explained. In Chapter 23 Christopher
Ruff provides one explanation by careful application ofbiomechanical principles and par-
ticularly his own analyses of the mechanical properties of the femoral shaft. In the medio-
lateral plane, the hip of A.L. 288-1 (Lucy) is unlike modem humans in its relatively short
femur, long femoral neck, wide biacetabular diameter, cortical thickness of the femoral
shaft and other features. Some of these peculiarities are consistent with the model describ-
ing modem human hips, but not all. If the pelvis is tilted up slightly on the unsupported
side during stance phase, however, the predictions from the model fit much more reason-
ably. This also fits with the elongated foot observed in A. afarensis.
Unanimity of opinion on all aspects of interpretation is not achieved here. Careful
reading reveals many differences of method and conclusion. This is a measure of health
and vitality of the field, of course. Add the fact that the fossil record samples only a tiny
portion of our family's history to appreciate the open field before us. We can delight in
what new discoveries and approaches will bring.

REFERENCES
Begun DR, Ward CV, and Rose MD (\ 997) Events in hominoid evolution. In DR Begun, CV Ward, and MD Rose
(eds.): Function, Phylogeny, and Fossils: Miocene Hominoid Evolution and Adaptation. New York: Plenum
Press, pp. 389-415.
Introduction to Part IV 335

Johanson DC, Taieb M, and Coppens Y (1982) Pliocene hominids from the Hadar Formation, Ethiopia
(1973-1977): Stratigraphic, chronologie, and paleoenvironmental contests, with notes on hominid mor-
phology and systematics. Am. J. Phys. Anthropol. 57:373--420.
Kinzey WG (1992) Dietary and dental adaptations in the Pithecinae. Am. J. Phys. Anthropol. 88:499-514.
Tuttle RH (1974) Darwin's apes, dental apes, and descent of man - normal science in evolutionary anthropology.
Curro Anthropol. J5:389-398.
 18

AFROPITHECUS, PROCONSUL, AND THE

PRIMITIVE HOMINOID SKELETON

Carol V Ward

Departments of Anthropology and Pathology and Anatomical Sciences

107 Swallow Hall
University of Missouri
Columbia, Missouri 65211

1. INTRODUCTION

Many distinctive synapomorphies of modem apes (and humans) are found in the
postcranial skeleton. These characters reflect a basic adaptation, variably developed and
practiced among modem species, to forelimb-dominated arboreal locomotion, including
climbing, brachiation, and/or forelimb suspension (e.g., Cartmill and Milton, 1977;
Fleagle et aI., 1981; Hunt, 1992; Keith, 1923; Stem, 1971; Stem et aI., 1977). The mor-
phological pattern shared by modem hominoids has led to the general assumption that lo-
comotor divergence was an initial hallmark of the hominoid lineage, setting them apart
from their monkey-like forbears. As more is learned about the earliest hominoids, how-
ever, paleontologists realize that not all apes share a similar pattern ofpostcranial anatomy
and locomotor behavior, and that the suite of features seen in extant apes evolved in a mo-
saic fashion over the course of hominoid evolutionary history (reviews and references in
Begun et aI., 1997a).
Hundreds of hominoid fossils have been discovered from the Miocene of Africa,
Asia and Europe. The earliest of these occur in East Africa, and comprise several genera.
Many of these taxa are known from postcranial elements, allowing us to assess their mor-
phological similarities to and differences from their modem relatives. A recent cladistic
analysis of 240 cranial and postcranial characters of nine fossil and four extant hominoid
taxa (Begun et aI., 1997b; Figure 1) suggests that the early part of the hominoid lineage
was characterized by taxa that lacked many derived traits of the postcranial skeleton found
in extant apes (reviews and references in Rose, 1997; Ward, 1997). The combination of
features typical for many early Miocene genera consists of hominoid synapomorphies,
primitive retentions, and unique characters. This combination is unique among anthro-
poids (Rose, 1983, 1993, 1994), and reflects a generalized arboreal locomotor adaptation,

337
338 c. V. Ward

characterized by more frequent pronograde postures and a lesser emphasis on suspensory

activities than in modem apes.
The generalized pattern of postcranial anatomy and inferred locomotor behavior
characterized most early to middle Miocene hominoids, except perhaps for the enigmatic
Morotopithecus (Gebo et al., 1997), for which comparatively little fossil evidence is
known. Only later do the beginnings of extant hominoid-like forelimb-dominated arboreal
adaptations begin to appear, and only in the later Miocene did apes begin to take on their
modem postcranial form.
Despite the postcranial similarity among the earliest hominoids, these apes appear to
have had an array of dietary and ecological craniofacial adaptations (Harrison, 1982; Kay,
1977; Kay and Ungar, 1997; Walker et al., 1994). The degree of postcranial similarity cou-
pled with craniofacial difference among early Miocene genera equals or exceeds that
found in any subsequent period of hominoid evolutionary history. In no two taxa among
early apes is this pattern of intergeneric postcranial similarity and craniofacial diversity
more apparent than Afropithecus and Proconsul.
Because they are so well known, Afropithecus and Proconsul provide an excellent
opportunity to compare cranial and postcranial adaptations in two early Miocene apes.
This paper reviews the postcranial similarities between these genera, and contrasts this
with their very different craniodental adaptations. This contrast between cranial and post-
cranial diversity in Afropithecus and Proconsul illustrates the complex nature of mosaic
evolution in primates, and provide a basis for understanding the early evolution of the
hominoid lineage.

2. THE FOSSILS
For many years, Proconsul has been the best-known genus of Miocene ape, known
from virtually all body parts, and for multiple individuals of four species: P. africanus
from Koru and other Tinderet localities (Hopwood, 1933), P heseloni from site R1l4 on
Rusinga Island (Walker et aI., 1993) and other Kisingiri localities, P. nyanzae from type
locality RI-3 (Clark and Leakey, 1951) and other sites on Rusinga Island and from Mfan-
gano Island, and P. major from Songhor (Clark and Leakey, 1951) and other Tinderet lo-
calities (review and references in Walker, 1997). All of these sites date to the early
Miocene, around 18 million years ago.
Proconsul is known from a nearly complete skull, as well as many maxillary,
mandibular and dental specimens. As a result, its cranial and postcranial adaptations are
fairly well-understood. Proconsul species show little postcranial variability, despite their
size differences (but see Nengo and Rae, 1992), so their functional morphology is consid-
ered here at the generic level.
In 1986, Leakey and Leakey published the discovery of new Afropithecus turkanen-
sis fossils from Kalodirr in northern Kenya. Although Afropithecus is known from four
sites, Kalodirr, Buluk, Moruorot and Locherangan, postcrania are only known from
Kalodirr and Buluk (Leakey et aI., 1988). All Afropithecus sites date to the early Miocene,
16-18 million years ago, so they were broadly contemporaneous with Proconsul sites.
Among the Kalodirr fossils is a nearly complete facial and anterior cranial skeleton
(Leakey and Leakey, 1986; Leakey et aI., 1988). It is probably from a male, and was about
the size of a chimpanzee, as was one of the two best-known species of Proconsul, P. nyan-
zae. Females appear to have been smaller, but these smaller specimens may represent a
second species (Leakey and Walker, 1997; Leakey et aI., 1988). The smaller forms were
Afropithecus, Proconsul, and the Primitive Hominoid Skeleton 339

slightly larger than the best-known Proconsul species, P heseloni, which was about the
size of a siamang. Here, the two sizes are considered to represent males and females of A.
turkanensis (Leakey et aI., 1988).
Afropithecus is currently the second-best known postcranially of early Miocene
hominoids. Postcrania are known from both sexes, and from the sites of Buluk and
Kalodirr (Leakey and Walker, 1985; Leakey et aI., 1988). The Buluk postcranial specimen
is a phalanx. The Kalodirr postcranial remains include specimens associated with cranial
elements (parts of an ulna, fibula, and first, third and fourth metatarsals), as well as speci-
mens that are unassociated, preserving parts of numerous joint complexes of the upper and
lower limbs, including bones of the elbow, wrist, hand, fingers, ankle, foot and toes
(Leakey et aI., 1988). A detailed functional analysis of these bones is in preparation by M.
Rose, C. Ward, A. Walker and M. Leakey. The Afropithecus sample provides enough infor-
mation to reconstruct a fairly accurate picture oflocomotor behavior in this genus, and the
even more complete sample of Proconsul fossils provides an even better picture.
Another well-known primitive ape (Begun et aI., 1997b) is Kenyapithecus, from the
middle Miocene of Kenya. The sample of postcrania attributed to Kenyapithecus has been
expanding, and now includes most skeletal elements (McCrossin, 1994; McCrossin and
Benefit, 1997; McCrossin et aI., this volume), but many of these important fossils have
only recently been discovered and remain unpublished (Ward and Brown, 1996; Rose,
pers. comm.).
Afropithecus and Proconsul, along with Kenyapithecus and sometimes other genera,
have been interpreted to be basal members of the hominoid lineage, but were probably not
sister taxa (Figure 1) (reviews and references in Begun et aI., 1997b). This hypothesis is
supported by a recent phylogenetic analysis of 240 characters from the cranial and post-
cranial skeletons of nine fossil and four extant hominoid taxa (Begun et aI., 1997b). Al-
though the phyletic relationships of these taxa relative to one another are equivocal, their
position relative to extant hominoids is more stable; placing any of these taxa within the
extant hominoid clade increases tree length by 4% or more. It is therefore reasonable to
hypothesize that these three taxa represent an early radiation of stem hominoids that ex-
isted throughout the early and middle Miocene.
Recently, based on newly discovered and yet unpublished fossils, McCrossin (1997)
has claimed that Kenyapithecus may have shared more postcranial apomorphies with ex-

Figure 1. Hypothesis of hominoid relationships based on a cladistic analysis of 240 cranial and postcranial char-
acters (from Begun et aI., 1997b). The tree length is 446 steps and the consistency index is 63. The relationships
among taxa illustrated here differs from some other published hypotheses, but most phylogenies propose that Pro-
consul and Afropithecus were not sister taxa (reviews and discussions in Begun et aI., 1997a).
340 c. V. Ward

tant apes than do Proconsul or Afropithecus. McCrossin reports that new Kenyapithecus
fossils from Maboko Island exhibit derived postctanial features, such as a straighter hu-
meral shaft, than previously reported for this taxon. McCrossin interprets these features to
mean a more modern great ape-like locomotor adaptation for Kenyapithecus than is found
in Afropithecus or Proconsul, including an emphasis on forelimb-dominated arboreality.
This new interpretation contrasts with previous arguments for terrestrial adaptations in
Kenyapithecus, which is suggested by many postcranial features, including a retroflexed
humeral shaft (Benefit and McCrossin, 1995; McCrossin, 1994, McCrossin and Benefit,
1994, 1997). These apparently contrasting morphologies confuse functional and phyloge-
netic interpretation of Kenyapithecus, and so a more precise interpretation of its affinities
will have to wait until the new fossils are published. Nevertheless, Kenyapithecus appears
to be the closest sister taxon to extant hominoids, and is considered a middle Miocene
stem hominoid here (reviews and references in Begun et aI., 1997).
Afropithecus and Proconsul are strikingly similar in preserved parts of their postcra-
nial skeletons, exhibiting no evidence for substantial differences in locomotor adaptation.
Their faces, teeth and reconstructed dietary adaptations, however, are quite different. They
illustrate the pattern of mosaic evolution characterizing not only the early hominoid line-
age, but much of hominoid evolution. It appears that the generalized locomotor adapta-
tions of early hominoids permitted ecological and dietary specialization without
necessitating substantial postcranial change. It may be that the generalized nature of early
hominoid locomotor adaptations even facilitated the initial radiation of apes, with selec-
tion producing considerable postcranial modification later in hominoid evolution.

3. COMPARATIVE POSTCRANIAL ANATOMY OF

AFROPITHECUS AND PROCONSUL

3.1. Elbow
Afropithecus is known from a small piece of distal humeral articular surface, includ-
ing part of the capitulum and the entire width of the zona conoidea (Figure 2; Leakey et
ai., 1988). Although small, this fossil preserves one of the most functionally diagnostic re-
gions of the distal humerus (Morbeck, 1983; Rose, 1983, 1988, 1993a). Proconsul is
known from more complete distal humeri (Napier and Davis, 1959; Rafferty et aI., 1995;
Senut, 1980; Walker and Pickford, 1983; Walker and Teaford, 1988; Walker et aI., 1985).
Both Afropithecus and Proconsul resemble extant hominoids in having a well-defined,
deep zona conoidea medial to the capitular surface for articulation with the margins of the
radial head (Figure 2), although not as pronounced as in African apes. In contrast, most
monkeys tend to have a much flatter, broader zona conoidea. This difference between apes
and most monkeys in humeral morphology reflects two fundamentally different patterns of
habitual loading and mobility at the elbow (Napier and Davis, 1959; Rose, 1987, 1993a),
and the presence of a well-defined zona conoidea in Afropithecus and Proconsul suggests
that it arose early in the hominoid lineage.
Hominoids have fairly symmetrical radial heads that maintain similar amounts of
humeral contact during all phases of pronation-supination, reflecting the habitually varied
forelimb postures of these animals during climbing, hanging, and bridging arboreal loco-
motion. In addition, the rounded capitulum and deep zona conoidea of hominoids provide
joint surfaces capable of effectively resisting loads from directions other than relatively
simple axial loading. In contrast, cercopithecoids have an asymmetrical configuration of
Afropithecus, Proconsul, and the Primitive Hominoid Skeleton 341

A B

[;J!)
~
Pan

~.

,---- ..\
, I
Afropithecus
[j

Figure 2. The distal humerus and proximal ulna in extant and

fossil taxa. A) Distal view of left humeri, Afropithecus is
known from the fragment (KNM-WK 17008) shown here,
7 Proconsul

~
&5J ~
comparable areas on other taxa are illustrated. Only the area
for articulation with the radius is illustrated for Proconsul. B) ....... \ Papio
Anterior view of proximal left ulnae, Afropithecus (KNM-WK
18395) is missing the proximal end of the olecranon and the
distal end of the trochlear notch. Modified from Rose (1987).

the humeroradial joint that maximizes chondral contact and joint congruence between the
humerus and radius during pronation, the position in which the elbow is typically loaded
during palmi grade quadrupedalism, at the expense of articular contact in supinated pos-
tures (Rose, 1987).
The zona conoidea of Proconsul and Afropithecus resembles that of other homi-
noids, suggesting that they used their elbows in varied positions of pronation and supina-
tion during locomotion. Further support for this functional hypothesis comes from the rest
of the humeroradial joint of Proconsul, which more closely resembles that of extant apes
than most monkeys (Morbeck, 1983; Rose, 1983, 1993b, 1997).
A similar pattern of weight-bearing during all phases of forelimb pronation and supi-
nation may also be implied by the proximal ulna. Trochlear notches of Afropithecus and
Proconsul ulnae were more symmetrical than those of Old World monkeys with weak, but
more proximodistally-oriented, keels (Rose, 1987, 1993a, 1997).
The derived elbow morphology suggested by the fragmentary elbow joint remains of
Afropithecus, and seen more extensively in Proconsul and other fossil hominoids, resem-
bles that of extant apes in important ways, implying a similar pattern of varied elbow use
during locomotion. Afropitheclis and Proconsul, as with extant hominoids, appeared to
have used their elbows in a variety of postures, from pronation to supination, during loco-
motion. Although chimps and gorillas maintain a pronated hand posture during terrestrial
quadrupedalism, they employ varied elbow postures during arboreal climbing and bridg-
ing activities. The antiquity of this morphology suggests that adaptation to variable prona-
tion-supination postures of the elbow in hominoids evolved before advanced suspensory
adaptations (see below), and is sufficiently generalized to accommodate a range of loco-
motor specializations.
342 C. V. Ward

Proconsul, however, differed from extant hominoids in other ways, including having
a longer olecranon process. Although the olecranon process is broken in Afropithecus, the
preserved contour of its posterior ulna suggests that the olecranon process extended some-
what further proximally than is seen in extant apes, probably to the extent seen in Procon-
sul (Figure 2; Rose, 1993b, 1997). Among extant anthropoids, longer olecranons are
associated with habitually flexed elbows (Harrison, 1982; Rose, 1993). As the elbow
moves into extension, a proximally elongate olecranon results in more rapid shortening of
the triceps muscle, moving the triceps towards the end of its length-tension curve where it
would have less contractile power (Pitman and Peterson, 1989). A shorter olecranon al-
lows powerful contraction of the triceps muscle near full elbow extension. Proconsul, and
probably Afropithecus, appears to have been adapted for at least a moderate amount of ha-
bitual elbow flexion during locomotion, rather than the extended elbow postures of great
apes when arm-hanging or knuckle-walking. This suggests that although Proconsul, and
perhaps Afropithecus, was a habitual climber, it did not rely on arm-hanging as exten-
sively as extant apes.

3.2. Hand
The emphasis on a range of pronation-supination in apes and pronated hand postures
in monkeys is also seen in bones associated with the midcarpal joint. Jenkins (1981) dem-
onstrated structural adaptations of the extant hominoid wrist that allow significant midcar-
pal supination, an important adaptation for brachiation. These morphologies are expressed
most strongly in hylobatids and orangutans, but are also seen in chimpanzees, and to a
lesser extent gorillas. Midcarpal supination is permitted by an embrasure between the
capitate and trapezoid for the scaphoid/centrale that widens palmarly in apes. This partly
is a result of the laterally-facing facet for the scaphoid/centrale on the capitate in suspen-
sory apes. In monkeys, the dorsally-facing joint surface for the scaphoid/centrale results in
a narrowing of the embrasure for the scaphoid/centrale towards the palm, limiting move-
ment of these bones palmarly, and restricting midcarpal supination.
The scaphoid/centrale facet on capitates of Afropithecus and Proconsul resembles
that of most monkeys, and faces dorsolaterally, with evidence for only a small gap against
the trapezoid (Figure 3). Of course, without knowing what the other bones involved would
have looked like, any conclusions about joint function are tentative. The narrow, laterally-
facing capitate head, however, implies limited midcarpal supination in Afropithecus and
Proconsul, but probably free pronation at this joint. Because this wrist morphology is also
found in most monkeys, this pattern of restricted midcarpal supination is probably primi-
tive for catarrhines.
Because midcarpal supination is an important component of forelimb suspension in
extant apes, it appears that Afropithecus and Proconsul did not rely as heavily on forelimb-
dominated suspension during locomotion as do many extant apes or even Sivapithecus. In-
stead, they were probably habitually palmigrade. This monkey-like midcarpal anatomy,
however, would not affect their ability to climb. The only limitation it would impose, com-
pared with that seen in extant apes, is on midcarpal pronation, which tends to occur when
moving from branch to branch when hanging below a support by the forelimbs, and is only
extensively developed in the most arboreal extant apes, gibbons and orangutans.
The pattern of weight-bearing through the carpus of Afropithecus appears to have
differed from that of most extant catarrhines. The lunate exhibits an atypical pattern ofre-
lationships among some intercarpal joints. The lunate of Afropithecus, like that of Procon-
sul, is compressed proximodistally (Figure 3). The capitate and triquetral facets are almost
Afropithecus, Proconsul, and the Primitive Hominoid Skeleton 343

A Nasalis ProcOllsul Afropithecu Pan

Figure 3. Some carpal bones of ex-
tant and fossil taxa. A) Right trapezia
in palmar (above) and dorsal (below)
views, distal end is down. Afropi-
thecus and Proconsul share a sellar
metacarpal facet that extends onto the
dorsal aspect of the bone with extant
hominoids; this is a feature not found
on most monkey trapezia. Modified
from Rose (1992). B) Right capitates B
in dorsal view, distal end is down, lat-
eral side is right. Suspensory anthro-
poids are represented by Ateles. Mandrillus Procollsul Afropithecus Pan Aleles
Contour of lateral sides is generally
related to morphology of embrasure
between capitate and trapezoid. Figure
modified from Rose (1984). C) Right
lunates in palmar view, radial facet is
down, triquetral facet upper left, capi- c
tate facet upper right. Papio ProcOlIsul Afropilhecus Pan

parallel to the radial facet. This shape is only seen in certain New World monkeys today
(Rose, 1993b), and may be primitive for hominoids, if not all catarrhines. Because a joint
surface must remain normal to transarticular loads, the flattened lunate shape implies that
weight could be effectively transferred through the lunate from both the capitate and tri-
quetral to the radius, when the wrist is neutral with respect to radial and ulnar deviation.
The angle between the triquetra1 and radial facets is slightly more oblique in cercopi-
thecids, and much more so in apes.
The triquetral facet on great ape lunates is set roughly normal to the radial facet, and
almost parallel to the scaphoid surface (Figure 3C). It seems that in extant hominoids
weight may be transferred to the radius primarily through the capitate and then lunate
when the wrist is neutral with respect to pronation and supination, and through the tri-
quetral to the lunate to the scaphoid when the wrist is adducted. The carpal joint configu-
ration seen in Miocene apes and New World monkeys indicates a common pattern of load
transmission through the carpus that differs from that of other anthropoids, and probably
represents the primitive condition for catarrhines.
Extant apes have altered their wrist even further by proximodistal elongation of the
hamate allowing increased mobility between the triquetral and hamate (Sarmiento, 1988;
Lewis, 1989). They also have reduced the length of the ulnar styloid process, further in-
creasing the capacity for ulnar deviation by providing more space between the ulna and car-
pus (Lewis, 1969, 1972, 1974, 1989). In most monkeys, weight is transferred via bony
contact directly from the triquetral to the ulna during palmigrade postures, and it may be
that relatively little passes directly through the triquetrolunate joint to the radius, explaining
the oblique orientation of this joint surface with respect to the radial lunate surface in cer-
copithecoids (Beard et ai., 1986; Cartmill and Milton, 1977; Lewis, 1989; Sarmiento, 1988).
Articular facets for the ulnar styloid are visible on the pisiform and triquetral of Proconsul,
344 c. V. Ward

suggesting that they also had direct bony contact here (Beard et aI., 1986). Although this
may indicate the capacity for weight transfer through the medial portion of the wrist, the
long, proximally-facing facet for the ulnar styloid on the pisiform may indicate a greater
range of ulnar deviation in Proconsul than is typical for most monkeys, except ate lines and
Alouatta. Proconsul shows signs of enhanced ulnar deviation when compared with cercopi-
thecoids, probably reflecting a greater reliance on climbing and varied limb postures during
arboreal locomotion than is typical for cercopithecoids, but did not have the capacity for ul-
nar deviation found in extant hominoids. Although no triquetra I or pisiform is known for
Afropithecus, because the Afropithecus lunate so closely resembles that of Proconsul, the
other carpal bones may have been similar morphologically and functionally.
The Afropithecus hand is also known from the second and fourth metacarpals, as
well as an associated first metacarpal and trapezium; these elements are also known for
Proconsul. Afropithecus and Proconsul have morphologically functionally similar trapezia
and first metacarpals. The joint between the trapezium and first metacarpal of these fossil
apes is sellar (Figure 3A) and functioned most like that of extant hominoids (Rafferty,
1990; Rose, 1992). Greater ranges of abduction-adduction and rotation were possible at
this joint in all living and fossil apes than in monkeys, as a result of greater curvature and
incongruity of the opposing joint surfaces in the hominoids (Rafferty, 1990). This similar-
ity among hominoids probably represents a synapomorphy, and reflects an habitual reli-
ance on enhanced grasping capabilities in even the earliest apes.
Although the grasping capabilities of Afropithecus and Proconsul resembled those of
extant hominoids, they appear to have had slightly more gracile pollical phalanges (Be-
gun, 1994) and metacarpals. Still, the pollical phalanx and metacarpals were more robust
than those of Old World monkeys, suggesting strength of manual grasping intermediate
between that of apes and cercopithecoids. As with the foot (see below), this suggests that
grasping was more important than for most monkeys, but was not as well-developed as in
extant apes, who rely more heavily on it when engaged in forelimb suspensory and bridg-
ing activities. Again, this is consistent with the hypothesis that early hominoids were fre-
quent climbers, but were not adapted for extensive forelimb suspension.
The second and fourth metacarpals are also informative. Hominoids have the meta-
carpal bases facing directly proximally, most likely correlated with the neutral or flexed
position of the carpometacarpal joints during climbing and, in the case of African apes,
knuckle-walking. Most monkeys, on the other hand, have the metacarpal bases facing
proximodorsally. Because joint surfaces must be oriented normal to transarticular loads, a
proximodorsal orientation probably reflects habitual extension at the carpometacarpal
joints, as occurs during palmigrade or mildly digitigrade progression. Afropithecus and
Proconsul resemble monkeys in the proximodorsal orientation of their metacarpal bases,
suggesting habitual palmi grady on large supports or when terrestrial. Combined with in-
formation from the thumb, this evidence suggests that fairly powerful manual grasping
was an important adaptation in these apes, and that Afropithecus and Proconsul were
adapted for palmi grade quadrupedality as well as climbing.

3.3. Foot
The ankle of Afropithecus is not only known from a nearly complete fibula, but from
a nearly complete talus. These elements resemble those known for Proconsul nyanzae.
Great apes have an asymmetrical talar trochlea with a shallower groove than do most
monkeys. The tali of Proconsul, Afropithecus, and most other Miocene apes are more sym-
metrical with deeper trochleas than those of extant apes (Figure 4B), which would have
Ajropithecus, Proconsul, and the Primitive Hominoid Skeleton 345

Pan Ajropithecus Proconsul Cercopithecus

Figure 4. Talocrural bones of extant and fossil 8

taxa. A) Anterior view of left fibulae. B) Dorsal
view of right tali. See text for discussion.

provided less conjunct rotation of the foot during flexion and extension, and suggests
more habitual plantigrady during locomotion.
Evidence of hallucal size and morphology of Afropithecus comes from the fibula, ta-
lus and the first metatarsal itself. The Afropithecus fibular shaft was more gracile than
those of extant apes relative to the size of its talar facet, but was more robust than those of
most monkeys (Figure 4A). A large fibula implies the presence of well-developed hallucal
flexor muscles. A strong flexor hallucis longus is also indicated by the pronounced groove
on the posterior margin of the talar trochlea (Figure 4B). _
In addition, the talar facet on the fibula is set at an angle relative to the fibular shaft
(Figure 4A). This condition is found in extant apes and the most arboreally adapted mon-
keys, including atelines and the fossil colobine Rhinocolobus turkanensis. An angled ta-
lofibular joint provides a joint surface contact area that is oriented obliquely with respect
to the talar trochlear surface, perhaps indicating weight-bearing when the foot is in everted
postures, such as when it is grasping a limb, and/or the knee joint is abducted when the
foot is in a plantigrade position. All of these postures occur when climbing and require the
presence of a grasping hallux. At least as importantly, talar facet orientation probably also
reflects the relative size of the hallucal flexors, and their angle of pull relative to the fibu-
lar shaft.
The first metatarsal of Afropithecus and Proconsul was not as large as that of apes,
but more robust than that of monkeys (Figure 5A). The insertion site of peroneus longus
is pronounced, implying a strong muscle and correspondingly strong hallucal adduction.
Proconsul halluces are long relative to those of comparably-sized monkeys, and are more
strongly built, and the hallucal terminal phalanges are broad relative to those of other an-
thropoids (Begun et aI., 1994; Strasser, 1993). Robust hallucal metatarsals and phalanges
imply fairly powerful grasping abilities in Afropithecus, especially when combined with
evidence of powerful hallucal flexors. The cuneiform facet has a definite concave medial
portion, suggesting habitual abduction, though not as strong as in most extant apes.
While hallucal size and strength in Afropithecus and Proconsul appear to have exceeded
those of extant monkeys, they were not as well-developed as in extant apes. The se-
samoid grooves on the head of the first metatarsal are pronounced, as in monkeys. This
morphology suggests that hallucal grasping was probably not as strong as in extant apes,
and that the feet were frequently used in plantigrade postures. Most likely, feet were used
346 c. V. Ward

Pan Ajropithecus Proconsul Papio

Figure 5. Metatarsals of extant and fossil

B taxa. A) Medial view of left first metatar-
sals, distal is down. B) Medial view of left
fourth metatarsals, distal is down. See text
for discussion.

in grasping during climbing, but probably not extensively in bridging or suspensory ac-
tivities.

3.4. Phalanges

Many phalanges are known for Afropithecus, and even more for Proconsul. The pha-
langes of Proconsul have been described and analyzed in detail (Begun, 1994). The Afro-
pithecus phalanges do not differ appreciably from those of Proconsul, so Begun's
conclusions about the locomotor repertoire of Proconsul, based on phalangeal morphol-
ogy, are equally valid for Afropithecus.
Proconsul manual and pedal phalanges exhibit only mild curvature and relatively
weak secondary shaft characteristics, suggesting that these early Miocene apes did not rely
on suspensory postures as extensively as do extant apes (Begun, 1994). Instead, they were
probably above-branch arboreal quadrupeds. Again, like other parts of the postcranium,
the Afropithecus and Proconsul phalanges suggest that these apes were well-adapted
climbers, but probably not as suspensory as extant apes. In addition, metacarpal and pha-
langeal morphology suggests slightly hyperextended postures of the carpometacarpal and
metacarpophalangeal joints at least some of the time, probably on larger arboreal supports
and when terrestrial.
Extant hominoids exhibit clearly differentiated manual phalanges compared with
pedal phalanges, reflecting the divergent use of the hands and feet in climbing and suspen-
sion. Monkeys, on the other hand, tend to have morphologically similar manual and pedal
phalanges, reflecting similar use of the hand and foot in locomotion. There appears to
have been only minor differentiation between manual and pedal phalanges in Proconsul
(Begun, 1994), and in Afropithecus. The hint of a difference in Proconsul and Afropithecus
may indicate the beginnings of specialization for more forelimb-dominated climbing ac-
tivities in these early apes, but the derived condition found in extant apes had not yet
taken place. Again, Proconsul and Afropithecus phalangeal morphology reiterates the
functional scenario revealed in other parts of the skeletons.
Afropithecus, Proconsul, and the Primitive Hominoid Skeleton 347

4. LOCOMOTOR BEHAVIOR OF THE EARLIEST HOMINOIDS

AND EVOLUTION OF THE APE SKELETON
The known parts of the Afropithecus turkanensis skeleton are morphologically, and
presumably functionally, equivalent to those of Proconsul nyanzae and P. heseloni. In fact,
the postcrania known for the presumptive males of Afropithecus turkanensis are difficult
to distinguish morphologically from those of Proconsul nyanzae. Because parts of the el-
bow, hands, and feet are known for both genera, and show no appreciable differences, it is
reasonable to assume that the rest of their postcranial skeletons were similar as well.
These animals appear to have had similar locomotor adaptations. In a direct comparison of
similar body parts, no obvious functional distinctions can be made.
All of the information from these genera can be used to provide a picture of their lo-
comotor adaptations. It is clear that Afropithecus and Proconsul were primarily
pronograde quadrupeds. They both exhibit signs of habitual extension at the carpometa-
carpal, tarsometatarsal, and metapodial-phalangeal joints. There also are pronounced se-
samoid grooves on the metapodial heads, as in most monkeys, and only a modest amount
of differentiation between hand and foot phalanges, suggesting roughly similar use of the
hands and feet during locomotion.
Proconsul, being better known, provides additional evidence that it was primarily a
pronograde animal. Proconsul had a narrow thoracic cage, as seen in the cercopithecoid-
like curvature of the ribs, the narrow iliac blades, and the ventrally-situated lumbar verte-
bral transverse processes (Ward, 1993; Ward et aI., 1993). These morphologies suggest
that the scapulae were oriented in a roughly parasagittal position on the sides of the rib
cage, and that the glenohumeral joints were oriented ventrally. In this position, the
thoraco- and scapulohumeral muscles are aligned to produce primarily flexion-extension
movements of the limbs. Extant hominoids, along with atelines to a lesser extent, have
broader thoracic cages with laterally-facing shoulder joints, enabling their upper limb
muscles to produce more effective abduction-adduction movements, critical for many
forelimb-dominated climbing and suspensory behaviors. Proconsul was built more like
monkeys, and emphasized primarily flexion-extension of its limbs during propulsive
movements in locomotion.
Proconsul also had a long, flexible torso like most monkeys, in contrast to the short,
stiff backs of extant apes. Short backs decrease bending moments about the lower verte-
brae, and provide for effective attachment and function of latissimus dorsi to adduct the
humerus, an important movement during suspensory locomotion, without risking exces-
sive stresses on the lower back (Ward, 1993; Ward et aI., 1993). Because Afropithecus
limbs were so similar to those of Proconsul, it probably resembled Proconsul in axial mor-
phology and function as well.
Despite adaptations for habitual pronogrady, morphological adaptations to climbing
are clearly present in these earliest hominoids. Both of these genera exhibit numerous ad-
aptations to climbing, and to a more diverse array of limb postures during locomotion than
typifies Old World monkeys. The elbows of Afropithecus and Proconsul were adapted to
weight-bearing in a variety of pronation-supination postures, albeit habitually slightly
flexed elbow postures. The Proconsul hip and knee were similar, and lack adaptations to
loading in stereotypical postures seen in Old World monkeys, even when arboreal. Al-
though early hominoids exhibit better developed grasping capabilities of the hand and foot
and more mobile wrists and ankles than do most monkeys, they were not as well-adapted
for suspensory locomotion as are extant hominoids. Thus, early Miocene apes probably
did not engage in as much suspensory activities as their modem counterparts.
348 c. V. Ward

In fact, the mobile elbow, hip and knee, and moderately well-developed grasping
abilities of Proconsul and Afropithecus (where known), suggest that climbing was an impor-
tant component of their locomotor repertoire. An arboreal milieu necessarily has randomly
oriented supports, requiring diverse limb postures to negotiate effectively. Cercopithecoids
are able to negotiate among arboreal supports by above-branch quadrupedal ism, combined
with some leaping and bridging. It appears that the earliest apes increased their emphasis on
climbing and bridging behaviors, perhaps because of their relatively large body sizes. Male
Afropithecus individuals and Proconsul nyanzae were comparable to modem chimpanzees
in size. Because of this, they would not have been able to locomote as monkeys do in the
trees. Large individuals would have had to distribute their weight over a greater number of
supports to reach terminal branches where nuts or seeds might be located. They would also
have to bridge gaps and/or be terrestrial more frequently than many monkeys. Even the
smaller species, Proconsul heseloni, seems to have been adapted to increased climbing and
bridging, and decreased running and leaping, than is typical for most monkeys. Kelley
(1995) has linked tail loss with the need for alternate forms of balance in trees, and has ar-
gued that postcranial adaptations of early apes represents a solution to this problem. This
may be why Proconsul, and perhaps other early apes, lacked external tails (Ward et aI.,
1991). Certainly, early hominoids practiced a form of above-branch arboreality that in-
volved substantial limb mobility for weight distribution and balance. Many morphologies
shared by early and later hominoids may represent adaptations to this more generalized lo-
comotor repertoire than the derived, forelimb-dominated locomotor modes seen among ex-
tant species (Kelley, 1995). Only later in evolution did the shoulder joint and torso change
form to become adapted for forelimb suspension and forelimb-dominated climbing.
The generalized form of hominoid postcranial anatomy and locomotion exhibited by
Afropithecus and Proconsul seems to have been typical for most early Miocene apes.
Turkanapithecus, which is known from several postcranial bones, may have been slightly
better adapted for arboreal climbing than were Afropithecus and Proconsul, but still, it
shares their general body form in their preserved parts. Even Kenyapithecus shares most
early hominoid morphologies, and morphological characters suggestive of increased terre-
striality (McCrossin, 1994, 1997) represent relatively minor variations on the early homi-
noid theme. Among early to middle Miocene hominoids, only Morotopithecus appears to
differ much from the early hominoid pattern, and more closely resembles extant apes
(Gebo et aI., 1997; Sanders and Bodenbender, 1994; Walker and Rose, 1968; Ward, 1993).
Such a generalized set of adaptations for arboreal locomotion appears to have been
sufficiently flexible to allow early Miocene apes to exploit a variety of dietary niches. It
would not have made them less effective on the ground than in the trees, and it would
have permitted them to reach the terminal branches of trees to exploit resources there as
well. The fact that early apes did exploit an array of dietary niches is also illustrated
clearly by Afropithecus and Proconsul.

5. CRANIAL ANATOMY AND DIETARY ADAPTATIONS OF

AFROPITHECUS AND PROCONSUL

The overwhelming similarities between the postcranial anatomy of Proconsul and

Afropithecus contrast markedly with the disparity in their apparent dietary adaptation and
associated craniofacial structure.
Afropithecus appears to have been a sclerocarp forager, based on its craniofacial and
dental morphologies (Leakey and Walker, 1997; McCrossin and Benefit, 1997) . Afropi-
Afropithecus, Proconsul, and the Primitive Hominoid Skeleton 349

thecus mimics the suite of craniofacial and dental characters summarized by Kinzey
(1992) for pitheciins, and noted for Kenyapithecus (Benefit and McCrossin, 1995;
McCrossin, 1994; McCrossin and Benefit, 1994, 1997). Afropithecus canines are large,
laterally splayed, and have strong roots with short crowns, an arrangement that would al-
low incisor and canine functions to be uncoupled, so that the incisors can be used for crop-
ping foods and the canines for forceful puncturing (Kinzey, 1992). This evidence,
combined with the relatively low crowns and the apparent minimal sexual dimorphism of
Afropithecus canines, supports the argument that Afropithecus canines were used more for
food preparation than agonistic intraspecific display. These morphologies parallel those in
pitheciins, such as Chiropotes, which use their canines to puncture hard fruits or seeds
(Kinzey, 1992). Afropithecus, however, has large incisors such as those generally found in
frugivores (Kay, 1984). The central upper incisors are much larger than the lateral. The in-
cisors are procumbent, and the upper ones are set in a prognathic premaxilla and occlude
with the robust, styliform lower incisors.
The molar morphology of Afropithecus is consistent with its craniofacial form. Afro-
pithecus has short molar shearing crests, consistent with a diet of frugivory or seed preda-
tion (Leakey and Walker, 1997; McCrossin and Benefit, 1997). It also has thick molar
enamel with extremely heavy tooth wear. The pronounced frontal trigon diminishes in size
with age, a feature correlated with a powerful temporalis muscle (Leakey and Walker,
1997; Simons, 1987). In addition, the mandible is strongly buttressed, also indicating that
the masticatory apparatus of Afropithecus was powerful (Leakey and Walker, 1997). Taken
together, its craniofacial and dental morphology indicate that Afropithecus was adapted to
exploiting a diet of hard foods, much like pitheciins.
This reconstruction of diet in Afropithecus differs from that inferred for Proconsul,
which was probably a more generalized frugivore. Dental and craniofacial anatomy of
Proconsul is only well-known for two of the four species, P heseloni and P nyanzae. The
functional discussion here is based on these species.
Proconsul had moderate shearing crests on its molars (Kay, 1977; Kay and Ungar,
1997) implying a frugivorous diet. Molar microwear studies (Walker et aI., 1994) further
support this interpretation. The interpretation that Proconsul was a frugivore is supported
by its inferred brain-body size ratio (Walker et aI., 1993), which is larger than expected for
a folivore. Proconsul species are all characterized by a moderate degree of body size di-
morphism, about 1.3: I, similar to that found in Pan troglodytes (Walker et aI., 1993; Raf-
ferty et aI., 1995). Canine dimorphism, however, is strong, with males having large,
blade-like canines and females having small, conical ones (Walker, 1997). The canines are
more vertically implanted than in Afropithecus, and are in line with the postcanine tooth
row. Thus, the canines of Proconsul were probably more important for intraspecific dis-
plays than diet, unlike the condition found in Afropithecus.
Proconsul incisors are vertically implanted, and appear small relative to molar size
(Harrison, 1982). This size relationship, however, could be a result of the proposed relative
molar megadontia (Rafferty et aI., 1995; Ruff et aI., 1989; Teaford et aI., 1993) of Procon-
sul, rather than diminished incisor size. The Proconsul face and mandible are more gracile
than that of Afropithecus, and there is no frontal trigon. It appears that the masticatory appa-
ratus was not as well-developed in Proconsul, and the tooth wear was not as severe, sup-
porting the interpretation of a diet of softer fruits in Proconsul than found in Afropithecus.
This divergent craniofacial morphology of Afropithecus and Proconsul suggests
quite different ecological and dietary niches for these two hominoids. These very different
skulls, however, are associated with strikingly similar postcranial skeletons. This disparity
illustrates the mosaic nature of hominoid evolution. Clearly, the generalized form of arbo-
 350 C. V. Ward

real locomotion practiced by these two apes allowed them to exploit widely different food
sources without necessitating any evolutionary change in the postcranial skeleton.

6. SUMMARY AND CONCLUSIONS

Afropithecus are Proconsul are the two best-known genera of early Miocene apes.
They are broadly similar in size, and strikingly so in postcranial morphology. Still, while
they appear to have had similar locomotor adaptations, they were divergent in craniofacial
form and inferred dietary specialization. In no two hominoid genera is the disparity
greater between postcranial similarity and craniofacial and dental dissimilarity.
Because these genera appear to share no particular phyletic relationship, and be-
cause their skeletons are broadly similar to those of many other early to middle Miocene
apes, it is reasonable to assume that the pattern of postcranial anatomy they share was
close to the primitive hominoid condition. Many extant hominoid postcranial synapomor-
phies were already present in these earliest apes, but those that reflect the most specialized
adaptations to forelimb-dominated climbing and suspensory locomotion seen in extant
apes do not appear to have evolved until much later in hominoid evolution. The disparity
in craniofacial and dental form between Afropithecus and Proconsul suggests that early
hominoid loeomotor adaptations did not limit or constrain ecological diversification, and
may even have facilitated the initial hominoid radiation.

ACKNOWLEDGMENTS
I would like to thank the editors for inviting me to be a part of the Primate Locomo-
tion 1995 symposium and volume. I would also like to express my admiration and appre-
ciation of the late Warren Kinzey for his pioneering contributions to the field of primate
studies, which formed the basis for much of the work presented here. I would like to ex-
press my gratitude to Mike Rose for inviting me to work on the Afropithecus fossils, to
Alan Walker and Mark Teaford for inviting me to work on the Proconsul fossils, and to
Meave Leakey, the Government of Kenya and the directors and staff of the National Mu-
seums of Kenya for the opportunity to work on these and other fossils, and to the editors
and reviewer for their helpful suggestions. This research was supported by the National
Science Foundation and by the L. S. B. Leakey Foundation.

REFERENCES
Beard KC, Teaford MF, and Walker A (1986) New wrist bones of Proconsul africanus and P. nyanzae from
Rusinga Island, Kenya. Folia Primatol. 47:97-118.
Begun DR (1994) Comparative and functional anatomy of Proconsul phalanges from the Kaswanga Primate Site,
Rusinga Island, Kenya. J. Hum. Evol. 26:89--165.
Begun DR, Ward CV, and Rose MD, eds. (1997a) Function, Phylogeny and Fossils: Miocene Hominoid Evolution
and Adaptations. New York: Plenum Press.
Begun DR, Ward CV, and Rose MD (I 997b) Events in hominoid evolution. In DR Begun, CV Ward and MD Rose
(eds.): Function, Phylogeny and Fossils: Miocene Hominoid Evolution and Adaptations. New York: Ple-
num Press, pp. 389--415.
Benefit BR, and McCrossin ML (1995) Miocene hominoids and hominid origins. Ann. Rev. Anthropol.
24:237-256.
Cartmill M, and Milton K (1977) The lorisiform wrist joint and the evolution of "brachiating" adaptations in the
Hominoidea. Am. J. Phys. Anthropol. 47:249--272.
Afropithecus, Proconsul, and the Primitive Hominoid Skeleton 351

Clark WE Le Gros, and Leakey LSB (1951) The Miocene Hominoidea of East Africa. Br. Mus. Nat. Hist. Fossil
Mamm.Afr.1:1-117.
Fleagle J, Stern J Jr., Jungers W, Susman R, Vangor A, and Wells J (1981) Climbing: A biomechanicallink with
brachiation and with bipedalism. Symp. Zoo\. Soc. Lond. 48:359-375.
Gebo DL, MacLatchy LM, Kityo R, Deino A, Kingston J, and Pilbeam 0 (1997) A hominoid genus from the early
Miocene of Uganda. Science 276:401-404.
Harrison T (1982) Small-Bodied Apes from the Miocene of East Africa. Ph.D. Dissertation, University of London.
Hopwood AT (1933) Miocene primates from British East Africa. Ann. Mag. Nat. Hist. 11:96-98.
Hunt KD (1992) Social rank and body size as determinants of positional behavior in Pan troglodytes. Primates
33:347-357.
Jenkins FA (1981) Wrist rotation in primates: A critical adaptation for brachiators. Symp. Zoo\. Soc. Lond.
48:429-451.
Kay RF (1977) Diets of early Miocene African hominoids. Nature 268:628-630.
Kay RF, and Ungar PS (1997) Dental evidence for diet in some Miocene catarrhines with comments on the effects
of phylogeny on the interpretation of adaptation. In DR Begun, CV Ward and MD Rose (eds.): Function,
Phylogeny and Fossils: Miocene Hominoid Evolution and Adaptations. New York: Plenum Press, pp.
131-151.
Keith A (1923) Man's posture, its evolution and disorders. Brit. Med. J. 1:451-672.
Kelley J (1995) A functional interpretive framework for the early hominoid postcranium. Am. J. Phys. Anthropo\.
Supplement. 20: 125.
Kinzey WG (1992) Dietary and dental adaptations in the Pitheciinae. Am. J. Phys. Anthropo\. 88:499-514.
Leakey M, and Walker A (1997) Afropithecus - function and phylogeny. In DR Begun, CV Ward and MD Rose
(eds.): Function, Phylogeny and Fossils: Miocene Hominoid Evolution and Adaptations. New York: Ple-
num Press, pp. 225-239.
Leakey MG (1985) Early Miocene cercopithecids from Buluk, Northern Kenya. Folia Primatol. 44:1-14.
Leakey RE, and Leakey MG (1986) A new Miocene hominoid from Kenya. Nature 324: 143-145.
Leakey RE, Leakey MG, and Walker AC (1988) Morphology of Afropithecus turkanensis from Kenya. Am. 1.
Phys. Anthropol. 76:289-307.
Lewis OJ (1969) The hominoid wrist joint. Am. 1. Phys. Anthropol. 30:251-268.
Lewis OJ (I972) Osteological features characterizing the wrists of monkeys and apes, with a reconsideration of
this region in Dryopithecus (Proconsul) africanus. Am. J. Phys. Anthropol. 36:45-58.
Lewis OJ (1974) The wrist articulations of the Anthropoidea. In FA Jenkins (ed.): Primate Locomotion. New York:
Academic Press, pp. 143-169.
Lewis OJ (1989) Functional Morphology of the Evolving Hand and Foot. Oxford: Oxford University Press.
McCrossin ML (1994) The phylogenetic relationships, adaptations and ecology of Kenyapithecus. Ph.D. Disserta-
tion, University of California, Berkeley.
McCrossin ML (1997) New postcranial remains of Kenyapithecus and their implications for understanding the ori-
gins of hominoid terrestriality. Am. J. Phys. Anthropo!. Supplement. 24:164.
McCrossin ML, and Benefit BR (1994) Maboko Island and the evolutionary history of Old World monkeys and
apes. In RS Corruccini and RL Ciochon (eds.): Integrative Paths to the Past. Englewood Cliffs: Prentice
Hall, pp. 95-122.
McCrossin ML, and Benefit BR (1997) On the relationships and adaptations of Kenyapithecus, a large-bodied
hominoid from the middle Miocene of eastern Africa. In DR Begun, CV Ward and MD Rose (eds.): Func-
tion, Phylogeny and Fossils: Miocene Hominoid Evolution and Adaptations. New York: Plenum Press, pp.
241-267.
Morbeck ME (1983) Miocene hominoid discoveries from Rudablinya: Implications from the postcranial skeleton.
In RL Ciochon and RS Corruccini (eds.): New Interpretations of Ape and Human Ancestry. New York: Ple-
num Press, pp. 369-404.
Napier J, and Davis P (1959) The forelimb skeleton and associated remains of Proconsul africanus. Foss. Mamm.
Afr.16:1-70.
Nengo 10, and Rae TC (1992) New hominoid fossils from the early Miocene site of Songhor, Kenya. J. Hum.
Evo!. 23:423-429.
Pitman MI, and Peterson L (1989) Biomechanics of skeletal muscle. In M Nordin and VH Frankel (eds.): Basic
Biomechanics of the Musculoskeletal System. Philadelphia: Lea and Febiger, pp. 89-111.
Rafferty KL (1990) The functional and phylogenetic significance of the carpometacarpal joint of the thumb in an-
thropoid primates. M.A. Thesis, New York University.
Rafferty KL, Walker A, RuffCB, Rose MD, and Andrews PJ (1995) Postcranial estimates of body weight in Pro-
consul, with a note on distal tibia of P. major from Napak, Uganda. Am. J. Phys. Anthropol. 97:391-402.
 352 C. V. Ward

Rose MD (1983) Miocene hominoid postcranial morphology: Monkey-like, ape-like, neither, or both? In RL Cio-
chon and RS Corruccini (eds.): New Interpretations of Ape and Human Ancestry. New York: Plenum Press,
pp.405--417.
Rose, MD (1984) Hominoid postcranial specimens from the Middle Miocene Chinji Formation, Pakistan. J. Hum.
Evol. 13:503-516.
Rose MD (1987) Another look at the anthropoid elbow. J. Hum. Evol. 17:193-224.
Rose MD (1992) Kinematics of the trapezium-I st metacarpal joint in extant anthropoids and Miocene hominoids.
J. Hum. Evol. 22:255-266.
Rose MD (1993a) Functional anatomy of the primate elbow and forearm. In DL Gebo (ed.): Postcranial Adapta-
tion in Nonhuman Primates. De Kalb: Northern Illinois Press, pp. 70-95.
Rose MD (I 993b) Locomotor anatomy of Miocene hominoids. In DL Gebo (ed.): Postcranial Adaptation in Non-
human Primates. De Kalb: Northern Illinois Press, pp. 252-272.
Rose MD (1994) Quadrupedalism in Miocene hominoids. J. Hum. Evol. 26:387-41l.
Rose MD (1997) Functional and phylogenetic features of the forelimb in Miocene hominoids. In DR Begun, CV
Ward and MD Rose (eds.): Function, Phylogeny and Fossils: Miocene Hominoid Evolution and Adapta-
tions. New York: Plenum Press. pp. 79-100.
Ruff CB, Walker A, and Teaford MF (1989) Body mass, sexual dimorphism and femoral proportions of Proconsul
from Rusinga and Mfangano Islands, Kenya. 1. Hum. Evol. 18:515-536.
Sanders WJ and Bodenbender BE (1994) Morphological analysis of lumbar vertebra UMP 67-28. J. Hum. Evol.
26:203-237.
Sarmiento EE (1988) Anatomy of the hominoid wrist joint: its evolutionary and functional implications. Int. J. Pri-
matol. 9:281-345.
Senut B (1980) New data on the humerus and its joints in Plio-Pleistocene hominids. Coli. Anthrop. 1:87-93.
Simons EL (1987) New faces of Aegyptopithecus, early human forebear from the Oligocene of Egypt. J. Hum.
Evol. 16:273-290.
Stem JT (1971) Functional myology of the hip and thigh of cebid monkeys and its implications for the evolution
of erect posture. Bibl. Primatol. Basel: Karger 14: 1-318.
Stern JT, Wells JP, Vangor AK, and Fleagle JG (1977) Electromyography of some muscles of the upper limb in
Ateles and Lagothrix. Yrbk. Phys. Anthropol. 20:498-507.
Strasser E (1993) Kaswanga Proconsulfoot proportions. Am. J. Phys. Anthropol. Supplement. 16:19l.
Teaford MF, Walker A, and Mugaisi GS (1993) Species discrimination in Proconsul from Rusinga and Mfangano
Islands. In WH Kimbel and LB Martin (eds.): Species, Species Concepts and Primate Evolution. New
York: Plenum Press, pp. 373-392.
Walker A (1997) Proconsul- function and phylogeny. In DR Begun, CV Ward and MD Rose (eds.): Function, Phy-
logeny and Fossils: Miocene Hominoid Evolution and Adaptations. New York: Plenum Press, pp. 209-224.
Walker A, Falk D, Smith R, and Pickford M (1983) The skull of Proconsul africanus: Reconstruction and cranial
capacity. Nature 305:525-527.
Walker A, and Pickford M (1983) New postcranial fossils of Proconsul africanus and Proconsul nyanzae. In R
Ciochon and R Corruccini (eds.): New Interpretations of Ape and Human Ancestry. New York: Plenum
Press, pp. 325-35l.
Walker A, and Rose MD (1968) Fossil hominoid vertebra from the Miocene of Uganda. Nature 217:980-98l.
Walker A, and Teaford MF (1988) The Kaswanga Primate Site: An Early Miocene hominoid site on Rusinga Is-
land, Kenya. J. Hum. Evol. 17:539-544.
Walker A, Teaford MF, and Leakey RE (1985) New information concerning the RI14 Proconsul site, Rusinga Is-
land, Kenya. In J Else and P Lee (eds.): Proc. Xth Congress of the Int. Primatol. Soc., Vol. I: Primate Evo-
lution. Cambridge: Cambridge University Press, pp. 143-149.
Walker A, Teaford MF, Martin L, and Andrews P (1993) A new species of Proconsul from the early Miocene of
RusingaiMfangano Islands, Kenya. J. Hum. Evol. 25:43-56.
Walker A, Teaford MF, and Ungar PS (1994) Enamel microwear differences between species of Proconsul from
the early Miocene of Kenya. Am. J. Phys. Anthropol. Supplement. 18:202-203.
Ward CV (1993) Torso morphology and locomotion in Proconsul nyanzae. Am. J. Phys. Anthropol. 92:291-328.
Ward CV (1997) Functional and phyletic implications of the hominoid trunk and hindlimb. In DR Begun, CV
Ward and MD Rose (eds.): Function, Phylogeny and Fossils: Miocene Hominoid Evolution and Adapta-
tions. New York: Plenum Press, pp. 101-130.
Ward CV, Walker A, and Teaford MF (1991) Proconsul did not have a tail. 1. Hum. Evol. 21:215-220.
Ward CV, Walker A, Teaford MF, and Odhiambo I (1993) A partial skeleton of Proconsul nyanzae from Mfangano
Island, Kenya. Am. 1. Phys. Anthropol. 90:77-11l.
Ward S, and Brown B (1996) Forelimb of Kenyapithecus africanus from the Tugen Hills, Baringo District, Kenya.
Am. J. Phys. Anthropol. Supplement. 22:240.
 19

FOSSIL EVIDENCE FOR THE ORIGINS OF

TERRESTRIALITY AMONG OLD WORLD
HIGHER PRIMATES

Monte L. McCrossin,' Brenda R. Benefit,' Stephen N. Gitau,'

Angela K. Palmer,' and Kathleen T. Blue2

'Department of Anthropology
Southern Illinois University
Carbondale, Illinois 62901
2Department of Anthropology
University of Chicago
Chicago, Illinois 60637

1. INTRODUCTION

Preference for terrestrial substrates is one of the most significant adaptive differ-
ences between some members of the radiation of Old World higher primates and the an-
thropoids of the Neotropics (Le Gros Clark, 1959; Napier and Napier, 1967, 1985;
Fleagle, 1988; Martin, 1990). Adaptations for terrestriality are most conspicuous among
savanna baboons (Papio - Rose, 1977), geladas (Theropithecus - Jolly, 1967; Dunbar and
Dunbar, 1974), and humans (Napier, 1967). Varying degrees of semi-terrestriality and ter-
restriality are also present among the African great apes (Gorilla - Remis, 1995 and Pan -
Hunt, 1992; Doran, 1993) and some of the Asian colobines (Presby tis entellus - Ripley,
1967 and Rhinopithecus roxellana - Davison, 1982), guenons (Cercopithecus aethiops and
Erythrocebus patas - Hall, 1965), mandrills and drills (Mandril/us - Jouventin, 1975),
mangabeys (Cercocebus - Waser, 1984), and macaques (e.g., Macaca nemestrina - Caldi-
cott, 1986). In contrast, terrestrial adaptations are notably absent from the otherwise di-
verse adaptive array of New World anthropoids.
Substrate preference exhibits clear linkages with mode of locomotion among Old
World higher primates (Ashton and Oxnard, 1964; Rose, 1973; Fleagle, 1988). The posi-
tional behavior of tree-dwelling Old World monkeys ranges from fairly deliberate
pronograde quadrupedalism to more active and agile styles of arboreal quadrupedal ism
supplemented by leaping and arm-swinging (Ripley, 1967; Fleagle, 1977; Rollinson and
Martin, 1981). Ground-dwelling cercopithecoids, in contrast, are primarily cursorial quad-

353
 354 M. L. McCrossin et at.

rupeds when on the ground and pronograde quadrupeds when in the trees (Rose, 1977).
Modes of locomotion differ more dramatically between the arboreal hominoids of Asia
and the terrestrially adapted African hominoids. Gibbons progress from branch to branch
through brachiation (Fleagle, 1974) and orang-utans use all four limbs to clamber through
the trees (MacKinnon, 1974; Cant, 1987) while gorillas and chimpanzees knuckle-walk
over the ground and use a diverse repertoire of positional behaviors, including orthograde
climbing and arm-swinging while in the trees (Hunt, 1992; Doran, 1993; Remis, 1995).
Beyond these obvious correlates between substrate preference and mode of locomo-
tion, terrestrially adapted Old World monkeys and apes have been observed to differ from
their arboreal relatives in several aspects of their ecology (tolerance and occasionally even
preference for more open environments, larger day ranges and home-ranges, greater geo-
graphic distribution, and greater niche separation among sympatric primate species), be-
havior (more allomothering, more aggressive anti-predator behaviors, and foraging
strategies directed toward harvesting and consuming a more diverse spectrum of foods),
social organization (more precocial infants, greater likelihood of having multi-male
groups, and larger group size), and anatomy (increased body size and higher levels of sex-
ua
dimorphism in canine height and body weight) (DeVore, 1963; Crook and Gartlan,
1966; Eisenberg et aI., 1972; Clutton-Brock and Harvey, 1977). A few of these changes in
ecology and behavior, especially a shift toward preference for open environments and an
eclectic (omnivorous) foraging strategy, have long been implicated in the origin of bipedal
Hominidae (Jolly, 1970; Lovejoy, 1981; Wolpoff, 1982; White, 1995).
Many of the distinctions drawn between ground-dwelling and tree-dwelling catarrhi-
nes have also been observed in comparisons between sympatric semi-terrestrial and arbo-
real lemurs at Antserananomby, Madagascar (Sussman, 1974). According to Sussman
(1974), the semi-terrestrial ring-tailed lemur (Lemur catta) differs from the arboreal brown
lemur (L. fulvus rufus) ecologically in its use of more arid habitats, behaviorally in its
more varied diet, and socially in its larger group size. The existence of these same differ-
ences in both arboreal and terrestrial strepsirhines and catarrhines tends to support view-
ing these contrasts as either direct causes or consequences of terrestriality, rather than the
result of common inheritance due to close phylogenetic relationships. Among some Afri-
can cercopithecines, however, the influence of phylogenetic heritage seems to over-ride
the effects wrought by differences of substrate preference. Rowell (1966) and Struhsaker
(1969) have demonstrated that comparisons between arboreal and terrestrial cercopi-
thecines tend not to support the correlations between terrestriality and larger group size
seen in some other primate groups.
Although there is an extensive literature seeking to explain the origins of terrestrial
bipedalism in early hominids (Jolly, 1970; Lovejoy, 1981; Wolpoff, 1982; White, 1995),
the change from life in the trees to life on the ground among non-human catarrhines has
received much less attention. While the evolutionary history of some other fundamental
aspects of primate adaptation are fairly well understood, such as diurnal and nocturnal ac-
tivity patterns among living primates (Martin, 1979), it is not known how many times ter-
restrial adaptations have been independently acquired in the evolutionary history of
cercopithecoids (Napier, 1970; Rollinson and Martin, 1981; Andrews and Aiello, 1984;
Benefit, 1987; Pickford and Senut, 1988). Similarly, the antiquity of terrestrial adaptations
among large-bodied Miocene hominoids and its relevance for understanding the emer-
gence of knuckle-walking and bipedalism among African apes and humans are debated
(Washburn, 1968; Conroy and Fleagle, 1973; Tuttle, 1974; Gebo, 1992; McCrossin, 1997).
Most research concerning catarrhine adaptations for life on the ground has focussed
upon the functional anatomy of living species (e.g., Jolly, 1967; Tuttle, 1967). The com-
Fossil Evidence for the Origins of Terrestriality among Old World Higher Primates 355

parative anatomy of modern primates provides information concerning osteological corre-

lates of particular positional behaviors. Osteological correlates of locomotor pattern are
crucial for reconstructing the substrate preferences of fossil primates (e.g., Oxnard, 1963;
Jolly, 1967; Knussman, 1967; Tuttle, 1967). Neontological comparisons, however, cannot
address the contextual circumstances surrounding the origins of catarrhine terrestriality.
It is most fortunate, therefore, that recent advances in our understanding of the fossil
record of early Old World monkeys and apes provide much new data concerning the ori-
gins of terrestriality among catarrhines. Consideration of these new data, some presented
here for the first time, provide new ideas concerning the timing, ecological context, and
adaptive correlates of the origins ofterrestriality among Old World higher primates. In this
paper, we attempt to clarify the origins of terrestriality among catarrhines by reviewing
postcranial indicators of semi-terrestriality, paleoenvironmental setting, considerations of
body size, and reconstruction of the dietary adaptations of Victoriapithecus and Kenyapi-
thecus.

1.1. Suggested Causes of the Transition from Arboreality to

Terrestriality
Three major factors have been advanced as primary causes of the shift from life in
the trees to life on the ground among Old World higher primates. First, it is widely thought
that the transition from arboreality to terrestriality was spurred on by a widespread change
in environments (Brain, 1981). The arboreality of the archaic catarrhines of the Oligocene
(Conroy, 1976; Fleagle, 1983) and the formative hominoids of the early Miocene (Napier
and Davis, 1959; Rose, 1993) is inextricably linked with the fact that they inhabited
densely forested environments (Bowen and Vondra, 1974; Andrews and Van Couvering,
1975; Van Couvering 1980; Bown et al. 1982). The terrestrial adaptations of many of the
cercopithecoids (Jolly, 1967; Birchette, 1981; Ciochon, 1993) and the early hominids
(Lovejoy, 1981) of the Plio-Pleistocene of eastern Africa are widely regarded as resulting
from colonization of the patchily vegetated savannas. Second, the shift from life in the
trees to life on the ground is correlated with an increase in body size, with terrestrial pri-
mates tending to be larger than their arboreal relatives (Andrews, 1983; Fleagle, 1985;
Doran, 1993). Third, a fundamental change in diet and foraging strategy is commonly
thought to have been a primary motive force in the advent of terrestriality, especially
among hominoids (Jolly, 1970; Coursey, 1973; Szalay, 1975; White, 1995). The first
ground-dwelling catarrhines are thought to have exploited terrestrial food sources such as
grass seeds (Jolly, 1970), tuberous roots (Coursey, 1973), and animal carcasses (Szalay,
1975), in contrast to the concentration of arboreal primates upon foods available in the
trees, primarily fruit and leaves.
It is unclear, however, to what degree occupation of open environments, increased
body size, and consumption of terrestrial foods might be consequences rather than causes
of life on the ground. The fossil record provides evidence that allows separation of cause
from consequence in factors associated with the advent of terrestriality among Old World
higher primates.
The earliest clear evidence of terrestrial adaptations among Old World higher primates
is seen in two genera from the middle Miocene site ofMaboko Island in Kenya (Benefit and
McCrossin, 1989; Feibel and Brown, 1991; McCrossin and Benefit, 1994), the cercopithe-
coid Victoriapithecus (Von Koenigswald, 1969; Benefit, 1987, 1993, 1994; Benefit and
McCrossin, 1991, 1993, 1997; McCrossin and Benefit, 1992) and the large-bodied homi-
noid Kenyapithecus (Le Gros Clark and Leakey, 1951; Leakey, 1962, 1967; McCrossin and
356 M. L. McCrossin et al.

Benefit, 1993, 1997; McCrossin, 1994a). The fossil primate community from Maboko Is-
land also includes two bushbabies -Komba and a new genus (McCrossin, 1990, 1992a), two
small-bodied primitive catarrhines - Simiolus and cf. Limnopithecus (Benefit, 1991), and an
oreopithecid - Mabokopithecus (Von Koenigswald, 1969; McCrossin, 1992b; Benefit et aI.,
1998). Victoriapithecus has also been identified at the early Miocene site of Napak in
Uganda (Pilbeam and Walker, 1968; Pickford et aI. 1986) as well as from two other sites
from the middle Miocene of Kenya: Baragoi (Ishida, 1986) and Nyakach (Pickford and
Senut, 1988). Kenyapithecus also occurs at the Kenyan middle Miocene sites of Fort Ternan
(Leakey, 1962), Majiwa and Kaloma (Pickford, 1982a), Nyakach (Pickford, 1985), Baragoi
(Ishida et aI. 1984) and Kipsaramon (Brown et aI., 1991).

2. TAXONOMY AND PHYLOGENETIC RELATIONSHIPS OF

VICTORIAPITHECUS AND KENYAPITHECUS

Von Koenigswald (1969) named two species of Victoriapithecus, V. macinnesi and V.

leakeyi, based on a small sample of cercopithecoid dentognathic and postcranial remains
from Maboko Island. Because of the retention of a crista obliqua on its upper molars, von
Koenigswald (1969) suggested that Victoriapithecus was a basal cercopithecid and as-
signed the genus to a distinct subfamily (Victoriapithecinae), ancestral to both Cercopi-
thecinae and Colobinae.
Delson (1973, 1975) reassessed the Maboko Island cercopithecoids and suggested
that Victoriapithecus macinnesi was an early colobine while V. leakeyi was an early cer-
copithecine. Delson (1975) perceived colobine-like arboreal adaptations in a distal
humerus (KNM-MB 3) and an intermediate phalanx (KNM-MB 93) from Maboko and at-
tributed them to V. macinnesi. Delson (1975) referred all other cercopithecoid postcranial
remains then known from Maboko to V. leakeyi, describing cercopithecine-like terrestrial
adaptations for another distal humerus (KNM-MB 19), two proximal ulnae (KNM-MB 2
and 32), a calcaneum (KNM-MB 16), a proximal phalanx (KNM-MB 12) and three inter-
mediate phalanges (KNM-MB 13,21, and 22). Analysis of the greatly expanded sample of
dentognathic remains of Victoriapithecus (Pickford, 1984; Benefit and McCrossin, 1989),
however, has indicated that only one species of cercopithecoid, V. macinnesi, is present at
Maboko Island (Benefit, 1987, 1993, 1994) and Delson (Strasser and Delson, 1987) has
subsequently abandoned his hypothesis that Colobinae and Cercopithecinae are repre-
sented in the craniodental assemblage from Maboko. Reassessment of the range of vari-
ation in Victoriapithecus postcranial remains (Harrison, 1989), including several new
specimens collected on Maboko from 1982 to 1984 (Pickford, 1986), also support the ex-
istence of a single species (but see Senut, 1986a and Pickford and Senut, 1988 for argu-
ments in support of two species). The absence of numerous craniodental synapomorphies
of Colobinae and Cercopithecinae demonstrates that Victoriapithecus belongs to the sister-
taxon (Victoriapithecidae) of Cercopithecidae (Benefit, 1987, 1993, 1994; Benefit and
McCrossin, 1991, 1993, 1997).
The taxonomy of Kenyapithecus is a matter of long dispute. The genus was pro-
posed by Leakey (1962) based on material from Fort Ternan, with K. wickeri as the type
species. Subsequently, Leakey (1967) transferred other specimens from Maboko Island,
including the type specimen of Sivapithecus africanus (Le Gros Clark and Leakey, 1951),
to the genus as K. african us.
Although this taxonomic arrangement has been confirmed by most workers who
have studied the original fossil samples (Ishida et aI., 1984; Ishida, 1986; Brown et aI.,
Fossil Evidence for the Origins of Terrestriality among Old World Higher Primates 357

1991; McCrossin and Benefit, 1993, 1994, 1997; McCrossin, 1994a), there have been re-
peated suggestions of generic distinctions between Kenyapitheeus wiekeri and K. afri-
eanus (Simons and Pilbeam, 1978; Pickford, 1985; Andrews, 1992; Harrison, 1992).
According to Pickford (1985), K. wiekeri differs from K. afrieanus in having a palatal
process that is more highly arched, a maxillary sinus that is positioned at a higher level, an
anterior root of the zygomatic arch that originates at a higher level above the molar alveo-
lar plane and flares more markedly, a palatal process that is more highly arched, a maxil-
lary sinus that is positioned at a higher level, an II that lacks a distinct lingual pillar, a
more vertically oriented canine socket, a p 3 with one (rather than two labial roots), a p4
with weaker transverse ridges, and M I.2 without a lingual cingulum. Detailed metric and
morphological examination of these features, however, demonstrates that the suggested
differences between K. wiekeri and K. afrieanus do not exceed levels of variation ob-
served within extant hominoid genera (McCrossin, 1994a; Benefit and McCrossin, 1994;
McCrossin and Benefit, 1997). In addition, K. wiekeri and K. afrieanus share several de-
rived features compared with other Miocene hominoids (McCrossin, 1994a; McCrossin
and Benefit, 1997), including a distinctively proclined mandibular symphysis (McCrossin
and Benefit, 1993) and lower molar roots that extend inferiorly almost to the basal margin
of the mandibular corpus (Brown, 1997).
The phylogenetic relationships of Kenyapitheeus have also elicited much interest.
Influential suggestions of the phylogenetic relationships of Kenyapitheeus include affini-
ties to the great ape and human clade in general (Greenfield, 1979; Ward and Pilbeam,
1983; Andrews, 1985; McCrossin and Benefit, 1993) and to Homo specifically (Leakey,
1962, 1967; Andrews, 1971; Walker and Andrews, 1973; Simons and Pilbeam, 1978). An-
other recent suggestion is that Kenyapitheeus, Sivapitheeus, and Dryopitheeus are mem-
bers of an archaic radiation of apes antecedent to the last common ancestor of living
hominoids (McCrossin, 1994a; McCrossin and Benefit, 1994, 1997; Benefit and
McCrossin, 1995; Pilbeam, 1997). Finally, it has recently been suggested that Kenyapi-
thee us exhibits craniodental and postcranial features indicating that it was the earliest
known member of the African great ape and human clade (Ishida, 1986; Brown and Ward,
1988; McCrossin, 1997). A phylogenetic tree showing the inferred relationships of Vieto-
riapitheeus and Kenyapitheeus is shown in Figure 1.

3. SKELETAL INDICATORS OF SEMI-TERRESTRIALITY

Despite numerous pervasive differences in postcranial morphology indicative of

their respective cercopithecoid and hominoid affinities (von Koenigswald, 1969; Harrison,
1989; McCrossin and Benefit, 1992; McCrossin, 1994a), many skeletal indicators of semi-
terrestriality are shared by both Vietoriapitheeus and Kenyapitheeus. Functional implica-
tions of these features have been considered in greater detail elsewhere (McCrossin,
1992b, 1994a; McCrossin and Benefit, 1994, 1997) and will only be briefly reviewed
here.

3.1. Victor;ap;thecus
The proximal humerus of Vietoriapitheeus (Figure 2) was until recently known only
from an isolated articular end, KNM-MB 12044 (Senut, 1986a; Harrison, 1989). In isola-
tion, this specimen was interpreted as indicating that the articular surface of the humeral
head extended farther proximally than the greater tubercle (Senut, 1986a; Harrison, 1989).
358 M. L. McCrossin et al.

A egyptopithecus

Victoriapithecus

Colobinae Figure 1. Branching diagram of the phylogenetic re-

lationships of Victoriapithecus and Kenyapithecus to
Cercopithecinae other Old World higher primates. The sister-group re-
lationship of Victoriapithecus to Cercopithecidae is
Proconsul supported by synapomorphic cercopithecoid features
of the cranium, facial skeleton, dentition, and post-
Hylobates
cranium and by the fact that living colobines and cer-
Sivapithecus copithecines share several additional features,
especially in the deciduous dentition and permanent
Pongo molars, not seen in Victoriapithecus (von Koenig-
swald, 1969; Benefit, 1987, 1993, 1994; Benefit and
Kenyapithecus McCrossin, 1991, 1993, 1997; Harrison, 1989;
McCrossin and Benefit, 1992). Affinity of Kenyapi-
Gorilla thecus to the African great ape and human clade is
supported by the subnasal pattern, the morphology of
Pan
dP3' the humerofemoral index, the volar slant of the
Ardipithecus inferior radioulnar joint, the presence of a metacarpal
torus, and the articular morphology of the calcaneum,
Australopithecus cuboid, and navicular (Ishida, 1986; Brown and
Ward, 1988; McCrossin and Benefit, 1993;
Homo McCrossin, 1994a, 1997; McCrossin et a\., 1998).

:,'."~ ~
r••l \ .1

y
'\.~. "

...: .., !-
--1r"=-'''''

a b c

~
.
,)'{ ........
.•.r •.
J ' -\

.'~"'-
.... .1'<
- •.•• ~\y ',.
. . ~ : .
o
eM ,J:,'ft!
'. '; So; ' .

h
J ,

d e

Figure 2. Isolated left proximal humerus of Victoriapithecus (KNM-MB 12044) in anterior (a), posterior (b),
proximal (c), lateral (d), and medial (e) views.
Fossil Evidence for the Origins of Terrestriality among Old World Higher Primates 359

Senut (1986a) and Harrison (1989) suggested that this perceived configuration related to a
degree of shoulder mobility and arboreality for Victoriapithecus comparable to that seen in
living arboreal guenons and colobines. Because the proximal disposition of the humeral
head and greater tubercle are determined in part by the curvature of the humeral shaft
(McCrossin, 1994a), however, it is not possible to assess this feature with proximal
humerus specimens isolated from their shafts, including KNM-RU 17376 (Proconsul afri-
canus or Dendropithecus macinnesi; Gebo et aI., 1988), KNM-MB 21206 (Mabokopi-
thecus; McCrossin, 1992b), and GSP 28062 (?Catarrhini indet.; Rose, 1989) as well as
KNM-MB 12044 (v. macinnesi; Senut, 1986a; Harrison, 1989).
An almost complete humerus of Victoriapithecus recently collected from Maboko Is-
land reveals that the greater tubercle actually extends above the surface of the humeral
head (Figure 3). This proximal extension of the greater tubercle is due in large part to the
curvature of the humeral shaft, especially in the region of the deltopectoral crest. In extant
hominoids, colobines (except some individuals of Presby tis entellus as well as the extinct
species Mesopithecus pentelici and Cercopithecoides williamsi; Birchette, 1981, 1982),
guenons (except Cercopithecus aethiops and Erythrocebus patas) and New World mon-
keys, the humeral head extends above the greater tubercle (McCrossin, 1994a: Figure 28).
In terrestrial cercopithecoids, such as Erythrocebus, Papio, and Theropithecus, the greater
tubercle extends high above the humeral head. Possible allometric influences were exam-
ined by Birchette (1982: 151) who concluded that proximal extension of the greater tuber-
cle above the humeral head appears not to be related to "mechanical demands placed on
supraspinatus by the increase in limb weight associated with a general increase in body

1IIIIIuul
Figure 3. Right humerus (missing capitulum and part of trochlea) of Victoriapithecus (KNM-MB 21809) in lateral
view. Curvature of the shaft in the deltopectoral region results in proximal extension of greater tubercle slightly
above the articular surface. (The scale bar equals one centimeter.)
360 M. L. McCrossin et al.

weight" and thus is not a function of large body size in Erythrocebus, Papio, and Theropi-
thecus.
Jolly (1967) suggested that elevation of the greater tubercle above the humeral head
is an adaptation to terrestrial walking and running. This arrangement serves to resist pas-
sive retraction of the arm that results from weight-bearing at the glenohumeral joint and
enhances forceful protraction of the arm by providing for elongation of the moment arm of
m. supraspinatus (Jolly, 1967). Electromyographic studies, however, indicate that the most
important function of m. supraspinatus may be to stabilize the shoulder during the stance
phase of quadrupedal ism (Larson and Stem, 1992). The lower lesser tubercle and shorter
moment arm of m. supraspinatus in platyrrhines, hominoids, and arboreal cercopithecoids
is related to greater mobility at the glenohumeral joint because it results in faster (albeit
weaker) protraction of the humerus (Jolly, 1967; Birchette, 1982). The greatest degree of
humeral head projection, seen in atelines, Hylobates, and Pongo, allows for extreme mo-
bility and rotatory capabilities at the glenohumeral joint.
The distal humerus of Victoriapithecus (Figure 4) exhibits a combination of features,
all of which seem to be shared derived similarities with living cercopithecoids (Pickford
and Senut, 1988; Harrison, 1989). The distal humerus articular surface of Victoriapithecus
(Figure 4) resembles modern cercopithecoids in that, overall, it is narrow, the trochlea is
narrower than the capitulum, and a strong trochlear keel is present medially (Pickford and
Senut, 1988; Harrison, 1989). The broad and somewhat flattened capitulum of Victoriapi-
thecus (Pickford and Senut, 1988; Harrison, 1989) indicates that primary transference of
weight is through the humeroradial joint while the narrow trochlea and strong medial tro-
chlear keel reflect emphasis on restriction of the humeroulnar joint to hinge-like flexion
and extension motions (Napier and Davis, 1959; Jenkins, 1973).
One of the most distinctive features of the Victoriapithecus distal humerus is the
strong posterior inclination of the medial epicondyle (Pickford and Senut, 1988). Delson
(1973, 1975) claimed that the variation in medial epicondyle inclination seen in the hu-
meral remains of Victoriapithecus supported assigning one specimen to an arboreal colo-
bine and another specimen to a terrestrial cercopithecine. Harrison (1989: 18), however,

Figure 4. Left distal humeri of Victoriapithecus

(left column: KNM-MB 3; right column: KNM-
MB 19) in posterior (top row) and distal (bottom
row) views. The strong medial trochlear keel and
11I1I1i1l11 11111111111 posterior orientation of the medial epicondyle in
KNM-MB 19 are some of the most compelling in-
dicators of semi-terrestriality for Victoriapithecus.
The more medially oriented medial epicondyle of
KNM-MB 3 was once regarded as indicating that it
represented a more arboreal taxon, but the differ-
ence in medial epicondyle orientation seen in
KNM-MB 3 and 19 does not exceed intraspecific
variation in living species. (The scale bars equal
11111111111 one centimeter.)
Fossil Evidence for the Origins of Terrestriality among Old World Higher Primates 361

has shown that the variation in Victoriapithecus medial epicondyle inclination "does not
exceed the variation seen in extant catarrhine species."
In hominoids and platyrrhines, the medial epicondyle is large and medially directed,
reflecting the premium placed on digital grasping (Fleagle and Simons, 1982). In contrast,
in cercopithecoids, especially the semi-terrestrial and terrestrial cercopithecines, the me-
dial epicondyle is abbreviated and posteromedially oriented (Jolly, 1967). This shortening
and posterior reflection of the medial epicondyle in ground-dwelling cercopithecines has
been related to a reduction in the mass of the carpal and digital flexors, which take their
origin from the entepicondyle, and to an increase in the moment arm of m. pronator teres
around the axis of pronation (Jolly, 1967; Birchette, 1982).
The proximal ulna of Victoriapithecus (Figure 5) is dominated by a moderately long
and dorsally deflected ulnar olecranon (von Koenigswald, 1969; Harrison, 1989). This
combination of a moderately long and retroflexed olecranon is seen among semi-terrestrial
and terrestrial cercopithecoids (Jolly, 1967). Arboreal cercopithecoids usually have longer
and straighter olecranons (Oxnard, 1963; Jolly, 1967). Angles of olecranon retroflexion
differ greatly among extant cercopithecoids, with means between 25 (Procolobus) and 37
degrees (Rhinopithecus) for modem colobine genera (Birchette, 1982: Table 25) while in-
dividuals of Papio range from 40 to 60 degrees (Jolly, 1972).
The olecranon is the site of insertion of m. triceps brachii, the major extensor of the
forearm, and "the triceps is at its greatest mechanical advantage when its line of action lies
farthest from the axis of the elbow joint, i.e., when it is perpendicular to the axis of the
olecranon" (Birchette, 1982:241). Thus, retroflection of the olecranon may help to en-
hance the thrust produced by m. triceps brachii when the elbow is in postures approaching
maximum extension (Jolly, 1967).

Figure 5. Right proximal ulna of Victoriapithecus

(KNM-MB 32) in lateral (left) and medial (right)
views. The slight dorsal angulation of the olecranon
process with respect to the longitudinal axis of the ul-
nar shaft is a resemblance to living semi-terrestrial
Old World monkeys. (The scale bars equal one centi-
meter.) 1,,"1,,"\ 11111111111
 362 M. L. McCrossin et al.

The radial neck is relatively very short, indicating that "Victoriapithecus was prob-
ably capable of rapid locomotion on level surface substrates" (Harrison, 1989:25). The
length of the radial neck corresponds to the moment arm of m. biceps brachii and conse-
quently relates to the speed and forcefulness of elbow flexion (Napier and Davis, 1959).
Among arboreal anthropoids, a premium is placed on the force produced by m. biceps
brachii in elbow flexion during hoisting and climbing and the radial neck is relatively
long (Conroy, 1976). Terrestrial cercopithecoids, in contrast, rely upon rapid flexion of the
elbow during the recovery phase of cursorial quadrupedalism and the radial neck is rela-
tively short (Jolly, 1967). Recently, Reno et al. (1997:197) have questioned the proposed
correlation between radial neck length and substrate preference because total length of the
radius is correlated to body mass and "lengthening of the radius attendant to increasing
body mass is differentially expressed by distal (rather than proximal) growth of the ra-
di us". According to Reno et al. (1997: 197), "any association of relative radial neck length
with substrate preference such as that reported in monkeys" by Harrison (1989) "merely
reflects selection on increased body mass in a terrestrial environment". Radial neck length
among anthropoids, however, does not appear to be wholly dictated by allometric effects.
When indexed against the minimum diameter of the radial head, the radial necks of arbo-
real anthropoids are relatively long while those of semi-terrestrial and terrestrial cercopi-
thecoids are relatively short (Harrison, 1989: Fig. 11).
The ischial body of Victoriapithecus is relatively long, as in semi-terrestrial and ter-
restrial cercopithecoids (McCrossin and Benefit, 1992:280, Table 1, Fig. 1). The length of
the ischium corresponds to the lever arm of the ischiocrural musculature, upon which the

11111111111

Figure 6. Right femur (lacking distal epiphysis) of Victoriapithecus (KNM-MB 20230) in anterior view. The low
femoral neck-shaft angle seen in this specimen indicates mainly parasagittal movements of the thigh in adducted
postures. (The scale bar equals one centimeter.)
Fossil Evidence for the Origins of Terrestriality among Old World Higher Primates 363

torque produced by the hamstrings depends (Waterman, 1929; Leutenegger, 1970). In ad-
dition, a raised line from the origin of m. gemellus extends continuously along the ischial
body of Victoriapithecus, from the ischial spine to the tuberosity (McCrossin and Benefit,
1992). This configuration is also seen in semi-terrestrial cercopithecoids such as Cercopi-
thecus aethiops and Erythrocebus patas while markings from the origin of m. gemellus are
more proximally restricted in arboreal Old World monkeys such as Lophocebus albigena
and Presby tis rubicundus.
Until recently, the femur of Victoriapithecus was known only from four very poorly
preserved fragments of the femoral head, one preserving a small portion of the femoral
neck (Harrison, 1989). An almost complete femur of Victoriapithecus, lacking only the
distal epiphysis, was recently collected from Maboko Island (Figure 6). The previously
collected fragments and the new femur show that the femoral head of Victoriapithecus is
relatively small, its articular surface continues onto the posterior and proximal sides of the
femoral neck, the femoral neck is quite short, and the neck angle is low. Although shared
by both tree-dwelling and ground-dwelling cercopithecoids, these features are associated
with greater efficiency of parasagittal movements of the thigh (Jenkins and Camazine,
1977) and may be related to adaptations for walking and running over level substrates.
The entocuneiform of Victoriapithecus exhibits a first metatarsal facet that is mainly
confined to the distal end and extends for only a short distance medially (Harrison, 1989).
This morphology is also seen in living terrestrial cercopithecoids, such as Erythrocebus,
Papio, and Theropithecus, and is associated with a limited range of hallucial abduction
(Jolly, 1967). Arboreal cercopithecoids, in contrast, have a much more medially extensive
first metatarsal facet on the entocuneiform, allowing a great range of hallucial abduction
(Napier and Davis, 1959).
Preliminary indications from study of the greatly expanded sample of cercopithecoid
pedal specimens from Maboko Island (Strasser, 1997:222) re-affirm that the functional at-
tributes linking Victoriapithecus to living Old World monkeys are "related to stabilizing
the foot at the upper and lower ankle joints, increased pronation of the forefoot and re-
duced hallucial abduction". The morphocline polarity of five cercopithecine-like charac-
ters in the foot skeleton of Victoriapithecus are unresolved and "may reflect a shared
habitus rather than a phyletic link" (Strasser, 1997:222 - her emphasis).
Delson (1973, 1975) divided Victoriapithecus intermediate phalanges into two cate-
gories. One specimen, KNM-MB 93, perceived as being long and slender, was referred to
V. macinnesi and regarded as representing an arboreally adapted colobine (Delson, 1973,
1975). Three other specimens (KNM-MB 13, 21 and 22), described as being relatively
short and robust, were attributed to V. leakeyi, a species that Delson (1973, 1975) claimed
was a terrestrially adapted cercopithecine. Reassessment of these and additional speci-
mens, however, reveals that the range of variation seen in Victoriapithecus does not ex-
ceed that seen in living Old World monkey species (Harrison, 1989). In fact, the
"relatively short and stout" proportions of the phalanges "indicate that Victoriapithecus
was suited to moving effectively on a terrestrial substrate, and implies that terrestrial digi-
tigrady may have been an important component of its locomotor repertoire" (Harrison,
1989:38).
In summary, the postcranial skeleton of Victoriapithecus exhibits numerous indica-
tors of semi-terrestriality and few, if any, adaptations for the degree of arboreality seen
among arboreal guenons and colobines. This interpretation differs from that proposed by
Delson (1973, 1975), who presented one set of cercopithecoid postcrania from Maboko Is-
land as belonging to an arboreally adapted colobine and another group of specimens as be-
longing to a terrestrially adapted cercopithecine. Although Senut's (1986a; Pickford and
364 M. L. McCrossin et aL

Senut, 1988) and Harrison's (1989) studies did not support Delson's (1973, 1975) division
of the Maboko Island cercopithecoids into colobines and cercopithecines, they also per-
ceived the presence of both arboreal and terrestrial adaptations in the sample of Victoriapi-
thecus postcrania. Senut (1986a) interpreted these differences as indicating the presence of
two species. Harrison's (1989) results, in contrast, were seen as supporting Benefit's
(1987) contention that only one cercopithecoid species, V. macinnesi, is present at Maboko
Island. Harrison's (1989:50) perception of arboreal adaptations in the shoulder anatomy of
Victoriapithecus led him to speculate that Victoriapithecus "may have been capable of ag-
ile climbing and clambering in the multi-dimensional small-branch milieu of the upper
canopy". As discussed previously, however, the isolation from the humeral shaft of the
KNM-MB 12044 proximal humerus does not allow interpretation ofthe disposition of the
greater tubercle and the kinematics of m. supraspinatus of Victoriapithecus. While we
have long recognized that the semi-terrestriality of Victoriapithecus involved utilization of
both arboreal and terrestrial substrates and resources (Benefit, 1987; McCrossin and Bene-
fit, 1992, 1994), we see no evidence from the postcranial skeleton for agile arboreal
climbing similar to that of arboreal guenons and colobines (contra Harrison, 1989).

3.2. Kenyapithecus
Semi-terrestrial adaptations of the limb skeleton of Kenyapithecus from Maboko Is-
land and Fort Ternan have only recently been recognized (McCrossin, 1994a,b; McCrossin
and Benefit, 1992, 1994, 1997; Benefit and McCrossin, 1995). Previously, postcranial re-
mains of Kenyapithecus from Maboko Island and Fort Ternan, as well as from Baragoi,

Figure 7. Left humerus of Colobus guereza, Pan panis-

cus, and Kenyapithecus composite reconstruction (includ-
ing the following specimens: KNM-MB 24729, left
proximal humerus; BMNH M. 16334, left humerus shaft;
KNM-FT 2751, right distal humerus- reversed) in anterior
view. The humeral shaft and proximal end of Kenyapi-
thecus from Maboko Island were collected in 1933 and
Colobus Pan Kenyapithecus 1992, respectively. A jagged reciprocal pattern of breakage
allows them to be conjoined. (The scale bar equals five
I I I I centimeters. )
Fossil Evidence for the Origins of Terrestriality among Old World Higher Primates 365

Figure 8. Left humerus of Colobus guereza, Pan

paniscus, and Kenyapithecus composite reconstruc-
tion (including the following specimens: KNM-MB
24729, left proximal humerus; BMNH M. 16334,
left humerus shaft; KNM-FT 2751, right distal
humerus - reversed) in posterior view. The greater
tubercle extends proximally farther than the articu-
lar surface of the humeral head in Kenyapithecus. Colobus Pan Kenyapithecus
(The scale bar equals five centimeters.) I I I

have been interpreted as indicating Proconsul-like arboreal quadrupedal adaptations (An-

drews and Walker, 1976; Rose, 1983, 1993; Harrison, 1992; Rose et ai., 1996).
The most compelling features indicative of semi-terrestriality for Kenyapithecus
concern the humerus (Figures 7-8). The head of the humerus is directed posteriorly as in
most anthropoids, including terrestrially adapted cercopithecoids. This pattern of gleno-
humeral articulation is most effective in the predominantly rectilinear protraction and re-
traction of the humerus employed by all quadrupedal primates, but most efficiently by the
cursorial ground-dwelling cercopithecines. The posterior orientation of the humeral head
of Kenyapithecus differs from the medial orientation of the humeral head seen in extant
hominoids and atelines. Medial orientation of the humeral head is most extreme in Gorilla
and Pan, and enables maintenance of semi-pronated forearm postures during pronograde
quadrupedal ism (Larson, 1988).
Like cercopithecoids and most ceboids, the humeral shaft of Kenyapithecus col-
lected from Maboko Island in 1933 (Le Gros Clark and Leakey, 1951) is flexed approxi-
mately one-third of the way down its length, due to strong development of the
deltopectoral crest. Napier and Davis (1959) suggested that flexion of the humeral shaft is
associated with quadrupedal modes of locomotion among primates. As preserved, how-
ever, the shaft of another Kenyapithecus humerus collected from Maboko Island in 1996 is
quite straight in medial view (Figure 9), like that of living hominoids (McCrossin, 1997).
The greater tubercle of Kenyapithecus (Figure 10) extends slightly farther superiorly
than the level of the articular surface of the humeral head (McCrossin, 1994a; McCrossin
and Benefit, 1997). The articular surface of the humeral head of Kenyapithecus is also
366 M. L. McCrossin et al.

Figure 9. Right humeral shaft of Kenyapithecus collected from Maboko Island in 1996 (middle), compared with
humeri of Papio (left) and Homo (right) in medial view. As preserved, this Kenyapithecus specimen shares a
straight humeral shaft with modern hominoids, including humans, rather than the flexed humeral shaft of cercopi-
thecoids.

quite flattened proximally. Among agile arboreal anthropoids, especially suspensory forms
such as Ateles, Lagothrix, and Hylobates, the proximal surface of the humeral head is
strongly domed, forming a ball-like articulation for the glenoid cavity of the scapula.
In order to gauge the overall phenetic resemblance of the Kenyapithecus proximal
humerus to other anthropoids, a principal components analysis was performed using five
natural logarithm transformed variables (McCrossin, 1994a). The first component ac-
counts for 88% of the variance and is derived from the first four variables (proximodistal
height of the head, mediolateral breadth of the head, lesser tubercle diameter, and depth of
the bicipital groove) while the breadth of the bicipital groove is the primary determinant
of the second component, which accounts for 8% of the variance (McCrossin, 1994a: Fig.
32). A plot of the first and second principal components (Figure 11) shows Kenyapithecus
positioned closest to terrestrial and semi-terrestrial cercopithecoids such as Mandrillus
Fossil Evidence for the Origins of Terrestriality among Old World Higher Primates 367

(including both the mandrill and drill), Papio, Presby tis entellus, Theropithecus gelada,
and Macaca nemestrina.
Distally, the humeral medial epicondyle of Kenyapithecus also resembles that of Vic-
toriapithecus and terrestrially adapted cercopithecoids in being very short and posteriorly
oriented (Figure 12). The angle of posterior reflection of the medial epicondyle seen in K.
wickeri (54 degrees) is most comparable to values for Erythrocebus patas (mean = 51 de-
grees, mean ± I s.d. = 46-57 degrees; Fleagle and Simons, 1982: Table 2) and Macaca
mulatta (mean = 56 degrees, range = 40--{)9 degrees; Harrison, 1989: Fig. 6). The medial
epicondyle is more medially oriented in other Oligo-Miocene non-cercopithecoid catarrhi-
nes, including Aegyptopithecus zeuxis (23-27 degrees; DPC 1026, DPC 1275, CGM
40123, CGM 40855), Pliopithecus vindobonensis (13 degrees; Individual I; Zapfe, 1960),
Simiolus enjiessi (26 degrees; KNM-WK 17009), and Dendropithecus macinnesi (28 de-
grees; KNM-RU 1675) (Fleagle and Simons, 1982: Table 1; Rose et a!., 1992: Table 3).
Comparable values are seen in arboreal quadrupeds such as Cebus apella (mean = 35 de-
grees; mean ± 1 s.d. = 30-39 degrees) and Alouatta seniculus (mean = 14 degrees, range =
12-44 degrees) (Fleagle and Simons, 1982: Table 2; Harrison, 1989: Fig. 6). It is reason-
able, therefore, to reconstruct the ancestral catarrhine condition for medial epicondyle ori-
entation as spanning the range from approximately 10 to 40 degrees, based on
commonality of distribution among platyrrhines (as an outgroup), propliopithecids, pliopi-
thecids, small-bodied primitive catarrhines, and arboreally adapted cercopithecoids. Di-
vergently derived conditions are seen in the more medially directed medial epicondyles
reconstructed for the last common ancestor of living hominoids (with angles of approxi-
mately 0 to 10 degrees) and the higher medial epicondyle angles that were independently
evolved by terrestrially adapted cercopithecoids and Kenyapithecus. Consequently, the
moderately medioposteriorly angled medial epicondyle of Dryopithecus brancoi (35 de-
grees for Rud 53; Rose et a!., 1992: Table 3) reflects conservative retention of arboreal
quadrupedal ism like that seen in Proconsul while the retroflexed medial epicondyle of
Kenyapithecus is a derived condition, related to terrestrial adaptations.
The olecranon process of a proximal ulna of Kenyapithecus (KNM-MB 24737) is
moderately long, most like that of cercopithecines (Figure 13), and unlike the extremely
reduced olecranon of modern hominoids (McCrossin, 1994a:147-153). The olecranon ap-
pears to be somewhat retroflexed judging from the posteriorly directed sloping of the ante-
rior border from the anconeal process to the insertion scar of m. triceps brachii. In anterior
view, the superior margin of the articular surface for the humeral trochlea is asymmetrical,
with a greater proximal extension present on the lateral side of the anconeal process.
The presence of a greater amount of articular surface on the lateral (than medial)
side of the anconeal process in Kenyapithecus is shared with terrestrial cercopithecoids
while arboreal cercopithecoids exhibit mediolateral symmetry of the superior margins of
the trochlear surface (Birchette, 1982:252-253; Fig. 13). In terrestrial cercopithecoids, the
lateral side of the anconeal process articulates with a posterior extension of articular sur-
face on the lateral margin of the olecranon fossa of the distal humerus. This lateral articu-
lar buttressing serves to resist lateral displacement at the elbow joint.
A distal radius of Kenyapithecus collected from Maboko Island in 1997 provides the
first knowledge of the inferior radioulnar joint for a large-bodied hominoid from the mid-
dle to late Miocene (McCrossin et a!., 1998). Although critical to reconstruction of fore-
limb kinematics and phylogenetic affinities (Jenkins and Fleagle, 1975; O'Connor, 1975;
McHenry and Corruccini, 1983), the inferior radioulnar joint of Miocene catarrhines has
previously been known only for Proconsul from Site R114 on Rusinga Island (Napier and
Davis, 1959) and Pliopithecus from Neudorf an der March (Zapfe, 1960). The distal ra-
 ....
~

(, ,
Colobus Pall Xenyapithecus

I I I I I I
Figure 10. Humeral head projection above greater tubercle x lOO/greater tubercle diameter (McCrossin, I 994a). Some details of sample composition are as follows: Presbytis com-
bines P. entellus (N = 2), P. cristatus (N = 10), and P. rubicundus (N = I); Cercopithecus includes C. aethiops (N =6), C. mitis (N = 9), C. ascanius (N = 2). C. cephus (N = I). C. di-
ana (N = I). C. lhoesti (N = I), C. neglectus (N = I). and C. nictitans (N= I). The vertical line indicates the median value. the boxed area encloses the inter-quartile range, the
horizontal line extends to 1.5 times the inter-quartile range, and the small boxes are values which lie beyond 1.5 times the inter-quartile range. The greater tubercle of Kenyapithecus
projects slightly above the level of the humeral head. most like cercopithecoids such as Lophocebus, Presbytis entellus. Macaca, and Cercopithecus aethiops. ~
r
,.,3:
(")
C1
~
5'
~
,...
'"
 ~
HUMERUS head proj. x 100/gr. tub. diam. E:
I:"l
a:<
.
:I
::6
d'
.,
I ;.
Colobus 17 ...,o
Presbytis 13
-.----c==o-
I --[]- dQO
So
Nasalis 13 -c:=o- '"So
Pygathrix 2 I CD ;'
Cercopito 22 I--- a:::t
Erythro. 2 I I !:
4-
Macaca 13 a>
-c:r=:f 51
Q
Cercocebus 2 ~
Lophocebus 4 --f
o
I- is:
Mandrillus 5 -l I I ~
.,
Papio 12 ----1 f- I is:
I :c
Theropith. I ~
...,
Hylobates 30 I --t
I =l'
pongo 5 I 0 9'
a>
I ;-
Gorilla 13 -c=:r=J- '"
I
Pan 15 --c:TI-
Kenyapith. 1
+
Figure 10, (continued)
....
~
 370 M. L. McCrossin et al.

dius of Kenyapithecus strongly resembles modem hominoids, especially Gorilla and Pan,
in several features (McCrossin et aI., 1998). The medial margin of the Kenyapithecus dis-
tal radius forms a dorsoventrally crescentic and concave articular embrasure for the distal
ulna (McCrossin et aI., 1998). The similarly crescentic inferior radioulnar joint of living
hominoids allows an extensive range of forearm pronation and supination (O'Connor,
1975). Despite erosion, the dorsal ridges of the distal radius appear to have been strongly
developed (McCrossin et aI., 1998). According to Tuttle (1974: Fig. 8), the dorsal ridges
of the distal radius "are implicated in the close-packed position" of the radioscaphoid joint
of Pan troglodytes. The radius of Kenyapithecus appears to exhibit the marked volar slant
of the distal end that serves to limit dorsiflexion of the wrist during knuckle-walking and
distinguishes Gorilla and Pan from ceboids, Pliopithecus, cercopithecoids, Proconsul, Hy-
lobates, and Pongo (Napier and Davis, 1959; Zapfe, 1960; Jenkins and Fleagle, 1975;
O'Connor, 1975; McHenry and Corruccini, 1983). In addition, the articular surfaces for
the scaphoid and lunate are deeply concave (McCrossin et aI., 1998), unlike the relatively
flat surfaces of platyrrhines, Pliopithecus, cercopithecoids, Proconsul, Hylobates, and
Pongo (Napier and Davis, 1959; Zapfe, 1960; Tuttle, 1974).
A complete third metacarpal of Kenyapithecus (Figure 14) has a strong transverse
dorsal ridge, or metacarpal torus, adjacent to the distal articulation for the proximal pha-
lanx. A similar transverse dorsal ridge is seen on the second through fifth metacarpals of
Gorilla and Pan, but is absent from homologues of Hylobates and Pongo (Tuttle, 1974:
Fig. 7). Because of its presence on the second through fifth metacarpals of Gorilla and
Pan, the metacarpal torus is thought to be a derived feature that is a potential indicator of
a knuckle-walking pattern of locomotion in fossil taxa (Tuttle, 1967). During knuckle-
walking, the metacarpal torus prevents hyperextension at the metacarpophalangeal joint
(Tuttle, 1967, 1974). The presence of a transverse dorsal ridge on the third metacarpal of
Kenyapithecus appears to be indicative, therefore, of postures of the metacarpophalangeal
joint similar to those employed by Gorilla and Pan.
In plantar view, the medial portion of the entocuneiform facet of a left first metatar-
sal of Kenyapithecus (KNM-MB 24728) is quite flat (Figure 15). A similarly flat configu-
ration is also seen in Papio hamadryas, but the medial portion of the entocuneiform facet
is proximally recurved in Colobus guereza and Pan troglodytes (Figure 15). The absence

Figure 11. Plot of a principal components analysis of the proximal humerus, based on five dimensions trans-
formed to their natural logarithms: I) depth of the bicipital groove, 2) superoinferior height of the head, 3) medio-
lateral breadth of the head, 4) lesser tubercle diameter, and 5) breadth of the bicipital groove (McCrossin, I 994a).
Abbreviations for plotted means of taxa (and their sample sizes) are as follows: Ag = Aegyptopithecus zeuxis
(OPC 1275; Fleagle and Simons, 1982), Al = Alouaffa (N = 8), At = Ateles (N = 3), Ca = Cercopithecusaethiops
(N = 6), Cb = Cebuella (N = I), Cc = Cercocebus (N = 2), Ch = Chiropotes (N = I), Cj = Cacajao (N = 2), CI =
Colobus (N = 17), Cs = Cebus (N = 3), Cx = Cercopithecus spp. (C. ascanius. N = 2; C. cephus, N = I; C. diana.
N = I; C. lhoesti, N = 1; C. mitis, N = 9; C. neglectus, N = I; C. nictitans. N = 1), Er = Erythrocebus (N = 2), Gr =
Gorilla (N = 13), HI = Hylobates (N = 30), Kn = Kenyapithecus (KNM-MB 24729 and BMNH M. 16334), Lp =
Lophocebus (N = 4), Md = Mandrillus (N = 5), Mf= MacacaJascicularis (N = 9), Mn = Macaca nemestrina (N =
5), Mp = Miopithecus (N = 2), Ns = Nasalis (N = 13), Ny = cf. Mabokopithecus (KNM-MB 21206; McCrossin,
1992a), Pe = Presbytis entellus (N = 2), Pg = Pygathrix (N = 2), PI = Pliopithecus vindobonensis (DE 304;
Ginsburg and Mein, 1980), Pn = Pan (N = 15), Po = Pongo (N = 5), Pp = Papio (N = 12), Px = Presbytis spp. (P.
cristatus, N = 10; P. rubicundus. N = 1), Ru = Dendropithecus macinnesi or Proconsul aJricanus (KNM-RU
17376; Gebo et a\., 1988), Sg = Saguinus (N = 1), Sm = Saimiri (N = 2), Th = Theropithecus (N = I). Kenyapi-
thecus lies closest to semi-terrestrial and terrestrial forms such as Mandril/us, Papio, and Presbytis entellus.
 M
\j
Nasalis Papio Pan Kenyapithecus

I I I I I I

Plot of First Two Principal Components

2.1

1.6

1.1
Dr-

-
N
C
GI
c
0 0.6
Q.
E
0
0

I'a
0.1

/ .. '"f!~
.son .r;•

I1r Lp. )"

'~
.r;b
-0.4

~Er '~
-0.9
~\." Hd

-5.6 -3.6 -1.6 0.4 2.4 4.4

Component 1
 CM
;j

~
~

Colobus Macaca Pan Kenyapithecut;

I I I I I I
Figure 12. Retroflection of the medial epicondyle of the humerus in degrees (McCrossin, I 994a). The vertical line is the mean and the box encloses the mean ± I standard deviation
for Erythrocebus patas, Macacafascicularis, and M. nemestrina (data are from Fleagle and Simons, 1982). The vertical line is the median and the box encloses the entire range for
Colobus polykomos, Presbytis rubicundus, Miopithecus talapoin, Cercopithecusaethiops, C. mitis, C. nictitans, Macaca mulatta, Lophocebus albigena, Papio anubis, Theropithecus
gelada, Hylobates lar, Pongo pygmaeus, Gorilla gorilla, and Pan troglodytes (data for all but C. aethiops and C. nictitans from Harrison, 1989). Data for Aegyptopithecus zeuxis,
Pliopithecus vindobonensis, Simiolus enjiessi, Dendropithecus macinnesi, Proconsul africanus, and Dryopithecus brancoi are from Rose et al. (1992). The posteriorly directed me-
dial epicondyle of Kenyapithecus finds its closest match with semi-terrestrial and terrestrial forms such as Erythrocebus, Macaca, and Theropithecus.

rs:
r
:::
~
iS'
~
:--
""
 "!l

~
HUMERUS medial epicondyle retroflection t"'l
~
is:
.~
d'
.,
c. poly. 11 ..:r
P. rubi. 8
C. aeth. 36 • 3:5l
C. mitis 7
C. nict. 2 [JJ
a.
M. tala. 3 r-----.--, .,;;l
E. patas 6
•• i:::!.
M. fasc. 6 ,------,---, !!.
M. mulatta 6 ~
10
M. Neme. 6 I aco
L. albi. 3 JS
P. anubis 5 r- r o
T. gelada 7 5:
H.lar 4 CI:J ~
.,
P.pygm. 5 I 5:
G. gorilla 5 r- =--- -----] I
P. trog. 5 ____:::J
r§:
=
!!j
Aegypto. 1 ."
:::!.
Pliopith 1
•I••
a
10
Simiolus 2 c:::r::::::::J
~
Dendro. 2
••I
Proconsul 2 [JJ

Dryopith. 1
•
Kenyapith. 1
-t
Figure 12. (continued)
tH
i:;1
374 M. L. McCrossin et al.

Colobus Macaca

@ \,\
,I'
~I\
~'I,
'II ' ' ,/~-::::\
/,,"-
I \ , :
I \ I
I I

Pan Kenyapithecus

I I I I I I
Figure 13. Left proximal ulna of Colobus guereza, Macaea nigra, Pan paniscus, and Kenyapithecus africanus
(KNM-MB 24737 - dashed lines indicate reconstructed coronoid process and distal portion of trochlear notch) in
medial view. (The scale bar equals five centimeters.)

of articular surface wrapping medially around the entocuneiform may indicate that the
hallux of Kenyapithecus was habitually adducted as in baboons during terrestrial quadru-
pedalism.
A left cuboid of Kenyapithecus (Figure 16) shows derived resemblances to African
apes and humans in several features (McCrossin, 1997). The cuboid of Kenyapithecus is
relatively short and strongly wedged and the proximal end possesses a moderately well
developed peg-like process for articulation with a reciprocal concavity on the distal sur-
face of the calcaneum (McCrossin, 1997). In many respects, the cuboid of Kenyapithecus
differs from primitive character states seen in the cuboids of propliopithecids, Proconsul,
and Sivapithecus (Langdon, 1986).
Fossil Evidence for the Origins of Terrestriality among Old World Higher Primates 375

Figure 14. Right third metacarpal of Kenyapithecus collected

from Maboko Island in dorsal (left) and medial (right) views. The
strongly developed metacarpal torus of Kenyapithecus (arrows in-
dicate position) is a point of resemblance to Gorilla and Pan and
suggests resistance of hyperextension at the metacarpophalangeal
joint. (The scale bar equals one centimeter.)

An intermediate phalanx of Kenyapithecus (KNM-MB 28393) is relatively short and

stout with little dorsoventral curvature of the shaft and well marked flanges on the ventral
surface for insertion of the digital flexors (Figure 17). A bivariate plot of proximodistal
length and proximal mediolateral breadth (Figure 18) shows that the intermediate phalanx
of Kenyapithecus is relatively short and robust, like that of semi-terrestrial catarrhines, es-

Yl-l\
~
Colobus Papio

Figure 15. Proximal end of left first

metatarsal of Colobus guereza, Papio
hamadryas, Pan troglodytes, and Ken-
yapithecus africanus (KNM-MB Kenyapithecus
Pan
24728) in plantar view. Arrows indi-
cate proximal recurvature of en-
tocuneiform facet in Pan and
comparatively flat surface in Kenyapi-
theel/so (The scale bar equals five cen-
timeters.) I I I I I I
376 M. L. McCrossin et al.

Figure 16. Left cuboid, fourth metatarsal, and fifth metatarsal of Papio (left), Kenyapithecus (middle), and Homo
(right) in dorsal view. The cuboid of Kenyapithecus resembles Gorilla, Pan, and Homo in that, overall, it is rela-
tively short (proximodistally), strongly wedged and the proximal peg (for articulation with the calcaneum) is mod-
erately large.

Colobus Pan Kenyapithecus

I I I

Figure 17. Intermediate phalanx of Colobus guereza, Pan troglodytes, and Kenyapithecus aJricanus (KNM-MB
28393) in ventral view. The intermediate phalanx of Kenyapithecus closely resembles that of Pan in terms of its
robusticity, curvature, and the development of secondary shaft features, especially the ridges for the phalangeal
flexors. (The scale bar equals two centimeters.)
Fossil Evidence for the Origins of Terrestriality among Old World Higher Primates 377

18
/
;'
;'
gorill. /
;'
;'
/
15 ;' /
/ ;' /
/
;' /
/ /
/
/
/
/
/
12 /
/

6
;' • gib60n
,r' /
L'
.;'~
r.d colc;mus
/

bJ-u. /
/

o 10 20 30 40 50
proximo-distallength (mm)

Figure 18. Plot of mean proximodistallength (mm) and mean proximal mediolateral breadth (mm) of intermedi-
ate phalanges (McCrossin, 1994a). The relatively short and robust intermediate phalanges of semi-terrestrial and
terrestrial primates, including the vervet monkey (Cercopithecus aethiops, N =46), the patas monkey (Erythroce-
bus patas, N = 17), savanna baboons (Papio anubis, N = 22; P. cynocephalus, N = 23; P. hamadryas, N = 2), the
mandrill (Mandril/us sphinx, N =2), the bonobo (Pan paniscus, N =2), the chimpanzee (Pan troglodytes, N = 18),
and the gorilla (Gorilla gorilla, N =\3) fall close to or above the regression line. In contrast, the relatively long
and gracile intermediate phalanges of the blue monkey (Cercopithecus mitis, N = 30), the red colobus (Colobus
badius, N = 16), the guereza (Colobus guereza, N = 30), the douc langur (Pygathrix nemaeus, N =2), the probos-
cis monkey (Nasalis larvatus, N =5), the white-handed gibbon (Hylobates lar, N = 14) and the orangutan (Pongo
pygmaeus, N = 6) fall below the regression line. The intermediate phalanges of Victoriapithecus are most similar
to those of the vervet monkey while that of Kenyapithecus is intermediate between those of bonobos and chimpan-
zees.
378 M. L. McCrossin et al.

pecially Pan. The intermediate phalanges of arboreal catarrhines, such as red colobus,
douc langurs, gibbons, and orang-utans, are relatively longer and more gracile.

3.3. Summary
In summary, several indicators of semi-terrestriality are shared by Victoriapithecus
and Kenyapithecus. The most conspicuous of these are the proximal extension of the
greater tubercle of the humerus, the posterior orientation of the humeral medial epicon-
dyle, the retroflection of the ulnar olecranon process, the distally restricted nature of the
articulation between the entocuneiform and the hallucal metatarsal, and the robusticity of
the intermediate phalanges. Overall, these similarities reflect shared emphases on stability
and forceful protraction of the shoulder, a premium on rapid flexion and extension at the
elbow, and a reduction of cheiridial grasping.
In addition to these similarities, Victoriapithecus and Kenyapithecus exhibit aspects
of morphology that reveal fundamental differences in their styles of locomotion. In virtu-
ally every aspect of its postcranial anatomy, Victoriapithecus resembles living semi-terres-
trial cercopithecoids, especially vervet monkeys. Kenyapithecus, in contrast, exhibits
numerous features indicative of greater joint mobility than is seen in cercopithecoids and
many of these features are shared with living hominoids. Hominoid-like features of Ken-
yapithecus include the relative size of the humeral head, the gracility of the humeral shaft,
the breadth of the humeral trochlea, the development of the humeral lateral trochlear keel,
the depth of the zona conoidea, the crescentic inferior radioulnar joint, and the angle of
the femoral neck (McCrossin, 1994a; McCrossin et aI., 1998). These features may relate to
scansoria1 activities, including forelimb-dominated movements such as hoisting, stable
pronation and supination of the forearm throughout the entire range of elbow flexion and
extension, as well as hindlimb participation in vertical climbing. Intriguingly, Kenyapi-
thecus also shares several distinctive postcranial features with Gorilla and Pan. The most
important of these features concern aspects of the design of the inferior radioulnar joint
and the metacarpophalangeal joint that appear to correspond to anatomical complexes re-
lated to knuckle-walking.

4. PALEOENVIRONMENT

There have been numerous attempts to reconstruct the environment of the middle
Miocene of eastern Africa based on faunal and floral evidence from Maboko Island and
Fort Ternan (Andrews and Walker, 1976; Andrews and Nesbit Evans, 1979; Van Couver-
ing, 1980; Nesbit Evans et aI., 1981; Shipman et aI., 1981; Pickford, 1983, 1987; Bonne-
fille, 1984, 1985; Shipman, 1986; Pickford and Senut, 1988; Retallack et aI., 1990;
Cerling et aI., 1991; Retallack, 1992; Solounias and Moelleken, 1993). The most impor-
tant contributions to understanding the paleoenvironments at these sites are the original
descriptions offauna and flora at Maboko Island (e.g., Macinnes, 1936, 1942; Whitworth,
1958; Cifelli et aI., 1986; Thomas, 1985; Winkler, 1994) and Fort Ternan (e.g., Lavocat,
1964, 1988, 1989; Hooijer, 1968; Churcher, 1970; Gentry, 1970; Hillenius, 1978; Pick-
ford, 1982b; Thomas, 1984; Denys and Jaeger, 1992; Retallack, 1992; Tong and Jaeger,
1993, Van der Made, 1996).
Unfortunately, there has been an oversimplifying tendency in the secondary litera-
ture to equate the paleoenvironment (abiotic and biotic contexts) of Maboko Island and
Fort Ternan with the paleoecology (e.g., diet, locomotion, and habitat preference) of Ken-
Fossil Evidence for the Origins of Terrestriality among Old World Higher Primates 379

yapithecus (Andrews and Nesbit Evans, 1979; Nesbit Evans et aI., 1981). One paleoenvi-
ronmental reconstruction even claims to impose restrictions on the possible phylogenetic
affinities and substrate preferences of Kenyapithecus (Cerling et aI., 1991). In addition,
identification of fauna by non-specialists has led to confusion concerning the paleoenvi-
ronment of Maboko Island and Fort Ternan. For example, Andrews et ai. (1981) list the
presence of the anomalurid (scaly-tail "flying squirrel") Zenkerella at Maboko, ostensibly
as an indicator of forest. Re-examination of the Maboko rodent assemblage, however, has
shown that no anomularids are present at the site (Winkler, 1994). In fact, the specimen
from Maboko identified as Zenkerella by Andrews et ai. (1981) is actually a bathyergid
(Cifelli et aI., 1986). Thus, Andrews et ai. (1981) used a subterranean rodent to indicate
the presence of a tree-top milieu at Maboko (Cifelli et aI., 1986).
Faunal lists ofMaboko Island and Fort Ternan (Harrison, 1992: Table 5) also contain
multiple errors and omissions concerning the primates, carnivores, rodents, lagomorphs,
and artiodactyls actually known from these sites (Lavocat, 1964, 1988; Thomas, 1984;
Cifelli et aI., 1986; Gentry, 1990; Benefit, 1991; McCrossin, 1992a, 1994a; Denys and
Jaeger, 1992; Tong and Jaeger, 1993, Van der Made, 1996). Some recent speculations con-
cerning the paleoenvironment, biostratigraphy, and paleobiogeography of Maboko Island
and Fort Ternan (Harrison, 1992) are based, therefore, on an unacceptably inaccurate
foundation.
The paleoenvironments of Maboko Island and Fort Ternan are interesting because
both sites register the effects of sweeping changes from the primarily forested environ-
ments that prevailed during the early Miocene in western Kenya. Many changes in the
fauna and flora probably reflect the emergence of more open country environments that
were effected by greater seasonality of climate than existed in the early Miocene (Van
Couvering and Van Couvering, 1976; Shipman et aI., 1981; Pickford, 1983; Benefit, 1987;
Pickford and Senut, 1988; McCrossin and Benefit, 1994). Global [G] and sub-Saharan Af-
rican [SSA] first appearances include hippopotamids [G], bovids [SSA], climacocerid gi-
raffoids [G], and the amebelodont gomphothere Choerolophodon [SSA] at Maboko Island
and C4 grasses [G], ostriches [G], advanced giraffoids (i.e., Giraffokeryx) [SSA], and
hyaenids [SSA] at Fort Ternan (MacInnes, 1936, 1942; Arambourg, 1945; Crusafont and
Aguirre, 1971; Andrews and Walker, 1976; Hamilton, 1978; Tassy, 1979; Thomas, 1979,
1985; Pickford, 1982b; Gentry, 1990; Retallack et aI., 1990).
Because antelopes make their first appearance in sub-Saharan Africa at Maboko Is-
land (Whitworth, 1958; Thomas, 1979) and are numerically dominant at Fort Ternan (Gen-
try, 1970), their adaptations are of particular relevance to understanding the
paleoenvironment of the middle Miocene of eastern Africa. At least four different ante-
lopes are known from Maboko Island, including a boselaphine resembling Eotragus, an
antilopine (cf. Gazella), an indeterminate caprine, and Nyanzameryx (Whitworth, 1958;
Gentry, 1970; Thomas, 1979, 1985). Thomas (1985) referred Nyanzameryx to the Family
Climacoceridae, but examination of the type specimen indicates that it is a bovid rather
than a giraffoid. The frontal appendages of Nyanzameryx, in particular, are hom cores
rather than ossicones as they are separated from the frontal squama by a pedicle and the
surface superior to the pedicle is vesiculated (unlike the continuously smooth cortical sur-
face ofgiraffoid ossicones). Five bovids are known from Fort Ternan (Gentry, 1970, 1990;
Thomas, 1979, 1984): two boselaphines - cf. Eotragus and Kipsigicerus labidotus, an an-
til opine or aepycerotine - cf. Gazella or Aepyceros, and two caprines - Hypsodontus tany-
ceras and Caprotragoides potwaricus.
Gentry (1970) demonstrated a series of morphological distinctions of the postcranial
skeleton between highly cursorial antelopes inhabiting open country, such as savanna
380 M. L. McCrossin et af.

grasslands, and slower moving antelopes living in more closed environments, such as
woodland and forest. In particular, Gentry focussed on clear distinctions in the structure of
the proximal and distal femur between open country and closed country antelopes. In his
analysis, Gentry demonstrated a series of cursorial adaptations for life in open country
habitats for the caprine Hypsodontus tanyceras and retention of adaptations for life in
woodland and forest habitats for the boselaphine Kipsigicerus labidotus. Specifically, H.
tanyceras exhibits cursorial adaptations of the femur, most notably an articular surface of
the head with "an anteroposteriorly long lateral part" and a distal end in which "the medial
side of the patellar fossa projects strongly anteriorly" (Gentry, 1970:279-280, Fig. 12b).
Cursorial adaptations similar to those of H. tanyceras are also seen in the distal femur
anatomy of Caprotragoides potwaricus. Using a specimen from Nyakach, Thomas
(1985:82, PI.2-3) has shown that strong projection of the medial keel of the patellar
groove is evident in C. potwaricus. Dental microwear analysis demonstrates that H. tany-
ceras was a grazer (Shipman et aI., 1981) and K. labidotus was a grazer or mixed feeder
(Solounias and Moelleken, 1993). Assessment of correlations between muzzle shape and
dietary preference (Janis and Ehrhardt, 1988) also indicates grazing or mixed feeding for
K. labidotus (Solounias and Moelleken, 1993).
Andrews and Walker (1976:299, 302) suggested that a wide range of environments,
including tracts of "open and lightly wooded country" as well as "evergreen forest" might
have been present near Fort Ternan. The existence of "moderately dense bushland" at Fort
Ternan was based on the presence of an ostrich and a springhare, Megapedetes (Andrews
and Walker, 1976:301). The presence of woodland and forest was supported by the identi-
fication of a leaf impression of Sterculiaceae, "a family of mostly woodland trees and
shrubs", gastropods suggestive of "woodland" (Burtoa) and "evergreen forest" (Maizania,
Homorus), an elephant shrew (Rhynchocyon), a lorisine, and a scaly-tail "squirrel" (Para-
nomalurus) (Andrews and Walker, 1976:299-302).
Shipman and colleagues (1981) suggested that a perceived difference in quality of
preservation between remains attributed to "Ramapithecus" (= Kenyapithecus) wickeri
and "Dryopithecus" (= Proconsul) nyanzae by Andrews and Walker (1976) indicated that
the former lived in savanna and woodland environments near the site while the latter lived
in forest environments farther away, on the slopes of the Tinderet volcano. The validity of
this idea has been cast into doubt by the demonstration that the specimens attributed to
"Dryopithecus" nyanzae by Andrews and Walker (1976) probably represent Kenyapi-
thecus wickeri (Pickford, 1985).
Terrestrial gastropod assemblages have been interpreted as indicating the presence
of "semi-arid woodland with gallery forest" at Maboko Island and "upland humid wood-
land" at Fort Ternan (Pickford and Senut, 1988:43). Further support for Gentry's (1970)
proposal of a grassy woodland environment at Fort Ternan comes from the identification
of grass pollen and phytoliths representing at least three species (Bonnefille, 1984, 1985;
Retallack et aI., 1990; Retallack, 1992).

5. BODY SIZE

Body size is an integral aspect of primate adaptation, with general correlations to

diet and locomotion (Fleagle and Mittermeier, 1980; Kay, 1984; Fleagle, 1985; Smith and
Jungers, 1997). In addition, knowledge of body mass is critical to understanding important
trends such as encephalization and sexual dimorphism (Radinsky, 1974, 1982; Fleagle et
aI., 1980). Furthermore, the distribution of body size is a fundamental attribute of primate
Fossil Evidence for the Origins ofTerrestriality among Old World Higher Primates 381

faunas and allows appreciation of differences in the structure of primate guilds and com-
munities through time (Fleagle, 1978, 1985; Mihlbachler et aI., 1996). Finally, dimensions
that are isometrically correlated with body weight are the most appropriate independent
variables for assessment of allometric trends. Consequently, a variety of methods for esti-
mating body weights from tooth and limb bone measurements have been developed and
applied to many fossil primates (Gingerich, 1977; Gingerich et aI., 1982; Conroy, 1987;
Dagosto and Terranova, 1992).
As is the case with many fossil primates, including Aegyptopithecus zeuxis and Pro-
consul africanus (Conroy, 1987), body weight estimates derived from the dentitions of
Victoriapithecus and Kenyapithecus yield greater values than those derived from postcra-
nial remains. Regressions of first molar area reconstruct a body weight of approximately
5.0 to 5.5 kg for Victoriapithecus (Gingerich et aI., 1982; Conroy, 1987). Postcranial di-
mensions indicate that Victoriapithecus individuals ranged from 2.0 to 5.5 kg, with esti-
mates of average female body weight at 3.0 kg and average male body weight at 4.5 kg
(Harrison, 1989).
A series of 28 logarithm-transformed molar and postcranial dimensions have been
used to estimate the body weight of Kenyapithecus (McCrossin, 1994a: Table 44). Esti-
mates based on postcranial measurements vary greatly, from a minimum of 17.3 kg based
on the mean of seven astragalar dimensions to a maximum of 36.2 kg based on the di-
mensions of the intermediate phalanx (McCrossin, 1994a: Table 44). Nevertheless, there
is close agreement between mean estimates based on molar dimensions, 31.8, and mean
estimates based on seven postcranial elements, 28.4 kg (Figure 19). In light of the high
levels of intraspecific variation (mainly due to sexual dimorphism) of body weight seen
among modern catarrhine species, these estimates are not very different from those of
23.0 kg and 18.4 kg derived from the midshaft and head diameters, respectively, of the
femur collected from Maboko Island in 1933 (Ruff et aI., 1989). Apparently, therefore,
Kenyapithecus was slightly larger than the mandrill (Mandril/us sphinx), the largest liv-
ing Old World monkey, but somewhat smaller than the bonobo (Pan paniscus), the small-
est living great ape.
Body weights in the size range of Victoriapithecus and Kenyapithecus are not en-
countered among insectivorous primates, which are substantially smaller, but are more
commonly associated with fruit- and leaf-eating species (Kay, 1984). Although overlap-
ping the size ranges of semi-terrestrial and terrestrial species, the body weights of Victo-
riapithecus and Kenyapithecus are also well within the range of arboreal primates. In
terms of locomotion, large arboreal primates tend to engage less frequently in bouts of
leaping and instead rely upon bridging gaps between branches by suspension or clamber-
ing (Fleagle and Mittermeier, 1980; Fleagle, 1985). However, most primates in the size
range of Kenyapithecus spend at least part of their lives on the ground, sometimes simply
to cross patches of open ground between clumps of trees, but more commonly in foraging
for foods, such as grass corms and fallen fruit (Napier, 1967).
The wide range of body weights estimated from various postcranial elements may
corroborate the high degree of sexual dimorphism inferred for Victoriapithecus and Ken-
yapithecus on the basis of two size categories of the upper canine (Benefit, 1987; Pick-
ford, 1985; McCrossin, 1994a). Sexual dimorphism of catarrhine canine height and body
size has been linked to social systems involving male-male competition (Leutenegger and
Kelly, 1977; Leutenegger, 1978; Plavcan and van Schaik, 1997). Among some African cer-
copithecins, however, the correlation between sexual dimorphism and social organization
is not straightforward (Gautier-Hion and Gautier, 1985) and it is possible that the greater
body size and possession of large canines seen among male cercopithecins may be more
 382 M. L. McCrossin et al.

body weight (kg)

8 16 24 32 40 48

Kp. african us

Kp. wickeri

Md. sphinx

Ns. larvatus

Pan paniscus

Pp. anubis

Pp. cynoceph.

Pp. hamadryas

Pro entellus

Tp. gelada

Figure 19. Estimated body weights of Kenyapithecus wickeri and K. africanus compared with body weights of
some extant catarrhines. Estimate for K. wickeri is from regression of M' area to body weight in a sample of 49
extant primates (McCrossin, I 994a: Table 44). Estimates for K. africanus come from regressions to body weight
of M' area, M, area, the mean of five humerus dimensions, the mean of two ulna dimensions, the mean of three fe-
mur dimensions, the mean of two patella dimensions, the mean of seven astragalus dimensions, the mean of six
pollical proximal phalanx dimensions, and one intermediate phalanx dimension (McCrossin, I 994a: Table 44). All
regressions have r-squared values greater than 0.90. Body weights of male and female Mandril/us sphinx, Nasalis
larvatus, Pan paniscus, Papio anubis, P. cynocephalus, P. hamadryas, Presby tis entellus, and Theropithecus
gelada are from Fleagle (1988). For K. africanus estimates, the plotting conventions are as in Figure 10. The verti-
cal line indicates the mean and the box encloses the weights of females and males for extant catarrhines. Body
weight estimates of Kenyapithecus exhibit a range similar to that seen in living catarrhines of similar body weight.
Overall, Kenyapithecus body weight appears to have overlapped the range from the largest of living Old World
monkeys (Mandril/us sphinx) to the smallest of living great apes (Pan paniscus).

closely related to solitary defense against predators in social systems characterized by fe-
male philopatry (Rowell and Chism, 1986). According to DiFiore and Rendall (1994), the
female philopatry of most living cercopithecoids is a derived pattern of social dispersal for
catarrhines. Thus, the body size and canine sexual dimorphism inferred for Victoriapi-
thecus, together with results from paleodemography (Benefit, 1994), may indicate that the
distinctive social organization of modern cercopithecoids, involving male transfer out of
the natal group, had evolved by the middle Miocene.
Fossil Evidence for the Origins of Terrestriality among Old World Higher Primates 383

The social organization of living hominoids are more diverse than those of extant cer-
copithecoids. The body size and canines of Hylobates are monomorphic and gibbons defend
their territories against interlopers of the same sex. Living great apes exhibit strong sexual di-
morphism in canine length and body size. Orang-utans have a distinct social organization, in-
volving few interactions between adult males and females, while gorillas and chimpanzees
are more gregarious and dispersal patterns always involve female transfer from the natal
group. Within this context, the high degree of sexual dimorphism seen in the canine and body
size of Kenyapithecus may reflect male-male competition but this is by no means certain.

6. DIET

Although most models of cercopithecoid origins predicted facultative folivory for

Old World monkeys (Jolly, 1966; Napier, 1970; Delson, 1973, 1975; Andrews, 1981; Te-
merin and Cant, 1983), analysis of molar shear crests indicates that Victoriapithecus
mainly consumed hard fruits (Kay, 1977; Benefit, 1987, 1990). A frugivorous diet for Vic-
toriapithecus is also indicated by the relatively broad proportions of the upper central inci-
sors (Benefit, 1987, 1990, 1993) and by the anatomy of the skull, particularly the strength
and position of markings for m. temporalis (Benefit and McCrossin, 1991, 1993, 1997)
(Figure 20).
Previous reconstructions of the dietary adaptations of Kenyapithecus focussed on its
thick-enameled molars, be!ween which hard food items were presumed to have been
ground (Andrews, 1971; Walker and Andrews, 1973; Andrews and Walker, 1976; Simons
and Pilbeam, 1978; Kay, 1981). Models for the diet of Kenyapithecus that placed an em-
phasis on the role of molar grinding often involved an expectation of reduction in upper
canine and lower incisor size. Although Kenyapithecus has been characterized as having
small upper canines (Yulish, 1970; Conroy, 1972; Andrews and Walker, 1976), it is now
recognized that the upper canines are sexually dimorphic, with smaller specimens (such as
KNM-FT 46; Leakey, 1962) representing females and larger specimens (such as KNM-FT
39, attributed to "Dryopithecus" (= Proconsul) nyanzae by Andrews and Walker, 1976) be-
longing to males (Pickford, 1985; McCrossin, 1994a; McCrossin and Benefit, 1997). Si-

Figure 20. Skull of Victoriapithecus

macinnesi (KNM-MB 29100; Benefit
and McCrossin, 1997) in left lateral
view. (Scale bar equals three centime-
ters.) Indicators of frugivory are marked
by arrows (in clockwise order, from the
top): anterior convergence of strongly
marked temporal lines, relatively short
shearing crests on molars, and relatively
long premaxilla with broad incisors
(known from isolated specimens) (Bene-
fit, 1987, 1990, 1993; Benefit and
McCrossin, 1991, 1993, 1997).
384 M. L. McCrossin et al.

Figure 21. Juvenile mandible of Kenyapithecus afi'i-

canus (KNM-MB 20573; McCrossin and Benefit,
1993) in medial view, showing cross-section of
mandibular symphysis. (The scale bar equals two
centimeters.) The strong procumbency of the lateral
incisor is a resemblance to pitheciines and indicates
that Kenyapithecus used a specialized anterior denti-
tion to consume hard fruits and nuts (McCrossin and
I I I Benefit, 1993, ) 994, 1997; McCrossin, 1994a).

mons and Pilbeam (1978:149, 152) confidently concluded that "Ramapithecus" (= Ken-
yapithecus) wickeri "had remarkably small lower incisors" that "were relatively unimpor-
tant in food preparation", but this inference was based on observation of empty alveoli
rather than knowledge of the actual size of the lower incisors.
More recent attempts to reconstruct the diet of Kenyapithecus include new insights
concerning the functional morphology of the maxilla, upper incisors, canines and premo-
lars, mandible (Figure 21), and lower incisors (McCrossin and Benefit, 1993, 1994, 1997;
McCrossin, 1994a). The maxilla of Kenyapithecus is distinctive in that it exhibits a promi-
nent canine jugum and a deep post-canine fossa together with an anteriorly positioned root
of the zygomatic process (Leakey, 1962, 1967; McCrossin, 1994a; McCrossin and Benefit,
1997). Based on comparisons of isolated specimens, the upper lateral incisors of Kenyapi-
thecus are substantially narrower (mesiodistally), thinner (labiolingually), and lower-
crowned than the upper central incisors (McCrossin, 1994a). The upper canines of
Kenyapithecus are quite robust and are inferred to have been implanted in a pattern involv-
ing external rotation of the root (Ward and Pilbeam, 1983; McCrossin, 1994a). Compared to
Proconsul and the small-bodied primitive catarrhines of the African early and middle Mio-
cene, the upper premolars of Kenyapithecus are large relative to the size of the first upper
molar (McCrossin, 1994a; McCrossin and Benefit, 1997). The mandibular corpus of Ken-
yapithecus is robust and marked by strong lateral buttressing while the mandibular sym-
physis is markedly proC\ined and dominated by the presence of a massive inferior transverse
torus (McCrossin and Benefit, 1993). Strain gauge tests of primate mandibles indicate that
the inferior transverse torus resists anteroinferiorly directed bending moments during incisal
biting (Hylander, 1984). The lower incisors of Kenyapithecus are strongly procumbent and
narrower (mesiodistally), thicker (labiolingually), and higher-crowned than those of extant
apes (McCrossin and Benefit, 1993; McCrossin, 1994a).
This suite of features, especially the presence of markedly heteromorphic upper inci-
sors, tusk-like upper canines, enlarged upper premolars, and procumbent lower incisors
(McCrossin and Benefit, 1993, 1997; McCrossin, 1994a), is quite similar to adaptations for
seed-predation seen in the pitheciines-the saki (Pithecia), the bearded saki (Chiropotes)
and the uakari (Cacajao) (Mittermeier and van Roosmalen, 1981; van Roosmalen et aI.,
1988; Kinzey, 1992; Kinzey and Norconk, 1993). The presence of these features in Ken-
yapithecus appears to be related to a diet involving consumption of hard fruits and nuts
(McCrossin and Benefit, 1993, 1994, 1997; McCrossin, 1994a; Benefit and McCrossin,
1995). An earlier stage in the trend toward a pitheciine-like dentognathic apparatus is re-
corded for Afropithecus (McCrossin, 1994a; McCrossin and Benefit, 1993, 1994, 1997).
Afropithecus possesses enlarged upper premolars and a deep post-canine fossa in combina-
Fossil Evidence for the Origins of Terrestriality among Old World Higher Primates 385

Figure 22. Scanning electron micrographs of micro wear on Facet

9 (hypoconid) of the second lower molar in Victoriapithecus
(KNM-MB 24530 - top; magnified 220 times) and Kenyapithecus
(KNM-MB 20200 - bottom; magnified 150 times). Both Victo-
riapithecus and Kenyapithecus show the extensively pitted mi-
crowear that is characteristic of a diet of hard fruits, in contrast to
the heavily scratched enamel, and inferred folivory. of sympatric
small-bodied primitive catarrhines (Simiolus) and oreopithecids
(Mabokopithecus) (Palmer et aI., 1998).

tion with externally rotated and tusk-like canines while catastrophic exposure of dentine on
the molars of Afropithecus finds its match only in pitheciihes among primates, fossil and
modem (McCrossin, 1994a; McCrossin and Benefit, 1993, 1994, 1997).
Dental microwear analyses corroborate reconstructions of preference for hard fruits,
seeds, and nuts by Victoriapithecus and Kenyapithecus based on functional assessments of
craniodental morphology (Palmer et aI., 1998). Both Victoriapithecus and Kenyapithecus
resemble hard object feeders in terms of the high percentage of microwear features that
are pits (Palmer et aI., 1998; Figure 22). A more specialized mode of hard object feeding
for Kenyapithecus may be indicated by the fact that its microwear pits are much wider
than those of Victoriapithecus (Palmer et a!., 1998). In contrast, ostensibly folivorous pri-
mates from Maboko, such as Simiolus and Mabokopithecus, exhibit a great number of par-
allel scratches (Palmer et a!., 1998).

7. CONCLUSIONS
Due to their integral phylogenetic relationships (Figure 1), the presence ofsemi-ter-
restrial adaptations in Victoriapithecus and Kenyapithecus are especially significant for
understanding the origins of terrestriality among cercopithecoids and hominoids.
386 M. L. McCrossin et al.

Because parsimony indicates that morphology and adaptations shared by victoriapi-

thecids and either, or both, cercopithecid subfamilies are most likely to be primitive for
Cercopithecoidea (Benefit, 1987, 1993, Harrison, 1989; Benefit and McCrossin, 1991,
1993, 1997; McCrossin and Benefit, 1992), a predominantly semi-terrestrial substrate
preference probably characterized ancestral Old World monkeys. The possibility that
semi-terrestriality was the primitive substrate preference of cercopithecids (Benefit, 1987;
Pickford and Senut, 1988; Strasser, 1988; Harrison, 1989) is also corroborated by the fact
that many fossil colobines, including Mesopithecus (von Beyrich, 1861) and Cercopithe-
coides (Birchette, 1981), were also adapted for life on the ground. Consequently, the ar-
boreality of most living colobines may be a comparatively recent adaptive novelty
(Benefit, 1987; Pickford and Senut, 1988; Strasser, 1988; Harrison, 1989) rather than the
conservative retention of an ancestral arboreality once reconstructed for cercopithecids
(Jolly, 1970; Napier, 1970).
Consideration of the paleoenvironment, body size, and diet of Victoriapithecus al-
lows us to separate cause from consequence in the origin of terrestrial adaptations among
Old World monkeys. Unlike the open grassland habitats currently occupied by the most
terrestrially adapted cercopithecoids, such as savanna baboons and geladas, the middle
Miocene environment reconstructed for Maboko Island shows that the semi-terrestriality
of ancestral Old World monkeys probably originated in a woodland milieu (Benefit, 1987;
Pickford and Senut, 1988). Although "increased body mass in a terrestrial environment" is
expected among ground-dwelling cercopithecoids (Reno et ai., 1997: 197), body size esti-
mates of Victoriapithecus show that an increase in body size did not influence the earliest
phases of the transition from life in the trees to life on the ground among Old World mon-
keys. Thus, the adoption of a diet of hard fruits and seeds in seasonally dry woodlands by
Victoriapithecus may have been the driving force behind the origin of semi-terrestriality in
cercopithecoids (Benefit, 1987, 1990).
According to Rose (1993 :269), "obvious morphological correlates that are indicators
of terrestrial locomotor modes" have not been "demonstrated convincingly for any Mio-
cene hominoids" but "such a demonstration-for known or yet-to-be found Miocene
hominoids-will be essential for an understanding of their paleoecology, and of hominid
and African-ape evolution". Indeed, it has recently been claimed that postcranial remains
"from Maboko Island .. .indicate little change from the generalized arboreal quadrupedal-
ism present in the early Miocene hominoids like Proconsul" (Andrews, 1992:643).
Although substantially augmenting knowledge of torso and limb structure, addi-
tional postcranial material of Proconsul (Walker and Pickford, 1983; Beard et ai., 1986;
Langdon, 1986; Ruff et ai., 1989; Ward et ai., 1993, 1995) has done little to modify the
original conclusion that it was a generalized arboreal quadruped (Napier and Davis, 1959).
The postcranial anatomy, locomotor pattern and substrate preference of Dryopithecus are
less clearly understood. Suspensory locomotion, especially orangutan-like positional be-
haviors, have recently been suggested for Dryopithecus (Moya-Sola and Kohler, 1996),
but these claims rest partly on reconstructed, rather than known, intermembral propor-
tions. Moreover, the posteromedial orientation of the humeral medial epicondyle, the re-
tention of a bony joint between the ulna and wrist, and the weak development of the
hamulus on the hamate indicate that Dryopithecus lacked the specialized digital grasping
and wrist mobility of suspensory hominoids and instead was primarily an arboreal quadru-
ped, perhaps with some enhanced adaptations for climbing compared with Proconsul. A
broad array of "generalized, varied, mostly arboreal" locomotor patterns have also been
reconstructed for Sivapithecus (Rose, 1984:503), but recent interpretations of the elbow
and carpometacarpal joints may also indicate some terrestrial behaviors (Senut, 1986b;
Fossil Evidence for the Origins of Terrestriality among Old World Higher Primates 387

Spoor et aI., 1992). The pervasive semi-terrestrial adaptations inferred here for Kenyapi-
thecus are in marked contrast, therefore, to the apparently "generalized morphology and
underlying generalized capabilities" usually reconstructed for Miocene catarrhines such as
Proconsul, Dryopithecus, and Sivapithecus (Rose, 1983:415).
Contrary to expectations of a linkage to the spread of African savanna caused by the
Messinian Event (Hsu et aI., 1973) approximately 5 million years ago (Brain, 1981), the
shift from life in the trees to life on the ground among apes and humans was initiated
about 15 million years ago (McCrossin, 1994a; Benefit and McCrossin, 1995). With re-
gard to environmental context, this change in substrate preference first occurred in an Af-
rican ape inhabiting mainly woodland, not savanna, environments. Although an-increase in
body size is widely thought to be directly correlated with terrestrial adaptations among
apes and humans (Napier, 1967; Andrews, 1983; Fleagle, 1985), body size estimates for
Kenyapithecus fall within the range of both tree-dwelling and ground-dwelling catarrhi-
nes. Colonization of savanna environments and attainment of body size in excess of that
seen in arboreal primates may have played a role, therefore, in the evolution of more spe-
cialized terrestrial adaptations among humans and gorillas, but seem not to be involved
with the origin of semi-terrestriality in Hominoidea.
Evidence from Kenyapithecus (McCrossin and Benefit, 1993, 1997) indicates that a
change to a pitheciine-like diet and sclerocarp foraging strategy may have been the most
profound impetus for the origin of terrestrial adaptations among apes and humans. Unlike
other platyrrhines that almost always move about in trees, PUhecia, Chiropotes, and Caca-
jao often descend to the ground (Ayres, 1986, 1989; Walker, 1996). Ayres (1986, 1989) and
Walker (1996) have suggested that utilization of terrestrial substrates by Pithecia and Caca-
jao may be related to their sclerocarpivory and attempts to locate their preferred foods. As
noted by Walker (1996:355), "in the habitats of Chiropotes and Cacajao, the various vegeta-
tion types diverge greatly in height", "emergent trees are often used for feeding", and "this
necessitates frequent ascent and descent for traveling between trees". Although Cacajao is
primarily a canopy primate (Walker, 1996), during the driest part of the year Ayres (1986)
observed uakaris feeding on the ground approximately 30 to 40% of the time. The terrestri-
ality of Cacajao may be permitted by the scarcity of predators in the seasonally flooded for-
ests (Ayres, 1986), but uakaris "occur in large groups, typically in the middle to upper
canopy, with numerous individuals to watch for predators" (Walker, 1996:358).
The possibility of a causal relationship between sclerocarp foraging and the advent of
terrestriality is compelling because it exists in such distantly related anthropoids as the
uakaris of the flooded forests of Amazonia and Kenyapithecus, the large-bodied hominoid
best known from the middle Miocene ofMaboko Island. The terrestrial behaviors ofuakaris
are nascent and occur primarily in association with feeding activities and during locomotion
between trees. Indeed, the postcranial skeleton ofCacajao lacks any of the indicators ofter-
restrial locomotion seen among Old World higher primates, including Kenyapithecus
(McCrossin, I 994a). It is tempting to view the postcranial adaptations for life on the ground
seen in Kenyapithecus as the product of a dietary and foraging strategy involving seed-pre-
dation, perhaps originally evolving in an Afropithecus-like ancestor (McCrossin, 1994a;
McCrossin and Benefit, 1993, 1997) but without some of the autapomorphies of that taxon.
Postcranial remains of Afropithecus have been interpreted as being Proconsul-like and in-
dicative of pronograde arboreal quadrupedalism (Rose, 1993), but their fragmentary preser-
vation does not enable confident determination of substrate preference. In any event, the
shift from the trees to the ground among African apes and humans seems not to have been
initiated by the consumption of grass, tuberous roots, or animal carcasses, but by the har-
vesting of hard fruit and nuts, presumably fallen from trees, by Kenyapithecus.
388 M. L. McCrossin et at.

Some interesting insights emerge from a preliminary comparison of the environ-

ment, diet, and body size of Kenyapithecus and the earliest known hominids, Ardipithecus
ramidus from 4.4 million year old deposits at Aramis, Ethiopia (White et ai., 1994; Wolde-
Gabriel et ai., 1994) and Australopithecus anamensis from 3.5 to 4.2 million year old sedi-
ments at Kanapoi, Allia Bay, and Sibilot in Kenya (Leakey et ai., 1995). The dominant
mammals from Aramis are colobine monkeys and a kudu, Tragelaphus sp. (WoldeGabriel
et ai., 1994). The Aramis site is reconstructed as dense woodland based in large part on the
assumption that the colobines were arboreally adapted. One of the Aramis colobines, how-
ever, is tentatively allied to Colobinae sp. A (WoldeGabriel et ai., 1994). A skeleton of
Colobinae sp. A from Leadu has been analyzed by Ciochon (1993). His results show de-
tailed and pervasive resemblances to semi-terrestrially adapted forms, especially Presby tis
entellus. Thus, the reconstruction of a dense woodland environment at Aramis is partly
based on direct analogy to the substrate and habitat preferences only of living African
colobines rather than the morphology of the fossil taxa present. According to Wolpoff
(1996), Ardipithecus can be estimated to have weighed approximately 42 kg.
The most numerous mammal species at Kanapoi are Parapapio, a macaque-like cer-
copithecine, and Tragelaphus kyaloae, a kudu (Leakey et ai., 1995). These faunal ele-
ments are suggested to indicate dry and open woodland or bushland conditions with
gallery forest. Australopithecus anamensis may have weighed approximately 47-55 kg
(Leakey et ai., 1995).
Surprisingly, therefore, there are preliminary indications that the earliest hominids
remained denizens of woodland environments and that more open environments were not
colonized until substantially later on, perhaps 3.5 million years ago. Moreover, substantial
increases in body weight that accompany the first appearance of bipedalism seem to be
consequences rather than causes of life on the ground (McCrossin, 1997).
Explicit dietary reconstructions have not yet been made for Ardipithecus ramidus
and Australopithecus anamensis, but it is interesting to see the degree to which detailed
elements of a Kenyapithecus-like dentognathic morphology are retained by the earliest
known hominids. Similarities to Kenyapithecus are especially evident in the dP3 morphol-
ogy of Ardipithecus and the proclined long axis of the mandibular symphysis of the type-
specimen of Australopithecus anamensis (McCrossin and Benefit, 1993, 1994, 1997;
White et ai., 1994; Leakey et ai., 1995).
Whatever the phylogenetic relationships of Kenyapithecus ultimately prove to be,
the fossil record has been very kind to us by providing a large-bodied hominoid caught in
the act of undergoing the transition from life in the trees to life on the ground. If Kenyapi-
thecus is an archaic avatar of modern great apes, then the ground-dwelling capabilities of
the living African apes and humans are an adaptive palimpsest, with ancient adaptations
for semi-terrestriality being obscured by the more recent and obvious specializations of
knuckle-walking and bipedalism. On the other hand, if Kenyapithecus turns out to be spe-
cially related to African apes and humans, then it would appear that the first steps on a
long evolutionary path toward the committed terrestrial bipedalism of humans were taken
15 million years ago, by a creature possessing few, if any, of the features we expected of a
pioneer of life on the ground.
It is not currently known whether the middle Miocene advent of semi-terrestriality in
cercopithecoids and hominoids was synchronously accompanied by the pervasive constella-
tion of changes in ecology and behavior associated with life on the ground among living ca-
tarrhines. Nevertheless, this transition in substrate preference among early representatives
of the two major catarrhine clades represents an adaptive departure of tremendous magni-
Fossil Evidence for the Origins of Terrestriality among Old World Higher Primates 389

tude from New World anthropoids and almost all prosimians, with profound repercussions
for the subsequent evolutionary career of Old World monkeys, apes, and humans.

ACKNOWLEDGMENTS

We gratefully acknowledge the Office of the President of the Republic of Kenya and
the National Museums of Kenya for permission to conduct excavations on Maboko Island.
We thank M.G. Leakey and the curatorial staff of Paleontology Division of the National
Museums of Kenya, especially M. Muungu, E. Mbua, F. Kyalo, A. Ibui, and N. Malit. We
extend our thanks to our field crew on Maboko Island, especially B. Onyango, C. Omondi
Agak, J. Onyango Miumi, and C. Obote Odhiambo. We thank the following agencies for
financial support: National Science Foundation, L.S.B. Leakey Foundation, Fulbright Col-
laborative Research Program, Wenner-Gren Foundation for Anthropological Research, Ro-
tary International Foundation, Office of Research Development and Administration of
Southern Illinois University, the Boise Fund of Oxford University, and the Lowie Fund of
the University of California at Berkeley. We thank T. Gatlin for illustrations and three
anonymous reviewers for their helpful comments. Finally, we thank E. Strasser, J. Fleagle,
A. Rosenberger, and H. McHenry for giving us the opportunity to participate in the con-
ference at the University of California at Davis and contribute to this volume.

REFERENCES
Andrews P (1971) Ramapithecus wickeri mandible from Fort Ternan, Kenya. Nature 228:537-540.
Andrews P (1981) Species diversity and diet in monkeys and apes during the Miocene. In CB Stringer (ed.): As-
pects of Human Evolution. London: Taylor and Francis, pp. 25-61.
Andrews P (1983) The natural history of Sivapithecus. In RL Ciochon and RS Corruccini (eds.) New Interpreta-
tions of Ape and Human Ancestry. New York: Plenum Press, pp. 441-464.
Andrews P (1985) Family group systematics and evolution among catarrhine primates. In EDelson (ed.): Ances-
tors: The Hard Evidence. New York: AR Liss, pp. 14--22.
Andrews P (1992) Evolution and environment in the Hominoidea. Nature 360:641-646.
Andrews P, and Aiello L (1984) An evolutionary model for feeding and positional behavior. In OJ Chivers, BA
Wood, and A Bilsborough (eds.): Food Acquisition and Processing in Primates. New York: Plenum Press,
pp. 429-466.
Andrews P, and Nesbit Evans E (1979) The environment of Ramapithecus in Africa. Paleobiology 5:22-30.
Andrews P, and Van Couvering JAH (1975) Paleoenvironments in the East African Miocene. Contrib. Primatol.
5:62-103.
Andrews P, and Walker A (1976) The primate and other fauna from Fort Ternan, Kenya. In GLI Isaac and ER
McCown (eds.): Human Origins: Louis Leakey and the East African Evidence. Menlo Park, CA: WA Ben-
jamin, pp. 279-304.
Andrews P, Meyer G, Pilbeam DR, Van Couvering JA, and Van Couvering JAH (1981) The Miocene fossil beds of
Maboko Island, Kenya: Geology, age, taphonomy, and paleontology. 1. Hum. Evol. 10:35-48.
Arambourg C (1945) Anancus osiris, un nouveau Mastodonte du Pliocene inferieur d'Egypt. Bull. Soc. Geol. Fr.
15:479-495.
Ashton EH, and Oxnard CE (1964) Locomotor patterns in primates. Proc. Zool. Soc. Lond. 142: 1-28.
Ayres JM (1986) Uakaris and Amazonian Flooded Forest. Ph.D. Dissertation, University of Cambridge.
Ayres JM (1989) Comparative feeding ecology of the uakari and bearded saki, Cacajao and Chiropotes. 1. Hum.
Evol. 18:697-716.
Beard KC, Teaford MF, and Walker A (1986) New wrist bones of Proconsul africanus and P. nyanzae from
Rusinga Island, Kenya. Folia Primatol. 47:97-118.
Benefit BR (1987) The Molar Morphology, Natural History, and Phylogenetic Position of the Middle Miocene
Monkey Victoriapithecus. Ph.D. Dissertation, New York University.
 390 M. L. McCrossin et al.

Benefit BR (1990) Fossil evidence for the dietary evolution of Old World monkeys. Am. J. Phys. Anthropol.
81:191.
Benefit BR (1991) The taxonomic status of Maboko small apes. Am. J. Phys. Anthropol. 12:50--51.
Benefit BR (1993) The permanent dentition and phylogenetic position of Victoriapithecus from Maboko Island,
Kenya. J. Hum. Evol. 25:83-172.
Benefit BR (1994) Phylogenetic, paleodemographic, and taphonomic implications of Victoriapithecus deciduous
teeth. Am. J. Phys. Anthropol. 95:277-331.
Benefit BR, and McCrossin ML (1989) New primate fossils from the middle Miocene of Maboko Island, Kenya. J.
Hum. Evol. 18:493-497.
Benefit BR, and McCrossin ML (1991) Ancestral facial morphology of Old World higher primates. Proc. Natl.
Acad. Sci. USA 88:5267-5271.
Benefit BR, and McCrossin ML (1993) The facial anatomy of Victoriapithecus and its relevance to the ancestral
cranial morphology of Old World monkeys and apes. Am. J. Phys. Anthropol. 92:329--370.
Benefit BR, and McCrossin ML (1994) Comparative study of the dentition of Kenyapithecus africanus and K.
wickeri. Am. J. Phys. Anthropol. Suppl. 18:55.
Benefit BR, and McCrossin ML (1995) Miocene hominoids and hominid origins. Annu. Rev. Anthropol.
24:237-256.
Benefit BR, and McCrossin ML (1997) Earliest known Old World monkey skull. Nature 388:368--371.
Benefit BR, Gitau SN, McCrossin ML, and Palmer AK (1998) A mandible of Mabokopithecus clarki sheds new
light on oreopithecid evolution. Am. J. Phys. Anthropol. Suppl. 26: I 09.
Birchette MG (1981) Postcranial remains of Cercopithecoides. Am. J. Phys. Anthropol. 54:20 I.
Birchette MG (1982) The Postcranial Skeleton of Paracolobus chemeroni. Ph.D. Dissertation, Harvard University.
Bonnefille R (1984) Cenozoic vegetation and environments of early hominids in East Africa. In RO Whyte (ed.):
The Evolution of the East Asian Environment, Volume 2: Palaeobotany, Palaeozoology, and Palaeoanthro-
pology. Hong Kong: University of Hong Kong, pp. 579--612.
Bonnefille R (1985) Evolution of the continental vegetation: The paleobotanical record from East Africa. S. Afr. J.
Sci. 81:267-270.
Bowen BE, and Vondra CF (1974) Paleoenvironmental interpretations of the Oligocene Gabal el Qatrani Forma-
tion, Fayum Depression, Egypt. Annals of the Geological Survey of Egypt 4:115-138.
Bown TM, Kraus MJ, Wing SL, Fleagle JG, Tiffany B, Simons EL, and Vondra CF (1982) The Fayum forest revis-
ited. J. Hum. Evol. 11:603-632.
Brain CK (1981) Hominid evolution and climatic changes. S. Afr. J. Sci. 77: 104-105.
Brown B (1997) Miocene hominoid mandibles: Functional and phylogenetic perspectives. In DR Begun, CV
Ward, and MD Rose (eds.): Function, Phylogeny, and Fossils: Miocene Hominoid Evolution and Adapta-
tions. New York: Plenum, pp. 153--171.
Brown B, and Ward S (1988) Basicranial and facial topography in Pongo and Sivapithecus. In JH Schwartz (ed.)
Orang-utan Biology. London: Oxford University Press, pp. 247-260.
Brown B, Hill A, and Ward S (1991) New Miocene large hominoids from the Tugen Hills, Baringo District,
Kenya. Am. J. Phys. Anthropol. Suppl. 12:55.
Butler PM (1984) Macroscelidea, Insectivora and Chiroptera from the Miocene of East Africa. Palaeovertebrata
14: 117-200.
Caldicott JO (1986) An ecological and behavioral study of the pig-tailed macaque. Contrib. Primatol. 21: 1-259.
Cant JG (1987) Effects of sexual dimorphism in body size on feeding postural behavior of Sumatran orangutans
(Pongo pygmaeus). Am. J. Phys. Anthropol. 74:143-148.
Cerling TE, Quade J, Ambrose SH, and Sikes NE (1991) Fossil soils, grasses, and carbon isotopes from Fort Ter-
nan, Kenya: Grassland or woodland? J. Hum. Evol. 21:295-306.
Churcher CS (1970) Two new upper Miocene giraffids from Fort Ternan, Kenya, East Africa: Palaeotragus pri-
maevus n. sp. and Samotheriumafricanum n. sp. In LSB Leakey and RJG Savage (eds.): Fossil Vertebrates
of Africa, Vol. 2. London: Academic Press, pp. 1-106.
Cifelli RL, Ibui AK, Jacobs LL, and Thorington RW (1986) A giant tree squirrel from the late Miocene of Kenya.
J. Mamm. 67:274-283.
Ciochon RL (1993) Evolution of the cercopithecoid forelimb: Phylogenetic and functional implications from mor-
phometric analyses. Univ. Calif. Publ. Geol. Sci. /38:1-251.
Clutton-Brock TH, and Harvey PH (1977) Primate social organization and ecology. Nature 250:539--542.
Conroy GC (1972) Problems in the interpretation of Ramapithecus: With special reference to anterior tooth reduc-
tion. Am. J. Phys. Anthropol. 37:41-47.
Conroy GC (1976) Primate postcranial remains from the Oligocene of Egypt. Contrib. Primatol. 8: 1-134.
Conroy GC (1987) Problems of body-weight estimation in fossil primates. Int. J. Primatol. 8: 115-137.
Conroy GC, and Fleagle JG (1972) Locomotor behavior in living and fossil pongids. Nature 237: I 03-1 04.
Fossil Evidence for the Origins ofTerrestriality among Old World Higher Primates 391

Coursey DG (1973) Hominid evolution and hypogeous plant foods. Man 8:634-635.
Crook JH, and Gartlan JS (1966) On the evolution of primate societies. Nature 210:1200-1203.
Crusafont M, and Aguirre E (1971) A new species of Percrocuta from the middle Miocene of Kenya. Abh. Hess.
L.-Amt. Bodenforsch. 60:51-58.
Dagosto M, and Terranova CJ (1992) Estimating the body size of Eocene primates: A comparison of results from
dental and postcranial variables. Int. J. Primatol. 13:307-344.
Davison GWH (1982) Convergence with terrestrial cercopithecines by the monkey Rhinopithecus roxellanae. Fo-
lia Primatol. 37:209-215.
Delson E (1973) Fossil Colobine Monkeys of the Circum-Mediterranean Region and the Evolutionary History of
the Cercopithecidae (Primates, Mammalia). Ph.D. Dissertation, Columbia University.
Delson E (1975) Evolutionary history of the Cercopithecidae. Contributions to Primatology 5: 167-217.
Denys C, and Jaeger J-J (1992) Rodents of the Miocene site of Fort Ternan (Kenya). First part: Phiomyids, bathy-
ergids, sciurids and anomalurids. N. Jb. Geol. Paleont. Abh. 185:63-84.
DeVore I (1963) A comparison of the ecology and behavior of monkeys and apes. In SL Washburn (ed.): Classifi-
cation and Human Evolution. Chicago: Aldine, pp. 301-319.
DiFiore A, and Rendall D (1994) Evolution of social organization: A reappraisal for primates by using phyloge-
netic methods. Proc. Natl. Acad. Sci. USA 91:9941-9945.
Doran DM (1993) Sex differences in adult chimpanzee positional behavior: The influence of body size on locomo-
tion and posture. Am. 1. Phys. Anthropol. 91:99-116.
Dunbar RIM, and Dunbar EP (1974) Ecological relations and niche separation between sympatric terrestrial pri-
mates in Ethiopia. Folia Primatol. 2/:36--60.
Eisenberg JF, Muckenhim NA, and Rudran R (1972) The relation between ecology and social structure in pri-
mates. Science 176:863-874.
Feibel CS, and Brown FH (1991) Age of the primate-bearing deposits on Maboko Island, Kenya. 1. Hum. Evol.
2/:221-225.
Fleagle JG (1974) Dynamics of a brachiating siamang (Hylobates (Symphalangl/s) syndactylus). Nature
248:259-260.
Fleagle JG (1977) Locomotor behavior and skeletal anatomy of sympatric Malaysian leaf-monkeys (Presby tis ob-
scura and Presbytis melalophos). Yrbk. Phys. Anthropol. 20:440-453.
Fleagle JG (1978) Size distributions ofliving and fossil primate faunas. Paleobiology 4:67-76.
Fleagle JG (1983) Locomotor adaptations of Oligocene and Miocene hominoids and their phyletic implications. In
RL Ciochon and RS Corruccini (eds.): New Interpretations of Ape and Human Ancestry. New York: Ple-
num Press, pp. 301-323.
Fleagle JG (1985) Size and adaptation in primates. In WL Jungers (ed.): Size and Scaling in Primate Biology. New
York: Plenum Press, pp. 1-19.
Fleagle JG (1988) Primate Adaptation and Evolution. New York: Academic Press.
Fleagle JG, and Mittermeier RA (1980) Locomotor behavior, body size, and comparative ecology of seven Suri-
nam monkeys. Am. J. Phys. Anthropol. 52:301-314.
Fleagle JG, and Simons EL (1982) The humerus of Aegyptopithecus: A primitive anthropoid. Am. J. Phys. Anthro-
pol. 59:175--193.
Fleagle JG, Kay RF, and Simons EL (1980) Sexual dimorphism in early anthropoids. Nature 287:328-330.
Gautier-Hion A, and Gautier J-P (1985) Sexual dimorphism, social units and ecology among sympatric forest
guenons. In J Ghesquiere, RD Martin, and F Newcombe (eds.): Human Sexual Dimorphism. London: Tay-
lor and Francis, pp. 61-77.
Gebo DL (\992) Plantigrady and foot adaptation in African apes: Implications for hominid origins. Am. J. Phys.
Anthropol. 89:29-58.
Gebo DL, Beard KC, Teaford MF, Walker A, Larson SG, Jungers WL, and Fleagle JG (1988) A hominoid proxi-
mal humerus from the early Miocene of Rusinga Island, Kenya. 1. Hum. Evol. 17:393-40 I.
Gentry A (1970) The Bovidae of the Fort Ternan fossil fauna. In LSB Leakey and RJG Savage (eds.) Fossil Verte-
brates of Africa, Vol 2. London: Academic Press, pp. 243-324.
Gentry AW (1990) Ruminant artiodactyls of Pasalar, Turkey. J. Hum. Evol. 19:529-550.
Gingerich PD (1977) Correlation of tooth size and body size in living hominoid primates, with a note on relative
brain size in Aegyptopithecus and Proconsul. Am. J. Phys. Anthropol. 47:395--398.
Gingerich PD, Smith BH, and Rosenberg K (1982) Allometric scaling in the dentition of primates and prediction
of body weight from tooth size in fossils. Am. J. Phys. Anthropol. 58:81-100.
Ginsburg L, and Mein P (1980) Crouzelia rhodanica, nouvelle espece de primate Catarrhinien et essai sur la posi-
tion systematique des Pliopithecidae. Bull. Mus. Nat., Paris 2:57--85.
Greenfield LO (1979) On the adaptive pattern of"Ramapithecus". Am. J. Phys. Anthropol. 50:527-548.
 392 M. L. McCrossin et al.

Hall KRL (1965) Behavior and ecology of the wild patas monkey, Erythrocebus palas, in Uganda. J. Zool.
/48:15-87.
Hamilton WR (1978) Fossil giraffes from the Miocene of Africa and a revision of the phylogeny of the Giraf-
foidea. Phil. Trans. Roy. Soc. Lond. 283:165-229.
Harrison T (1989) New postcranial remains of Victoriapithecus from the middle Miocene of Kenya. J. Hum. Evol.
18:3-54.
Harrison T (1992) A reassessment of the taxonomic and phylogenetic affinities of the fossil catarrhines from Fort
Ternan, Kenya. Primates 33:501-522.
Hillenius D (1978) Notes on chameleons IV. A new chameleon from the Miocene of Fort Ternan, Kenya
(Chamaeleonidae, Reptilia). Beaufortia 28:9-15.
Hooijer DA (1968) A rhinoceros from the late Miocene of Fort Ternan, Kenya. Zoologische Mededelingen
43:77-92.
Hsu KJ, Ryan WBF, and Cita MB (1973) Late Miocene desiccation of the Mediterranean. Nature 242:240-244.
Hunt K (1992) Positional behavior of Pan troglodytes in the Mahale Mountains and Gombe Stream National
Parks, Tanzania. Am. J. Phys. Anthropol. 87:83-105.
Hylander W (1984) Stress and strain in the mandibular symphysis of primates: A test of competing hypotheses.
Am. J. Phys. Anthropol. 64:1--46.
Ishida H (1986) Investigation in northern Kenya and new hominoid fossils. Kagaku 56:220-226.
Ishida H, Pickford M, Nakaya H, and Nakano Y (1984) Fossil anthropoids from Nachola and Samburu Hills, Sam-
buru District, Northern Kenya. Afr. Stud. Monogr. (Kyoto Univ.) Suppl. 2:73-86.
Janis CM, and Ehrhardt D (1988) Correlation of relative muzzle width and relative incisor width with dietary pref-
erence in ungulates. Zool. 1. Linnean Soc. 92:267-284.
Jenkins FA (1973) The functional anatomy and evolution of the mammalian humero-ulnar articulation. Am. J.
Anat. /37:281-298.
Jenkins FA, and Camazine SM (1977) Hip structure and locomotion in ambulatory and cursorial carnivores. J.
Zool., Lond. 18/:351-370.
Jenkins FA, and Fleagle JG (1975) Knuckle-walking and the functional anatomy of the wrists in living apes. In
RH Tuttle (ed.): Primate Functional Morphology and Evolution. The Hague: Mouton, pp. 213-227.
Jolly CJ (1966) Introduction to the Cercopithecoidea with notes on their use as laboratory animals. Symp. Zool.
Soc. Lond. 17:427--457.
Jolly CJ (1967) The evolution of baboons. In H Vagtborg (ed.): The Baboon in Medical Research, Volume 2.
Austin: University of Texas Press, pp. 23-50.
Jolly CJ (1970) The seed eaters: A new model of hominid differentiation based on a baboon analogy. Man 5:5-26.
Jolly CJ (1972) The classification and natural history of Theropithecus (Simopithecus) (Andrews, 1916), baboons
of the African Plio-Pleistocene. Bull. Brit. Mus. (Nat. Hist.), Geol. 22:1-122.
Jouventin P (1975) Observations sur la socio-ecologie du mandrill. La Terre et La Vie 29:493-532.
Kay RF (1977) Post-Oligocene evolution of catarrhine diets. Am. J. Phys. Anthropol. 47:141-142.
Kay RF (1981) The nut-crackers: A new theory of the adaptation of the Ramapithecinae. Am. J. Phys. Anthropol.
55:141-151.
Kay RF (1984) On the use of anatomical features to infer foraging behavior in extinct primates, In PS Rodman and
JGH Cant (eds.): Adaptations for Foraging in Nonhuman Primates. New York: Columbia University Press,
pp.21-53.
Kinzey WG (1992) Dietary and dental adaptations in the Pitheciinae. Am. J. Phys. Anthropol. 88:499-514.
Kinzey WG, and Norconk MA (1993) Physical and chemical properties of fruit and seeds eaten by Pithecia and
Chiropotes in Surinam and Venezuela. Int. J. Primatol. 14:207-227.
Knussman R (1967) Humerus, Ulna und Radius der Simiae. Bibl. Primatol. 5: 1-399.
Langdon JH (1986) Functional morphology of the Miocene hominoid foot. Contrib. Primatol. 22:1-225.
Larson SG (1988) Subscapularis function in gibbons and chimpanzees: Implications for interpretation of humeral
torsion in hominoids. Am. 1. Phys. Anthropol. 76:449--462.
Larson SG, and Stem JT (1992) Further evidence for the role of supraspinatus in quadrupedal monkeys. Am. J.
Phys. Anthropol. 87:359-363.
Lavocat R (1964) Fossil rodents from Fort Ternan, Kenya. Nature 202: 1131.
Lavocat R (1988) Un rongeur Bathyergide nouveau remarquable du Miocene de Fort Ternan (Kenya). C. R. Acad.
Sci. Paris 306: 1301-1304.
Lavocat R (1989) Osteologie de la tete de Richardus excavans Lavocat, 1988. Palaeovertebrata 19:73-80.
Le Gros Clark WE (1959) The Antecedents of Man: An Introduction to the Evolution of the Primates. Edinburgh:
Edinburgh University Press.
Le Gros Clark WE, and Leakey LSB (1951) The Miocene Hominoidea of East Africa. Fossil Mammals of Africa
/:1-117.
Fossil Evidence for the Origins of Terrestriality among Old World Higher Primates 393

Leakey LSB (1962) A new lower Pliocene fossil primate from Kenya. Ann. Mag. Nat. Hist. 4:689--696.
Leakey LSB (1967) An early Miocene member of Hominidae. Nature 2 I 3: 155-163.
Leakey MG, Feibel CS, McDougall I, and Walker A (1995) New four-million-year-old hominid species from
Kanapoi and Allia Bay, Kenya. Nature 376:565-571.
Leutenegger W (1970) Das Becken der rezenten Primaten. Morph. Jahrb 115:1-101.
Leutenegger W (1978) Scaling of sexual dimorphism in body size and breeding system in primates. Nature
272:610-611.
Leutenegger W, and Kelly JT (1977) Relationship of sexual dimorphism in canine size and body size to social, be-
havioral and ecological correlates in anthropoid primates. Primates 18: 117-136.
Lovejoy CO (1981) The origin of man. Science 2//:341-350.
Macinnes DG (1936) A new genus of fossil deer from the Miocene of Africa. 1. Linn. Soc. Lond. Zoo I.
39:521-530.
Macinnes DG (1942) Miocene and post-Miocene Proboscidea from East Africa. Trans. Zool. Soc., Lond.
25:33-103.
MacKinnon J (1974) The behavior and ecology of wild orang-utans (Pongo pygmaeus). Anim. Behav. 22:3-74.
Martin RD (1979) Phylogenetic aspects of prosimian behavior. In GA Doyle and RD Martin (eds.): The Study of
Prosimian Behavior. New York: Academic Press, pp. 45-77.
Martin RD (1990) Primate Origins and Evolution: A Phylogenetic Reconstruction. Princeton NJ: Princeton Uni-
versity Press.
McCrossin ML (1990) Fossil galagos from the middle Miocene of Kenya. Am. 1. Phys. Anthropol. 81:265-266.
McCrossin ML (1992a) A new species of bushbaby from the middle Miocene of Maboko Island, Kenya. Am. J.
Phys. Anthropol. 89:215-233.
McCrossin ML (1992b) An oreopithecid proximal humerus from the middle Miocene of Maboko Island, Kenya.
Int. J. Primatol. 13:659--677.
McCrossin ML (I 994a) The Phylogenetic Relationships, Adaptations, and Ecology of Kenyapithecus. Ph.D Dis-
sertation, University of California at Berkeley.
McCrossin ML (1994b) Semi-terrestrial adaptations of Kenyapithecus. Am. J. Phys. Anthropol. Suppl.
/8:142-143.
McCrossin ML (1997) New postcranial remains of Kenyapithecus and their implications for understanding the ori-
gins of hominoid terrestriality. Am. J. Phys. Anthropol. Suppl. 24: 164.
McCrossin ML, and Benefit BR (1992) Comparative assessment of the ischial morphology of Victoriapithecus
macinnesi. Am. J. Phys. Anthropol. 87:277-290.
McCrossin ML, and Benefit BR (1993) Recently discovered Kenyapithecus mandible and its implications for great
ape and human origins. Proc. Natl. Acad. Sci. USA 90: 1962-1966.
McCrossin ML, and Benefit BR (1994) Maboko Island and the evolutionary history of Old World monkeys and
apes. In RS Corruccini and RL Ciochon (eds.): Integrative Paths to the Past: Paleoanthropological Ad-
vances in Honor of F. Clark Howell. Englewood Cliffs, NJ: Prentice-Hall, pp. 95-124.
McCrossin ML, and Benefit BR (1997) On the relationships and adaptations of Kenyapithecus, a large-bodied
hominoid from the middle Miocene of eastern Africa. In DR Begun, CV Ward, and MD Rose (eds.): Func-
tion, Phylogeny and Fossils: Miocene Hominoid Evolution and Adaptations. New York: Plenum, pp.
241-267.
McCrossin ML, Benefit BR, and Gitau SN (1998) Functional and phylogenetic analysis of the distal radius of
Kenyapithecus, with comments on the origin of the African great ape and human clade. Am. J. Phys. An-
thropol. Suppl. 26:158--159.
McHenry HM, and Corruccini RS (1983) The wrist of Proconsul africanus and the origin of hominoid postcranial
adaptations. In RL Ciochon and RS Corruccini (eds.): New Interpretations of Ape and Human Ancestry.
New York: Plenum, pp. 353-367.
Mihlbachler MC, McCrossin ML, and Benefit BR (1996) Body size distribution and the evolution of African pri-
mate community structure. Am. J. Phys. Anthropol. Suppl. 22: 168.
Mitterrneier RA, and Van Roosmalen MGM (1981) Preliminary observations on habitat utilization and diet in
eight Surinam monkeys. Folia Primatol. 36:1-39.
Moya-Sola S, and Kohler M (1996) A DryopitheclIs skeleton and the origins of great ape locomotion. Nature
379:156--159.
Napier JR (1967) Evolutionary aspects of primate locomotion. Am. 1. Phys. Anthropol. 27:333-342.
Napier JR (1970) Paleoecology and catarrhine evolution. In JR Napier and PH Napier (eds.): Old World Monkeys.
London: Academic Press, pp. 55-95.
Napier JR, and Davis PR (1959) The forelimb skeleton and associated remains of Proconsul africanus. Fossil
Mammals of Africa 16: 1--69.
Napier JR, and Napier PH (1967) A Handbook of Living Primates. New York: Academic Press.
 394 M. L. McCrossin et al.

Napier JR, and Napier PH (1985) The Natural History of the Primates. Cambridge, MA: The MIT Press.
Nesbit Evans EM, Van Couvering JAH, and Andrews P (1981) Palaeoecology of Miocene sites in Western Kenya.
1. Hum. Evol. 10:99-116.
Norconk MA (1996) Seasonal variation in the diets of white-faced and bearded sakis (Pithecia pithecia and Chiro-
potes satanas) in Guri Lake, Venezuela. In MA Norconk, AL Rosenberger, and PA Garber (eds.): Adaptive
Radiations ofNeotropical Primates. New York: Plenum Press, pp. 403--423.
O'Connor BL (1975) The functional morphology of the cercopithecoid wrist and inferior radioulnar joints, and
their bearing on some problems in the evolution of the Hominoidea. Am. 1. Phys. Anthropol. 43: 113-122.
Oxnard CE (1963) Locomotor adaptations in the primate forelimb. Symp. Zool. Soc. Lond. 10: 165-182.
Palmer AK, Benefit BR, McCrossin ML, and Gitau, SN (1998) Paleoecological implications of dental microwear
analysis for the middle Miocene primate fauna from Maboko Island, Kenya. Am. 1. Phys. Anthropol.
Suppl. 26: 175.
Pickford M (1982a) New higher primate fossils from the middle Miocene deposits at Majiwa and Kaloma, western
Kenya. Am. J. Phys. Anthropol. 58:1-19.
Pickford M (1982b) On the origins of Hippopotamidae together with descriptions of two new species, a new genus
and a new subfamily from the Miocene of Kenya. Geobios 16:193-217.
Pickford M (1983) Sequence and environments of the lower and middle Miocene hominoids of western Kenya. In
RL Ciochon and RS Corruccini (eds.): New Interpretations of Ape and Human Ancestry. New York: Ple-
num Press, pp. 421--439.
Pickford M (1984) Kenya Palaeontology Gazetteer, Volume I - Western Kenya. Nairobi: National Museums of
Kenya, Department of Sites and Monuments Documentation.
Pickford M (1985) A new look at Kenyapithecus based on recent discoveries in western Kenya. 1. Hum. Evol.
14:113-143.
Pickford M (1987) Fort Ternan (Kenya) paleoecology. 1. Hum. Evol. 16:305-309.
Pickford M, and Senut B (1988) Habitat and locomotion in Miocene cercopithecoids. In F Bourliere, A Gautier-
Hion, and J Kingdon (eds.): A Primate Radiation: Evolutionary History of the African Guenons. Cam-
bridge: Cambridge University Press, pp. 35-53.
Pickford M, Senut B, Hadoto 0, Musisi J, and Kariira C (1986) Nouvelle decouvertes dans Ie Miocene inferieur
de Napak, Ouganda Oriental. C. R. Acad. Sci. Paris 302:47-52.
Pilbeam DR (1997) Research on Miocene hominoids and hominid origins: The last three decades. In DR Begun,
CV Ward, and MD Rose (eds.): Function, Phylogeny, and Fossils: Miocene Hominoid Evolution and Adap-
tations. New York: Plenum Press, pp. 13-28.
Pilbeam DR, and Walker A (1968) Fossil monkeys from the Miocene of Napak, northeast Uganda. Nature
220:657-660.
Plavcan JM, and van Schaik C (1997) Interpreting hominid behavior on the basis of sexual dimorphism. J. Hum.
Evol. 32:345-374.
Radinsky L (\974) The fossil evidence of anthropoid brain evolution. Am. J. Phys. Anthropol. 41: 15-27.
Radinsky L (1982) Some cautionary notes on making inferences about relative brain size. In E Armstrong and 0
Falk (eds.): Primate Brain Evolution: Methods and Concepts. New York: Plenum Press, pp. 29-37.
Remis MJ (\995) Effects of body size and social context on the arboreal activities oflowland gorillas in the Cen-
tral African Republic. Am. 1. Phys. Anthropol. 97:413--434.
Reno PL, McCollum MA, and Lovejoy CO (1997) Anthropoid radial neck length and its implications for hominid
locomotor behavior. Am. 1. Phys. Anthropol. Suppl. 24: 197.
Retallack GJ, Dugas DP, and Bestland EA (1990) Fossil soils and grasses ofa middle Miocene East African grass-
land. Science 247: 1325-1328.
Retallack GJ (1992) Middle Miocene fossil plants from Fort Ternan (Kenya) and evolution of African grasslands.
Paleobiology 18:382--400.
Ripley S (1967) The leaping of langurs: A problem in the study of locomotor behaviors. Am. 1. Phys. Anthropol.
26: 149-170.
Rollinson JMM, and Martin RD (1981) Comparative aspects of primate locomotion, with special reference to ar-
boreal cercopithecines. Symp. Zoo I. Soc. Lond. 48:377--427.
Rose MD (1973) Quadrupedalism in primates. Primates 14:337-358.
Rose MD (1977) Positional behavior of olive baboons (Papio anubis) and its relationship to maintenance and so-
cial activities. Primates 18:59-116.
Rose MD (I983) Miocene hominoid postcranial morphology: Monkey-like, ape-like, neither, or both? In RL Cio-
chon and RS Corruccini (eds.): New Interpretations of Ape and Human Ancestry. New York: Plenum Press,
pp. 405--417.
Rose MD (1984) Hominoid specimens from the middle Miocene Chinji Formation, Pakistan. 1. Hum. Evol.
/3:503-516.
Fossil Evidence for the Origins of Terrestriality among Old World Higher Primates 395

Rose MD (1989) New postcranial specimens of catarrhines from the middle Miocene Chinji Formation, Pakistan:
Descriptions and a discussion of proximal humeral functional morphology in anthropoids. J. Hum. Evol.
8:131-162.
Rose MD (1993) Locomotor anatomy of Miocene hominoids. In D Gebo (ed.): Postcranial Adaptation in Nonhu-
man Primates. DeKalb: Northern Illinois University Press, pp. 252-272.
Rose MD, Leakey MG, Leakey RE, and Walker AC (1992) Postcranial specimens of Simiolus enjiessi and other
primitive catarrhines from the early Miocene of Lake Turkana, Kenya. J. Hum. Evol. 22: 171-237.
Rose MD, Nakano Y, and Ishida H (1996) Kenyapithecus postcranial specimens from Nachola, Kenya. African
Study Monographs (Kyoto University) Supplementary Issue 24: I-56.
Rowell TE (1966) Forest living baboons in Uganda. J. Zool., Lond.149:344-364.
Rowell TE, and Chism J (1986) Sexual dimorphism and mating systems: Jumping to conclusions. In M Pickford
and B Chiarelli (eds.): Sexual Dimorphism in Living and Fossil Primates. Florence: II Sedicesimo, pp.
\07-111.
Ruff CB, Walker A, and Teaford MF (1989) Body mass, sexual dimorphism and femoral proportions of Proconsul
from Rusinga and Mfangano Islands, Kenya. J. Hum. Evol. 18:515-536.
Senut B (1986a) Nouvelle decouvertes de restes post-craniens de primates Miocenes (Hominoidea and Cercopithe-
coidea) sur Ie site Maboko au Kenya occidental. C. R. Acad. Sci. Paris 303:1359-1362.
Senut B (I986b) New data on Miocene hominoid humeri from Pakistan and Kenya. In JG Else and PC Lee (eds.):
Primate Evolution. Cambridge: Cambridge University Press, pp. 151-161.
Shipman P (1986) Paleoecology of Fort Ternan reconsidered. J. Hum. Evol. 15: 193-204.
Shipman P, Walker A, Van Couvering JAH, and Hooker PJ (1981) The Fort Ternan hominoid site, Kenya: Geol-
ogy, taphonomy, and paleoecology. J. Hum. Evol. 10:49-72.
Smith RJ, and Jungers WL (1997) Body mass in comparative primatology. J. Hum. Evol. 32:523-559.
Simons EL, and Pilbeam DR (1978) Ramapithecus (Hominidae, Hominoidea). In VJ Maglio and HBS Cooke
(eds.): Evolution of African Mammals. Cambridge, MA: Harvard University Press, pp. 147-153.
Solounias N, and Moelleken SMC (1993) Tooth microwear and premaxillary shape of an archaic antelope. Lethaia
26:261-268.
Spoor CF, Sondaar PY, and Hussein ST (1992) A new hominoid hamate and first metacarpal from the late Miocene
Nagri Formation of Pakistan. J. Hum. Evol. 21:413-424.
Strasser E (1988) Pedal evidence for the origin and diversification of cercopithecid clades. J. Hum. Evol.
17:225-245.
Strasser E (1997) A cladistic analysis of the cercopithecid foot. Am. J. Phys. Anthropol. Suppl. 24:222.
Strasser E, and Delson E (1987) Cladistic analysis of cercopithecid relationships. J. Hum. Evol. 16:81--99.
Struhsaker TT (1969) Correlates of ecology and social organization among African cercopithecines. Folia Prima-
tol. 1l:80-118.
Sussman RW (1974) Ecological distinction in sympatric species of Lemur. In RD Martin, GA Doyle, and AC
Walker (eds.): Prosimian Biology. London: Duckworth, pp. 75-108.
Szalay FS (1975) Hunting-scavenging protohominids: A model for hominid origins. Man 10:420-429.
Tassy P (1979) Les proboscideans (Mammalia) du Miocene d' Afrique orientale: Resultats preliminaires. Bull Soc.
Geol. France 21:265-269.
Temerin LA, and Cant JGH (1983) The evolutionary divergence of Old World monkeys and apes. Am. Nat.
122:335-351.
Thomas H (1979) Les bovides Miocenes des rifts est-Africains: Implications paleobiogeographiques. Bull. Soc.
Geol. France 21:295-299.
Thomas H (1984) Les bovides ante-Hipparions des Siwaliks Inferieurs (Plateau du Potwar, Pakistan). Mem. Soc.
Geol. France 145:1-65.
Thomas H (1985) Les Giraffoidea et les Bovidae miocenes de la Formation Nyakach (Rift Nyanza, Kenya). Pa-
laeontographica (A) 183:64-89.
Tong H, and Jaeger J-J (1993) Muroid rodents from the middle Miocene Fort Ternan locality (Kenya) and their
contribution to the phylogeny ofmuroids. Palaeontographica Abt. A 229:51-73.
Tuttle RH (1967) Knuckle-walking and the evolution of hominoid hands. Am. J. Phys. Anthropol. 26: 171-206.
Tuttle RH (1974) Darwin's apes, dental apes and the descent of man: Normal science in evolutionary anthropol-
ogy. Curro Anthropol. 15:389-398.
Van Couvering JAH (1980) Community evolution in East Africa during the Late Cenozoic. In AK Behrensmeyer
and AP Hill (eds.): Fossils in the Making: Vertebrate Taphonomy and Paleoecology. Chicago: University of
Chicago Press, pp. 272-298.
Van Couvering JAH, and Van Couvering JA (1976) Early Miocene mammal fossils from East Africa: Aspects of
geology, faunistics and paleoecology. In GLI Isaac and E McCown (eds.): Human Origins: Louis Leakey
and the East African Evidence. Menlo Park, CA: WA Benjamin, pp. 155-207.
396 M. L. McCrossin et aL

Van der Made J (1996) Albanohyus, a small Miocene pig. Acta Zoologica Cracoviensia 39:293-303.
Van Roosmalen MGM, Mittermeier RA, and Fleagle JG (1988) Diet of the bearded saki (Chiropotes satanas chi-
ropotes): A neotropical seed predator. Am. J. Primatol. 14: 11-35.
von Beyrich H (1861) Uber Semnopithecus pentelicus. Phys. Abh. K. Akad. Wiss., BerI.1860:1-26.
von Koenigswald GHR (1969) Miocene Cercopithecoidea and Oreopithecoidea from the Miocene of East Africa.
Fossil Vertebrates of Africa 1:39-51.
Walker A, and Andrews P (1973) Reconstruction of the dental arcade of Ramapithecus wickeri. Nature
224:213-214.
Walker A, and Pickford M (1983) New postcranial fossils of Proconsul ajricanus and Proconsul nyanzae. In RL
Ciochon and RS Corruccini (eds.): New Interpretations of Ape and Human Ancestry. New York: Plenum
Press, pp. 325-351.
Walker SE (1996) The evolution of positional behavior in the saki-uakaris (Pithecia, Chiropotes, and Cacajao). In
MA Norconk, AL Rosenberger, and PA Garber (eds.): Adaptive Radiations of Neotropical Primates. New
York: Plenum Press, pp. 335-367.
Ward CV, Walker A, Teaford MF, and Odhiambo I (1993) Partial skeleton of Proconsul nyanzae from Mfangano
Island, Kenya. Am. J. Phys. Anthropol. 90:77-111.
Ward CV, RuffCB, Walker A, Teaford MF, Rose MD, and Nengo 10 (1995) Functional morphology of Proconsul
patellas from Rusinga Island, Kenya, with implications for other Miocene-Pliocene catarrhines. J. Hum.
Evol. 29:1-19.
Ward SC, and Pilbeam DR (1983) Maxillofacial morphology of Miocene hominoids from Africa and Indo-Paki-
stan. In RL Ciochon and RS Corruccini (eds.): New Interpretations of Ape and Human Ancestry. New
York: Plenum Press, pp. 211-238.
Waser P (1984) Ecological differences and behavioral contrasts between two mangabey species. In PS Rodman
and JGH Cant (eds.): Adaptations for Foraging in Non-human Primates. New York: Columbia University
Press, pp. 211-238.
Washburn SL (1968) Speculations on the problems of man's coming to the ground. In B Rothblatt (ed.): Changing
Perspectives on Man. Chicago: University of Chicago Press, pp. 191-206.
Waterman HC (1929) Studies on the evolution ofthe pelvis of man and other primates. Bull. Am. Mus. Nat. Hist.
58:585-642.
White TO (1995) African omnivores: Global climatic change and Plio-Pleistocene hominids and suids. In E Vrba
(ed.) Paleoclimate and Evolution, With Emphasis on Human Origins. New Haven: Yale University Press,
pp.369-384
White TO, Suwa G, and Asfaw B (1994) Australopithecus ramidus, a new species of early hominid from Aramis,
Ethiopia. Nature 371:306-312.
Whitworth T (1958) Miocene ruminants of East Africa. Fossil Mammals of Africa 15: I-50.
Winkler AJ (1994) Middle Miocene rodents from Maboko Island, western Kenya: Contributions to understanding
small mammal evolution during the Neogene. J. Vert. Paleo. 14:53A.
Woldegabriel G, White TO, Suwa G, Renne P, de Heinzelin J, Hart WK, and Heiken G (1994) Ecological and tem-
poral placement of early Pliocene hominids at Aramis, Ethiopia. Nature 371:330-333.
Wolpoff MH (1982) Ramapithecus and hominid origins. Curro Anthropol. 23:50 I-51 O.
Wolpoff MH (1996) Human Evolution. New York: McGraw-Hili.
Yulish SM (1970) Anterior tooth reduction in Ramapithecus. Primates 11:255-263.
Zapfe H (1960) Die Primatenfunde aus der miozanen Spaltenfullung von Neudorf an der March (Oevinska Nova
Ves), Tschechoslowakei. Mit anhang: Der Primatenfund aus dem Miozan von Klein Hadersdorf in Nied-
erosterreich. Schweiz. Palaeontol. Abh. 78:1-293.
 20

ECOLOGICAL MORPHOLOGY OF
A USTRALOPITHECUS AFARENSIS
Traveling Terrestrially, Eating Arboreally

Kevin D. Hunt

Department of Anthropology
Indiana University
Bloomington, Indiana 47405

1. INTRODUCTION

Kinzey (1976, 1977, 1978) was quick to appreciate the utility of integrating system-
atic ecological research with study of positional behavior and morphology, a synthetic
area of scholarly pursuit now distinguished by its own appellation, ecological morphology
(Wainwright and Riley, 1994). Primatologists so often focus on food and food-gathering
behaviors as keys to understanding primate anatomy because primate activity budgets are
dominated by feeding. Across habitats ranging from thicket woodland to closed canopy
forest, chimpanzees consistently dedicate half of their activity budget to feeding (Table I).
The next most common activity, "resting," might as well be called "digesting." As a sim-
plifying first-assumption, an ecological perspective takes the view that the hominoid body
is a food-getting machine, and ignores the presumably lesser selective roles played by in-
traspecific aggression and predator avoidance (though predation may be more important
for smaller primates: van Schaik, 1983; van Schaik and van Hooff', 1983; van Schaik et
ai., 1983a, b; van Schaik and van Noordwijk, 1989). This view holds as significant the rar-
ity of predation on hominoids (Cheney and Wrangham, 1986), and that intraspecific
agonism is less a threat to survival than is starvation.
For those who find this perspective comfortable, feeding hypotheses (Du Brul, 1962;
Jolly, 1970; Tuttle, 1975, 1981; Rose, 1976, 1984, 1991; Stem, 1976; Wrangham, 1980;
Tuttle et ai., 1991; Hunt, 1994a) for the origin of bipedalism (and/or the divergence of
apes and humans) are particularly appealing. Jolly's seed eating hypothesis (1970) main-
tained that the prevalence of grass seeds in dry habitats selected for a suite of human char-
acteristics. The small diameter and even distribution of grass seeds, he argued, demanded
sustained bipedalism during postural collection. Shuffling bipedalism was seen as a high-

397
398 K. D. Hunt

Table 1. Activity budgets for Gombe, Mahale, and Kibale chimpanzees

Gombe Gombe* Mahale Kibale (Ngogo)
Activity (Hunt, 1989) (Wrangham, 1977) (Hunt, 1989) (Ghiglieri, 1984)
Food 49.0 55.7 46.9 57.3
Rest, socialize 36.2 30.3 40.3 31.5
Travel 14.8 13.9 12.8 11.1
·Males only

profile variation on gelada scooting, advantageous for moving between food resources
while sustaining a higher reach. The manipulation of small-diameter seeds selected for
considerable manual dexterity, preadapting hominids for tool use.
Savanna baboons were observed to be bipedal when feeding on small food items,
though grass seeds themselves were not common (Rose, 1976, 1984, 1991). Wrangham
(1980) noted that gathering fruit from bushes, in particular, elicited bipedalism among
chimpanzees. He offered a locomotor corollary to Rose's small-object feeding hypothesis:
bipedal locomotion saves energy by eliminating the action of raising the upper body to
feed bipedally after walking between resources quadrupedally. Jolly agreed, suggesting
that specialization on small-diameter fruits, rather than seeds, might have been the critical
selective pressure that resulted in the evolution of bipedalism (Jolly and Plog, 1987). The
small-object postural feeding hypothesis (Rose, 1976, et seq.) derived from the research of
Jolly, Rose and Wrangham postulates that bipedalism evolved as a terrestrial feeding pos-
ture advantageous for reaching into trees, and that bipedal locomotion evolved to reduce
energy costs when traveling between densely packed feeding sites.
Recently, Hunt (1994a, 1996) added an arboreal component to the small-object pos-
tural feeding hypothesis, noting that small-object feeding elicits bipedalism in trees as
well as on the ground. The small diameter of supports in the small trees in which smaIl-di-
ameter fruits tend to be found appears to encourage bipedalism arboreally as much as fruit
diameter. The data from which these conclusions were made are examined below in more
detail than is available in Hunt (1992a, 1996).

2. METHODS
I observed chimpanzees for 571 hours at the Mahale Mountains National Park and
for 130 hours at the Gombe Stream National Park (Hunt, 1992a). Sixteen thousand three
hundred and three instantaneous, 2-minute focal observations were made on 26 well-ha-
bituated prime adults spanning all social ranks. Twenty-five positional behavior variables
were monitored, including positional mode, behavioral context of the mode, and a number
of feeding parameters. Two thousand eighty seven observations were made on Gombe ba-
boons over 83 hours using identical methods.
I identified one of 65 locomotor or postural modes in a target animal at each 2-minute
point. I recorded positional behavior, location in canopy, height, size of supports, proximity
to others, and food type. Because the chimpanzees were well-habituated, I was able to make
observations during all hours of the day, in all contexts, and with no decrease in quality of
observation when individuals were on the ground.
Bipedalism was defined as posture or locomotion in which it was judged that more
than half of the body weight was borne by the hind limbs in compression. If neither fore-
limb nor any other part of the body other than the hind feet touched a support, I labeled
Ecological Morphology of Australopithecus afarensis 399

the mode "unassisted bipedalism." When a forelimb or other body part supported some
but less than half of the body weight, I called the mode "assisted bipedalism." I put behav-
iors for which I judged that more than half of the weight was borne by an abducted fore-
limb in an "arm-hanging" mode; arm-hanging is not part of this analysis.

3. RESULTS

3.1. Contexts of Chimpanzee Bipedalism

Bipedalism was not a common chimpanzee behavior. Ninety-seven instances of
bipedalism among 21 individuals were sampled in 700 hours of observation. No two ob-
servations were made in consecutive time-points. Ad libitum observations (i.e., observa-
tions on non-target individuals, observations between time-points) were used as
supplemental evidence, but tables and figures include only systematic observations (i.e.,
focal individual, time-point samples). By far, the most common context of bipedalism was
feeding (Figure 1).
"Move in patch" typically occurred when an individual moved from one harvesting
perch in a fruit tree to another within the same tree. Not uncommonly, a target individual
did not even cease chewing during the move. If this behavior is pooled with "feed," which
seems reasonable, a full 80% of chimpanzee bipedalism was in the context of feeding. The
feeding function of bipedalism is dramatically illustrated by comparing bipedal contexts to
the daily activity budget. Whereas half of the typical chimpanzee day is occupied with
feeding (Table 1), over 80% of bipedalism was in a feeding context. Bipedalism was
clearly a feeding mode both on the ground and in the trees (Figure 2). Rose found that
bipedalism among baboons had the same feeding function (Rose, 1976).

~~-------------------------------------------,

70
60
50

% 40
30

20
10

o
feed move In big play scan respond make bed respond display hold copulate unknown
patch to threat to caD intant

Figure 1. Contexts in which bipedalism was observed in Tanzanian (Gombe and Mahale) chimpanzees. Of97 ob-
servations of bipedalism, feeding and moving within a feeding patch constituted 80.4% of all bipedal behavior.
400 K. D. Hunt

90,------------------------------------,
80+---------------------
70
60
50
%
40
30
20
10

o
terrestrial arboreal
I_ feed 0 other 1
Figure 2. Percentage of bipedalism engaged in during feeding (dark columns) versus other activities (light col-
umns) on the ground (left) and in trees (right). Note that bipedalism functioned as a feeding posture whether chim-
panzees were on the ground or in the trees (N=97).

Perhaps the greatest surprise is that after feeding, the next two most common con-
texts of bipedalism are "beg" and "play" (Figure 1), two behaviors that have not been im-
plicated as selective pressures for the evolution of bipedality (Rose, 1991). Chimpanzees
stood bipedally occasionally to get a better view of their environment ("scan"); such scan-
ning was sampled twice, totaling 4.1 % of bipedal behavior. "Respond to threat" (4.1 %)
describes behavior in which an individual stood bipedally, often with one or both fore-
limbs lightly contacting a tree, to be ready to climb the trunk quickly if a displaying male
or an altercation drifted in their direction. Chimpanzees were not uncommonly bipedal to
free their hands to manipulate foliage while making a nest, or night bed (4.1 %). Chimpan-
zees sometimes stood bipedally to listen to distant vocalizations (not sampled here, but
observed ad libitum), and while responding to calls with their own vocalizations (2%).
Display, though impressive when it occurs, is actually a rare behavior among chimpanzees
(2%). Observations reported here do not include those made in feeding camp, where domi-
nance displays in contests for provisioned food, or in response to unusually large social
groupings, have allowed the frequent filming of bipedalism. Chimpanzees often support
clinging newborns with a forelimb when walking, but utilizing both forelimbs for support
and thereby requiring bipedalism ("hold infant") was rare (2%). Bipedal infant-carrying is
probably so rare because infants become capable of gripping mother's fur within days of
birth, after which little support is needed. I suspect, based on ad hoc observations, that
mothers walk little the first few days after giving birth, reducing the need for helping a
newborn to cling during its helpless phase. "Copulate" and "unknown" are rare contexts
for chimpanzee bipedalism as well (2% each).

3.2. Postural versus Locomotor Bipedalism

Although postural behavior is more common than locomotor behavior in chimpan-
zees-85% of chimpanzee positional behavior is postural (Table 2)--this balance does not
hold for bipedalism, where the postural component predominates. Postural bipedalism
Ecological Morphology of Australopithecus afarensis 401

Table 2. Proportion of locomotion vs. posture

among chimpanzees
Population N Posture Locomotion
Mahale 11,471 84.8 15.2
Gombe 2,910 83.3 16.7

made up 95% of bipedal observations (Figure 3). This proportion is identical to that re-
ported by Rose (1976) for baboons.

3.3. Is Bipedalism a Terrestrial Behavior?

Bipedalism is sometimes assumed to be a behavior naturally elicited by terrestriality.
Chimpanzee behavior refutes this assumption. Chimpanzees were actually more likely to
be bipedal when arboreal than when terrestrial; 61 % of all bipedal behavior was arboreal
(Figure 4). Nor is bipedalism limited to large branches in the tree core. Nearly 30% of
bipedalism was observed among the terminal branches (Figure 4).
Although terrestriality did not elicit bipedalism, some might expect locomotor
bipedalism to have been more common terrestrially than arboreally. That was not the
case. Over 80% of locomotor bipedalism was arboreal. Of the 6 bipedal locomotor bouts
in the sample, 5 were arboreal. Chimpanzees were moving arboreally within a feeding
patch in 4 of 6 observations (Figure 1). The fifth observation was making a night nest.
The sole terrestrial observation of bipedal locomotion was in the context of "playing"
(Figure 1).

3.4. Forelimb-Suspension and Bipedalism Are Linked

Bipedalism was stabilized by an arm-hanging-like (=unimanual forelimb-suspen-
sion) support from a forelimb nearly 60% of the time (Figure 5). It appears that the inher-
ent instability of bipedalism (Kummer, 1991) is amplified arboreally by the small, flexible
nature of the support. While bipedal posture has two advantages-it increases the height

100~-------------------------,
90
80
70
60
% 50
40
30
Figure 3. Percent of postural bipedalism
compared to locomotor bipedalism. Al- 20
though locomotion typically makes up 15% 10
of all positional behavior, of 97 observa-
tions of bipedalism only 6 (6%) were loco- o
motor. posture locomotion
 402 K. D. Hunt

40
35
30 Figure 4. Percent of total bipedal bouts
observed on the ground versus in the tree
25
core (central part of tree), versus in the
%20 tenninal branches (within I m of tree
edge). Note that only 38% of bipedality
15 was terrestrial. It was arboreality, not ter-
restriality, that most commonly elicited
10 bipedalism among chimpanzees. Notably,
bipedalism was nearly as common among
5 tiny twigs in the terminal branches as it
was in the tree core or on the ground. The
0 common assumption that a stable sub-
terrestrial tree core terminal branches strate elicits bipedalism is clearly wrong.

of the reach and does not require a "gear-change" expense> (sensu Wrangham, 1980) when
moving-it requires additional stabilization arboreally to be an effective harvesting pos-
ture. Assisted bipedalism involved stabilizing the body with the forelimb in a fully ab-
ducted forelimb-suspension gestalt.
The link between forelimb-suspension and bipedalism was most powerful among the
terminal branches of trees. A forelimb oriented as in arm-hanging (though bearing less
weight) stabilized bipedal posture in 93% of observations among terminal branches
(N=27). Bipedalism occurred on significantly smaller branches than did other postures
(12.2 cm vs. 15.0 cm, Mann-Whitney U test, U=123,620, P=O.OOOl, NI.2=64,5375), prob-
ably because small trees offer few large branches stable enough for sitting or unassisted
bipedal standing. In the tree core, or central part of the tree, 52% (N=23) of bipedalism
was assisted by a forelimb. Arboreal bipedal locomotion was relatively rare (4.1 % of all
bipedal episodes), and consisted exclusively of short-stride-Iength shuffling.
Terrestrial bipedalism was less often assisted, presumably because the firm substrate
did not require an extra stabilizing contact point. Terrestrial postural bipedalism was unas-
sisted nearly 2/3 of the time, whereas bipedalism was assisted 3/4 of the time arboreally
(Figure 6). During terrestrial gathering both hands were often used to harvest fruits. Not
infrequently, one hand was used to pull down and hold an otherwise inaccessible fruit-
bearing limb, so that the posture became a terrestrial arm-hanging-bipedalism. A few ter-
restrial bipedal bouts were locomotor bipedalism in the context of moving between
feeding sites at the same tree (4.1%). That is, short-distance within-site shuffling, rather
than long distance travel, was the most common context for locomotor bipedalism.
Fruit was the most common food resource harvested by bipedal chimpanzees (Fig-
ure 7). Among other foods harvested bipedally, only ants might be called common. Ma-
nipulation of an ant-dipping tool often required both hands, and therefore sites from which
chimpanzees harvested ants tended to be those in which they could work while bipedal.

* That is, changing postural or locomotor mode entails some expense. In this case, the cost of raising or lowering
the torso is an energetic expense that must be balanced against the increased energy expenditure required to loco-
mote using an inefficient mode (bipedalism). If the difference between the energy expended to locomote
bipedally versus quadrupedally is less than the energy expended to lower the torso to allow quadrupedal locomo-
tion, and then raise the torso to a bipedal posture at a new feeding site, there is selection for the individual to lo-
comote bipedally.
Ecological Morphology of Australopithecus a/arensis 403

60

50

Figure 5. Frequency of assisted versus 40

unassisted bipedalism. Nearly 60% of
bipedalism in chimpanzees was assisted
by a forelimb, most often in an forelimb- % 30
suspension-like manner. These figures do
not include observations in which the 20
forelimb was judged to be bearing more
than half the body weight, a posture la-
beled arm-hanging (=forelimb-suspen- 10
sion). Supported forelimb-suspension
made up another 3.6% of all positional be-
havior (N= II ,393).
o
assiSlad unasslSlad

3.5. Characteristics of Foods Harvested BipedaUy

Bipedalism occurred both terrestrially and arboreally when chimpanzees fed from
Garcinia huillensis, Harungana madagascarensis, Monanthotaxis poggei and Grewia sp.
Together these four species of trees constituted 27% of all bipedal feeding episodes and
48% of the bipedal episodes in which the plant material being eaten could be identified. At
Mahale, Garcinia huillensis rarely reaches 15 m in height. It lives in forest edge habitats

50,--------------------------------,
45
40
35
30
%25
20
15
10
5
o
,r-.-
arboreal
- -O
assIIIad - Ull8Sliltld
----"
terresIriaI

Figure 6. Frequency of assisted (=stabilized by a forelimb) and unassisted (support by hind limbs only) bipedal-
ism in trees versus on the ground. Bipedalism is mostly unassisted on the ground, and mostly assisted in the trees.
Unassisted bipedalism should be preferred because it allows a higher feeding rate, since both hands can be used
for gathering (Rose. 1976). The higher frequency of assisted bipedalism among terminal branches is presumed to
be due to the instability of bipedalism on small-diameter branches.
 404 K.D. Hunt

60.0

SO.O

40.0

% 30.0

20.0

10.0

0.0
fruit ants leaf minerai not seen meat stems shoots blossoms

Figure 7. Food items harvested bipedally by chimpanzees. Note that aside from fruits, only ants are commonly
harvested bipedally.

and lake-side environments. Its fruits are approximately 2 cm in diameter. Harungana

madagascarensis is a many-branched understory tree ranging from several meters to 12 m
in height. It is widely distributed in East Africa, typically in forest edge environments
(Hamilton, 1982). Its fruits are 3-5 mm in diameter. It is typically found in monospecific
stands (pers. obs.). Monanthotaxis poggei is a 1-2 m tall shrubby plant found in forest
edge environments. It often occurs in dense stands. The fruits are approximately 1 cm in
diameter. Grewia is a 5 m open-forest tree with I cm fruits.
All of these trees are short, forest-edge or open-forest trees. The fruits of each spe-
cies are small (-2 cm, 0.4 cm, 1 cm and 1 cm respectively). The trees occur more often in
monospecific stands than trees chimpanzees harvested fruits from in the more closed-for-
est part of their range. Three of the four (not Garcinia) are dry, fibrous, difficult to masti-
cate fruits.
Although I did not sample bipedal gathering both arboreally and terrestrially in any
other tree, other sniall trees with small fruits elicited bipedalism either arboreally or terre-
strially much more commonly than did large trees. Bipedal food collecting was signifi-
cantly more common among small (mature height of::;; 15 m) trees (Figure 8) with small
fruits (44 vs. 8, Fisher's Exact test, P<O.OOI, X2=27.8, df=l), suggesting that fruit diameter
and tree height are the critical factors eliciting bipedalism. It is difficult to distinguish be-
tween the effects of small trees and small fruits, since all but one small tree also had small
(::;; 2 cm) fruit. When plant-foods gathered during bipedal bouts were identified, 28 of 33
fruits (85%) were::;; 2 cm in diameter (Figure 9).

3.6. Why Do Chimpanzees Climb Trees?

Chimpanzee bipedalism and other behaviors might be argued to be poor predictors
of behavior among the more terrestrial hominids. Chimpanzees may forage in trees be-
Ecological Morphology of Australopithecus afarensis 405

~.----------------------------.

30

25

20
observations
15

10

Figure 8. Although chimpanzees fed in 5

small trees (N= 1439) and large trees
(N=1536) almost equally, most observa- o
tiops of bipedalism were in small trees. small tree large tree

cause they prefer arboreality, whereas australopithecines might have preferred terrestrial-
ity. Chimpanzee foraging data suggest that chimpanzees go into trees not because they
prefer them, but because their most preferred foods are found there. Early hominids may
have been similarly constrained.
Chimpanzees spent 70% of the time they were in trees feeding. Furthermore, chim-
panzees entered terminal branches almost exclusively to feed; feeding constituted nearly
90% of terminal branch activity (Figure 10).
Because different chimpanzees spend differing amounts of time in trees, we may
draw some generalizations about arboreality that can provide a model for canopy use in
protohominids. Social rank, body size and sex appear to determine arboreality in chimpan-
zees (Hunt, 1992b, 1994b). Larger chimpanzees spent less time in trees than did smaller
chimpanzees. In a multiple regression that included social rank, body size, and canopy
height, with social rank factored out, large males positioned themselves lower in the can-
opy (R=0.30, P<0.0002, N=6,600).
When like-rank males of different body sizes are divided into 2 classes, large males
fed from smaller tree species than did small males (Figure 11; large and small individuals

90r-------------------------~
80
70
60
50
% 40
30
20
Figure 9. Frequency of bipedalism com- 10
pared when eating small versus medium-
sized and large fruits (N=33). Small fruits o
are associated with bipedalism. small fruits medium, large fruits
 406 K. D. Hunt

~.----------------------------------,

80
70
60
50
%
40
30
20
10
o
tenninal branches tree core terrestrial
I_ feed (n:5112) 0 other (n=n46) 1

Figure 10. Frequency of feeding in different canopy locations. Chimpanzees do most of their feeding in trees, and
they enter terminal branch sites almost exclusively to feed. Apparently there is little food on the ground that chim-
panzees prefer.

40,-------

35 +-----1
30 +-----1

25 +-----1

% 20 +---1

15

10

o
large trees small trees ground fruit
I_ large males (n:589) 0 small males (11=338) 1

Figure 11. Comparison of feeding site preferences for large and small males, matched for social rank. Large males
preferentially fed in small trees and on the ground. Large males minimized climbing, hypothetically because feed-
ing sites lower in the canopy were more valuable to them.
Ecological Morphology of Australopithecus afarensis 407

80~------------------------------,

70+-----------------------
60+----------------------
50+----------------------
% 40 +-------------

30+-------------
20+-------
10+------
o
terminal branches tree core terrestrial
• large males (n::3181) 0 small males (n=1562)

Figure 12. Comparison of canopy location between large and small males. matched for social rank. Large males
spent more time on the ground.

were matched for social rank). Matched for social rank, large males spent more time on
the ground than did small males (Figure 12). Matchedfor social rank, large males climbed
less often, 0.8% of all behavior versus 3.2% for small males (X 2=6.65, df= 1, P=O.O 1,
N=494, 476). The likeliest explanation for this pattern is that arboreal positional behavior
is less demanding for smaller males.
Because these observations are limited to males, sex and infant care did not contrib-
ute to the differences. Individual males were matched for social rank, which means that
body size alone caused these differences in canopy use. Comparisons of male and female
chimpanzees show the same results, though it is not possible to adjust for infant-care and
social rank differences. Females spend more time in the trees (Figure 13). It is likely that
body size contributes to female arboreality by making arboreality less energetically de-
manding. Arboreality may also be predator-avoidance strategy for females, though the evi-
dence from males suggests not.

4. AUSTRALOPITHECINE ECOLOGY: CRANIODENTAL

EVIDENCE

Craniofacial shape, robusticity of the masticatory apparatus, cusp morphology, tooth

size, enamel thickness, and incisor size are among the evidence that has been brought to
bear on reconstructing australopithecine diets, most notably in a synthesis by Kay (1985).
Although the correlation between each of these variables and diet is low, if considered to-
gether they provide a detailed model of australopithecine diet.
Molar dental microwear has not been examined in Australopithecus a{arensis, leav-
ing A. africanus microwear as an admittedly unsatisfactory stand-in (Walker, 1981; Tea-
ford and Walker, 1984; Grine and Kay, 1988; Kay and Grine, 1988; Teaford, 1994). The
408 K. D. Hunt

50,-------------------------------,

~+-------------------------

40+-------------~

35+-------
30+-------
% 25 +-------
20 +-----1
15
10
5
o
terminal branches tree core terrestrial
,_ male (n=1896) D female (n=l406) ,

Figure 13. Comparison of canopy location in males and females. Females are more arboreal and spend more time
among the terminal branches.

similarity of dentition of the two (Tobias, 1980), however, makes this unsatisfactory proxy
more acceptable. Walker (1981) interpreted the Pan- or Mandrillus-like microwear of both
A. africanus and A. robustus as meaning they were frugivores. Kay and Grine (1988)
found that in both microwear feature width and relative frequency of pits versus scratches
suggested the same thing. Molar microwear feature width falls between Alouatta palliata
and Cebus nigrivittatus (Kay and Grine, 1988). Pit:scratch frequency comparisons place
them between orangutans and chimpanzees (Kay and Grine, 1988). With the exception of
howlers, these primates eat fruit at least 57% of the time (Table 3). Howlers concentrate
on leaves. All four species include at least 10% leaves in their diet. With the exception of
pith/herbs for chimpanzees, other dietary items are uncommon.
Pit frequencies for Australopithecus fall intermediate between orangutans and chim-
panzees suggesting a low frequency of seed-eating and oral nut-cracking (Kay and Grine,
1988; Table 3). Orangutans open nuts orally, though it is not clear how common the be-

Table 3. Primate diets

Species Insects Leaf Meat Fruit Piths/herbs Flowers Bark Other
Pan troglodytes' 5.6 10.3 1.0 57.0 22.6 0.7 0.0
Alouatta pallia/a] 0.0 64.0 0.0 12.0 0.0 18.0 0.0 0.0
Pongo pygmaeus 3 1.0 26.0 0.0 58.0 13.0 2.0
Cebus spp.' 20.0 15.0 0.0 65.0 0.0 0.0 0.0
Mean 7.0 29.0 0.3 48.0 6.0 5.0 3.0 1.0
(1989); feeding time, based on 3,891 feeding records of Pall tlVglodytes schll'eill/imhii at Mahale.
I Hunt
"Glander (1978); feeding time.
'Rodman (1984); feeding time, Kutai, Kalimantan 40,022 min. observations.
4Hladik and Hladik (1969); dry weight of stomach contents; records for Cebus Iligrivittallis are 100% fruit
(Fooden, 1964).
Ecological Morphology of Australopithecus afarensis 409

havior is; chimpanzees rarely crack nuts orally. Other evidence suggests that nut-cracking
was not common in australopithecines. Peters (1987) argued that a small modelled gape in
australopithecines and a relative lack of crenulations suggest low levels of nut-cracking.
Although some nut-cracking cannot be ruled out, a small gape and unpitted teeth suggest
that A. africanus and A. robustus were not specialized for this dietary item. A high fre-
quency of pitting is also found in animals that eat foods that have adhering grit (Teaford
and Walker, 1984; Teaford, 1994). The low frequency of pitting in australopithecine mo-
lars makes the utilization of roots and tubers and other below-ground resources with ad-
hering grit unlikely.
Macrowear adds an exclamation point to indications of some leaf-eating in australo-
pithecines. In A. afarensis the incisors are unevenly worn so that the occlusal surface has
an undulating appearance in frontal view (Puech et ai., 1984). Such wear results from
stripping, such as pulling a twig through the mouth to remove leaves.
An objective but somewhat crude estimate of the australopithecine diet might be
gained by simply averaging the frequency of dietary items in primates that have mi-
crowear most similar to australopithecines (Table 3). Frugivory is suggested, with leaves
as an important secondary component of the diet.
Dental microwear of A. africanus narrows the possible australopithecine diet some-
what from what is indicated by dental morphology.t Among primates, summed (i.e., II and
12) incisor size is correlated with food object size and are larger in frugivores than gram-
nivores or folivores (Hylander, 1975; Kay and Hylander, 1978). Primates that eat small
fruits have small incisors. Fruits require incisal penetration of an often fibrous and/or
abrasive husk; leaves do not. The extra wear on incisors cause by fruit processing requires
larger incisors to maintain a functioning incisal edge in older animals. A regression of the
natural logarithm of maxillary incisor width and log body weight for 57 primates yields a
regression line with good separation between principally frugivorous cercopithecines and
principally folivorous colobines (Hylander, 1975; Kay and Hylander, 1978). Australo-
pithecines fall below both the regression line for apes alone (Kay 1985) and for pooled
primates (Kay and Hylander, 1978; using McHenry, 1992 body weights).
The two earliest hominid species are not represented by lateral incisors, and cannot
be directly compared to Kay and Hylander's (1978) results. Both, however, have liS quite
near the mean of A. afarensis (10.63 mm, White et aI., 1981). The I' of Ardipithecus
ramidus is 10.0 mm (White et aI., 1994), and the II of A. anamensis is 10.5 mm (Leakey et
ai., 1995). The small early australopithecine incisor dimensions as expressed in their rela-
tion to the primate and ape regression line suggest that they consumed smaller food items
than extant primate frugivores, including orangutans and chimpanzees.
Australopithecus afarensis molar areas fall above the regression line of cheek tooth
area compared to body mass (log-log plot) for hominoids (Kay, 1985). The residual is
nearly equidistant between orangutans and A. africanus (McHenry, 1984; Kay, 1985), and
well above the values for chimpanzees. Among extant primates, species that open nuts and
seeds and/or eat hard-to-masticate fibrous fruits have large molars (orangutans and capu-
chins, Kay, 1981, 1985).
Australopithecines have thick enamel (Kay, 1983), even thicker than the thickest-
enamelled of extant primates (i.e., mangabeys, capuchins, and orangutans). Thick enamel
is correlated with consumption of hard and brittle foods (Teaford, 1985) such as seeds,

t Newly discovered fossils (Ardipithecus ramidus, White et aI., 1995; A. anamensis, Leakey et aI., 1995) are yet to
be thoroughly evaluated, but where possible they are discussed. Otherwise, discussion centers on the more abun-
dant and better studied fossi Is of A. afarensis and A. africanus.
410 K. D. Hunt

nuts or hard-husked fruits. Thick enamel also occurs among mammals that consume foods
that are abrasive, or foods that have adhering grit, as subterranean items do. Chimpanzees
and gorillas have thin enamel with well-developed shearing blades (Kay, 1981), a mor-
phology correlated with folivory and believed to function to finely comminute leaves into
a fine, more digestible gruel. Because australopithecines have thicker enamel and shorter
shearing blades than do chimpanzees and gorillas (Kay 1983), their diet must have con-
tained considerably fewer leaves and piths than does the diet of African apes.
Compared to apes, A. afarensis has taller, more squared-off, more robust zygomae
(White et aI., 1981). The diameter of the mandibular corpus is also greater. A forward shift
of the zygomae and thickening of the mandibular corpus mean that australopithecines could
produce greater occlusal forces than those of extant apes, or could sustain high masticatory
pressures for longer (Jolly, 1970). Feeding on hard-to-masticate items is suggested. Reduced
prognathism suggests a reduced gape for australopithecines compared to apes.
If we consider all of this evidence at once, a rather precise modelling of the australo-
pithecine diet is possible. Small incisors and the small gape, compared to living apes, sug-
gest that australopithecines consumed smaller food items than do African apes or
orangutans. Molar microwear suggests that australopithecines rarely engaged in nut-crack-
ing, and that they did not consume below-ground foods. Their microwear suggests a
frugivorous diet supplemented with piths or leaves. Their low-cusped molars with short
shearing blades, however, offer contrary evidences, suggesting a lower proportion of piths
and leaves. Large, thickly enameled molars suggests a specialization on fibrous fruits. A
robust face suggests hard to masticate food items, again implying a specialization on fi-
brous fruits. Australopithecine craniodental fossils unambiguously suggest a diet high in
small-diameter, fibrous fruits. The proportion ofleaves in the diet is less clearly indicated,
but it seems likely they were an important dietary item, though one clearly less important
than fruits.

5. AUSTRALOPITHECINE POSITIONAL BEHAVIOR

5.1. Postcranial Evidence for Arm-Hanging.

The australopithecine torso is broad, shallow and cone-shaped (Schmid, 1983,

1991), the glenoid fossa cranially oriented (Robinson, 1972; Stem and Susman, 1983), the
cross sectional area of the vertebral column quite small (Robinson, 1972; Jungers, 1988),
and the brachial index chimpanzee-like (Kimbel et aI., 1994). These features are adapta-
tions to arm-hanging (Hunt, 1991a).
The australopithecine wrist is mobile (McHenry, 1991a), an adaptation that reduces
stress on the wrist when suspending from vertical supports (Hunt, 1991a). Thin twigs bend
to vertical when chimpanzees hang from them, necessitating a mobile wrist to maintain
grip.

5.2. Postcranial Evidence for Vertical Climbing

The bicipital groove in which the biceps tendon rests is large (Robinson, 1972;
Lovejoy et aI., 1982), implying a large biceps muscle. The supracondylar ridge (proximal
attachment of the extensor carpi radialis and brachioradialis muscles) is huge (White et
aI., 1993). The large muscles implied by these skeletal features suggest an ability to per-
Ecological Morphology of Australopithecus afarensis 411

form a powerful pull-up action, seen in extant hominoids most often during vertical climb-
ing (Hunt, 1991 b).
The convex joint surface of the A. afarensis medial cuneiform indicates a rudimen-
tary ablity to abduct the first toe (Stern and Susman, 1983; Deloison, 1991; pers. obs.)
contra Latimer and colleagues (Latimer et ai., 1982). Australopithecus afarensis also has
long, curved toes (Tuttle, 1981; Stern and Susman, 1983) and an antero-posteriorly short,
rounded lateral femoral condyle (in the smaller specimens; Tardieu, 1983; McHenry, 1986,
1991 a). A strongly developed fibular groove for the tendon of the peroneus longus muscle
suggests ape-like great-toe flexion (Tuttle, 1981; Deloison, 1991), as might be used to grip
branches when standing arboreally or climbing. Alternatively, a robust peroneus longus
muscle may stabilize a more mobile foot, or support a ligamentously poorly supported
arch; if so, bipedalism would be that much less energetically efficient, since muscular sup-
port would be necessary for toe-off, rather than a non-energy-consuming ligamentous sup-
port. A plantar set, or at least greater mobility (Latimer and Lovejoy, 1990), of the ankle
allows full plantarflexion of the foot. Gombe and Mahale chimpanzees plantarflexed their
feet when they used their toes grip a branch to support body weight with the hind limb in
tension. Curved pedal phalanges (Tuttle, 1981) and a third pedal digit longer than the first
or second (Stern and Susman, 1983) are gripping adaptations. Such pedal gripping, espe-
cially with the lateral 4 toes only, is used by Tanzanian chimpanzees during arm-hanging
to increase stability among slender terminal branches. Climbing adaptations in the hin-
dlimb are not limitied to the foot. A long moment arm for the hamstrings (Stern and Sus-
man, 1983) increases the power of hip extension, implying a better climbing adaptation
than in modern humans.

5.3. Postcranial Evidence for Both Arm-Hanging and Vertical Climbing

The deltoid tuberosity is large and laterally flaring, suggesting a large deltoid (White
et ai., 1993; Kimbel et ai., 1994), and the coracoid process (proximal attachment of biceps
brachii and distal attachment of pectoralis minor muscle) is large (Robinson, 1972). Chim-
panzees use the deltoid to raise the arm during vertical climbing and when reaching out to
pluck fruits when arm-hanging.
Australopithecine fingers are curved and have large flexor sheath ridges (Stern and
Susman, 1983). The fingers are more human-like than ape-like in length; but the thumb is
short (=chimpanzee-like) with a chimpanzee-like articulation (Tuttle, 1981; McHenry,
1991a). Arm and leg length proportions are intermediate between those of modern humans
and chimpanzees, even when the diminutive stature of the fossils is considered (Jungers,
1982, 1991). These features are adaptations to powerful gripping of cylindrical surfaces,
such as occurs in chimpanzees only during arm-hanging and vertical climbing (Hunt,
1992a).

5.4. Postcranial Evidence for Bipedalism

In both general morphology and detail, the pelvis and the hind limb morphology of
A. afarensis and later hominids indicate bipedalism (Johanson and Edey, 1981; Lovejoy,
1988; Latimer and Lovejoy, 1989, 1990; McHenry, 1991a). The lumbar vertebrae are lor-
dotic (Abitbol, 1987), the sacral alae are expanded, and the pelvis has a very human ge-
stalt (Lovejoy, 1988; McHenry, 1991a). The femur has a deep patellar groove, and at least
some specimens have an elliptical lateral condyle. The calcaneus is essentially modern
 412 K. D. Hunt

(Latimer and Lovejoy, 1989). The great toe is robust and the foot has well developed
transverse and longitudinal arches (Latimer and Lovejoy, 1990; Langdon et aI., 1991).

5.5. Morphology at Odds with Refined Locomotor Bipedalism

Other features in A. afarensis suggest a bipedalism that is not as refined as that of
modem humans. Although the os coxae are human-like in appearance, A. afarensis has
smaller sacro-iliac ligaments than modem humans (Stem and Susman, 1983). The width
of the AL 288-1 pelvis is proportionally greater than the femoral neck length, suggesting
that when compared to modem humans, A. afarensis had a greater joint reaction force at
the hip and a lower mechanical advantage for muscles that prevent the hip from collapsing
when one foot is off the ground (Jungers, 1991). A relatively small acetabulum/femoral
head in australopithecines compounds the stresses caused by wide hips, creating even
more stress in the hip joint.
Wide hips also cause the moment arm of the body weight of A. afarensis to be in-
creased over that of modem humans, increasing the stress on the diaphyseal/femoral neck
junction when bearing weight (Hunt, 1994a) and decreasing energetic efficiency during
walking by requiring greater muscular activity (Jungers, 1991). The extraordinarily wide
hips of AL 288 (Berge and Kazmierczak, 1986; Rak, 1991) and STS 14 (Robinson, 1972)
are not obstetric adaptations (Rosenberg and Trevathan, 1995; Stoller, 1995). The broad
hips of these hom in ids are due mostly to an unusually broad pelvic inlet. The pelvic index
averaged 77.6 for modem human females (Tague and Lovejoy, 1986), but is 57.6 in AL
288-1. Although a large biacetabular breadth in modem humans is a necessary adaptation
for giving birth to large-headed offspring, cephalopelvic reconstruction of AL 288-1
shows a considerable gap between the fetal head and pelvic inlet walls opposite the
acetabula (Figure 14). In other words, the pelvis of australopithecines is much broader
than could possibly be necessary for parturition.
Other features suggest reduced locomotor competence as well. Australopithecus
afarensis has quite short hind limbs for its weight and height, suggesting greater energy
expenditure per unit distance traveled (Jungers, 1991). The lumbar vertebrae and lum-
bosacral articular surface of other australopithecines are small, whether in proportion to

Figure 14. Reconstructed pelvis of A. afarensis (redrawn from Lovejoy and Tague, 1986 and from Lovejoy, 1988)
with a hypothetical fetal head superimposed. The pelvis is much broader than necessary for parturition. Broad hips
and short hind limbs are hypothesized to function to lower the center of gravity. A low center of gravity confers
greater balance, which is particularly valuable when standing or moving on small and therefore unstable arboreal
supports.
Ecological Morphology of Australopithecus afarensis 413

body weight, hip width or nearly any other measure that has been attempted (Robinson,
1972; Jungers, 1988, 1991; McHenry, 1991 a; Rak, 1991).

6. DISCUSSION

We can bring four lines of evidence to bear on the question of australopithecine

bipedalism: (1) australopithecine craniodental data; (2) australopithecine postcranial anat-
omy; (3) the contexts that elicit bipedalism in living animals (chimpanzees in this case);
and, (4) differences between large and small chimpanzees.
Molar feature width and low pitting argue strongly against nut-cracking or speciali-
zation on below-ground resources for australopithecines. Fruit and leaves are most consis-
tent with their microwear, but these data conflict with dental morphology. Thick enamel,
low molar shearing quotients, large molar areas and robust craniofacial morphology indi-
cate a diet high in hard items, but not particularly high in leaves or piths. Because below-
ground resources are precluded by the microwear data, and because nut-cracking is not
strongly indicated, hard-husked and/or fibrous fruits that require extensive mastication fit
best with the microwear data. Small incisor breadths indicate an australopithecine spe-
cialization on small diameter food items (Rose, 1976).
If australopithecines were frugivores, as these craniodental data suggest, they could
forsake arboreality only if they lived in a habitat where most fruits were found in bushes
shorter than about 2 m. Is there evidence that australopithecines had forsaken arboreality?
In a word, no.
Above the waist, australopithecines are intermediate in morphology between chim-
panzees and humans. Chimpanzees are adapted to unimanual suspension (arm-hanging)
and vertical climbing. The positional repertoire of A. afarensis might then be recon-
structed as containing about half as much arboreal activity as chimpanzees. Curved, pow-
erful fingers accord well with an arm-hanging adaptation. Short fingers suggest that the
supports from which A. afarensis suspended themselves were of smaller diameter than
those commonly used by extant chimpanzees. Small trees with small branches are indi-
cated as the most common feeding sites for australopithecines.
Below the waist, australopithecines had short limbs, long, curved toes, a more grip-
ping great toe than that of modern humans, and extraordinarily wide hips. Short lower
limbs bring the most massive part of the body, the torso, closer to the substrate, thereby
lowering the center of gravity. Wide hips, especially in combination with the cone-shaped
torso, lowers the center of gravity by increasing the mass of the lower portion of the torso,
essentially allowing internal organs to rest lower in the torso. Both these features increase
arboreal competence by making balancing on branches easier. Among apes, gripping toes
are frequently used to stabilize the body. The strong toes of A. afarensis may have been
used to grip small twigs during arboreal bipedal gathering.
Smaller joint surface areas and a less robustly ligamented pelvis than in modern hu-
mans indicate that A. afarensis was more prone to fatigue or injury during powerful and
sustained bipedal locomotion. Bipedal carrying, with its imposition of greater stress on the
musculoskeleton, must have been less common than it is in modern humans (Jungers,
1988, 1991; Hunt, 1994a). Australopithecus afarensis probably engaged in less sustained
bipedal walking and less bipedal carrying than modern humans do.
Bipedalism is neither a locomotor behavior nor strictly a terrestrial behavior among
chimpanzees; over 90% of Mahale chimpanzee bipedalism was postural, and more than
half of all their bipedal episodes were arboreal. Among Mahale chimpanzees, bipedalism
414 K. D. Hunt

was overwhelmingly a feeding posture; 80% of all bipedalism occurred during feeding.
Other behaviors that have been hypothesized to have exerted selective pressure on pre-
bipedal protohominids to become bipedal--searching for predators, threatening, hunt-
ing--each constituted <5% of bipedal episodes among Mahale chimpanzees.
Chimpanzees fed bipedally from short trees with small fruits. Bipedal standing allowed
individuals to reach higher in trees when feeding terrestrially, to bring more fruits within their
reach. It also allows faster gathering (Jolly, 1970; Jolly and Plog, 1987). By using both hands,
a flow of fruits can be maintained that takes full advantage of chimpanzee chewing capacities.
This is particularly important for small fruits, since it is picking, not chewing, that constrains
feeding rate. Terrestrial gathering is practical only when feeding from short trees or bushes.
Even the lowest branches of trees> 15 m were too high for the fruit to be reached. In some
trees (e.g., Harungana) chimpanzees held the tree lower with one forelimb while they fed
with the other. The effect was a terrestrial arm-hanginglbipedalism.
Small trees elicited bipedalism when they were fed from arboreally as well, but for dif-
ferent reasons. The small diameter of the branches discouraged sitting as a collecting posture.
The small twigs in these trees bent to vertical under the weight of chimpanzees, making the
substrate impossible to sit on. Among the terminal branches of these trees chimpanzees used
as many contact points as possible. By extending the hip and fully abducting the arm they
maximized the reach so that a larger number of branches were within grasp.
Short-stride-length arboreal movement and terrestrial bipedal shuffling are advanta-
geous for collecting fine-grained resources such as small evenly distributed fruits in small
trees. Frequent short distance (i.e., -1 m) travel is necessary when harvesting small fruits
both in the trees and on the ground. Feeding sites are depleted quickly so that postures that
allow a switch to locomotion with little energy cost are preferred (Wrangham, 1980). Sitting
postures during feeding are often engaged in on small-diameter supports, yet a squatting
shuffling (gelada-style) to an adjacent branch is impossible. When bearing weight, terminal
branch supports are often bent to such a degree that they are a meter or more lower than oth-
erwise. Movement to adjacent small branches can be accomplished more effectively with
bipedal locomotion, since it allows a hindlimb to be raised to the as yet unweighted level of
the adjacent branch. Bipedal locomotion on small substrates is much like walking on an ex-
tremely soft cushion. Legs must be raised very high with each step. It is not practical for a
harvester to avoid the costs of small branch collecting by breaking off a branch and retreat-
ing to a more stable perch, since anyone branch contains too little food to constitute a meal.
The combination of greater stability and easier transition to locomotion makes bipedalism a
favored collecting strategy among small branches common among smaller trees.
Among chimpanzees, large individuals spent more time on the ground, less time
among the terminal branches, and fed from smaller trees. Because vertical climbing is dis-
proportionally more expensive for large animals (Taylor et aI., 1972), the most straightfor-
ward interpretation of these data is that large individuals must balance the benefits of
climbing (obtaining high-quality foods) with the costs of vertical ascent (a larger energy
expenditure per unit weight, compared to smaller individuals). Because climbing is more
expensive for big chimpanzees, arboreal foods are less valuable. As large-bodied primates
to begin with, we might expect that compared to other primates chimpanzees prefer to
minimize climbing, and large chimpanzees make more compromises to minimize climbing
than do small chimpanzees. This leads to the expectation that adult male australopithe-
cines should forage on the ground most often, since their body weight was nearly double
that of females (McHenry, 1991 b). Females might be expected to be more arboreal.
Ecological Morphology of Australopithecus afarensis 415

7. CONCLUSION

The synthesis of chimpanzee ecology and australopithecine functional morphology

yields a postural feeding hypothesis that suggests that australopithecines were semi-arbo-
real, postural bipeds that specialized on gathering small, hard-husked fruit in short-
statured trees.
Large males hypothetically found it more advantageous to gather terrestrially,
whereas females were about half as arboreal as chimpanzees. Short hind limbs and wide
hips are splendid adaptations for tree movement because they lower the center of gravity,
increase stability, and improve climbing mechanics.
Bipedalism is hypothesized to have evolved initially in conjunction with arm-hang-
ing, as a feeding posture effective for collecting small diameter fruits from small trees.
Refinements to allow efficient bipedalism in the pelvis and hind limb of A. afarensis sug-
gest that their terrestrial locomotion was fully bipedal, albeit less efficient than the
bipedalism of modern humans. This suggests that their mode of travel on the ground was
fully upright bipedalism.
Chimpanzees tend to travel terrestrially, and to feed arboreally. The woodland and
open-forest habitat suggested for australopithecines would have required even more terres-
trial movement than is found among extant chimpanzees. The ecological morphology of
australopithecines suggests that they traveled bipedally terrestrially, but fed arboreally us-
ing a number of chimpanzee-like suspensory behaviors in addition to sitting and bipedal
standing.

ACKNOWLEDGMENTS

I am grateful to the National Science Foundation, the Wenner-Gren Foundation for

Anthropological Reserarch, the California State University, Sacramento Foundation and the
University of California, Davis for sponsoring the Primate Locomotion - 1995 conference,
out of which this paper grew. I thank the editors of this volume, E. Strasser, J. Fleagle, A.
Rosenberger, and H. McHenry for inviting me to speak at the conference and to write this
manuscript. I also thank the anonymous reviewers and the editors for their comments on the
manuscripts, which have improved it substantially. Research was aided by the L.S.B.
Leakey Foundation, NSF BNS-86-09869, Richard W. Wrangham, and Harvard University.

REFERENCES
Abitbol M (1987) Evolution of the lumbosacral angle. Am. J. Phys. Anthropol. 72:361-372.
Berge C, and Kazmierczak J-B (1986) Effects of size and locomotor adaptations on the hominid pelvis: Evaluation
of australopithecine bipedality with a new multivariate method. Folia Primatol. 46: 185--204.
Cheney DL, and Wrangham RW (1986) Predation. In BB Smuts, DL Cheney, RM Seyfarth, RW Wrangham, and
TT Struhsaker (eds.): Primate Societies. Chicago: University of Chicago Press, pp. 227-239.
Deloison Y (1991) Les Australopitheques marchaient-ils comme nous? In Y Coppens and B Senut, (eds.): Orig-
ine(s) de la Bipedie chez les Hominides, Cah. Paleoanthrop. Paris: Editions du CNRS, pp. 177-186.
Du Brul EL (1962) The general phenomenon of bipedalism. Am. Zool. 2:205--208.
Fooden J (1964) Stomach contents and gastro-intestinal proportions in wild-shot Guinan monkeys. Am. 1. Phys.
Anthropol. 22:227-232.
Gantt DG (1983) The enamel of Neogene hominoids: Structural and phyletic implications. In RL Ciochon and RS
Corruccini (eds.): New Interpretations of Ape and Human Ancestry. New York: Plenum Press, pp.
249-298.
416 K. D. Hunt

Ghiglieri MP (1984) The Chimpanzees of Kibale Forest. New York: Columbia University Press.
Glander KE (1978) Howling monkey's feeding behaviour and plant secondary compounds: A study of strategies.
In GG Montgomery (ed.): The Ecology of Arboreal Folivores. Washington: Smithsonian Inst. Press, pp.
561-573.
Grine FE, and Kay RF (1988) Early hominid diets from quantitative image analysis of dental microwear. Nature.
333:65-68.
Hamilton AC (1982) Environmental History of East Africa: A Study of the Quaternary. London: Academic Press.
Hladik A, and Hladik CM (1969) Rapports trophiques entre vegetation et primates dans la foret de Barro Colorado
(Panama). Terre et Vie 23:25-117.
Hunt KD (1989) Positional behavior in Pan troglodytes at the Mahale Mountains and Gombe Stream National
Parks, Tanzania. Ph.D. Dissertation., University of Michigan.
Hunt KD (l99Ia) Positional behavior in the Hominoidea. Int. J. Primato!' 12:95-118.
Hunt KD (199Ib) Mechanical implications of chimpanzee positional behaviour. Am. J. Phys. Anthropo!.
86:521-536.
Hunt KD (1992a) Positional behaviour of Pan troglodytes in the Mahale Mountains and Gombe Stream National
Parks, Tanzania. Am. J. Phys. Anthropo!. 87:83-107.
Hunt KD (I 992b) Social rank and body weight as determinants of positional behavior in Pan troglodytes. Primates
33:347-357.
Hunt KD (1994a) The evolution of human bipedality: Ecology and functional morphology. J. Hum. Evo!.
26: 183-202.
Hunt KD (1994 b) Body size effects on vertical climbing among chimpanzees. Int. J. Primato!. 15:855-865.
Hunt KD (1996) The postural feeding hypothesis: An ecological model for the evolution of bipedalism. S. Afr. J.
Sci. 92:77-90.
Hylander WL (1975) Incisor size and diet in anthropoids with special reference to Cercopithecidae. Science
189: 1095-1098.
Johanson DC, and Edey M (1981) Lucy: The Beginnings of Humankind. New York: Simon and Schuster.
Jolly CJ (1970) The seed-eaters: A new model of hominid differentiation based on a baboon analogy. Man. 5: 1-26.
Jolly CJ and Plog F (1987) Physical Anthropology and Archeology. New York: Knopf.
Jungers WL (1982) Lucy's limbs: Skeletal allometry and locomotion in Australopithecus afarensis. Nature.
297:676-678.
Jungers WL (1988) Relative joint size and hominoid locomotor adaptations with implications for the evolution of
hominid bipedalism. J. Hum. Evo!. 17:247-265.
Jungers WL (199\) A pygmy perspective on body size and shape in Australopithecus afarensis (AL 288-1,
"Lucy"). In Y Coppens and B Senut (eds): Origine(s) de la Bipedie chez les Hominides, Cah. Paleoanthrop.
Paris: Editions du CNRS, pp. 215-224.
Kay RF (1981) The nut-crackers - A new theory of the adaptations of the Ramapithecinae. Am. J. Phys. Anthro-
po!. 55:141-151.
Kay RF (1985) Dental evidence for the diet of Australopithecus. Ann. Rev. Anthropo!. 14: 15-41.
Kay RF, and Grine FE (1988) Tooth morphology, wear and diet in Australopithecus and Paranthropus from south-
ern Africa. In F Grine (ed.): Evolutionary History of the Robust Australopithecines. Chicago: Aldine, pp.
427-447.
Kay RF, and Hylander WL (1978) The dental structure of mammalian folivores with special reference to Primates
and Phalangeroidea (Marsupialia). In GG Montgomery (ed.): The Biology of Arboreal Folivores. Washing-
ton, D.C.: Smithsonian Inst. Press, pp. 173-191.
Kimbel WH, Johanson DC, and Rak Y (1994) The first skull and other new discoveries of Australopithecus
afarensis at Hadar, Ethiopia. Nature 368:449-451.
Kinzey WG (1976) Positional behavior and ecology in Callicebus torquatus. Yrbk. Phys. Anthropo!. 20:468-480.
Kinzey WG (1977) Diet and feeding behaviour of Callicebus torquatus. In TH Clutton-Brock (ed.): Primate Ecol-
ogy. London: Academic Press, pp. 127-152.
Kinzey WG (1978) Feeding behaviour and molar features in two species oftiti monkey. In DJ Chivers and J Her-
bert (eds.): Recent Advances in Primatology, Vo!. I. London: Academic Press, pp. 373-385.
Kummer B (1991) Biomechanical foundations of the development of human bipedalism. In Y Coppens and B
Senut (eds): Origine(s) de la Bipedie chez les Hominides, Cah. Paleoanthrop. Paris: Editions du CNRS, pp.
1-8.
Langdon JH, Bruckner J, and Baker HH (1991) Pedal mechanics and bipedalism in early hominids. In Y Coppens
and B Senut (eds): Origine(s) de la Bipedie chez les Hominides, Cah. Paleoanthrop. Paris: Editions du
CNRS, pp. 159-167.
Ecological Morphology of Australopithecus afarensis 417

Latimer B (1991) Locomotor adaptations in A ustralopithecus afarensis: The issue of arboreality. In Y Coppens and
B Senut (eds.): Origine(s) de la Bipedie chez les Hominides, Cah. Paleoanthrop. Paris: Editions du CNRS,
pp.169-176.
Latimer B, and Lovejoy CO (1989) The calcaneus of Australopithecus afarensis and its implications for the evolu-
tion ofbipedality. Am. J. Phys. Anthropol. 78:369-386.
Latimer B, and Lovejoy CO (1990) Hallucal tarsometatarsal joint in Australopithecusafarensis. Am. J. Phys. An-
thropol. 82: 125-133.
Latimer B, Lovejoy CO, Johanson DC, and Coppens, Y (1982) Hominid tarsal, metatarsal and phalangeal bones
recovered from the Hadar formation: 1974-1977 collections. Am. J. Phys. Anthropol. 53:701-719.
Leakey MG, Feibel CS, McDougall I, and Walker AC (1995) New four-million-year-old hominid species from
Kanapoi and Allia Bay, Kenya. Nature 376:565-571.
Lovejoy CO (1981) The origin of man. Science. 211:341-350.
Lovejoy CO (1988) The evolution of human walking. Sci. Am. 259: 118-125.
Lovejoy CO, Johanson DC, and Coppens Y (1982) Hominid upper limb bones recovered from the Hadar Forma-
tion: 1974-1977 collections. Am. 1. Phys. Anthropol. 57:637-649.
McHenry HM (1986) The first bipeds: A comparison of the A. afarensis and A. africanus postcranium and implica-
tions for the evolution of bipedalism. J. Hum. Evol. 15:177-191.
McHenry HM (1991 a) First steps? Analyses of the postcranium of early hominids. In Y Coppens and B Senut
(eds): Origine(s) de la Bipedie chez les Hominides, Cah. Paleoanthrop. Paris: Editions du CNRS, pp.
133-141.
McHenry HM (199Ib) Sexual dimorphism in Australopithecus afarensis. J. Hum. Evol. 20:21-32.
McHenry HM (1992) Body size and proportions of early hominids. Am. J. Phys. Anthropol. 87:407-431.
Peters CR (1987) Nut-like oil seeds: Food for monkeys, chimpanzees, humans and probably ape-men. Am. J.
Phys. Anthropol. 73:333-363.
Puech P-F, and Albertini H (1984) Dental microwear and mechanisms in early hom in ids from Laetoli and Hadar.
Am. J. Phys. Anthropol. 65:87-92.
Rak Y (1991) Lucy's pelvic anatomy: Its role in bipedal gait. J. Hum. Evol. 20:283-290.
Robinson IT (1972) Early Hominid Posture and Locomotion. Chicago: University of Chicago Press.
Rodman PS (1984) Foraging and social systems of orangutans and chimpanzees. In PS Rodman and JGH Cant
(eds.): Adaptations for Foraging in Nonhuman Primates: Contributions to an Organismal Biology of
Prosimians, Monkeys, and Apes. New York: Columbia University Press, pp. 134-160.
Rose MD (1976) Bipedal behaviour of olive baboons (Papio anubis) and its relevance to an understanding of the
evolution of human bipedalism. Am. J. Phys. Anthropol. 44:247-261.
Rose MD (1984) Food acquisition and the evolution of positional behaviour: The case of bipedalism. In DJ
Chivers, BA Wood, and A Bilsborough (eds.): Food Acquisition and Processing in Primates. New York:
Plenum Press, pp. 509-524.
Rose MD (1991) The process ofbipedalization in hominids. In Y Coppens and B Senut, (eds): Origine(s) de la
Bipedie chez les Hominides, Cah. Paleoanthrop. Paris: Editions du CNRS, pp. 37-48.
Rosenberg KR, and Trevathan W (1995) Bipedalism and human birth: The obstetrical dilemma revisited. Evol.
Anthropol. 4:161-168.
Sarmiento EE (1988) Anatomy of the hominoid wrist joint: Its evolutionary and functional implications. Int. J. Pri-
matol. 9:281-345.
Schmid P (1983) Eine Reconstrucktion des skelettes von A.L. 288--1 (Hadar) und deren konsequenzen. Folia Pri-
matol. 40:283-306.
Schmid P (1991) The trunk of the Australopithecines. In Y Coppens and B Senut, (eds): Origine(s) de la Bipedie
chez les Hominides, Cah. Paleoanthrop. Paris: Editions du CNRS, pp. 225-234.
Stem IT, Jr (1976) Before bipedality. Ybk. Phys. Anthropol. 19:59-68.
Stem IT, Jr, and Susman RL (1983) The locomotor anatomy of Australopithecus afarensis. Am J. Phys. Anthropol.
60:279-317.
Stoller M (1995) The Obstetric Pelvis and Mechanism of Labor in Nonhuman Primates. Ph.D. Dissertation, The
University of Chicago.
Tague RG, and Lovejoy CO (1986) The obstetric pelvis of A.L. 288-1 (Lucy) J. Hum. Evol. 15:237-255.
Tardieu C (1983) L'articulation du genou. Analyse morpho-fonctionelle chez les primates. Application aux
hominides fossiles. Paris: Centre National de la Recherche Scientifique.
Taylor CR, Caldwell SL, and Rowntree VJ (1972) Running up and down hills: Some consequences of size. Sci-
ence 178:1096-1097.
Teaford MF (1985) Molar microwear and diet in the genus Cebus. Am. J. Phys. Anthropol. 66:363-370.
Teaford MF (1994) Dental microwear and dental function. Evol. Anthropol. 3: 17-30.
418 K. D. Hunt

Teaford MF, and Walker AC (1984) Quantitative differences in dental microwear between primate species with
different diets and a comment on the presumed diet of Sivapithecus. Am. J. Phys. Anthropol. 64: 191-200.
Tobias PV (1980) A survey and synthesis of the African hominids of the late Tertiary and early Quaternary periods.
In L Konigsson (ed.): Current Argument on Early Man. Oxford: Pergamon, pp. 86--113.
Tuttle RH (1975) Parallelism, brachiation and hominoid phylogeny. In WP Luckett and FS Szalay (eds.): The Phy-
logeny of the Primates: A Multidisciplinary Approach. New York: Plenum, pp. 447-480.
Tuttle RH (1981) Evolution of hominid bipedalism and prehensile capabilities. Phil. Trans. Roy. Soc. 292:89--94.
Tuttle RH, Webb DM, and Tuttle NI (1991) Laetoli footprint trails and the evolution of hominid bipedalism. In Y
Coppens and B Senut (eds.): Origine(s) de la Bipedie chez les Hominides, Cah. Paleoanthrop. Paris: Edi-
tions du CNRS, pp.187-198.
van Schaik CP (1983) Why are diurnal primates living in groups? Behaviour 87: I 2(}"'1 43.
van Schaik CP and van Hooff J (1983) On the ultimate causes of primate social systems. Behaviour 85:91-117.
van Schaik CP, van Noordwijk MA, Boer RJ de, Tonkelaar I den. (1983) The effect of group size on time budgets
and social behaviour in wild long-tailed macaques. Behav. Ecol. Sociobiol. 13: 173-181.
van Schaik CP, van Noordwijk MA, Warsono B, and Sutriono E (1983) Party size and early detection of predators
in Sumatran forest primates. Primates. 24:211-221.
van Schaik CP, and van Noordwijk MA (1989) The special role of male Cebus monkeys in predation avoidance
and its effect on group composition. Behav. Ecol. Sociobiol. 24:265--276.
Wainwright PC, and Reilly SM (1994) Ecological Morphology: Integrative Organismal Biology. Chicago: Univer-
sity of Chicago Press.
Walker AC (1981) Diet and teeth---ilietary hypotheses and human evolution. Phil. Trans. Roy. Soc. London B.
292: 57--{)4.
White TD, Johanson DC, and Kimbel WH (1981) Australopithecusafricanus: Its phyletic position reconsidered.
S. Afr. J. Sci. 77:445-470.
White TD, Suwa G, and Asfaw B (\ 994) Australopithecus ramidus, a new species of early hominid from Aramis,
Ethiopia. Nature 371:306--312.
White TD, Suwa G, Hart WK, Walter RC, WoldeGabriel G, de Heinselin J, Clark JD, Asfaw B, and Vrba E (1993)
New discoveries of Australopithecus at Maka in Ethiopia. Nature 366:261-265.
Wrangham RW (\ 977) Feeding behaviors of chimpanzees in Gombe National Park, Tanzania. In TH Clutton-
Brock (ed.): Primate Ecology. London: Academic Press, pp. 503-538.
Wrangham RW (1980) Bipedal locomotion as a feeding adaptation in gelada baboons, and its implications for
hominid evolution. J. Hum. Evol. 9:329--331.
 21

TIME AND ENERGY: THE ECOLOGICAL

CONTEXT FOR THE EVOLUTION OF
BIPEDALISM

Robert A. Foley and Sarah Elton

Human Evolutionary Biology Research Group

Department of Biological Anthropology
University of Cambridge
Downing Street
Cambridge, CB2 3DZ England

1. INTRODUCTION

This paper is concerned with explanations for the evolution of bipedalism. Its gen-
eral point is a very simple one--that the occurrence of bipedalism is context specific. The
pattern of hominid evolution, as much as that of any other lineage, reflects the costs and
benefits of the wayan animal is structured and behaves, and this ratio is entirely depend-
ent upon when and where it is occurring. Historically the context for bipedalism has been
the general characteristics of the environment--savanna grasslands, open environments,
patchy woodland versus forest. This remains important, but here we shall add a new con-
sideration, that of time budgets, which provides a more specific ecological context for
considering the energetics of bipedalism.
The paper will first discuss the various contexts that do need to be taken into ac-
count when considering the origins of bipedalism. It extends a model developed earlier
(Foley, 1992), based on a cost-benefit analysis, which suggested that the advantages of
bipedalism should be placed into the specific ecological context of the Pliocene hominids.
That context was specifically the effect of more arid, seasonal and open environments
with more patchy and dispersed resources. The major change predicted in the model was
an increase in day range length, and that the energetics of bipedalism are directly linked to
increased ranging area (Foley, 1992). In quantitative terms, it was suggested (Foley, 1992:
Figure 5.3) that bipedalism allowed a 50 kg hominid to exploit a day range of 16 km for
the same energy as a male chimpanzee (45 kg) uses for the maximally observed day range
length of 10 km (Rodman, 1984), a result that Leonard and Robertson (1997) have re-
cently replicated. The earlier model will be developed here to incorporate a new element,

419
 420 R. A. Foley and S. Elton

time. Although the energetic advantages of bipedalism allow longer day ranges, the exten-
sion of a day range length has to occur in the context of a finite resource, that of time, and
in particular hours of daylight (Foley, 1995). Time, as Dunbar (1992) has argued, is the
hidden constraint in behavioral ecology. To model correctly the energetics of bipedalism
in an ecological context it is therefore necessary to construct a time budget of daily activi-
ties. This is done here, with particular emphasis on how such activities are distributed ter-
restrially and arboreally. This distribution is important as it is not just the energetic
advantages that accrue from being bipedal as a terrestrial primate, but also the costs of
climbing. The model of daily energetics indicates that a very substantial part of the day
has to be spent terrestrially before the benefits of bipedalism exceed the costs. In the final
part of the paper, the implications for reconstructions of scenarios for the origins of
bipedalism are discussed.

2. ADAPTATION AS PROBLEM-SOLVING: THE COSTS AND

BENEFITS OF BIPEDALISM

Ecological context is highly specific, not just in terms of the habitat, but also the ac-
tivities of the animal concerned. Two species of primate can live in the same environment
but be morphologically different because their diets, activities and time budgets vary. The
same trait could thus be both adaptive or maladaptive, depending upon circumstances, and
no adaptation is for all times and places. The lack of context is perhaps the greatest dispar-
ity between what might be called the pre-modem synthesis and more recent approaches to
hominid evolution. The essence of a neo-Darwinian approach is that traits are advanta-
geous in particular circumstances, rather than in absolute terms. This implies that any fea-
ture of an organism, be it behavioral or morphological, will be an evolutionary response to
precise settings of time and place. Or, to put it another way, that such traits are not only
advantageous in particular environments, they are also disadvantageous in others. To ex-
plain the evolution of features such as bipedalism during the course of hominid evolution
thus requires a detailed consideration of the contexts in which it is likely to occur.
Within modem evolutionary biology there are two broad conceptual frameworks
that provide the basis for attempting to place adaptive evolutionary events into context.
The first of these is what might be called the optimal problem-solving approach (Maynard
Smith, 1978; Foley, 1987). Features evolve because they solve the problems of survival
faced by organisms in particular environments, and selection is the evolutionary mecha-
nism that leads to what might be called problem-solving optima. This approach to a wide
ranging series of adaptive problems in hominid evolution has been developed extensively
(F oley, 1987). Complementary to this idea is that of cost-benefit algorithms. Adaptations
may be treated as benefits to an organism. All such adaptive traits, however, impose costs
on an organism. These costs may be associated with development or maintenance; they
may be energetic or structural; they may be behavioral. They may be direct costs or oppor-
tunity costs-that is, costs that come into account because the evolution of one feature
will reduce the opportunities for other evolutionary changes. Cost-benefit analysis has
proved to be very productive as a means of analyzing evolutionary strategies, particularly
in terms of foraging behavior, where direct observation and measurement of energetic and
time inputs and outputs is possible. Both problem-solving and cost-benefit analysis are, in
effect, ways of making more practicable the basic Darwinian premise that features that de-
velop over the course of evolution, features that are selected for, provide specific advan-
tages-adaptations-in particular contexts.
Time and Energy: The Ecological Context for the Evolution of Bipedalism 421

Bipedalism is a classic example. Among mammals bipedalism is very rare, and as a

specialized adaptation in primates occurs uniquely in hominids. The changes in the mus-
culoskeletal system that result from bipedalism are very extensive, and it has been argued
that bipedalism is the basis for many other non-locomotory changes in hominid evolution.
The benefits that arise from bipedalism range from increased manual dexterity, enhanced
neural and general thermoregulation, decreased locomotor energetic expenditure, greater
anti-predatory vigilance, intra-specific display, etc. Such an array of benefits can at times
seem at odds with the rarity of bipedalism among primates. Only when the costs of
bipedalism are considered--that is, the disadvantages--does it become clearer why
bipedalism may both have evolved among hominids and, which is equally important, not
evolved among other species. The costs, in tum, are as context dependent as the benefits.
Discussions of bipedalism have generally focused on two types of advantage that
may accrue. The first of these is the direct locomotor advantage, that is, the energetic
benefits that arise from being able to walk upright habitually. There has been considerable
debate as to whether these advantages exist (Taylor and Rowntree, 1973; Rodman and
McHenry, 1980; Steudel, 1994; Isbell and Young, 1996) and, as a result, a second stream
of explanations relate to the non-locomotory advantages. In particular, it has been argued
that bipedalism leads to greater manual dexterity (Darwin, 1871), vigilance, and ther-
moregulatory efficiency (Newman, 1970; Wheeler 1984, 1985, 1991). These indirect ad-
vantages will not be incorporated here. Instead, in modeling the direct energetic costs and
benefits two novel themes in particular will be developed. The first is that at least some of
the costs imposed by bipedalism arise from a reduced ability to climb. The second is that
bipedalism, as with any locomotor behavior, occurs in the context of a daily schedule of
activities, and models should therefore be built around such a time-budgeting set of pa-
rameters.
Application of cost benefit analysis to problems in long term evolution is bound to
be more problematic than analyses ofliving populations. Direct observation in the field or
in the laboratory of actual energetic costs is obviously an important part of any such
analysis, and with animals long extinct this is not possible. The alternative that will be
used here will be to build a model using data and observations drawn from living spe-
cies-in particular, chimpanzees and humans. The limitations of such an approach should
be kept in mind.

3. THE CONTEXTS OF BIPEDALISM

Several contexts for assessing the costs and benefits of bipedalism can be described.

3.1. Taxonomic Context

There is now a relatively rich fossil record for hominids within Africa covering the
Plio-Pleistocene, representing a number of distinct taxa. The evidence they provide illus-
trates the diversity of hominid locomotor strategies.

3.1.1. Ardipithecus ramidus. This is the earliest known hominid (> 4.4 Myr), for
which only limited postcranial material has yet been published (White et aI., 1994). On the
basis of described material it cannot be ascertained whether A. ramidus was more bipedal
than extant apes, and it may be that this taxon either does not belong within the hominid
lineage or that it represents a form prior to any major changes in locomotor anatomy.
422 R. A. Foley and S. Elton

3.1.2. Australopithecus anamensis. This hominid is also over 4.0 Myr, and has a
relatively primitive cranial morphology (Leakey et ai., 1995). The associated tibia, how-
ever, has elements that link it with other early hominids, which may indicate some level of
bipedalism.

3.1.3. Australopithecus afarensis. This is the most extensively known of the early
hominids; two lines of evidence indicate that it is a relatively bipedal hominid. The post-
cranium of AL-288, especially the pelvis, indicates more habitual bipedalism (Lovejoy,
1979), while the footprint trail at Laetoli also suggests a bipedal hominid. These observa-
tions lead to the conclusion that at least some level of bipedalism was present in the early
hominids by shortly after 4.0 Myr, but there is considerable debate as to whether this is
universal among hominids, is fully established, precludes other locomotor behavior, and is
different from that found in modern humans (Lovejoy, 1980; Jungers, 1982; Senut and
Tardieu, 1985). The most likely consensus is that some hominids were active bipedally on
the ground for some of the time, but that their overall body size and shape still shared a
number of features with African apes (see papers by Hunt and Tuttle et ai., this volume).

3.1.4. Australopithecus africanus. In overall morphology this taxon is likely to be

relatively similar to or less like that of modern humans than A. afarensis. Interpretation
has been strongly influenced by the complete innominate Stsl4, which shows clear simi-
larities with later bipedal hominids. A recent discovery of a partial foot showing a diver-
gent hallux has led to some questioning of the extent of bipedalism in A. africanus,
although this position is complicated by uncertainty concerning the taxonomic integrity of
specimens assigned to this taxon (Clarke, 1985).

3.1.5. Australopithecus robustus and Allies. There is little doubt that these more ro-
bust and later « 2.7 Myr) taxa are at least as bipedal as the earlier australopithecines, and
many elements of their dietary adaptations have indicated a predominantly terrestrial way
of life (Susman and Brain, 1988; Grine, 1989).

3.1.6. Homo habiUs. The evidence of the Olduvai specimens has generally led peo-
ple to suppose that early Homo was fully bipedal, but this position has been called into
question to some extent by reinterpretations of the OH36 ulna and the postcranial anatomy
of OH62, both of which are relatively ape-like in morphology (Aiello and Dean, 1990).

3.1.7. Homo ergaster. The partial skeleton WTl5000 from West Turkana provides
the most complete evidence for hominid postcrania for the Plio-Pleistocene, and although
there are a number of small differences, nonetheless it is clear that by 1.6 Myr at least one
lineage, as represented by H. ergaster, is fully adapted to bipedalism in ways that are not
dissimilar to modern humans (Walker and Leakey, 1993).
In overall terms, therefore, the paleontological evidence shows that the context for
considering the evolution of bipedalism lies during the Plio-Pleistocene in sub-Saharan
Africa. At the beginning of the Pliocene, the remains from Aramis might be taken to indi-
cate essentially non-bipedal or partially bipedal hominids. The early australopithecines are
best thought of as being ape-like in morphology, but with a considerably greater set of ad-
aptations for bipedalism than the living African apes. The later australopithecines are
probably fully terrestrial and bipedal, but with some adaptive differences from later Homo,
while early Homo itself shows a mixed suite of adaptations. Full human bipedalism was
present by 1.6 Myr. It should be stressed that the recognition of multiple taxa changes the
Time and Energy: The Ecological Context for the Evolution of Bipedalism 423

way we should look at bipedalism. It is likely that during the Pliocene, when the hominids
were diversifying, so too was their locomotor behavior. The presence of evidence for
bipedalism in one taxon does not necessarily imply that it was universal among all.

3.2. Paleoenvironmental Context

The paleoenvironmental context of these hominids has itself been subject to consid-
erable discussion. An earlier consensus that all the hominids existed in relatively open en-
vironments has been somewhat eroded in recent years (see Reed, 1997, for a recent
summary). Earliest hominid sites from the Middle Awash and Hadar have all been inter-
preted as showing at least some tree cover, and recently the same interpretation has been
applied to the southern African sites. The earliest hominids are now considered to have oc-
cupied habitats that were not fully forest, but would certainly have contained at least some
tree cover. The amount of tree cover is likely to have varied considerably from location to
location and through time. There is little doubt that some hominids after 2.0 Myr were liv-
ing in fully open, semi-arid environments. Overall, the paleoenvironmental context for the
evolution of bipedalism is thus likely to have been a mixed tropical African environment:
savanna in the sense that grass would have been extensive, but also with significant levels
of tree and bush cover, and in all likelihood in relatively close proximity of water (see Ta-
ble 1). These environments are likely to have resources that were highly seasonal in distri-
bution, patchily and unevenly distributed, and occurring both in trees and on the ground.
In addition, the level of thermal stress is likely to have been considerable.

3.3. Phylogenetic Context

All evolutionary change is a function of the interaction between the existing compo-
nents of a species and its current environment. Initial phylogenetic conditions thus play an
important role in determining the nature of evolutionary change, for the existing pheno-
type provides the framework for the level of costs and benefits imposed by any particular
strategy. Phylogenetic context is almost certainly one of the reasons why baboons and
hominids have such divergent locomotor strategies despite very similar habitats, for the
costs and benefits of changing the existing phenotype would have been very different
(Foley, 1987). With regard to bipedalism, there is a great deal of uncertainty about what
might be the phylogenetic starting conditions. There is virtually a complete absence of ap-

Table 1. Reconstructions of the environments of early hominid taxa in Africa.

Reconstructions based primarily on Reed (1997)

Taxon Habitats
A. ramidus Woodland
A. anamensis ?
A. afarensis Closed to open woodland, with edaphic
grasslandlshrubland/deltaic floodplain/water and trees present
A. africanus Woodland and bushland with riverine forest
A. aethiopicus Bushland to open woodland/edaphic grassland
A. robustus Open or wooded-bushed grassland/edaphic grassland
A. boisei Woodland to scrublandledaphic grassland
H. habilis + H. rudolfensis Open grasslands/dry shrublands/edaphic grasslands
H. ergaster Shrublandlriparian woodlandlarid open landscapes
424 R. A. Foley and S. Elton

propriately situated fossil material, so what is known is largely estimated from modern hu-
mans, later fossil hom in ids and extant apes. It has been proposed that the last common an-
cestor of the hominids and African apes was variously a knuckle-walker, brachiator or
generalized clamberer/climber (see Hunt, this volume, for a discussion of this problem).
While it is difficult to test for these various models, it would seem more probable that the
ancestors of the first hominids either approximated the more generalized locomotor be-
havior of Pan or were even less specialized.

3.4. Behavioral Context

Bipedalism is usually treated as an anatomical problem by paleoanthropologists, but
at heart the problem is behavioral. Bipedalism is a means to an end, that end being the
avoidance of predators, the acquisition of mates and food, and the balancing of an overall
energy budget. Bipedalism will evolve when the costs of moving in this way are less than
the benefits. The key context is, therefore, the way in which an animal spends its day; time
is the 'hidden constraint in primate behavioral ecology' (Dunbar, 1992). It is essential to
realize that the key context in which bipedalism must be assessed is how a hominid would
have budgeted its day under different ecological, phylogenetic and energetic contexts.
Dunbar (1992) has developed a model in which a primate day is partitioned up into four
activities-feeding, traveling, resting and socializing. Each of these is essential for the sat-
isfactory functioning of a population. Feeding and traveling time are the highest priority,
but there are also limits on how much resting and socializing time can be varied in a typi-
cal activity budget. For example, among savanna baboons the range of time budgets is:
feeding = 23-56%, traveling = 17-36%, resting = 5-60%, and socializing = 22-57%.
Teleki (1989, quoted in Williamson, 1997) gives for chimpanzees from Gombe a feeding
time of 42.8%, traveling time of l3.4%, resting time of 18.9%, and 24.9% socializing
time. Williamson (1997) gives for both species of Pan and for different environments
ranges of 29.7-67% for feeding, l3-27.5% for traveling, and 30--43% for resting/socializ-
ing. When considering the energetics of bipedalism, it is for this scale of activity budget
that energetic efficiency has to be considered.
The background for modeling the costs and benefits of the evolution of bipedalism
can thus be approximated as that of a generalized ape living in a relatively well-treed en-
vironment in Africa. This ape would be capable of both arboreal and terrestrial foraging
and movement, and would have to acquire food efficiently across a patchy and hot envi-
ronment. The problem is to model the energetics of such an animal under different eco-
logical conditions, to determine when it may pay to become a more 'committed biped'.
The novel element is placing this setting into the framework of a daily time budget such as
would be expected for a typical social primate.

4. THE MODEL
There are two components to the model, one related to time budgeting and one to
energetics. The first component is a partitioning of a 12 hour day into percentage time
spent feeding and traveling. For baboons and chimpanzees these two activities may ac-
count for up to half of a day (Dunbar, 1992; Isbell and Young, 1996; Williamson, 1997).
As feeding and traveling are the most energetically expensive activities, and the most af-
fected by locomotor abilities, the time spent on these activities is varied in the model.
Both feeding time and traveling time are varied between a minimum of 20% and a maxi-
Time and Energy: The Ecological Context for the Evolution of Bipedalism 425

Table 2. Parameters used in the model, showing range of values and coefficients
applied in the simulations
Parameter Organism Value/range
Length of day (tropical environments) 720 min.
Body weight 30 kg
Feeding time range! 20-40%
Traveling time range! 20-40%
Resting/socializing time range! 40--60%
Energetic costs offeeding on ground (kJ/min)2 modem human 5.15
early biped 5.67
quadrupedal ape 6.44
Energetic costs offeeding in trees (kJ/min)J modem human 9.41
early biped 9.41
quadrupedal ape 7.06
Energetic costs of traveling on ground (kJ/min) modem human4 10.29
early biped 5 11.32
quadrupedal ape 6 12.86
Energetic costs of traveling in the trees (kJ/min) modem human 7 18.81
early biped 7 18.81
quadrupedal ape 8 14.11
Energetic costs of resting and socializing activities (all 2.91
habitats and species) (kJ/min)9
!Time ranges taken from Dunbar (1992).
2Energetic cost of feeding on ground estimated at half the terrestrial traveling rate.
JEnergetic cost of feeding in trees estimated at half the arboreal traveling rate.
4Energetic cost of traveling on the ground calculated using data obtained Elton et al. (in press; see Table 3).
5Energetic cost of traveling on the ground calculated as 10% more costly than for the modem human.
6Energetic cost of traveling on the ground calculated as 25% more costly than for the modern human.
7Energetic cost of traveling in the trees calculated using data from Elton et al. (in press; see Table 3).
"Energetic cost of traveling in the trees estimated as 75% of the modem human cost.
9Resting and socializing costs estimated using the modern human standing cost, using data from Elton et
al. (in press; see Table 3).

mum of 40% (Dunbar, 1992; Williamson, 1997; Table 2). A control value of 25% for feed-
ing time and 20% for traveling time is used when the other is being varied. A case can be
made that social behavior may also be sensitive to pattern of locomotion (see, for exam-
ple, Jablonski and Chaplin, 1993), but this is not pursued here. A further constraint on the
model is that it only explores strategies that reduce the costs of locomotion in particular
contexts; ways in which energetic benefits through access to additional resources may ac-
crue have not been considered, although a case may be made that this was a significant
factor in the evolution of bipedalism (Hunt, 1996).
The second component of the model is the energetics of the hominids and chimpan-
zees as a comparative animal for a quadrupedal ape. There are considerable data available
on human energy expenditure (see Ulijaszek, 1995, for a recent summary). There are vir-
tually none available on the energy costs of climbing, however, and therefore an experi-
mental study was carried out on the energetics of standing (which is used here as a
surrogate for resting/socializing), walking and climbing in a mixed sex sample (see Elton
et aI., in press, for details; Table 3). Data on chimpanzee energy expenditure are scarce,
most of which returns to an early study by Taylor and Rowntree (1973) on a juvenile
chimpanzee. There are no data on climbing costs in chimpanzees, although Caldwell et al.
(1972) found that moving vertically was around double the energetic costs of traveling
horizontally.
426 R. A. Foley and S. Elton

Table 3. Energetics of the activities used in the model. These estimates

are derived from a mixed sex study involving controlled
exercise (Elton et aI., in press)
Standing Walking Climbing
Model kJ/minlkg kJ/minlkg kJ/minlkg
Energy per kg 0.097 0.343 0.627
Small biped (30 kg) 2.91 10.29 18.81

In the absence of unequivocal data the following estimates were used. Standing,
walking and climbing costs were calculated in kJ per minute per kg using the results of the
study by Elton et al. (in press; Table 3). Energy expenditure during feeding time was cal-
culated at half the walking rate for terrestrial feeding, and half the climbing rate for arbo-
real feeding. This rate was selected to reflect the probability that feeding would involve
some movement on whatever substrate was being used. Energy expenditure during travel-
ing time was estimated using the walking rate for terrestrial travel, and the climbing rate
for arboreal travel. Energy expenditure during resting and socializing time was estimated
for modern humans using the rates for energy expenditure whilst standing. Overall daily
(12 hour) energy expenditure is the sum of time spent in each activity. In the model, the
percentage of time spent on the ground was varied from 20% to 100%. This affected over-
all energy expenditure as traveling and feeding time spent in the trees were more energeti-
cally expensive than that on the ground.
Three types of 'creature' were modeled: one that basically has the same rate of ex-
penditure as a modern human, using the rates described above; a less efficient, early bi-
ped; and, a chimpanzee-like quadruped. The climbing costs of the less-efficient biped
were set at the "modern human" rate, and the cost of bipedalism in this creature was 10%
more than that of the modern human (Table 2). The chimpanzee-like quadruped's terres-
trial efficiency was less than that of a biped, following Rodman and McHenry's (1980) re-
sults showing that chimpanzees are around a third less efficient on the ground than are
humans. A more conservative estimate of a 25% increase in energy expenditure was used
here. Conversely, it was assumed that this creature would have been more efficient in the
trees than a modern human, so the climbing rate for the chimpanzee was estimated at 75%
of the modern human rate. In the model described here a body size of 30 kg was used.
This is a relatively small body size, and is appropriate for some of the very earliest homi-
nids and for chimpanzee females. Increased body size would, in these models, not affect
the outcome other than by increasing overall energy expenditure across all model crea-
tures (but see Steudel,1994, for a discussion of the allometric factors relating to locomo-
tion in primates). The main parameters of the model are shown in Table 2.

4.1. Results
Applying the model to the creatures described above shows how differences in ac-
tivity patterns affect energy expenditure, and these can be used to explore the effects of
time budgets on the costs and benefits of bipedalism.

4.1.1. Feeding Time. As discussed above, in practical terms, using baboon and chim-
panzee analogues, a large social primate can expect to feed for about 20% of its time when
conditions are relatively good and to increase feeding time to as much as 40% when food
is either of poor quality, takes time to process, or is hard to find. Figure 1 shows the effect
Time and Energy: The Ecological Context for the Evolution of Bipedalism 427

6000 20% feeding time Early biped (30 kg)

5500
Modern human (30 kg)
Quadrupedal ape (30 kg)
5000

4500

..........................
4000 .........................
.•....•...•.•. :.::::..
3500

3000+-------r------.-------.------.
20 40 60 80 100
6000 30% feeding time
5500

::::~"-.~-~--~--~-~-"-~-."~-".~-- ~-".~"~ ~~~-~-~~--------

.. ... ..... ... .. .. .. ... ....

4000 ~-<::-:::.:::::
3500

3000+-----~r------r------~-----,
20 40 60 80 100

6000 40% feeding time

5500

5000

4500

4000

3500

30004-------r-----~-------r------,

20 40 60 80 100

% Time on ground
Figure t. Modeled daily energy expenditure of bipeds and quadrupedal apes. The vertical axis shows total energy
expenditure across all activities (feeding, traveling, resting, socializing). The horizontal axis shows increasing per-
centage of time spent on the ground. As time spent terrestrially increases, overall energy expenditure decreases,
but that of bipeds at a faster rate. All graphs use 20% travel time. The top graph shows the situation when 20% of
the time is spent feeding (relatively good conditions); the middle graph when 30% is spent feeding, and the bottom
when 40% is spent feeding (very harsh conditions).
428 R. A. Foley and S. Elton

of increasing feeding time for the three model creatures, with traveling time held constant
at 20%. As can be seen, for all three (modem humans, early bipeds, and chimpanzee-like
quadrupeds), the costs offoraging decrease with increasing time spent on the ground. This
reflects the greater costs of arboreal feeding in the model. When most of the feeding (and
proportionately, travel) is done in the trees, then the chimpanzee has a considerable ener-
getic advantage. As the percentage of time spent on the ground increases the gap between
the bipeds and the quadrupeds closes, and eventually the bipeds have an energetic advan-
tage. From the point of view of the context for the evolution of bipedalism, the question is,
at what stage does the cross-over occur. The answer is when more than 60% of activities
are spent on the ground for a modem biped, and more than 70% for a less efficient biped.
This is the case for a relatively inactive biped (20% feeding time), and increasing the
amount of feeding time raises the overall levels of energy expenditure, but has little effect
on the point at which the cross"over occurs relative to an equally active quadruped. Even a
less efficient biped will have an energetic advantage when the time it spends on the
ground exceeds 75%.

4.2. Traveling Time

The effects in relation to increasing traveling time are similar (Figure 2). The transi-
tion from an advantage to quadrupeds to an advantage to bipeds again occurs when over
60% time is spent on the ground for modem human levels of efficiency, and at 75% for a
less efficient biped. Similarly, as travel time increases (now holding feeding time constant
at 20%), overall levels of energy expenditure rise for all model creatures.

4.3. Integrated Patterns

What these results indicate is that, other things being equal, the critical zone, in
terms of time budgets, for the advantage shifting from a quadruped to a biped occurs when
terrestrial activity lies between 60 and 80% (Figure 3). Bringing in other factors may bring
this value down, but the transition point does appear to be relatively stable for the range of
daily activities a typical social primate is likely to employ.

5. DISCUSSION

It should be remembered that these are models based on limited energetic data and a
firm application of the principle of uniformitarianism; nonetheless, some interesting impli-
cations arise from considering bipedalism in the context of time budgets.
The first of these is that the results hinge on there being an advantage to bipedalism
as a mode of locomotion on the ground, and an increased cost when arboreal activity is in-
volved. The first of these has been relatively well documented (Taylor and Rowntree,
1973; Rodman and McHenry, 1980; Foley, 1992; Steudel, 1994; Leonard and Robertson,
1997). Climbing for a biped is energetically expensive, but at this stage we do not know
whether this is very much greater than for a chimpanzee-like animal. Biomechanical prin-
ciples would seem to indicate that chimpanzees would be able to climb more efficiently,
and they certainly have major advantages in terms of speed, opportunity costs, and risk of
accidents. Given as an assumption the additional climbing costs associated with bipedal-
ism, it is interesting, if not unexpected, to note that higher levels of terrestrial activity pro-
mote bipedalism.
Time and Energy: The Ecological Context for the Evolution of Bipedalism 429

8000 20% traveling time Early biped (30 kg)

7000 Modern human (30 kg)

Quadrupedal ape (30 kg)
6000

5000 .....""..~.::::.::::.::::.:::::.::::::::::.:::::.---
4000 ..................:::..::::.::::.::::...::::..

3000+-----.----r-----r----.
20 40 60 80 100

--
~
8000 30% traveling time
7000

6000
...•::::..~.:::::.::::.:::::.::::.:::::.::::.:::::.::::.--
5000 r-------..:::::::...:::.;;;..;..::
.....~.....-.
.. _::::-:-_--
.............::.::.:....
4000

3000+----,---,----,---,
20 40 60 80 100
8000 40% traveling time

::::,-..-.~-. .~-. .-.~-. ~-..-.~-..-~-.~-. .-~-. ~=. .~~~. ~:;,-~-~~~------

.....................::.---- ..............
5000 ..•..•...•..•....•••.

4000

3000+----.------..----r------.
20 40 60 80 100
% Time on ground
Figure 2. Modeled daily energy expenditure of bipeds and quadrupedal apes. The vertical axis shows total energy
expenditure across all activities (feeding, traveling, resting, socializing). The horizontal axis shows increasing per-
centage of time spent on the ground. As time spent terrestrially increases, overall energy expenditure decreases,
but that of bipeds at a faster rate. All graphs use 25% feeding time. The top graph shows the situation when 20%
of the time is spent traveling (relatively good conditions); the middle graph when 30% is spent traveling, and the
bottom when 40% is spent traveling (very harsh conditions).

The significance of the model lies in the fact that when time budgets are considered,
some level of quantification is possible. A time budget sets constraints on the level of ener-
getic advantage that is required. Although other things may have varied in the past in ways
that we cannot comprehend today, ultimately all diurnal primates are limited by what they
can achieve in a twelve hour period of daylight. A social primate must spend time maintain-
430 R. A. Foley and S. Elton

8000 -

:; 7000
~
CII

~ 6000
"C
c:
CII
a.
)(
CII 5000
>.
~ tlmlts ~ •.
CII
c: Or ve/}' Inact/v
w 4000 eande'"
'Jlclentb/
Peel

3000 -+----....--------,-----,--------r
20 40 60 80 100

% Time on ground

Figure 3. Summary graph. The upper limit for the quadrupedal ape is set by a large male chimpanzee under con-
ditions of high energy expenditure, the lower limit by a small female chimpanzee under low energy expenditure
conditions. The two lines for the biped set the outer limits to energy expenditure. The critical zone, derived from
the model discussed in text, is where the switch to bipedalism is most likely to be beneficial.

ing social relationships, and an overly active animal is likely both to run into energy deficit
and to increase the probability of accidents and predation, and hence there are limits to the
amount of time that can be spent feeding and traveling. Those limits set the real ecological
constraints within which the costs and benefits of bipedalism should be considered.
This model specifies a link between the habitat and the behavior of early hominids
that may go some way towards resolving the discussions about bipedalism and habitat.
The clearest result shown here is that bipedalism does have a significant advantage over
quadrupedal ism, but only when a very considerable amount of time is spent on the ground.
As long as the ancestral populations were feeding and traveling in the trees for as much as
40% of the time, then bipedalism will be more of a cost than a benefit. In other words, it
would take a predominantly terrestrial life to lead to bipedalism, unless other factors came
into play. Furthermore, traveling time has a greater affect than feeding time on the switch
to bipedalism with increased terrestriality, and thus if traveling time is high and feeding
time low, then bipedalism might develop if most of the traveling is on the ground, as
might be the case in scattered woodland.
What implications does this model have for our thinking on the evolution of bipedal-
ism among Plio-Pleistocene hominids? The first is that the major change is a behavioral
one. The strategy of feeding on the ground is likely to be independent of bipedalism for a
considerable period of time. As long as the ancestral populations were still feeding at least
half the time in the trees, then bipedalism will be more of a cost than a benefit. By the
time bipedalism does evolve then it is likely that the hominids would have been well es-
tablished as very competent terrestrial primates. Although evolutionary change can occur
very rapidly, it is likely that a long history of terrestrial activity may precede the first evi-
dence of bipedalism in the fossil record.
The second implication is that it is traveling time that is critical here. The model did
not explore in detail minor variations in time spent differentially on the ground when feed-
Time and Energy: The Ecological Context for the Evolution of Bipedalism 431

ing or traveling, but nonetheless it is the traveling time that is likely to be energetically
critical. What might be significant is not the relative amounts of time, but the overall in-
crease in the percentage of time spent both feeding and traveling at the expense of resting
and socializing time. Dunbar (1992) has shown that in hotter and drier environments ba-
boons will spend more time traveling, and that this is time taken from that available for
resting or socializing. Thus, a hominid living in a more arid environment would spend
more time traveling in search of food and traveling between dispersed patches. Overall,
these populations are likely to be more active and to have larger day ranges than forest
dwelling apes. The hypothesis that the energetic advantages of bipedalism are brought into
play primarily by increased day range lengths was proposed by one of us (Foley, 1991),
and has been supported by some recent reanalyses (Leonard and Robertson, 1997). These
time budget models support the hypothesis that the energetic advantages of bipedalism are
especially important in the context of increased day range lengths (Foley 1992), when
time stress can become a critical problem. It is interesting to note that a maximum day
range length of around 16 km arises from the day length model (Foley, 1992), the time
budget model presented here, and Leonard and Robertson's (1997) analyses.
A third implication of some importance is that while bipedalism may be the primary
observable change at the divergence of hominids from the other African apes, it may be a
consequence of another and equally significant change. Terrestriality is less energetically
costly than arboreality, but it still takes time to forage and, furthermore, food is likely to
be more widely dispersed. The key change might therefore be the ability to spend more
time in feeding and traveling behavior, in order to be generally more active across the
course of a day (Foley, 1995). This higher level of activity is the sort of pressure that
might also integrate well with the idea that early hominids have very high thermoregula-
tory costs and that thermoregulation has played a major role in shifting hominid anatomy
and physiology (Wheeler, 1984, 1985, 1991 a,b).
Finally, it may be worthwhile returning to the larger time scale of hominid evolution
to discuss the locomotor and paleoenvironmental evidence for Plio-Pleistocene hominids
in the light of these time budget models. As discussed earlier, the very early Pliocene
hominid, Ardipithecus ramidus, has not been published in sufficient detail to determine its
locomotor repertoire, but there is convincing evidence that the other three taxa of early
hominids----A. anamensis, A. afarensis, and A. africanus-were neither fully bipedal like
modern humans, nor exhibited chimpanzee-like quadrupedalism. They perhaps approxi-
mated the inefficient small-bodied biped modeled here. If so, it can be inferred that they
were likely to have been spending at least 65% of their time on the ground. It has been ar-
gued that they may have been living in substantially wooded environments, but while this
may have been the case, it is unlikely that this reflects specialized arboreality. Put the
other way, however, the level of bipedalism found in the early australopithecines is consis-
tent with as much as 35% of daily activity involving time spent in the trees. Later australo-
pithecines, with greater robusticity and megadontic specializations, in addition to
relatively good evidence associating them with more open habitats, are, on the basis of
these models, expected to be specialist bipeds, a conclusion consistent with the known
anatomical evidence. Homo ergaster is anatomically fully bipedal, and it would be in-
ferred to have been a complete terrestrial specialist with a time budget that would reflect
this. Perhaps the greatest remaining area of uncertainty is whether early representatives of
Homo, living around 2.5 Myr, would have still retained the pattern found in the earlier
australopithecines, or whether they already possessed the derived condition found in both
later Homo or the robust australopithecines. Certainly models of land use at Olduvai,
which have been linked to Homo habilis, imply a terrestrial way of life (Peters and
432 R. A. Foley and S. Elton

Blumenschine, 1995), but, as this paper has shown, terrestriality may be more widespread
than bipedalism.

6. CONCLUSIONS

Simulation models should always be treated with great caution, for they simulate our
hypotheses, not the past. Nonetheless, the model developed and discussed here does per-
haps provide some new insights into the evolution of bipedalism. Stress was laid on the
importance of context, for it is only in relation to very specific contexts that the costs and
benefits of any adaptation can be assessed. It was argued that the context for bipedalism
was ecological and, in particular, the way in which extinct hominoid populations deployed
their activities across a day-in other words, their time budget. By specifying the time
budget for a twelve hour set of activities it was possible to take into account not just the
benefits of bipedalism, but also the costs oflosing climbing ability.
The time and energy model examined how the percent of time spent on the ground
influenced the adaptive value of bipedalism in relation to the relative and absolute amount
of time spent feeding and traveling. The principal conclusion drawn was that at least 60%
of daily activities would have to be spent terrestrially before the energetic advantages of
bipedalism outweighed the loss of climbing ability. This result implies that, other things
being equal, hominids may well have had a long ancestry of terrestrial activity prior to the
evolution of bipedalism. Extensive foraging on the ground, and traveling larger distances
with greater day ranges, even in relatively closed habitats, may have been the ecological
heritage of the first hominids, and the essential pre-requisite for successful and ultimately
bipedal adaptation to drier and more open environments.

ACKNOWLEDGMENTS

We thank P.C. Lee for comments, Charles Fitzgerald for help with the computer
models, and RAF is grateful to Henry McHenry and Elizabeth Strasser for the invitation to
contribute to the conference in Davis. Elizabeth Strasser, Kevin Hunt and a number of
anonymous reviewers provided helpful comments on an earlier draft.

REFERENCES
Aiello LC, and MC Dean (1990) An Introduction to Human Evolutionary Anatomy. London: Academic Press.
Clarke RJ (1985) Alistralopitheclis and early Homo in southern Africa. In EDelson (ed.): Ancestors: the Hard Evi-
dence. New York: Alan Liss, pp. 171-177.
Darwin C (1871) Descent of Man and Selection in Relation to Sex. London: Murray.
Dunbar RI M ( 1992) Time: A hidden constraint on the behavioral ecology of baboons. Behav. Ecol. and Sociobiol.
31:35-49.
Elton S, Foley RA, and Ulijaszek SJ (in press) How much does it cost a human biped to climb and clamber? Ann.
Hum. BioI.
Foley RA (1987) Another Unique Species: Patterns of Human Evolutionary Ecology. Harlow: Longman.
Foley RA (1992) Evolutionary ecology of fossil hominids. In EA Smith and B Winterhalder (eds.): Evolutionary
Ecology and Human Behavior. Chicago: Aldine de Gruyter, pp. 131-164.
Foley RA (1995) Humans Before Humanity: An Evolutionary Perspective. Oxford: Blackwells Publishers.
Grine FE, Editor (1989) The Evolutionary History of the "Robust" Australopithecines. Chicago: Aldine de
Gruyter.
Time and Energy: The Ecological Context for the Evolution of Bipedalism 433

Hunt KD (1996) The postural feeding hypothesis - an ecological model for the evolution of bipedalism. S. Afr. J.
Sci. 92:77-90.
Hunt KD (this volume) Ecological morphology of Australopithecus afarensis: Traveling terrestrially, eating ar-
boreally. In E Strasser, JG Fleagle, AL Rosenberger, and HM McHenry (eds.): Primate Locomotion: Recent
Advances. New York: Plenum Press, pp. 397-418.
Isbell LA, and Young TP (1996) The evolution of bipedalism in hominids and reduced group-size in chimpanzees -
alternative responses to decreasing resource availability. J. Hum. Evo!. 30:389-397.
Jablonski NG, and Chaplin G (1993) Origin of habitual terrestrial bipedalism in the ancestor of the Hominidae. J.
Hum. Evo!. 24:259-280.
Jungers WL (1982) Lucy's limbs: Skeletal allometry and locomotion in Australopithecus afarensis. Nature
297:676-678.
Leakey MG, Feibel CS, McDougall I, and Walker A (1995) New four million year old hominid species from
Kanapoi and Allia Bay, Kenya. Nature 376:565-571.
Leonard WR, and Robertson ML (1997) Rethinking the energetics of bipedalism. Curro Anthropo!. 38:304-309.
Lovejoy CO (1979) A reconstruction of the pelvis of AL-288 (Hadar Formation). Am. 1. Phys. Anthropo!. 50:460.
Lovejoy CO (1980) Hominid origins: The role of bipedalism. Am. J. Phys. Anthropo!. 52:250.
Maynard Smith J (1978) Optimization theory in evolution. Ann. Rev. Eco!. Syst. 9:31-56.
Newman RW (1970) Why man is such a sweaty and thirsty naked animal: A speculative review. Hum. Bio!.
42:12-27.
Peters CR, and Blumenschine R (1995) Landscape perspectives on possible landuse patterns for early hominids in
the Olduvai Basin. J. Hum. Evo!. 29:321-362.
Reader J (1988) Missing Links. London: Pelican.
Reed KE (1997) Early hominid evolution and ecological change through the African Plio-Pleistocene. J. Hum.
Evo!. 32:289-322.
Rodman PS (1984) Foraging and social systems of orangutans and chimpanzees. In PS Rodman and JGH Cant
(eds.): Adaptations for Foraging in Non-Human Primates. New York: Columbia University Press, pp.
134-160.
Rodman PS, and McHenry HM (1980) Bioenergetics and origins of bipedalism. Am. J. Phys. Anthropol
52: I 03-\06.
Senut B, and Tardieu C (1985) Functional aspects of Plio-Pleistocene hominid limb bones: Implications for phylo-
geny and taxonomy. In EDelson (ed.): Ancestors: The Hard Evidence. New York: Alan Liss, pp. 193-20 I.
Steudel K (1994) Locomotor energetics and hominid evolution. Evo!. Anthropo!. 3:42-48.
Susman RL, and Brain TM (1988) New first metatarsal (SKX50 17) from Swartkrans and the gait of Paranthroplls
robustus. Am. J. Phys. Anthropol 77:7-16.
Taylor CR, and Rowntree VJ (1973) Running on two legs or four: Which consumes more energy. Science
179: 186-187.
Tuttle RH, Hallgrimsson B, and Stein T (this volume) Heel, squat, stand, stride: Function and evolution of hom i-
noid feet. In E Strasser, JG Fleagle, AL Rosenberger, and HM McHenry (eds.): Primate Locomotion: Re-
cent Advances. New York: Plenum Press, pp. 435--448.
Ulijaszek SJ (1995) Human Energetics in Biological Anthropology. Cambridge: Cambridge University Press.
Walker AC, and Leakey RE, Editors (1993) The Nariokotome Skeleton. Cambridge: Harvard University Press.
Wheeler P (1984) The evolution of bipedality and loss of functional body hair in hominids. J. Hum. Evo!.
13:91-98.
Wheeler P (1985) The loss of functional body hair in man: The influence of thermal environment, body form and
bipedality. J. Hum. Evo!. 14:23-28.
Wheeler PE (1991a) The influence of bipedalism on the energy and water budgets of early hominids. J. Hum.
Evo!. 21:117-136.
Wheeler PE (199Ib) The thermoregulatory advantages of hominid bipedalism in open equatorial environments:
The contribution of increased convective heat loss and cutaneous evaporative cooling. 1. Hum. Evo!.
21:107-116.
White TD, Suwa G, and Asfaw B (1994) Australopithecus ramidlls, a new species of early hominid from Aramis,
Ethiopia. Nature 371 :306-312.
Williamson D (1997) Hominid Socioecology. Ph.D. Dissertation, University College London.
 22

HEEL, SQUAT, STAND, STRIDE

Function and Evolution of Hominoid Feet

Russell H. Tuttle, I Benedikt Hallgrimsson,2 and Tamara Stein I

IDepartment of Anthropology
The University of Chicago
1126 East 59th Street
Chicago, Illinois 60627
2Department of Anatomy
University of Puerto Rico
GPO Box 5067
San Juan, Puerto Rico 00936

1. INTRODUCTION

Primate feet are remarkably diverse due to natural selection for a notable variety of
positional behavior in a wide spectrum of arboreal and terrestrial niches (Schultz, 1963).
Although positional behavior embraces both posture and locomotion, special features of
primate feet are customarily related almost exclusively to locomotor adaptations and be-
havior, with posture treated secondarily or not considered at all (Tuttle et aI., in press).
Based on comparative functional morphological studies of extant apes and Pliocene-Re-
cent hominids, we reason that squatting and bipedal standing were important components
of the selective complex that produced the human foot, which has been associated more
commonly with bipedal locomotion (Latimer and Lovejoy, 1989) instead of posture per se.

2. HEELS

Asian apes--orangutans and gibbons--have relatively modest heels, whereas hu-

mans and the African apes-gorillas, chimpanzees and bonobos-sport robust heels
(Schultz, 1963; Sarmiento, 1983, 1994; Figure 1). This difference is commonly ascribed to
the arboreality of Asian apes versus the terrestriality of African apes and people, with par-
ticular emphasis on locomotor differences among them (Tuttle, 1970, 1972).

435
436 R. H. Tuttle et al.

Figure 1. Medial views of left feet of an

orangutan (a), a chimpanzee (b), and a gorilla
(c). The African apes (b and c) have extensive
passive excursions of metatarsophalangeal
joints II-V (Tuttle, 1970), which would facili-
tate human-like bipedal locomotion if the toes
were reduced in length.

Our feet must bear the entire bodily mass; they are on the front line against ground
reaction forces when we stand, squat, crouch and move. Their peculiar morphology re-
flects the special demands that are placed upon them (Carrier et ai., 1994).
As we stand, our weight is evenly balanced between the feet (Nicole and Paul, 1988).
Within each foot, approximately half of each pedal load falls on the heel and the other half
Heel, Squat, Stand, Stride 437

a
! c

Figure 2. Bipedal right footprints of a gibbon (a), a chimpanzee (b), and a human who had never worn footgear
(e). Note the absence of a heel impression in the gibbon and the flat-footed impression of the chimpanzee versus
the arched plantigrade foot of the human.

is distributed among six contact points under the metatarsal heads. Two sesamoid bones be-
neath the first metatarsal head and the heads of the lateral four metatarsals share the load on
the distal foot during stance. The axis of balance of the foot bisects the heel and sole longi-
tudinally before passing between the second and third toes (Morton, 1935). Squatting can
shift weight from the pedal ball to the heel, thereby eccentrically loading it.
When we walk, the pedal load shifts medially so that the leverage axis falls between
the great and second toes (Morton, 1935). After heel strike, one's weight shifts to the ball
of the foot during the first third of the gait cycle (Nicol and Paul, 1988). Accordingly, hu-
man footprints are characterized by prominent heel and ball impressions (Figure 2). Be-
cause weight shifts medially during stance phase, the great toe is the final contact of the
foot before it swings clear of the ground. The hallucal pad is usually prominent in human
footprints, but commonly the lateral four toes leave fainter impressions. Because of the
medial longitudinal arch, much of the central sole is unimpressive in human footprints.
The lateral border of the sole may be evident, but lightly, due to the quick shift from the
heel to the medial ball as the foot acts as a compact lever.
Running is the supreme mechanical challenge to the ball and toes, especially when
we sprint quasi-digitigrade. Long, down-curved toes would be an impediment to running
hominids who might attempt to orient them strategically in the direction of travel.
The large human heel serves as a powerful lever for the triceps surae muscle, which
plantarflexes the foot at the ankle joint during bipedal walking and running (Basmajian
and De Luca, 1985). But, the robust heel probably also evolved to serve postural func-
tions: squatting and standing.

3. SQUATTING AND BIPEDALISM

African apes, particularly gorillas, also have robust heels (Schultz, 1963), yet they
rarely engage in bipedal locomotion. They squat for long periods of time, however, to for-
age and to rest (Tuttle and Watts, 1985; Remis, 1995).
438 R. H. Tuttle et aL

Squatting and bipedalism, versus sitting, increase the height that foragers can reach
overhead and keep the rump, thighs and legs off wet and otherwise uncomfortable sub-
strates. Moreover, squatters are better prepared for locomotion than were they sitting or
reclining. Accordingly, we should consider the possibility that robustness of the heel
evolved in our lineage in response to squatting, bipedal foraging, and short-distance
bipedal travel before early hominids were fully adapted to obligate bipedalism and long-
distance travel. In this scenario, the toes may have retained arboreal features while the
heel and ankle joint were more truly humanoid (Tuttle et aI., in press; Tuttle, submitted).

4. HOMINOID FOOT POSTURES

Most primates are semiplantigrade, a posture that is facilitated by mobile transverse

tarsal joints (Gebo, 1993a). They stand and walk with the proximal heel raised above the
substrate (Morton, 1935; Gebo, 1992). Accordingly, the African apes and people are ex-
ceptional in habitually exhibiting full contact of the plantar surface of the heel during
quadrupedal and bipedal walking and squatting.
Like catarrhine monkeys and prosimians, gibbons generally elevate the heel when
moving rapidly bipedally on the ground (Morton, 1935; Tuttle et aI., 1992; Gebo, 1993b),
and probably also on branches. In arboreal contexts, gibbons must grasp the substrate with
their toes, which is accompanied by flexed knees and hips and an elevated proximal heel
(Tuttle, 1972). Schmitt and Larson (1995) reported that 3 captive lar gibbons exhibited
mid-foot/heel contact-the heel touched the substrate after or simultaneously with the
midfoot-as they walked bipedally at unspecified speeds on a variety of branches of un-
specificed diameter. Moreover, 2 adult lar gibbons evidenced complete heel contact, fol-
lowing mid-foot strike, in >85% of cases as they walked bipedally on the ground in an
experimental setting (Schmitt and Larson, 1995). Contrarily, Tuttle et al. (1992) recorded
no heel contact among 12 grounded bipedal captive lesser apes (11 lar gibbons and 1 si-
abon, 2.5-9 years-old), which moved rapidly (1.175 ± 0.589 m/sec) on pressure-sensitive
paper runners (TMShutrak).
As usual, orangutans are unique and controversial (Tuttle and Cortright, 1988).
Gebo (1992, 1993b, 1996) claimed that although grounded orangutans exhibit heel-strike
plantigrady, they are not truly plantigrade sensu African apes and humans, while Meldrum
(1993) countered that they probably are. Schmitt and Larson (1995) reported that 5 ambu-
lating orangutans exhibited heel-strike at the end of swing phase.
The variability of facultative terrestrial foot postures among orangutans confounds
those who would classify them neatly. The combination of grossly elongate, ventrally
curved second-to-fifth toes and a puny hallux permits a spectrum of postures, ranging
from one in which only the lateral aspect of the sole and fifth digit are on the ground to
full plantigrade placement of the heel and extension of digits II-V, with the hallux serving
as a strut. When standing bipedally, the heels of captives may be respectably plantigrade,
or the subject may rise onto toe-tips so that the heel is free of the floor (Tuttle, 1970; Tut-
tle and Cortright, 1988; Figure 3).
Orangutans seem not to emulate the characteristic semiplantigrade posture of cer-
copithecoid monkeys and gibbons. Moreover, although, like orangutans, Visoke mountain
gorillas sometimes tightly flex their toes and invert their feet to walk on the lateral soles,
instead of being fully plantigrade (Tuttle and Watts, 1985), we do not exclude them from
the plantigrades.
Heel, Squat, Stand, Stride 439

Figure 3. Orangutan foot postures. Walking on the lateral

side of the foot and fifth toe (a, b) (note plantigrade heel in
a), standing bipedally with heels plantigrade, toes ex-
tended, and halluces widely abducted (c).

5. HEELS, SQUATTING, AND BIPEDALISM

Habitual terrestrial bipedalism is unique to humans among hominoid primates. Ac-

cordingly, the large heels and plantigrade postures of African apes cannot be explained as
adaptations to bipedalism.
There is no apparent feature of African pongid terrestrial quadrupedism that would
require a large heel and plantigrade posture. A prominent proximal calcaneus provides
purchase for the triceps surae muscle, which plantarflexes the ankle. Our EMG studies re-
vealed marked activity in the triceps surae muscle when chimpanzee and gorilla subjects
jumped to reach incentives overhead and while the chimpanzee stood bipedally on toe-
tips. This mechanism is probably important also during arboreal vertical climbing of both
species (Tuttle et aI., in press).
440 R. H. Tuttle et al.

The triceps surae muscle, however, showed only low or nil EMG potentials as the go-
rilla and chimpanzee stood and walked quadrupedally. Accordingly, we infer that the promi-
nent calcanean tubers of African apes are not primarily an adaptation for terrestrial
quadrupedism (Tuttle et ai., in press). Instead, selection for a secure squatting platform on the
ground and on large boughs may have been an important factor in the evolution of their feet.

6. BEHAVIOR
Watts found that, on average, Visoke mountain gorillas squat terrestrially 3l.6% of
feeding time and arboreally 3.8% of feeding time. They sit nearly twice as much (60%) as
they squat while feeding on the ground, but they squat predominantly while feeding in
trees. Indeed, a silverback never sat as he fed in trees, and squatting accounted for 86% of
arboreal feeding postures overall among the Visoke group. Since Visoke gorillas devote
approximately 60% of the average daily activity period to feeding, we may assume that
their heels bear considerable loads over the course of the day (Tuttle and Watts, 1985).
Watts noted that Visoke gorillas rarely run. Running is virtually restricted to play
and agonistic behavior. Bipedalism is also rare. Bipedally, they ran more than they walked
over short distances. Bipedal feeding and foraging are uncommon in trees and on the
ground. Silverbacks, however, stand bipedally on the ground to collect important food
sources overhead, while smaller gorillas climb for them (Tuttle and Watts, 1985).
At the Bai Hokou Study Site in the Central African Republic, western gorillas squat-
ted arboreally more in the wet season than in the dry season (Remis, 1995). During arbo-
real positional behavior, squatting accounted for 29% in females, 26% in lone males, and
18% in group males. In the dry season, arboreal squatting was rare: 3% of female posi-
tional behavior and 7% of group male positional behavior. Unlike Visoke gorillas, those at
Bai Hokou sat more than they squatted arboreally: in the wet season, females, 36%, lone
males, 40%, and group males, 57%, and in the dry season, females, 63%, and group males,
82%. Visibility was too poor for Remis (1995) to obtain reliable data on terrestrial squat-
ting, sitting and bipedalism.
Bai Hokou gorillas exibited low levels of arboreal bipedalism: 0% of arboreal posi-
tional behavior in the dry season; and in the wet season, females, 6%, lone males, 5%, and
group males, 3% of arboreal positional behavior.
Stein's (1995) 8-week (30-hr) focal-animal sampling (30-min intervals) and scan sam-
pling study (::;2 min each 10 min; Table 1) of positional behavior in a group of9 western go-
rillas at the Brookfield Zoo revealed frequencies of squatting that are similar to those from
the Bai Hokou Study Site. The Brookfield group included 1 silverback (38 yr), I black-
backed male (9 yr), 3 adult females (12-33 yrs), and 4 immature subjects (2.8---6 yrs). Most
of their postural behavior occurred on the horizontal cement floor and spacious platforms of
their holding area (21 hrs) or on the hilly floor and low, broad boughs in the exhibit (9 hrs).
We treated Stein's data on the Brookfield gorillas together regardless of sex or age
of the individual. To facilitate comparison of the Brookfield and Bai Hokou gorillas, we
combined Remis's (1995) data for each positional activity regardless of sex, social group-
ing or season, i.e., we recalculated frequencies of positional activity based on frequencies
and sample sizes in Remis's (1995) tables 8 and 10.
The Brookfield gorillas squatted 12% of the observation period versus 21 % by Bai
Hokou gorillas. While feeding, the Brookfield gorillas squatted 35% of the time, which is
quite similar to the frequency of squatting while feeding (39%) by Bai Hokou gorillas.
Further, Chi-square tests indicate that squatting among the Brookfield gorillas is signifi-
cantly associated with feeding behavior (X 2=46, N=229, P<O.OOI; Table 1).
Heel, Squat, Stand, Stride 441

Table 1. Frequencies of positional behaviors during all observations compared to during

only feeding behavior'
All behavior Feeding behavior
Observed group Squat Sit Bipedal Other Squat Sit Bipedal Other
Brookfield Zoo 12 53 6 29 34 59 03 6
Gorillas2 N=229 N=34
Bai Hokou 21 48 4 27 39 53 3 5
Gorillas4 N=1464 N=432
Tanzanian 29 346 0.4 37 44 43 NA 13
Chimpanzees 5 N=16303 N=4666
Visoke NA NA NA NA 35 60 0 5
Gorillas7 N=NA
I Frequencies are given in percentages (%).
20ata from Stein (1995). Chi-square tests indicate that squatting is significantly associated with feeding behav-
ior ()(2=46, P<O.OOO I).
3Because this is a zoo population, we would not expect bipedal feeding behavior as per Hunt (1991).
4 0ata from Remis (1995, Tables 8 and 10). Observations were of arboreal positional behavior only.
50ata from Hunt (1991, Table 2; 1992, Table I). Hunt's "sit-(in)" category was added to his "squat" category to
calculate our "squat" category.
6Because we used "sit-(in)" as squatting behavior, sitting behavior may be underestimated while squatting be-
havior may be overestimated.
7 0ata from Tuttle and Watts (1985, Table 3). Terrestrial and arboreal behaviors were combined.

Hunt's (1991) subjects squatted much less than they sat: 0,7% versus 62% of all po-
sitional behavior, respectively. Moreover, forelimb-assisted squatting occurred most often
on vertical and nearly-vertical supports. Nonetheless, his category "sit (in)," which en-
tailed flexion of the hind limbs and accounted for 28% of positional behavior, probably
often loaded their heels, in addition to their ischial tuberosities. Accordingly, we disagree
with Hunt (1992) that sitting produces little stress and that squatting is too infrequent to
have affected chimpanzee adaptive morphology. The plantigrade heels of chimpanzees
may relate to their sitting with acutely flexed hind limbs and squatting on the ground and
on boughs (Tuttle et aI., in press).
Hunt (1994) concluded that among wild adult Tanzanian chimpanzees bipedalism is
a feeding adaptation, and that it is overwhelmingly postural: 86% of their arboreal bipedal
activity occurred while feeding; 70% of their terrestrial bipedal activity occurred while
feeding; 95% of their bipedalism was postural, and only 5% was locomotor. They most
commonly employed bipedalism to feed on small fruits of low, open-forest trees by reach-
ing either from the ground or from a lower branch in the tree, Hunt (1994) noted that by
standing bipedally on the ground, chimpanzees increase their foraging rate because both
hands can harvest to keep the mouth full.
Most of the 5% locomotor bipedalism entailed shuffling between fresh feeding sites
under fruiting trees. The Tanzanian chimpanzees rarely employed bipedalism during social
display (1%) or to scan the environment (2%) (Hunt, 1994). Doran (1993) noted that Tal
Forest chimpanzees also employ forelimb-unassisted bipedalism chiefly during terrestrial
foraging, e.g., to gather and to carry nuts short distances to terrestrial cracking sites.

7. FOSSIL HOMINID HEELS

The earliest evidence of robust heels in a fully plantigrade hominid foot is the 3.5-
Ma Laetoli Site G bipedal trackways (Clarke, 1979; Leakey and Hay, 1979; Day and
442 R. H. Tuttle et al.

Figure 4. Cast of Laetoli 0-1137 (a), left print of a Peruvian Indian in mud (b), and soles of a 14-year-old male
Peruvian Indian who had never worn footgear (c).

Wickens, 1980; White, 1980; Capecchi, 1984; Drake and Curtis, 1987; Harris, 1987;
Leakey, 1987; Robbins, 1987; Tuttle, 1987).
In all discernible features, the pedal morphology of the Laetoli track-makers is like
that of Homo sapiens (Tuttle, 1996; Figure 4). Their foot indices, which indicate foot
length versus foot breadth, fit comfortably within a global sample of human foot indices
(Tuttle, 1987). The hallux is aligned with the lateral four toes, and the gap between it and
the second toe is quite human, particularly when compared to undeformed feet of persons
who have never worn footgear (Tuttle et ai., 1990; Feibel et ai., 1996).
The lateral four toes are arrayed relative to the hallux and to each other as in a mod-
em human foot, and none extends notably beyond the tip of the hallux. The toes of Laetoli
G hominids are approximately 30% of total foot length, which is not significantly different
from mean relative toe length of never-shod Peruvian Indians and Tanzanian Hadzabe
(Tuttle et ai., 1991; Musiba et ai., 1997).
The Laetoli G hominid prints evidence a medial longitudinal arch. Apparently, the
transfer of body weight during bipedal walking was quite human-from robust heel strike,
more lightly along the lateral sole, then more heavily medially across the ball of the foot
so that the brunt of toe-off was borne by the hallux, which, unlike the lateral toes, regu-
larly left prominent impressions in the substrate. Contrarily, African pongid prints are flat-
footed with the toes arrayed quite differently from those of human prints (Tuttle et ai.,
1992; Figure 2).
Mean widths of heel impressions for two (G-l and G-3) of the three individuals that
walked in the moist volcanic ash at Laetoli Site G are 65.8 ± 5.4 mm and 75.9 ± 5.1 mm,
respectively (Tuttle, 1987), which is notably larger than mean heel breadths of Hadzabe
adults (58.8 ± 6.2 mm) and juveniles (50.9 ± 6.1 mm) (Musiba et ai., 1997).
It is likely that heel size of the Laetoli hominids is somewhat exaggerated by the
pedal impressions. Apparently, G-l placed its heels quite deliberately in the substrate; its
heel impressions are generally quite deep (Tuttle, 1996; Deloison, 1991, 1992).
Three 3-Ma calcanei (AL 333-8; AL 333-37; AL 333-55) from Hadar, Ethiopia, al-
low morphological studies and mechanical modelling of heels in one species of Pliocene
hominid: Australopithecus afarensis (Latimer et ai., 1982; Latimer and Lovejoy, 1989).
Heel, Squat, Stand, St~de 443

Latimer and Lovejoy (1989) found that 2 calcanean tubers (AL 333-8, AL 333-55)
sport sufficient robusticity and minimum volume to attest to notable bipedality in Hadar
Australopithecusafarensis. Their minimum cross-sectional areas fall above those ofchim-
panzees and gorillas, the latter of which were certainly much heavier than were the Hadar
hominids, and at the lower extremity of a human range, based on calcanei of persons who
had worn western footgear.
Minimum volumes of the calcanean tubers, expressed as a product of minimum
cross-sectional area and length, sharply separate the Hadar hominids from chimpanzees
and female gorillas, on the one hand, and male gorillas and humans, on the other. Indices
of minimum calcanean volume to body weight place the Hadar calcanean tubers closer to
those of humans than to those of African apes (Latimer and Lovejoy, 1989).
Accordingly, one may fairly conclude that the bony heels of Hadar Australopithecus
afarensis are relatively robust and tend toward the human condition in this feature. More-
over, statements to the contrary (Susman et aI., 1984; Deloison, 1985) notwithstanding,
the 3 Hadar calcanean specimens evidence not only a medial plantar process but also a lat-
eral plantar process, which is characteristic of humans and is absent in apes (Latimer and
Lovejoy, 1989).
Were Hadar Australopithecus afarensis obligately bipedal on the ground? Most
probably, but this inference is more compellingly supported by other features of the pelvic
limb (Tuttle, 1981; Johanson et aI., 1982; Lovejoy et aI., 1982; Latimer et aI., 1987; La-
timer and Lovejoy, 1990a; Coppens, 1991; Jungers, 1991; Langdon et aI., 1991; Latimer,
1991; McHenry, 1991; Vancata, 1991) than by the calcanean tuber per se. Hadar hominid
calcanean morphology is consistent with this model, but it does not unequivocally com-
mand commitment to it, since the very features that would facilitate bipedalism also en-
able terrestrial and arboreal squatting. Nevertheless, we readily a~knowledge bipedal
standing and short distance walking by Hadar Australopithecus afarensis because of the
total morphological pattern of their pelvic limbs.
Whether Hadar Australopithecus afarensis walked like we do and whether they
could crouch, sprint and course quasi digitigrade like an athletic modern person are moot
puzzles that cannot be resolved by calcanean studies alone. Within the foot, there is highly
suggestive evidence that no matter how posturally bipedal the Hadar Australopithecus
afarensis may have been, the bipedal component of their locomotion was not as advanced
as those of Homo sapiens and probably also those of Pleistocene populations of Homo, or
perhaps even those of penecontemporaneous Australopithecus sensu lato, who lived in
more open habitats.
Although Lamy (1986) concluded from talonavicular morphology that, like other
eastern African species of Plio-Pleistocene Hominidae, Hadar Australopithecus afarensis
had longitudinally-arched feet, Gomberg (1985; Gomberg and Latimer, 1984) concluded
from calcaneonavicular morphology that they did not have humanoid pedal arches. This
disagreement recalls earlier controversy regarding whether Homo habilis (OH-8) and
hominids from Koobi Fora had relatively compact humanoid tarsal arches (Day and
Napier, 1964; Day and Wood, 1968; Day, 1976; Lamy, 1983; Wood, 1974a,b, 1976) or
more mobile pongoid transverse tarsal joints (Oxnard, 1972, 1973, 1984; Lisowski, 1976;
Oxnard and Lisowski, 1980). On balance, we express reasonable doubt that Hadar Aus-
tralopithecus afarensis had fully humanoid transverse tarsal joints.
Further distally, the feet of Hadar Australopithecus afarensis depart more tellingly
from those of Homo sapiens. Latimer and Lovejoy (1990b: 23) reported that the orienta-
tion of the basal articular surfaces of the proximal phalanges, the potential dorsoplantar
excursions of the metatarsophalangeal joints, and the shape and orientation of the metatar-
444 R. H. Tuttle et al.

sal heads of digital rays II-V not only confirm "a dramatic commitment to terrestrial
bipedality" in Hadar Australopithecus afarensis but also "contravene any significant pedal
grasping. "
Contrarily, via more refined, quantitative studies, Duncan et al. (1994) demonstrated
that the metatarsophalangeal joints of Hadar Australopithecus afarensis are not as human-
oid as had been claimed by Latimer and Lovejoy (1990b). Indeed, Hadar Australopithecus
afarensis fall midway between African apes and humans in orientation of the articular sur-
faces of their pedal proximal phalanges; the potential dorsoplantar excursions of their
metatarsophalangeal joints are greater than estimates by Latimer and Lovejoy (1990b);
and, actual orientation of the metatarsal heads are unmeasurable due to taphonomic dam-
age and incompleteness of the Hadar fossils (Duncan et aI., 1994). Moreover, because Pan
troglodytes and Homo sapiens share dorsally oriented metatarsal heads and both Pan go-
rilla and Pongo pygmaeus sport plantar orientations of the metatarsal heads Duncan and
coworkers (1994) rightly question the diagnostic value of this feature to indicate bipedal-
ism versus prehensility in the Hominoidea.
We all agree that the degree of dorsiflexion suggested by the metatarsophalangeal
joints of Hadar Australopithecus afarensis would facilitate bipedal locomotion (Tuttle,
1981; Latimer and Lovejoy, 1990b; Duncan et aI., 1994). But, the degrees of metatarso pha-
langeal plantarflexion evidenced by Hadar Australopithecus afarensis-whether they be the
stenotic guesstimates of Latimer and Lovej oy (1990b) or the wider excursions predicted by
Duncan et al. (1994 )-would facilitate pedal prehension of arboreal substrates, particularly
trunks, boughs and larger branches. Further, the permissive metatarsophalangeal dorsiflex-
ion of Hadar Australopithecusafarensis may have served them well during climbing, reach-
ing overhead while bipedal, and walking short distances bipedally (Tuttle, 1981).
In Hadar Australopithecus afarensis, the proximal phalanges and, to a lesser extent
the middle phalanges, of pedal digits II-V are curved downward (Tuttle, 1981); some sport
prominent ridges for attachment of fibrous flexor sheaths (Latimer et aI., 1982); and, their
heads face plantarly, which accentuates overall curvature of the articulated digits. Further,
the toes were probably relatively long in comparison with those of modem human feet
(McHenry, 1986, 1991; Tuttle, 1988; Latimer and Lovejoy, 1990b).
All in all, the second-to-fifth toes of Hadar Australopithecus afarensis possessed
prehensile capacities that would serve them well in climbing, squatting and standing
bipedally on arboreal substrates, particularly in the absence of a markedly prehensile hal-
lux. The full range of diameters of the substrates that Hadar Australopithecus afarensis
could have comfortably grasped cannot be estimated reliably because we do not know
whether their metatarsals were elongate, relatively short, or intermediate in length. Al-
though pongoid grasps of twigs and small branches may have been problematic for Hadar
Australopithecus afarensis, their toes, soles and heels attest to versatility vis-a-vis larger
arboreal supports, probably in a positional repertoire that is not emulated closely by any
extant hominoid species.

8. SCENARIO

Our Miocene arboreal hylobatian ancestors probably had small or modest heels,
though other features of the hind limbs and torso predisposed their terrestrial descendents
to bipedalism instead of quadrupedism (Tuttle, 1994). Hunt (1994) argued persuasively
that certain features of Pliocene hominids--represented by Australopithecus afaren-
sis-are adaptations to foraging bipedally on small fruits in low open-forest trees from ter-
Heel, Squat, Stand, Stride 445

restrial and low-branch vantage points. We would add that squatting to forage and to feed
on the ground, to dig, to process hard-shelled foods and during rest probably also selected
for prominent heels and plantigrade postures in our Pliocene ancestors, notably before the
evolution of long-distance bipedal walking and sustained running in open habitats. Ac-
cordingly, in the early phases of hominid calcanean enlargement, ground reaction forces
acting on the heel during squatting and bipedal foraging activities should be viewed as a
major factor that supplemented traction of the triceps surae muscle, particularly during ar-
boreal climbing.
During squatting, the gastrocnemius muscle was probably silent due to acute flexion
of the knee and the soleus was probably not needed to maintain balance even though the
ankle may have been dorsiflexed. Speculation on this point is moot since proportions of
the lower limb segments would determine, to some extent, the distribution of weight on
the heel versus distal areas of the foot.
Reduction of pedal digits II-V, which is essential for athletic human running, is not
evidenced by Hadar Australopithecus afarensis, but toe length and absence of down-
curved digits is consistent with fully human bipedal locomotion by hominids that made
the 3.5-Ma footprint trails at Laetoli Site G, Tanzania (Tuttle, 1985, 1996; Tuttle et a!.,
1991; Musiba et a!., 1997). These observations challenge to the conspecific status of
Hadar Australopithecus afarensis and the Laetoli printmakers.
Although the Laetoli footprints proffer no evidence to the contrary, we resist the
temptation to send them racing across the savanna. Instead, we await the discovery of pel-
vic limb skeletons and additional trackways before deciding whether and how to set them
on a faster course than that evidenced by the site G trackways (Tuttle, 1987, 1994, submit-
ted; Tuttle et a!., 1990, 1992). We somewhat more confidently state that their feet appear
to be better adapted for terrestrial walking, perhaps over notable distances, than those of
Hadar Australopithecus afarensis.
Unfortunately, studies of squatting facets on tali and distal tibiae will not resolve
questions on the role of squatting in early hominid evolution. Their absence is insufficient
to deny frequent squatting, and their presence is consistent with other positional behav-
iors, particularly in active creatures like the Plio-Pleistocene Hominidae (Trinkaus, 1995).
Precisely when a completely human foot that functioned like ours evolved is unan-
swerable because of the paucity of pedal specimens of early Homo spp. that preceded the
Neandertalians, whose robust, virtually human feet are well-documented and attest to a lo-
comotor repertoire basically like ours (Trinkaus, 1983). Like modern African apes, Holo-
cene humans living in many different cultures, and probably also their Plio-Pleistocene
predecessors, the Neandertalians, commonly engaged their large heels in prolonged bouts
of squatting (Trinkaus, 1975).

ACKNOWLEDGMENTS

This investigation was supported by NSF grants GS-3209, SOC75-02478 and BNS
8540290, a Public Health. Service research career development award (l-K04-
GM 1634 7~ 1) from the National Institutes of Health, the Guggenheim Foundation,
Brookfield Zoo, and NIH grant RR-00165 from the Division of Research Resources to the
Yerkes Regional Primate Research Center, which is fully accredited by the American As-
sociation of Laboratory Animal Care. We are especially grateful for the assistance of J.
Malone, E. Regenos, J. Perry, Dr. G.H. Bourne, Dr. EA. King, R. Pollard, S. Lee, R.
Mathis, 1. Roberts, Dr. M. Keeling, Dr. M. Vitti, and 1. Hudson.
446 R. H. Tuttle et aL

R. Tuttle thanks D.C. Johanson and M.D. Leakey for opportunities to study the
Hadar postcranial specimens and the Laetoli footprints, respectively, and colleagues and
staffers of the British Museum (Natural History); Institut royal des Sciences naturelles de
Belgique, Brussels; Musee de I'Homme, Paris; Museum of the Department of Antiquities,
Jerusalem; National Museum of Geology and Palaeontology, Zhagreb; National Museums
of Kenya, Nairobi; Peabody Museum, Harvard University, Cambridge, MA; Rheinisches
Landesmuseum, Bonn; Rockefeller Museum, Jerusalem; Tel Aviv University, Tel Aviv;
Transvaal Museum, Pretoria; University of New Mexico, Albuquerque; and University of
the Witwatersrand Medical School, Johannesburg, who facilitated his studies on Plio-
Pleistocene hominid specimens in their keeping.

REFERENCES
Basmajian JV, and De Luca CJ (1985) Muscles Alive, 5th ed., Baltimore: Williams and Wilkins.
Capecchi V (1984) Reflections on the footprints of the hominids found at Laetoli. Anthropologischer Anzeiger
42:81--86.
Carrier DR, Heglund NC, and Earls KD (1994) Variable gearing during locomotion in the human musculoskeletal
system. Science 265:651--653.
Clarke RJ (1979) Early hominid footprints from Tanzania. S. Af. J. Sci. 75:148-149.
Coppens Y (1991) L'evolution des hominides, de leur locomotion et de leurs environnnements. In Y Coppens and
B Senut (eds.): Origine(s) de la Bipedie chez les Hominides, Cah. Paleoanthrop. Paris: Editions du CNRS,
pp.295-301.
Day MH (1976) Hominid postcranial remains from the East Rudolf succession: A review. In Y Coppens, FC How-
ell, GL Isaac, and REF Leakey (eds.): Earliest Man and Environments in the Lake Rudolf Basin. Chicago:
University of Chicago Press, pp. 507-521.
Day MH, and Napier JR (1964) Hominid fossils from Bed I Olduvai Gorge, Tanganyika: Fossil foot bones. Nature
201:967-970.
Day MH, and Wickens EH (1980) Laetoli Pliocene hominid footprints and bipedalism. Nature 286:385-387.
Day MH, and Wood BA (1968) Functional affinities of the Olduvai hominid 8 talus. Man 3:440-455.
Deloison Y (1985) Comparative study of calcanei of primates and Pan-Australopithecus-Homo relationship. In PV
Tobias (ed.): Hominid Evolution: Past, Present and Future. New York: Liss, pp. 143-147.
Deloison Y (1991) Les Australopitheques marchaient-ils comme nous? In Y Coppens and B Senut (eds.): Orig-
ine(s) de la Bipedie chez les Hominides, Cah. Palt~oanthrop. Paris: Editions du CNRS, pp. 177-186.
Deloison Y (1992) Empreintes de pas il Laetoli (Tanzanie). Leur apport il une meilleure connaissance de la loco-
motion des Hominides fossiles. C. R. Acad. Sci., Ser. 11,315:103-109.
Drake R, and Curtis GH (1987) K-Ar geochronology of the Laetoli fossil localities. In MD Leakey and JM Harris
(eds.): Laetoli: A Pliocene Site in Northern Tanzania. Oxford: Clarenden Press, pp. 48-52.
Doran DM (1993) Sex differences in adult chimpanzee positional behavior: The influence of body size on locomo-
tion and posture. Am. J. Phys. Anthropol. 91:99-115.
Duncan AS, Kappelman J, and Shapiro LJ (1994) Metatarsophalangeal joint function and positional behavior in
Australopithecus afarensis. Am. J. Phys. Anthropol. 93:67--81.
Feibel CS, Agnew N, Latimer B, Demas M, Marshall F, Waane AC, and Schmid P (1996) The Laetoli hominid
footprints - a preliminary report on the conservation and scientific restudy. Evol. Anthropol. 4: 149-154.
Gebo DL (1992) Plantigrady and foot adaptation in African apes: Implications for hominid origins. Am. J. Phys.
Anthropol. 89:29-58.
Gebo DL (l993a) Functional morphology of the foot in primates. In DL Gebo (ed.): Postcranial Adaptation in
Nonhuman Primates. DeKalb: Northern Illinois University Press, pp. 175-196.
Gebo DL (I 993b) Reply to Meldrum. Am. J. Phys. Anthropol. 91:382-385.
Gebo DL (1996) Climbing, brachiation, and terrestrial quadrupedalism: Historical precursors of hominid bipedal-
ism. Am. J. Phys. Anthropol. 101:55-92.
Gomberg DN (1985) Functional differences of three ligaments of the transverse tarsal joint in hominoids. J. Hum.
Evol. 14:553-562.
Gomberg DN, and Latimer B (1984) Observations on the transverse tarsal joint of A. afarensis. Am. J. Phys. An-
thropol. 63: 164.
Heel, Squat, Stand, Stride 447

Harris JM (1987) Summary. In MD Leakey and JM Harris (eds.): Laetoli: A Pliocene Site in Northern Tanzania.
Oxford: Clarenden Press, pp. 524-531.
Hunt KD (1991) Mechanical implications of chimpanzee positional behavior. Am. J. Phys. Anthropol.
86:521-536.
Hunt KD (1992) Positional behavior of Pan troglodytes in the Mahale Mountains and Gombe Stream National
Parks, Tanzania. Am. J. Phys. Anthropol. 87:83-105.
Hunt KD (1994) The evolution of human bipedality: Ecology and functional morphology. J. Hum. Evol.
26: 183-202.
Johanson DC, Taieb M, and Coppens Y (1982) Pliocene hominids from the Hadar Formation, Ethiopia
(1973-1977): Stratigraphic, chronologic, and paleoenvironmental contexts, with notes on hominid mor-
phology and systematics. Am. J. Phys. Anthropol. 57:373-402.
Jungers WL (1991) A pygmy perspective on body size and shape in Australopithecus afarensis (AL 288-1,
"Lucy"). In Y Coppens and B Senut (eds.): Origine(s) de la Bipedie chez les Hominides, Cah. Paleoan-
throp. Paris: Editions du CNRS, pp. 215-224.
Lamy P (1983) Le systeme podal de certains hominides fossiles du Plio-Pleistocene d' Afrique de I'est: etude mor-
pho-dynamique. L' Anthropologie (Paris) 87:435-464.
Lamy P (1986) The settlement of the longitudinal plantar arch of some African Plio-Pleistocene hominids: A mor-
phological study. J. Hum. Evol. 15:31-46.
Langdon JH, Bruckner J, and Baker HH (1991) Pedal mechanics and bipedalism in early hominids. In Y Coppens
and B Senut (eds.): Origine(s) de la Bipedie chez les Hominides, Cah. Paleoanthrop. Paris: Editions du
CNRS, pp. 159--167.
Latimer B (1991) Locomotor adaptations in Australopithecus cifarensis: The issue of arboreality. In Y Coppens and
B Senut (eds.): Origine(s) de la Bipedie chez les Hominides, Cah. Paleoanthrop. Paris: Editions du CNRS,
pp.169--176.
Latimer BM, and Lovejoy CO (1989) The calcaneus of Australopithecus afarensis and its implications for the evo-
lution ofbipedality. Am. J. Phys. Anthropol. 78:369--386.
Latimer BM, and Lovejoy CO (1990a) Hallucal tarsometatarsal joint in Australopithecus afarensis. Am. J. Phys.
Anthropol. 82:125-133.
Latimer BM, and Lovejoy CO (1990b) Metatarsophalangeal joints of Australopithecusafarensis. Am. J. Phys. An-
thropol. 83:13-23.
Latimer BM, Lovejoy CO, Johanson DC, and Coppens Y (1982) Hominid tarsal, metatarsal, and phalangeal bones
recovered from the Hadar Formation: 1974-1977 collection. Am. J. Phys. Anthropol. 57:701-719.
Latimer BM, Ohman JC, and Lovejoy CO (1987) Talocrurual joint in African hominids: Implications for Aus-
tralopithecus afarensis. Am. J. Phys. Anthropol. 74:155-175.
Leakey MD (1987) The hominid footprints. Introduction. In MD Leakey and JM Harris (eds.): Laetoli: A Pliocene
Site in Northern Tanzania. Oxford: Clarenden Press, pp. 490-496.
Leakey MD, and Hay RL (1979) Pliocene footprints in the Laetolil Beds, northern Tanzania. Nature 278:317-323.
Lisowski FP, Albrecht GH, and Oxnard CE (1976) African fossil tali: Further multivariate morphometric studies.
Am. J. Phys. Anthropol. 45:5-18.
Lovejoy CO, Johanson DC, and Coppens Y (1982) Hominid lower limb bones recovered from the Hadar Forma-
tion: 1974-1977. Am. J. Phys. Anthropol. 57:679--700.
McHenry HM (1986) The first bipeds: A comparison of the A. afarensis and A. africanus postcranium and implica-
tions for the evolution of bipedalism. J. Hum. Evol. 15: 177-191.
McHenry HM (1991) First steps? Analyses of the postcranium of early hominids. In Y Coppens and B Senut
(eds.): Origine(s) de la Bipedie chez les Hominides, Cah. Paleoanthrop. Paris: Editions du CNRS, pp.
133-141.
Meldrum DJ (1993) On plantigrady and quadrupedalism. Am. J. Phys. Anthropol. 91: 379--381.
Morton OJ (1935) The Human Foot. New York: Columbia University Press.
Musiba CM, Tuttle RH, Hallgrimsson B, and Webb DM (1997) Swift and sure-footed on the savanna: A study of
Hadzabe gaits and feet in northern Tanzania. Am. J. Hum. BioI. 9:303-321.
Nicol AC, and Paul JP (1988) Biomechanics. In B Helal and D Wilson (eds.): The Foot, Vol. I. Edinburgh: Chur-
chill Livingstone, pp. 75-86.
Oxnard CE (1972) Some African fossil foot bones: A note on the interpolation of fossils into a matrix of extant
species. Am. J. Phys. Anthropol. 37:3-12.
Oxnard CE (1973) Form and Pattern in Human Evolution. Chicago: University of Chicago Press.
Oxnard CE (1984) The Order of Man. New Haven: Yale University Press.
Oxnard CEo and Lisowski FP (1980) Functional articulation of some hominoid foot bones: Implications for the
Olduvai (Hominid 8) foot. Am. J. Phys. Anthropol. 52:107-117.
448 R. H. Tuttle et al.

Remis M (I 995} Effects of body size and social context on the arboreal activities oflowland gorillas in the Central
African Republic. Am. J. Phys. Anthropol. 97:413--433.
Robbins LM (1987) Hominid footprints from Site G. In MD Leakey and JM Harris (eds.): Laetoli: A Pliocene Site
in Northern Tanzania. Oxford: Clarenden Press, pp. 497-502.
Sarmiento EE (l983) The significance of the heel process in anthropoids. Int. J. Primatol. 4: 127-152.
Sarmiento EE (1994) Terrestrial traits in the hands and feet of gorillas. American Museum Novitates, no. 3091, 56
pp.
Schmitt D, and Larson SG (1995) Heel contact as a function of substrate type and speed in primates. Am. 1. Phys.
Anthropol. 96:39-50.
Schultz AH (1963) Relations between the lengths of the main parts of the foot skeleton in primates. Folia Prima-
tol. 1: 150-171.
Stein TA (1995) Who's in Charge: Observations of Social Behavior in a Captive Group of Western Lowland Goril-
las. M.A. thesis. The University of Chicago.
Susman RL, Stem JT, Jr, and Jungers WL (1984) Arboreality and bipedality in the Hadar homionids. Folia Prima-
tol. 43: 113-156.
Trinkaus E (1975) Squatting among the Neandertals: A problem in the behavioral interpretation of skeletal mor-
phology. J. Arch. Science 2:327-351.
Trinkaus E (1983) The Shanidar Neandertals. New York: Academic Press.
Tuttle RH (1970) Postural, propulsive, and prehensile capabilities in the cheiridia of chimpanzees and other great
apes. In GH Bourne (ed.): The Chimpanzee, Vol. 2. Basel: Karger, pp. 167-253.
Tuttle RH (1972) Functional and evolutionary biology of hylobatid hands and feet. In DM Rumbaugh (ed.): Gib-
bon and Siamang, Vol. I. Basel: Karger, pp. 136-206.
Tuttle RH (1981) Evolution of hominid bipedalism and prehensile capabilities. Phil. Trans.Royal Soc., Lond. 8-
292: 89-94.
Tuttle RH (1985) Ape footprints and Laetoli impressions: A response to the SUNY claims. In PV Tobias (ed.):
Hominid Evolution: Past, Present and Future. New York: Liss, pp. 129-133.
Tuttle RH (1987) Kinesiological inferences and evolutionary impilications from Laetoli bipedal trails G-I, G-2/3,
and A. In MD Leakey and JM Harris (eds.): Laetoli: A Pliocene Site in Northern Tanzania. Oxford: Claren-
den Press, pp. 503-523.
Tuttle RH (1988) What's new in African paleoanthropology? Ann. Rev. Anthropol. 17:391-426.
Tuttle RH (1994) Up from electromyography: Primate energetics and the evolution of human bipedalism. In RS
Corruccini and RL Ciochon (eds.): Integrative Paths to the Past: Paleoanthropological Advances in Honor
ofF.e. Howell, New York: Prentice Hall, pp. 269-284.
Tuttle RH (1996) The Laetoli hominid G footprints. Where do they stand today? Kaupia 6:97-102.
Tuttle RH (submitted). Animalia, Homo, and the Kingdom of God. In TL Gilbert (ed.): The Epic of Creation: Sci-
entific and Religious Perspectives on our Origins.
Tuttle RH, and Cortright GW (1988) The positional behavior, adaptive complexes and evolution of Pongo
pygmaeus. In JH Schwartz (ed.): Orang-utan Biology. Oxford: Oxford University Press, pp. 311-330.
Tuttle RH, and Watts DP (1985) The positional behavior and adaptive complexes of Pan gorilla. In S Kondo (ed.):
Primate Morphophysiology, Locomotor Analyses and Human Bipedalism. Tokyo: Univeristy of Tokyo
Press, pp. 261-28.
Tuttle RH, Webb DM, and Baksh M (1991) Laetoli toes and Australopithecus afarensis. Hum. Evol. 6: 193-222.
Tuttle RH, Webb DM, Weidl E, and Baksh M (1990) Further progress on the Laetoli trails. J. Arch. Science
17:347-362.
Tuttle RH, Webb DM, Tuttle N I, and Baksh M (1992) Footprints and gaits of bipedal apes, bears and barefoot
people: Perspectives on Pliocene tracks. In S Matano, RH Tuttle, H Ishida, and M Goodman (eds.): Topics
in Primatology, Vol. 3: Evolutionary Biology, Reproductive Endocrinology and Virology. Tokyo: Univer-
sity of Tokyo Press, pp. 221-242.
Tuttle RH, Yant L, Hallgrimsson B, and Basmajian JV (in press) Hominoid heels. Proc.of the XVth Congress of
the International Primatological Society, Kuta Bali, Indonesia, August 4, 1994.
Vancata V (1991) The roots of hominid bipedality. In Y Coppens and B Senut (eds.): Origine(s) de la Bipedie chez
les Hominides, Cah. Paleoanthrop. Paris: Editions du CNRS, pp. 143-158.
White TD (1980) Evolutionary implications of Pliocene hominid footprints. Science 208: 175--176.
Wood BA (l974a) Evidence on the locomotor pattern of Homo from early Pleistocene of Kenya. Nature
251: 135--136.
Wood BA (1974b) A Homo talus ITom East Rudolf, Kenya. J. Anat. 117:203-204.
Wood BA (1976) Remains attributable to Homo in the East Rudolf succession. In Y Coppens, FC Howell, G LI
Isaac, and REF Leakey (eds.): Earliest Man and Environments in the Lake Rudolf Basin. Chicago: Univer-
sity of Chicago Press, pp. 490-506.
 23

EVOLUTION OF THE HOMINID HIP

Christopher Ruff

Department of Cell Biology and Anatomy

Johns Hopkins University School of Medicine
725 N. Wolfe Street
Baltimore, Maryland 21205

1. INTRODUCTION

The morphology of the hip region, and its functional implications, have figured
prominently in discussions of the origin and nature of hominid bipedality (Dart, 1949;
Broom and Robinson, 1950; Washburn, 1950; Le Gros Clark, 1955; Mednick, 1955;
Napier, 1964, 1967; Day, 1969, 1973; Robinson, 1972; Lovejoy et aI., 1973; McHenry,
1975; Wood, 1976; McHenry and Corruccini, 1978; Stern and Susman, 1983, 1991; Sus-
man et aI., 1984; Lovejoy, 1988; Berge, 1991; Jungers, 1991). During most of human
bipedal gait, the body is balanced over one lower limb (Inman et aI., 1981), a biomechani-
cal problem not faced by qlladrupeds. The solution to this problem has involved major
changes in the form of the human pelvis and proximal femur (as well as structures more
distal in the lower limb) from that of our primate quadrupedal contemporaries, and pre-
sumably ancestors (Le Gros Clark, 1959).
While there is little disagreement over what distinguishes the modern human hip
from that of monkeys and apes, there has been considerable debate over the extent to
which early hominids (i.e., australopithecines) approached the modern human condition in
this respect, in terms of both morphology and, especially, function (e.g., Napier, 1964;
Lovejoy et aI.; 1973; Stern and Susman, 1983; Lovejoy, 1988; Jungers, 1991). Despite
continuing differences in interpretation of the fossil evidence, however, two recurring
themes have emerged on which there is general agreement: a) morphology must be con-
sidered within its biomechanical context, and b) an appropriate "size" parameter must be
used in comparative analyses. In fact, as illustrated in the following example, these two is-
sues are actually closely related.
Many investigators have noted the apparently long femoral neck and small femoral
head characteristic of Australopithecus compared to modern humans (see McHenry and
Corruccini, 1978 and references therein). The most common method of quantifying these
differences has been to calculate ratios between head diameter or neck length and proxi-

449
450 C. Ruff

mal femoral shaft breadth (Napier, 1964; Day, 1969; Robinson, 1972; Wood, 1976). As
pointed out by Lovejoy (1975) and Wolpoff (1976), however, these kinds of ratios are
problematic, since it is impossible to distinguish to what extent the resulting variability is
due to differences in femoral shaft robusticity or to differences in head or neck dimen-
sions. Consequently, both authors (and Walker, 1973) recommended evaluating such di-
mensions relative to femur length rather than shaft breadth. Use offemur length in such a
role, however, is not without its own problems. We have previously argued that for
weight-bearing skeletal elements, body mass or some derivative of body mass is the most
appropriate measure of "size" against which to compare other dimensions (also see Jung-
ers, 1985; Ruff et aI., 1993). Due to systematic differences in body build, it is unlikely that
femur length (or even a power of length) showed the same proportionality to body mass in
australopithecines as it does in modern humans (Ruff, 1991; Aiello, 1992; Franciscus and
Holliday, 1992). Differences in body shape must also be taken into account when evaluat-
ing diaphyseal robusticity, even within Homo (Ruff et aI., 1993).
In fact, there is probably a direct functional relationship between certain proximal
femoral dimensions and shaft robusticity or strengthening. I have shown elsewhere (Ruff,
1995) that within modern humans the ratio of mediolateral to anteroposterior bending
strength of the proximal femoral shaft is significantly positively correlated with the ratio
of femoral neck length to femoral length. That is, as the femoral neck increases in length,
the proximal shaft becomes more buttressed in the M-L plane. This supported theoretical
predictions of a biomechanical model in which an increase in femoral neck length in-
creased M-L bending loads on the femoral shaft, particularly proximally. The same model
was used to explore differences in morphology between modern humans and early Homo.
Early Homo (H. erectus and "erectus-like") was shown to have a long femoral neck and
greatly increased M-L bending strength of the femoral shaft relative to modern humans, as
predicted by the model (biacetabular breadth of the pelvis was also predicted to be large in
early Homo). Interestingly, in terms of femoral shaft cross-sectional morphology, australo-
pithecines were shown to be intermediate between modern and earlier Homo.
In the present study this same biomechanical model of the hip is applied to the Aus-
tralopithecus afarensis A.L. 228-1 ("Lucy"), around which much of the more recent con-
troversy regarding hip functioning in australopithecines has centered (Stern and Susman,
1983; Lovejoy, 1988; Ruff, 1988; Berge, 1991; Jungers, 1991). In fact, using ostensibly
the same general approach applied to the same specimen, different investigators have
come to diametrically opposing conclusions regarding the ability of Lucy's hip muscula-
ture to function efficiently during gait (Lovejoy, 1988; Jungers, 1991)! It is shown here
that while A.L. 228-1 certainly was adapted to bipedality, it is very likely that she (and
perhaps at least early australopithecines in general) exhibited some subtle differences in
gait pattern from that of modern humans, as well as earlier Homo.

2. BIOMECHANICAL MODEL

McLeish and Charnley published a comprehensive study of lower limb loadings dur-
ing the single support phase of gait in 1970. They radiographed subjects standing on one
leg with the pelvis at various angles of inclination to the horizontal. Using a force plat-
form under the supporting foot and knowledge of the orientation of the abductor muscula-
ture from cadaver dissections, they were able to use basic Newtonian principles to
calculate the force vectors acting about the hip (for details see Ruff, 1995). Figure 1 is a
modified reproduction of one of their radiographic tracings of a subject standing with the
Evolution of the Hominid Hip 451

Modern Human

sending
t.\ornent
\~. rn)

Figure 1. Force vectors and resulting

femoral shaft mediolateral bending mo-
ments during one-legged stance in modem
humans. N, Newtons; m, meters; W, body W: 680N
weight; WL' weight of right lower limb; J:1476N
Ws.: weight superimposed on right hip (W - M: 898N
W L); M, gluteal abductor force; J, hip joint I: 284 N
reaction force; I, iliotibial tract force about
knee. All forces drawn to scale. Forces in
vector triangle doubled to better show de-
tails. (Ruff, 1995, Figure 5; modified from
Figure 8 in McLeish and Charnley, 1970.)

pelvis at _0.5° inclination (down on the nonsupport side), which is similar to the average
inclination during the midstance phase of gait (Eberhart et aI., 1954). From the subject's
body weight the actual magnitudes of the joint reaction force (1) and abductor force (M)
have been calculated, along with the force necessary in the iliotibial tract to stabilize the
knee (I). As shown previously (Ruff, 1995), the resulting magnitude ofI is reasonable and
acts as one check on the validity of the model.
To calculate the resulting mediolateral bending loads on the femur itself, the vectors
shown in Figure 1, calculated relative to the pelvis, need to be reversed in direction. When
this is done, M-L bending moments at any cross section in the femur may be derived using
the formula shown in Figure 2. The resulting bending moments along the femoral shaft are
shown on the left in Figure 1. These have also been shown to be reasonable in magnitude
and distribution (Ruff, 1995).
What happens when we apply the same type of analysis to A.L. 228-1? Figure 3 is a
tracing of A.L. 228-1 's pelvis and femur drawn from a photographic slide of the skeletal
reconstruction done by Dr. Peter Schmid (1983) and kindly made available to me by Dr.
Robert Martin. When checked against casts of the pelvis and left femur (supplied courtesy
of Dr. Bruce Latimer), all significant proportions (e.g., biacetabular breadth, biomechani-
cal femoral neck length, femoral length) are virtually identical in the figure: For the
analysis the body center of gravity (Ws) was kept in the same position relative to bi-
acetabular breadth, and the origin of the abductors (M) was maintained in the same posi-

• Reconstructions of the pelvis by Schmid (1983) and Tague and Lovejoy (I986}-the former the basis for the
photograph and the latter for the cast measured here--are slightly different, but the major dimensions of interest
for this study were found to be very similar in the two versions. It was not possible to use the recent new recon-
struction of the pelvis by Hausler and Schmid (1995) since the published photographs are not in the correct ori-
entation for the purposes of this study.
452 C. Ruff

Figure 2. Method of calculating femoral shaft bending mo-

Bending Moment at Section of Interest: ments from hip joint reaction and gluteal abductor forces.
Clockwise bending moments are considered to be positive (Ruff,
J y ' DJ - J x • DJ + My' DM + Mx' DM
x y x y 1995).

tion relative to bi-iliac (total pelvic) and iliac blade breadth as in the modem human of
Figure 1. (Note that the relative degree of sagittal or frontal orientation of the iliac blades
in A.L. 228-1 should have little if any effect on the model, since for calculation of M-L
moment arms all dimensions are projected into the coronal plane.) The precise position of
the knee joint in one-legged stance was determined by first calculating the length of the
tibia from the femur using a modem human African sample (Ruff, 1995) and assuming
that the distal articular surface of the tibia was centered under the whole body midline
(W). Because of significant distortion of the preserved proximal tibia (A.L. 228-1 aq) and
heavy reconstruction of the distal femur (A.L. 228-1ap) (Johanson et ai., 1982a), casts of
the better preserved knee joint of A.L. 129-la,b (Johanson and Coppens, 1976) were used
to establish the femoral-tibial angle at the knee. Slight differences in either tibia length or
this angle make very little difference in the final results. All vectors were scaled to the
same body weight (W) as in Figure 1, with the proportion of body weight superimposed
over the hip (Ws) assumed to be the same. Bending moments in the femur were calculated
as before.
The predicted change in magnitudes of hip joint reaction force (~J), abductor force
(~M), and bending moments at the subtrochanteric (~80% BM) and midshaft (~50% BM)
femur of A.L. 228-1 relative to the modem human model are shown in Figure 3 along with
the superimposed vector triangles of Figures 1 and 3 used to calculate these parameters.
The abductor force in A.L. 228-1 is predicted to rise by 27%, while the hip joint reaction
force rises by 12%. These increases are brought about by the very large increase in
Evolution of the Hominid Hip 453

A.L. 288-1

8J: +12%
8M: +27%
880% BM: +36%
850% BM: +39%
IfW: .64

Figure 3. Force vectors and femoral shaft bending moments in A.L. 228-1 compared to modem human of Figure
I. All forces drawn to scale. Hatched area in bending moment diagram shows increase in A.L. 228-1; in vector tri-
angle heavy lines are A.L. 228-1 and lighter lines are modem human (enlarged 2X). Numbers below triangle show
increases in J, M, and bending moments at 80% (subtrochanteric) and 50% (midshaft) locations, and the ratio of I
to W. All symbols as in Figure I.

biacetabular distance, which is only partially offset by the wide iliac flare; thus in order to
balance the pelvis, abductor force and joint reaction force must also increase (see Ruff,
1995 for further discussion). The predicted increases in mediolateral bending of the femo-
ral shaft (relative to body weight) are substantial- about 35-40%. The relative magni-
tude of the iliotibial tract force necessary to stabilize the knee (I) is larger than in the
modern human model, but still within reasonable limits (Ruff, 1995). t
So, if the modern human model applied to "Lucy", that is, if she walked the same
way we do, one would expect evidence for the following in her skeleton: a) increased but-
tressing of the ilium against the increased force of the gluteal abductors; b) a somewhat
larger femoral head to maintain hip joint stresses at about the same level under increased
joint force; and, c) markedly increased M-L bending strength of the femoral shaft to
counter much larger M-L bending moments. In order to properly evaluate these features,
though, they must be considered relative to her body mass or to another appropriate pa-
rameter related to body mass. The following section describes how body mass was derived
for A.L. 228-1 in this study.

t The force in the iliotibial tract would also probably reduce M-L bending of the femoral shaft (see Ruff, 1995),
but since this would also affect bending of the modem (as well as fossil--see below) Homo femur, the effect on
relative differences in M-L bending of the shaft should be minimal.
 454 C. Ruff

3. BODY MASS ESTIMATION

As discussed previously (Ruff et ai., 1993), in analyses of the relative size of post-
cranial skeletal features, it is difficult to avoid circular reasoning when estimating body
mass. For example, the joint size of weight-bearing skeletal elements bears a close rela-
tionship to body mass (Ruff, 1988; Jungers, 1990) and so has been used in the past to esti-
mate body mass in fossil hominids (Suzman, 1980; McHenry, 1991 b, 1992; Ruff et ai.,
1997). However, it is obviously not appropriate to employ this methodology when evaluat-
ing the relative size of a hindlimb joint itself, e.g., the femoral head! The same caveat ap-
plies to all mechanically related features, including shaft cross-sectional dimensions. An
alternative is to take a non-mechanical morphometric approach to the estimation of body
mass, in which preserved skeletal elements are used to directly assess the size and shape
of the body. We have used such an approach to estimate body mass in the early Homo
erectus KNM-WT 15000 (Ruff and Walker, 1993), as well as in other Pleistocene Homo
(Ruff, 1994; Ruff et ai., 1997).
The morphometric approach is based on a "cylindrical model" of the human body,
with stature as the height of the cylinder and bi-iliac breadth as the breadth (Ruff, 1991).
The volume of a cylinder is equal to 1tf2Z, where r is its radius (half ofbi-iliac breadth) and
Z is its length or height (stature). Since body density can be assumed to be approximately
constant and close to I, mass should be equivalent to volume.
Multiple regressions using bi-iliac breadth and stature predict body mass with a very
acceptable level of accuracy (standard errors of estimate of 3-6 kg) in a variety of modern
human populations (Ruff and Walker, 1993; Ruff, 1994). Estimates using this technique
are also concordant with those based on femoral head breadth regressions in Pleistocene
Homo (Ruff et ai., 1997). However, as noted previously (Ruff and Walker, 1993) the more
elliptical shape of the pelvis, and presumably the trunk as a whole, in A.L. 228-1 (Tague
and Lovejoy, 1986; Lovejoy, 1988) requires some adjustment in the method before it can
be applied to her body mass estimation. Previously we attempted to do this using internal
midplane anteroposterior breadth of the pelvis as an indicator of external A-P breadth of
the body, since true external A-P breadths for modern samples were not available, but also
realized its limitations as such an index (Ruff and Walker, 1993). Since then I have col-
lected data for true external A-P breadth of the pelvis, together with bi-iliac breadth, in
two osteological samples of modern humans, and so can apply a more appropriate correc-
tion factor for A.L. 228-1.
The two modern samples include an East African (Bantu) sample from the Kenya
National Museums in Nairobi (N=30 adults) and a sample of US whites from the Terry
Collection, Smithsonian Institution, Washington, DC (N=20 adults), each equally divided
between males and females. The two hip bones and sacrum were first articulated and held
together with several heavy rubber bands, and bi-iliac breadth measured using an
osteometric board. Then, with the pelvis in anatomical position (anterior edge of pubic
symphysis and anterior superior iliac spines in the same coronal plane), the A-P breadths
from the anterior-most points to the most posterior edges of the sacrum and the ilium (pos-
terior superior iliac spines) were measured, again using an osteometric board with the aid
of some larger movable end pieces. Because A.L. 228-1 's restored sacrum is probably
slightly distorted (flattened) (Tague and Lovejoy, 1986), and because sacral curvature var-
ies between modern males and females, only the A-P measurement to the posterior supe-
rior iliac spines is used here. The mean ratio of A-P to M-L (bi-iliac) pelvic breadths in the
African sample is 0.5082 (±0.0313 SD), and in the US white sample 0.5073 (±0.0532 SD).
The mean ratio of the pooled sample, rounded to 0.508 (±0.041 SD), is used here. A.L.
Evolution ofthe Hominid Hip 455

228-I's A-P/M-L ratio, measured in the same way, is 0.324, more than 4 standard devia-
tions from the pooled modern sample.
The formula for volume (=mass) of a cylinder with an elliptical cross-section, modi-
fied from the one given above, is 1tfli, where r) and r2 refer to the two breadths of the cyl-
inder and I is again its length. The modern reference sample used here to estimate AL.
228-1 's body mass here is a world-wide living sample of 56 sex/population means for
matched body mass, stature, and body breadth, including those for African Pygmies (Ruff,
1994).t Only bi-iliac breadth (not A-P breadth of the pelvis) was available as a body
breadth measurement for this sample. Thus, using the elliptical cylindrical model, the fol-
lowing procedure was carried out in order to apply these data to A.L. 228-1:

Body mass 1tf) ri

= 1I4nlDp2
= 0.785lD)D 2
= 0.785lD)2(D/D)

where D) and D2 refer to the diameters in the M-L and A-P planes, respectively. For the
modern sample, using the mean A-P/M-L ratio derived above, this body mass predictor be-
comes 0.785 . stature· (bi-iliac breadth)2 . 0.508; for AL. 228-1 it is 0.785 . stature· (bi-
iliac breadth)2 . 0.324. A.L. 228-1 's stature is taken as 107 cm (Jungers, 1988a) and her
bi-i1iac breadth, adjusted for the addition of 0.5 cm soft tissue: is 25.8 cm (Ruff, 1991),
giving a value of 18115 cm3 for this index.
Figure 4 is a plot of body mass against the morphometric body mass index in the
modern reference sample, and the predicted body mass of A.L. 228-1. Because AL. 228-1
lies well outside the range of modern values, including Pygmies, a Model II rather than
least squares equation through the modern data is used to estimate her body mass (Olivier,
1976; Jungers, 1988a). For reasons given elsewhere (Aiello, 1992; Ruffet aI., 1993), re-
duced major axis (RMA) analysis is used here.
The predicted body mass of A.L. 228-1 using this technique is 27.4 kg. This con-
firms her originally estimated body mass of 27-28 kg (Johanson and Edey, 1981; Latimer
et aI., 1987), and is also close to the midpoint of a range of previous body mass esti-
mates-23 to 30.4 kg--derived using a variety of techniques (Johanson and Edey, 1981;
McHenry, 1984; Latimer et aI., 1987; Jungers, 1988b,c, 1991; Ruff and Walker, 1993; Por-
ter, 1995). It is difficult to derive a true "confidence interval" for this estimate, since A.L.
228-1 lies well outside the range of modern values and the modern data points are them-

t An error in one of the primary sources of data used for the list in Ruff, 1994 has since been discovered: the body
mass for Aleut females reported by Laughlin in 1951 should have been 117 Ibs. (53.4 kg) rather than 177 Ibs.
(80.5 kg) (Laughlin, pers. comm.). I am indebted to Dr. Steven Churchill for drawing my attention to this.
o We have recently suggested that for Homo an addition for soft tissue of about I cm rather than 0.5 cm may be
more appropriate for individuals with bi-iliac breadths in this size range (Ruff et aI., 1997). However, the sacroil-
iac and pubic symphyseal joint sizes of A.L. 228-1 are much smaller than in Homo specimens with comparable
bi-i1iac breadths, and it is likely that soft tissue thickness (cartilage, fibrocartilage) within these joints was also
correspondingly thinner in A.L. 228-1. Thus, the original 0.5 cm correction may be more appropriate for this
specimen. If I cm is added to A.L. 228-1 's skeletal bi-iliac breadth, her body mass prediction index becomes
18824 cm 3, which gives an estimated body mass of 28.2 kg (see below), well within the error ranges plotted in
subsequent figures. It should also be noted that another recent reconstruction of A.L. 228-1 's pelvis (Hausler and
Schmid, 1995) appears to produce a narrower skeletal bi-iliac breadth (about I cm less than in other reconstruc-
tions), which could result in a lower body mass estimate, although this would also depend upon the A-P dimen-
sion of the reconstruction, impossible to measure from published photographs.
456 C. Ruff

80
RMA: y = 1.084x + 7.77
.836, SEE = 4.4 kg)
(r =
70

-
~
Ol
.........
60

(/)
(/)
co 50
~
>-
"0
0 40
en
• Modern Humans

30 A A.L. 288·1

20
10 20 30 40 50 60

Morphometric Body Mass Index (cm 3 11000)

Figure 4. Prediction of body mass in A.L. 228-1 from morphometric body mass index derived from stature and
body breadth (see text). Fifty-six modern human sex/population means derived from data in Ruff (1994). Reduced
major axis line drawn through modern sample.

selves not individuals but sex/population means. The standard error of estimate of the re-
gression equation through the moderns is 4.4 kg. In fact, all of the subsequent compari-
sons involve logarithmic transformations of body mass, which makes "confidence limits"
of this kind nonsymmetrical. For ease in plotting, the average of the upper and lower loge-
transformed differences of ±4.4 kg (±O.162) is given in these figures. This gives an actual
range of (nontransformed) values on the plots of about 23-32 kg, encompassing the range
of previous body mass estimates for A.L. 228-1 (see above). It seems very likely, there-
fore, that the range plotted actually includes Lucy's true body mass.

4. TESTING PREDICTIONS OF THE MODEL IN A.L. 228-1

4.1. Relative Femoral Head Size

One of the predictions of the biomechanical model of A.L. 228-1 (Figure 3) was that
she would exhibit a somewhat larger femoral head relative to body mass than in modern
humans. Figure 5 is log-log plot of femoral head superoinferior breadth against body mass
in a modern human sample and A.L. 228-1. The modern human sample consists of 46 East
African and 31 Pecos Pueblo Amerindian skeletons for which body mass was derived
from bi-iliac breadth and reconstructed stature using regression equations based on living
popUlations (Ruff et aI., 1997). (Stature in the Pecos sample was calculated using
Genoves' (1967) formulas for Mesoamericans, and in the East Africans using the equa-
Evolution ofthe Hominid Hip 457

3.9

-E
E
- . . ..
3.8 D (r = .727)

.+-'
r:::.
'0
m
Q) 3.7
tIl
rh
'0
3.6
m
Q)
:r: • Modern Humans
Ci3
'-
3.5 o Modern Humans
o (McHenry. 1991 b)
E D Early Homo
Q)
LL
3.4 A A.L. 288-1
c:
....J
•
3.3+--'--~--~-r--T-~--'--'--~--r--r--,
3.1 3.3 3.5 3.7 3.9 4.1 4.3

Ln (Body Mass) (kg)

Figure 5. Loge-transformed femoral head superoinferior breadth versus body mass in modern human individuals
(Pecos Pueblo Amerindians and East Africans), A.L. 228-1, and early Homo (from smallest to largest body mass,
KNM-ER 1481, 1472, and OH 28), and McHenry's (1991 b) four modern human means, including Pygmies (small-
est value). Both reduced major axis and least squares lines drawn through modern sample (McHenry's data not
used in calculating regression). Error bars around A.L. 228-1 represent approximately ±4.4 kg. See text for details.

tions given by Feldesman and Lundy (1988) for South African blacks.) These two samples
were used here in part because they have very different body builds, encompassing much
of the range of variation in body breadth/height present among modern humans (Ruff,
1995). In addition, the four modern human means for femoral head breadth and body mass
used by McHenry (McHenry, 1991b; 1992), which include data points for African Pyg-
mies and the small-bodied Khoisan, are plotted for comparison. Finally, three early Homo
(KNM-ER 1472, 1481 and OH 28) for which femoral head size can be measured or calcu-
lated (Ruff et aI., 1993) and body mass estimated morphometrically (Ruff and Walker,
1993; Ruff et aI., 1997) are included in the plot.+ Both RMA and least squares regression
lines through the modern East African-Pecos reference sample, extrapolated downwards to
A.L. 228-1, are shown.
What is most apparent from Figure 5 is that A.L. 228-1 has a femoral head size that
is about what would be predicted from her body mass, based on the scaling of these two

+ In Ruff et al. (1997) body masses of several early Homo specimens with intact femoral heads were estimated us-
ing equations based on femoral head size to body mass in modern humans, and are larger than the morphometric
estimates used here. The femoral head estimates are obviously inappropriate for the present analysis. The mor-
phometric estimates used here were derived following general procedures described in Ruff et al. (1997).
458 C. Ruff

parameters in modern humans. McHenry's (1991b) data points fall along the same general
trajectory as that of my modern human sample. Indeed, the fact that he obtained a body
mass estimate of26.7 kg for A.L. 228-1 based on femoral head size and a modern human
reference sample that is close to my morphometric estimate of 27.4 kg is indirect evidence
that her femoral head scaled at about the same proportion to body mass as in modern hu-
mans. Thus, similar to my conclusion based on a different comparison (Ruff, 1988), but
contrary to that of Jungers (Jungers, 1988c, 1991), Lucy did not have a relatively small
femoral head. However, the prediction of the biomechanical model was that she should
have had a relatively large femoral head, and there is no evidence from Figure 5 to sup-
port this. Interestingly, the three early Homo data points all lie in the upper range of the
modern humans, i.e., they appear to have somewhat enlarged femoral heads, although this
difference does not reach statistical significance (Ruff et aI., 1993). I have argued else-
where (Ruff, 1995) that hip joint reaction force in early Homo may have been slightly in-
creased over that of modern humans, largely due to increased biacetabular breadth, which
could explain this observation. This makes the non-deviation of femoral head size in A.L.
228-1, with her even more exaggerated biacetabular breadth, all the more striking.
It should also be noted that the analysis shown in Figure 5 intrinsically accounts for
non-isometric scaling of femoral head size with body mass within hominids. Specifically,
as shown previously (Ruff, 1988; McHenry, 1991 b; Ruff et aI., 1993), femoral head size is
positively allometric relative to body mass in modern humans (and earlier Homo; see Fig-
ure 5 and Ruff et aI., 1993), i.e., it increases in size faster than would be predicted to
maintain geometric similarity with body mass. We have argued that reduced major axis
analysis is the most appropriate method for evaluating such allometric relationships (Ruff
et aI., 1993); the RMA slope through the modern sample in Figure 5 is 0.578 (SE = 0.046),
highly significantly different from the isometric slope of 0.333 (P<O.OOI). Even the lower
least squares slope of 0.420 (same SE) suggests positive allometry (P<0.06). Lucy's femo-
ral head size is consistent with this allometric scaling within hominids.

4.2. Cross-Sectional Femoral Shaft Dimensions

Another prediction of the biomechanical model was that A.L. 228-1 would show
greatly increased mediolateral bending strength of the femoral diaphysis. Here not only
body mass but also relative activity level and muscle strength in general must be consid-
ered, since it is known that the cortical bone in diaphyses is highly plastic in response to
applied mechanical loads during life (Ruff and Runestad, 1992; Trinkaus et aI., 1994). The
A.L. 228-1ap femur has a break through it at approximately midshaft between its second
and third segments (Johanson et aI., 1982a). The endosteal boundary is clearly visible on
the distal surface of a cast of the fused first and second segments (housed in the Kenya
National Museums), which was photographed, digitized, and analyzed using SLICE
(Nagurka and Hayes, 1980). Other dimensions ofthis cast were found to be quite close to
those published for the original specimen (Johanson et aI., 1982b).
Figure 6 is a log-log plot of midshaft femoral cortical area in modern humans (East
Africans and Pecos Pueblo Amerindians), A.L. 228-1, and early Homo (KNM-ER 737,
1472, 1481, 1808, and OH 28) against estimated body weight. The modern and early
Homo cross-sectional data were derived as described elsewhere (Ruff et aI., 1993; Ruff,
1995); body weights for early Homo were derived as described above. Reduced major axis
and least squares regression lines are again plotted through the modern sample.
It is clear from Figure 6 that A.L. 228-1 had a very robust femoral diaphysis relative
to her body weight when compared to modern humans. Her midshaft cortical area is al-
Evolution ofthe Hominid Hip 459

6.3

o RMA: y = 1.375x + .259

6.1 o . •
'
• ,I'
I.

LS: y = .957x + 1.936

o o : ,I , ~'
E
---
E
t? 5.9 . ,•,1/·
e.
~
.." •
,~
:~
,
l~ (r = .696)

• r;;.
. /' • .
Q)
~..
< • e. .".
(U
o 5.7 . ... .. ,~:
'-'
'I'· •
t:
o
,,
, . ,": 7-
::.
--- .
()

,, " " , • Modern Humans

,
..
..
C ,
....J 5.5 , :
: o Early Homo

,
," :.:
I'
A A.L.288-1

5.3 +---.--r--..----..-......---,..--......--r---..--r--.....-.....,
3.1 3.3 3.5 3.7 3.9 4.1 4.3

Ln (Body Mass) (kg)

Figure 6. Loge-transformed femoral midshaft cortical area versus body mass in modem humans (Pecos Pueblo
Amerindians and East Africans), A.L. 228-1, and early Homo (from smallest to largest body mass, KNM-ER 1481,
1472, 737, OH 28, and KNM-ER 1808; data from Ruff et aI., 1993, 1997). Reduced major axis and least squares
lines drawn through modem sample. Error bars for A.L. 228-1 as in Figure 5.

most as large as that of the smallest modem humans in the sample, despite her body mass
being some 35% lower than any of the modem humans. The early Homo specimens are
also more robust than modem humans, as shown previously (Ruff et aI., 1993). This de-
viation is best viewed as a reflection of overall increased muscularity and activity level in
Plio-Pleistocene hominids (Ruff et aI., 1993), and it is not surprising that an early australo-
pithecine would exhibit as least as great an increase as later Homo. This means, however,
that to properly evaluate mediolateral bending strength of the femoral shaft, this overall
increased robusticity must be factored into the comparison. For this reason, ML bending
strength is compared to AP bending strength at midshaft as in previous analyses of early
Homo femora (Ruff, 1995). Anteroposterior bending strength should be a good index of
overall mechanical loading of the femur that is not directly related to changes in M-L
bending of the shaft. Measures of bending strength (section moduli) were derived from
cross-sectional second moments of area as described previously (Ruff, 1995).
Figure 7 shows M-L versus A-P bending strength in the modem human sample, A.L.
228-1 and early Homo. Because there was no need to estimate body mass in this compari-
son (and thus no need to calculate stature and body breadth), both modem and earlier sam-
ples of Homo could be increased (see figure caption). A.L. 228-1 falls at or below RMA
and least squares regression lines through the modem sample, indicating no increase in
relative M-L bending strength of her femoral shaft. In fact, her midshaft cross section is
almost circular, with nearly equal strengths in both planes. This is in stark contrast to the
early Homo sections, which are strongly asymmetrical and buttressed against relatively
high M-L bending loads. Again, as with relative femoral head size, the position of A.L.
460 C.Ruff

8.0
--
.J::

-
Cl RMA: y = .842 + .870x
c:
....
Q)
LS: y = 1.951 + .713x
CJ) 7.5 (r = .820)
Cl
c:
"0
c:
Q)
CO
7.0
....
CIS
Q)
ca
:§ • Modern Humans
"0

-
Q) 6.5 C Early Homo
:::! ~ A.L.288-1
c:
...J

6.0 +---.----r----r---.........-.,--......,...----,
6.3 6.7 7.1 7.5 7.9

Ln (Anteroposterior Bending Strength)

Figure 7. Loge-transformed mediolateral versus anteroposterior bending strength of the femoral midshaft in mod-
em humans (Pecos Pueblo Amerindians and East Africans), A.L.22S-I, and early Homo (from smallest to largest
A-P bending strength, Gesher Benot Ya'acov I, KNM-ER 14S1, Zhoukoudian 2, Zhoukoudian 5, KNM-ER 1472,
Zhoukoudian 4, OH 2S, Zhoukoudian 6, KNM-ER 737, KNM-ER IS0S, Zhoukoudian I, Ain Maarouf I, KNM-
ER S03; data from Ruff, 1995). Reduced major axis and least squares lines drawn through modem sample.

228-1 can not be attributed to purely allometric effects: two of the smaller early Homo
specimens with A-P bending strength close to that of A.L. 228-1 (KNM-ER 1481 and
Gesher Benot Ya'acov 1) have almost 50% greater M-L bending strength at midshaft (see
Figure 7).
I have previously noted that the femoral midshaft region is less reliable than the
proximal femoral shaft for inferring changes in hip structure (Ruff, 1995). Thus, it would
be desirable to carry out a similar analysis for a femoral subtrochanteric cross section. Un-
fortunately, no convenient natural break surface was available in this region for A:L. 228-
I. Therefore, as an approximation, A-P and M-L external subtrochanteric breadths were
used as general indices of A-P and M-L bending strengths of the proximal femoral shaft.
Fossil data, including that for A.L. 228-1, were derived from McHenry (1988), Weiden-
reich (1941), Day (1969), and my own measurements, and modern human data were de-
rived from a variety of sources (means for one African, six Amerindian, and four
European archaeological samples). As traditionally measured, subtrochanteric breadths are
more truly regarded as maximum and minimum breadths rather than M-L and A-P
breadths, respectively, with the difference between the two types of measurements varying
depending on femoral neck anteversion angle and other factors. These measurements also
do not, of course, reflect variation in internal structure such as cortical thickness. How-
ever, they should give some indication of proximal shaft shape that can be used to evalu-
ate general differences in morphology between groups (also see Ruff, 1987a).
 Evolution ofthe Hominid Hip 461

3.7

[J [J
RMA: y = -.151 + 1.12x
E 3.6 LS: y = .322 + .971x
E (r = .880)
..........
.L:
+-'
-0 3.5
co
Q)
'-
CO
co 3.4
'-
Q)
co
+-' • Modern Humans
(5 [J Early Homo
:0 3.3
Q) A A.L. 288-1
~ t:. other australopith.
c
....J
3.2 A

3.1 + - - - , . - - - , . - - - - . - - - - - , r - - - - , - - - - ,
2.8 2.9 3.0 3.1 3.2 3.3 3.4

Ln (Anteroposterior Breadth) (mm)

Figure 8. Loge -transformed mediolateral versus anteroposterior femoral subtrochanteric external breadths in mod-
em humans, A.L. 228-1 and other australopithecines, and early Homo. Early Homo data from Weidenreich, 1941;
Day, 1971; McHenry, 1988; and my reconstruction for KNM-ER 1808 (see Ruff, 1995); from smallest to largest
A-P breadth: KNM-ER 1481, 1472, Zhoukoudian 4, Zhoukoudian I, KNM-ER 1808,737, and 803. Australopithe-
cine data from McHenry, 1988; in addition to A.L.228-1, from smallest to largest A-P breadth: A.L. 128-1, KNM-
ER 1500, STS 14, OH 20, A.L.211-1, SK 97, SK 82, A.L.333-95, A.L.333W-40, A.L.333-3. Modem data points
are population sample means; see text for general provenience (the smallest modern data point is for an African
Bushman sample). Reduced major axis and least squares lines drawn though modem sample.

Figure 8 is a log-log plot of M-L versus A-P femoral subtrochanteric breadths in

modem humans, early Homo, and australopithecines, including A.L 228-1. The other aus-
tralopithecine data points, from McHenry's 1988 compilation, include both South and East
African specimens. The australopithecines fall in the same general distribution as the
modem human sample means. A.L. 228-1 is among the most platymeric (M-L ex-
panded/ A-P flattened) of them, with a positive deviation in M-L breadth from that pre-
dicted from A-P breadth of 8% relative to the least squares line through modem humans,
and 14% relative to the RMA line through modem humans. Thus, her proximal shaft is
moderately platymeric. However, the degree of platymeria present in the early Homo
specimens is noticeably greater, with an average deviation in M-L breadth of about 18%
from that predicted from A-P breadth using either method of line fitting through modem
humans. One early Homo specimen (KNM-ER 803) falls close to the modem regression
lines. However, its proximal shaft distal to the inferred position of the lesser trochanter
(which is not preserved) is quite weathered with large chips missing; thus, its A-P and M-
L breadths may not be reliable. Without this specimen, the average deviation of early
Homo from the modem regression lines is about 20%, or about 1.5-2.5 times greater than
that of A.L. 228-l.
462 C. Ruff

These results reinforce previous cross-sectional analyses for a somewhat more lim-
ited fossil sample (Ruff, 1995) that indicated that australopithecines on average have mod-
erately platymeric proximal femoral shafts, and early Homo have very markedly
platymeric shafts. The predicted increase in M-L bending of the femoral shaft in early
Homo, based on preserved femora and a range of possible pelvic morphologies, was actu-
ally less than that of A.L. 228-1 (Figure 3), yet early Homo shows much more evidence of
shaft buttressing against such increased M-L loading. This strongly suggests that the rela-
tionship between morphology of the hip and mechanical loading of the femoral shaft was
different in early Homo and A.L. 228-1.

4.3. Iliac Buttressing

The final prediction of the biomechanical model of A.L. 228-1 was that she would
exhibit evidence for increased gluteal abductor force relative to modern humans. The de-
velopment of an acetabulocristal buttress, or iliac pillar, is commonly viewed as a skeletal
indicator of increased abductor force in hominids (e.g., Lovejoy et ai., 1973). This feature
is apparently developmentally plastic (Rader and Peters, 1993) and thus would be ex-
pected to reflect the actual level of force exerted by the gluteal abductors across the iliac
blade.
The maximum breadth of the iliac crest, always occurring at the iliac tubercle, was
used here as an indicator of the size of the acetabulocristal buttress. Because of its devel-
opmental plasticity, and evidence for overall increased muscularity and activity level in
A.L. 228-1 (Figure 6), as with femoral shaft cross-sectional dimensions it is important to
consider iliac buttressing relative to another skeletal feature that incorporates overall in-
creased mechanical loading, but not specifically gluteal abductor loading. Here I chose the
minimum iliac crest breadth, always occurring posterior to the iliac tubercle, as such an in-
dicator.
Figure 9 shows iliac tubercle breadth plotted against minimum iliac crest breadth in
a sample of modern East Africans (similar data were not available for the Pecos sample),
A.L. 228-1, and two early Homo ilia (KNM-ER 3228 and Arago 44). A.L. 228-1 falls di-
rectly within the trajectory of the modern human data scatter (also note that both of her di-
mensions fall quite close to the lower end of this scatter, reinforcing the impression of her
general skeletal robusticity relative to body mass). The two early Homo ilia, in contrast,
fall above modern humans, especially KNM-ER 3228. Thus, while there is evidence for
increased gluteal abductor force in early Homo, there is no such evidence in Lucy's pelvis.

5. DISCUSSION

A summary ofbiomechanical predictions and observed morphologies for A.L. 228-1

is given in Table 1. In each instance, the prediction based on the biomechanical model of
the hip, assuming a gait pattern similar to that of modern humans (Figure 3), is not borne
out by her actual morphology. Somewhat paradoxically, the foregoing analysis has demon-
strated that A.L. 228-1 is in many respects very much like modern humans, with a femoral
head size, iliac tubercle, and femoral shaft cross-sectional shape well within modern hu-
man limits, relative to appropriate size measures. However, the overall morphology of her
hip region and its biomechanical consequences predict that she should not look like mod-
ern humans in these other respects. The fact that early Homo, who also appear to have de-
parted from modern humans in some of the same aspects of pelvic/femoral morphology,
Evolution of the Hominid Hip 463

28
[J KNM-ER 3228

- 25
E
E

-
.L:
"0
ro
Q)
.....
22
[J Arago

.
OJ
19
Q) "
(3
..... " "
....
Q)
.0
I-
::J 16
: :" " Modern Humans
[J Early Homo
o
ro A A.L. 288-1
13

10+-~--~~--~-r--r--r--r-~~r-~~--~-'

4 5 6 7 8 9 10 11

Minimum Iliac Crest Breadth (mm)

Figure 9. Iliac tubercle breadth versus minimum iliac crest breadth in modem East Africans, A.L.22S-I, and two
early Homo specimens.

including a wide interacetabular distance, long femoral neck, and possibly increased iliac
flare, show evidence consistent with biomechanical predictions of the model (Ruff, 1995),
strengthens the argument that A.L. 228-1, and possibly australopithecines as a whole, de-
viate from the model in some way. Stated in another way, if all else were equal, Lucy
would not be expected to look exactly like modern humans in the morphology of her
acetabulocristal buttress, femoral head, and femoral shaft shape, as is the case in early
Homo. The fact that she does indicates that all else was not equal.
What could account for A.L. 228-1 's deviation from the model predictions? One
possibility is suggested by other results of McLeish and Charnley's original study. They
found, not surprisingly, that a positive pelvic angle, that is, tilting of the pelvis up on the
nonsupport side, greatly reduced abductor and hip joint reaction forces during single-
legged stance. This is because a positive pelvic tilt also tilts the trunk over the support
limb, moving the superimposed center of gravity of the body towards the support hip joint
and greatly reducing the necessary balancing abductor moment (see Ruff, 1995: Figure 4).

Table 1. Predicted and observed morphology in A.L. 228-1 (relative to modern humans)
Biomechanical prediction Observed morphology
Moderately increased hip joint reaction force Little or no increase in relative hip joint size
Greatly increased gluteal abductor force No increase in relative iliac tubercle size
Greatly increased M-L bending offemoral shaft None to moderate increase in relative M-L
bending strength of femoral shaft
464 C. Ruff

Table 2. Effects of pelvic elevation on the nonsupport side in A.L. 228-1

A.L. 228-1 relative to modern
Pelvis Level Pelvis +4 Elevation
Hip joint reaction force +12% -4%
Gluteal abductor force +27% -5%
M-L bending proximal femur +36% +19%

In the same subject shown in my Figure I, they found that a positive pelvic angle of 13°
produced a decrease of more than 70% in abductor force and more than 40% in hip joint
reaction force compared to that for a level pelvis.
I performed the same type of analysis for A.L. 228-1, except that I tilted her pelvis
up on the nonsupport side by only 4° rather than 13°. Change in position of the body cen-
ter of gravity relative to the hip joint was estimated from results given by McLeish and
Charnley (1970) for three different positions of the pelvis in the same subject ("A", their
figures 8-10). The relationship between pelvic angle and body weight moment arm was
found to be virtually linear (R=0.999) over these three points, which ranged from -8.5° to
13° of pelvic tilt. Given the estimated position of the body center of gravity, the rest of the
analysis was carried out as in Figure 3. Results are shown in Table 2. With this slight de-
gree of pelvic elevation, predicted hip joint reaction force (relative to modems) falls from
+ 12% to -4%, abductor force falls from +27% to -5%, and mediolateral bending of the
proximal femur falls from +36% to + 19%. This in tum would predict approximately the
same sized femoral head and degree of iliac buttressing, and a moderate increase in M-L
buttressing of the femoral shaft relative to modems, which is precisely what A.L. 228-1
exhibits.
The same mechanism of pelvic elevation on the nonsupport side was observed by
Jenkins (1972) in young chimpanzees trained to walk bipedally. The amount of positive
pelvic inclination in the chimpanzees was of approximately the same order as the 4° arbi-
trarily chosen here (my measurements from his figure I). It is interesting that human pa-
tients with a painful hip demonstrate a similar phenomenon during weight-bearing on that
side, apparently to reduce joint reaction force (Poss and Sledge, 1981). In both this and in
so-called "gluteus medius limp" (resulting from weakness or paralysis of the gluteal ab-
ductors), lateral bending of the trunk over the support limb also occurs (Lehmann and De
Lateur, 1990). This same lateral bending of the trunk is evident in McLeish and Charnley'S
subject with the elevated pelvis (1970: Figure 9) and was almost certainly also occurring
in Jenkins' chimpanzee subjects, although this can not be determined from his published
figure (which ends at the pelvis).
All this is not meant to suggest that A.L. 228-1 walked either like a modem bipedal
chimpanzee or a modem human suffering from a gait abnormality. However, it does serve
to illustrate that the same general biomechanical principles used in the present analysis do
seem to operate in living animals that can be observed walking. It is possible that some
lateral bending of the trunk and some pelvic elevation on the nonsupport side were both
characteristic of A.L. 228-1 's gait pattern. In any case, these motions, if they were present,
must have begun during the late double-support phase of gait, i.e., prior to the assumption
of single-limb support, through increased action of the abdominal and back musculature,
as in modem humans with gluteal abductor weakness (Ducroquet et aI., 1968: Plate 95).
The fact that pelvic elevation occurs in bipedal chimpanzees, with their less well-devel-
Evolution ofthe Hominid Hip 465

oped hip abductor mechanism, illustrates that increased abductor force is not necessary to
produce this gait pattern.
If the nonsupport hip was tilted slightly upward during gait in AL. 228-1, this would
also have had the effect of providing somewhat more ground clearance for her swing limb.
Depending upon the particular reconstruction, AL. 228-1 has been described as having a
foot that was either at or slightly beyond the length predicted from her femur length in
modem human adults (Jungers and Stem, 1983; Susman et aI., 1984; White and Suwa,
1987). That is, relative to the probable length of her lower limb, she may have had a
slightly long foot. Based on analogies with modem children, who also have relatively long
feet, it has been proposed that she may have had to compensate for this during gait by in-
creasing "vertical excursions, velocities and accelerations of the more proximal joints and
limb segments" (Susman et aI., 1984:147) in order for the foot to clear the ground, result-
ing in a less efficient gait pattern. However, only a few degrees of pelvic elevation on the
swing side would have the same effect: 4° of elevation over AL. 228-1 's biacetabular
breadth of 122 cm would result in an increase in height above the ground of the nonsup-
porting hip of 8.5 cm, while 2° of elevation would add 4.3 cm of "clearance". Thus, in ad-
dition to lowering abductor and hip joint reaction forces on the stance limb, a positive
pelvic tilt could have facilitated movement of the swing limb during gait. This would have
been especially critical if the stance limb also exhibited more hip and knee flexion, as sug-
gested by some researchers (Stem and Susman, 1983; Preuschoft and Witte, 1991; Schmitt
et aI., 1996). A possible reduction in stride length in Australopithecus (Reynolds, 1987)
could also be related to increased lateral trunk flexion during gait.
It is interesting in this regard that Australopithecus possessed six lumbar vertebrae,
as opposed to the normal five found in modem humans (Robinson, 1972), since this
would have increased the lateral flexibility of the trunk. However, this vertebral pattern
is also found in early Homo (Latimer and Ward, 1993). Other mechanisms, including in-
creased rotational mobility of the trunk during gait, have also been proposed to explain
the retention of six lumbar vertebrae (Cartmill and Schmitt, 1996). The relationship be-
tween vertebral and hip morphology in hominid evolution is one that deserves further
study (Ruff, 1995).
It is clear that A.L. 228-1, and australopithecines in general, were adapted for
hipedality through a large number of alterations in skeletal morphology (see references at
the beginning of this chapter). In addition to those features already described (McHenry,
1991a) may be added the presence of a relatively robust lower limb--the femoral dia-
physis of A.L. 228-1 is very strong relative to her estimated body mass, much stronger
than in modem humans and perhaps stronger even than in early Homo (Figure 6). Since
great apes and modem humans fall very close to each other in lower limb cross-sectional
diaphyseal size relative to body mass (Ruff, 1987b) this implies that Lucy was applying
higher loads on her lower limb than great apes, i.e., that she was bipedal. (Modem humans
have less robust diaphyses in general because of reduced muscularity and activity levels;
Ruff, 1988; Ruff et aI., 1993.) However, if she did walk with a slightly elevated pelvis on
the swing side and/or laterally bent trunk over the support side, the additional muscular ef-
fort involved in tilting of the trunk over the stance limb would be expected to have made
this gait pattern less energetically efficient than in modem humans. This, in combination
with other features, such as relatively short lower limbs, may have limited her mobility on
the ground to some degree. This is consistent with the view that while a facultative biped,
she was likely not a long distance traveler (Rose, 1984; Hunt, 1994).
How far this interpretation can be extended to other australopithecines is difficult to
determine without similar quantitative analyses, which are made difficult by the fragmen-
466 C.Ruff

tary nature of most other specimens. However, recent reconstructions of the pelvis of STS
14, an Australopithecus africanus, are similar to that of A.L. 228-1, including having a
wide biacetabular breadth (Abitbol, 1995; Hausler and Schmid, 1995), and there are addi-
tional similarities in the proximal femur (Figure 8) and other areas of the lower limb
(McHenry, 1986; 1994) between A. afarensis and other species of Australopithecus. Thus,
it is possible that a slightly altered pattern of gait, and consequent restricted terrestrial mo-
bility, characterized australopithecines in general, as compared to later Homo.

6. SUMMARY

To interpret variations in structure of the hominid hip it is first necessary to estab-

lish a valid biomechanical model based on observations of living humans. Here such a
model is applied to the best preserved australopithecine pelvis and femur-A.L. 228-1
("Lucy")-during the single-legged stance phase of gait, a critical component of bipedal-
ity. The model predicts several structural consequences, including a large femoral head
and increased mediolateral buttressing of the femoral shaft and ilium, that are not borne
out in her skeleton. In contrast, the same model works well for predicting skeletal struc-
ture in both modern and earlier Homo (Ruff, 1995). The most parsimonious explanation
is that A.L. 228-1 did not walk exactly like Homo, thus violating the underlying assump-
tions of the model. One possibility is that instead of maintaining her pelvis level or
slightly depressed on the nonsupport (swing) side during gait, she slightly elevated her
nonsupport hip, reducing the abductor and hip joint reaction forces and mediolateral
bending of the femoral shaft on the stance limb. This would also have had the effect of
increasing ground clearance of the swing limb, which could have been advantageous if
her feet were relatively long and/or her stance limb was more flexed than in modern hu-
mans.
This analysis demonstrates the value of considering morphology within a
biomechanical context. It is also important, especially for small-bodied early hominids, to
carefully consider the methods by which "size" is controlled in such analyses. A.L. 228-1
shows many of the hallmarks of bipedality, including a femoral shaft that is quite robust
relative to her body mass, estimated here morphometric ally at about 27-28 kg. However, a
gait pattern characterized by increased lateral tilting of the trunk would probably have
made walking less efficient, and could have reduced mobility on the ground. A slightly al-
tered pattern of bipedal gait may have characterized australopithecines in general, with a
completely modern pattern only established with the appearance of Homo.

ACKNOWLEDGMENTS

I would like to thank Henry McHenry and Elizabeth Strasser for inviting me to par-
ticipate in this conference, Bob Martin for making available the slide of "Lucy's" recon-
structed skeleton and Bruce Latimer for providing casts of her reconstructed pelvis and
femur, Alan Walker and Erik Trinkaus for their collaboration in past and ongoing studies
of structural variation within Homo, and three anonymous reviewers for their thoughtful
comments. Data used in this study were collected in part through support from the Na-
tional Science Foundation.
Evolution ofthe Hominid Hip 467

REFERENCES
Abitbol M (1995) Reconstruction of the STS 14 sacrum and pelvis: AustralopithecusaJricanus. Am. J. Phys. An-
thropol. 96 :143--158.
Aiello LC (1992) Allometry and the analysis of size and shape in human evolution. 1. Hum. Evol. 22:127-148.
Berge C (1991) Quelle est la significance fonctionelle du pelvis tres large de AustralopithecusaJarensis (AL 228-
I)? In Y Coppens and B Senut (eds.): Origine(s) de 1a Bipedie chez les Hominides, Cah. Paleoanthrop.
Paris: Editions du CNRS, pp. 113--119.
Broom R, and Robinson JT (1950) Notes on the pelves of the fossil ape-men. Am. J. Phys. Anthropol. 8:489-494.
Cartmill M, and Schmitt D (1996) Pelvic rotation in human walking and running: Implications for early hominid
bipedalism. Am. 1. Phys. Anthropol. 22:81.
Dart RA (1949) Innominate fragments of Australopithecus prometheus. Am. 1. Phys. Anthropol. 7:301-333.
Day MH (1969) Femoral fragment of a robust australopithecine from Olduvai Gorge, Tanzania. Nature
221 :230--233.
Day MH (1973) Locomotor features of the lower limb in hominids. Symp. Zool. Soc. Lond. 33:29-51.
Ducroquet R, Ducroquet J, and Ducroquet P (1968) Walking and Limping. A Study of Normal and Pathological
Walking. Philadelphia: Lippincott.
Eberhart HD, Inman VT, and Bresler B (1954) The principle elements in human locomotion. In PE Klopsteg and
PD Wilson (eds.): Human Limbs and Their Substitutes. New York: McGraw-Hili, pp. 437-471.
Fe1desman MR, and Lundy JK (1988) Stature estimates for some African Plio-Pleistocene fossil hominids. 1.
Hum. Evol. 17:583--596.
Franciscus RG, and Holliday TW (1992) Hindlimb skeletal allometry in Plio-Pleistocene hominids with special
reference to A.L. 228-1 ("Lucy"). Bull. et Mem. de la Societe d' Anthropologie de Paris n.s. 4:5-20.
Genoves S (\ 967) Proportionality of the long bones and their relation to stature among Mesoamericans. Am. J.
Phys. Anthropol. 26:67-78.
Hausler M, and Schmid P (1995) Comparison of the pelves ofSts 14 and AL228-1: Implications for birth and sex-
ual dimorphism in australopithecines. 1. Hum. Evol. 29:363--383.
Hunt K (1994) The evolution of human bipedality: Ecology and functional morphology. 1. Hum. Evol.
26: 183--202.
Inman VT, Ralston HJ, and Todd F (1981) Human Walking. Baltimore: Williams and Wilkins.
Jenkins FA (1972) Chimpanzee bipedalism: Cineradiographic analysis and implication for the evolution of gait.
Science 178:877--879.
Johanson DC, and Coppens Y (1976) A preliminary anatomical diagnosis of the first Plio/Pleistocene hominid dis-
coveries in the Central Afar, Ethiopia. Am. J. Phys. Anthropol. 45:217-234.
Johanson DC, and Edey MA (1981) Lucy, The Beginnings of Humankind. New York: Simon and Schuster.
Johanson DC, Lovejoy CO, Kimbel WH, White TD, Ward SC, Bush ME, Latimer BM, and Coppens Y (1982a)
Morphology of the Pliocene partial hominid skeleton (A.L. 228-1) from the Hadar formation, Ethiopia.
Am. J. Phys. Anthropol. 57:403-451.
Johanson DC, Taieb M, and Coppens Y (\ 982b) Pliocene hominids from the Hadar Formation, Ethiopia
(1973--1977): Stratigraphic, chronologie, and paleoenvironmental contexts, with notes on hominid mor-
phology and systematics. Am. J. Phys. Anthropol. 57:373-402.
Jungers WL (\985) Body size and scaling of limb proportions in primates. In WL Jungers (ed.): Size and Scaling
in Primate Biology. New York: Plenum Press, pp. 345-381.
Jungers WL (\988a) Lucy's length: Stature reconstruction in Australopithecus aJarensis (A.L. 228-1) with impli-
cations for other small-bodied hominids. Am. J. Phys. Anthropol. 76:227-231.
Jungers WL (1988b) New estimates of body size in australopithecines. In FE Grine (ed.): Evolutionary History of
the "Robust" Australopithecines. New York: Aldine de Gruyter, pp. 115-125.
Jungers WL (1988c) Relative joint size and hominid locomotor adaptations with implications for the evolution of
hominid bipedalism. J. Hum. Evol. 17:247-265.
Jungers WL (1990) Scaling of postcranial joint size in hominoid primates. In FK Jouffroy, MH Stack, and C
Niemitz (eds.): Gravity, Posture and Locomotion in Primates. Florence: II Sedicesimo, pp. 87-95.
Jungers WL (1991) A pygmy perspective on body size and shape in AustralopithcusaJarensis (AL 228-1, "Lucy").
In Y Coppens and B Senut (eds.): Origine(s) de la Bipedie chez les Hominides, Cah. Paleoanthrop. Paris:
Editions du CNRS, pp. 215-224.
Jungers WL, and Stern JT (1983) Body proportions, skeletal allometry and locomotion in the Hadar hominids: A
reply to Wolpoff. 1. Hum. Evol. 12:673--684.
Latimer B, Ohman JC, and Lovejoy CO (1987) Talocrural joint in African hom in ids: Implications for Australopi-
theclts afarensis. Am. J. Phys. Anthropol. 74: 155-175.
468 C. Ruff

Latimer B, and Ward CV (1993) The thoracic and lumbar vertebrae. In A Walker and R Leakey (eds.): The Nariok-
otome Homo Erectus Skeleton. Cambridge: Harvard Univ. Press, pp. 266-293.
Le Gros Clark WE (1955) The os innominatum of the recent Ponginae with special reference to that of the Aus-
tralopithecinae. Am. J. Phys. Anthropo!. 13:19-27.
Le Gros Clark WE (1959) The Antecedents of Man. Edinburgh: Edinburgh University Press.
Lehmann JF, and De Lateur Bl (1990) Gait analysis: Diagnosis and management. In FJ Kottke and JF Lehmann
(eds.): Krusen's Handbook of Physical Medicine and Rehabilitation. Philadelphia: Saunders, pp. 108-124.
Lovejoy CO (1975) Biomechanical perspectives on the lower limb of early hominids. In RH Tuttle (ed.): Primate
Functional Morphology and Evolution. The Hague: Mouton, pp. 291-326.
Lovejoy CO (1988) Evolution of human walking. Sci. Am. 259:118-125.
Lovejoy CO, Heiple KG, and Burstein AH (1973) The gait of Australopithecus. Am. 1. Phys. Anthropo!.
38:757-780.
McHenry HM (1975) Biomechanical interpretation of the early hominid hip. 1. Hum. Evo!. 4:343-355.
McHenry HM (1984) Relative cheek-tooth size in Australopithecus. Am. 1. Phys. Anthropo!. 64:297-306.
McHenry HM (1986) The first bipeds: A comparison of the A. afarensis and A. africanus postcranium and implica-
tions for the evolution of bipedalism. 1. Hum. Evo!. 15:177-191.
McHenry HM (1988) New estimates of body weight in early hominids and their significance to encephalization
and megadontia in "robust" australopithecines. In FE Grine (ed.): Evolutionary History of the "Robust"
Australopithecines. New York: Aldine de Gruyter, pp. 133-148.
McHenry HM (1991 a) First steps? Analyses of the postcranium of early hominids. In Y Coppens and B Senut
(eds.): Origine(s) de la Bipedie chez les Hominides, Cah. Paleoanthrop. Paris: Editions du CNRS, pp.
133-141.
McHenry HM (1991 b) Sexual dimorphism in Australopithecus afarensis. 1. Hum. Evo!. 20:21-32.
McHenry HM (1992) Body size and proportions in early hominids. Am. 1. Phys. Anthropo!. 87:407-431.
McHenry HM (1994) Early hominid postcrania. Phylogeny and function. In RS Corruccini and RL Ciochon (eds.):
Integrative Paths to the Past (Advances in Human Evolution Series, Vo!' 2). Englewood Cliffs: Prentice
Hall, pp. 168-251.
McHenry HM, and Corruccini RS (1978) The femur in early human evolution. Am. J. Phys. Anthropo!.
49:473-488.
McLeish RD, and Charnley J (1970) Abduction forces in the one-legged stance. 1. Biomech. 3: 191-209.
Mednick LW (1955) The evolution of the human ilium. Am. J. Phys. Anthropo!. /3:203-216.
Nagurka ML, and Hayes WC (1980) An interactive graphics package for calculating cross-sectional properties of
complex shapes. J. Biomech. 13:59-64.
Napier JR (1964) The evolution of bipedal walking in the hominids. Arch. BioI. (Leige) 75 (Suppl.):673-708.
Napier JR (1967) The antiquity of human walking. Sci. Am. 216:56-66.
Olivier G (1976) The stature of Australopithecus. 1. Hum. Evo!. 5:529-534.
Porter AMW (1995) The body weight of AL 228-1 ("Lucy"): A new approach using estimates of skeletal length
and the body mass index. Int. 1. Osteoarch. 5:203-212.
Poss R, and Sledge CB (1981) Surgery of the hip in rheumatoid arthritis. In ED Harris, S Ruddy, and CB Sledge
(eds.): Textbook of Rheumatology, Vo!' II. Philadelphia: Saunders, pp. 1960-1972.
Preuschoft H, and Witte H (1991) Biomechanical reasons for the evolution of hominid body shape. In Y Coppens
and B Senut (eds.): Origine(s) de la Bipedie chez les Hominides, Cah. Paleoanthrop. Paris: Editions du
CNRS, pp. 59-77.
Rader WT, and Peters CR (1993) Hypertrophy of the acetabulocristal buttress in Homo sapiens. Am. 1. Phys. An-
thropo!. 92:149-153.
Reynolds TR (1987) Stride length and its determinants in humans, early hominids, primates, and mammals. Am. J.
Phys. Anthropo!. 72: 101-115.
Robinson JT (1972) Early Hominid Posture and Locomotion. Chicago: Univ. Chicago Press.
Rose MD (1984) Food acquisition and the evolution of positional behavior: The case of bipedalism. In OJ Chivers,
BA Wood, and A Bilsborough (eds.): Food Acquisition and Processing in Primates. New York: Plenum, pp.
509-523.
Ruff CB (1987a) Sexual dimorphism in human lower limb bone structure: Relationship to subsistence strategy and
sexual division oflabor. 1. Hum. Evo!. 16:391-416.
RuffCB (1987b) Structural allometry of the femur and tibia in Hominoidea and Macaca. Folia. Primato!' 48:9-49.
Ruff CB (1988) Hindlimb articular surface allometry in Hominoidea and Macaca, with comparisons to diaphyseal
scaling. J. Hum. Evo!. 17:687-714.
RuffCB (1991) Climate, body size and body shape in hominid evolution. J. Hum. Evo!. 21:81-105.
Ruff CB (1994) Morphological adaptation to climate in modern and fossil hominids. Yrbk. Phys. Anthropo!.
37:65-107.
Evolution of the Hominid Hip 469

Ruff CB (1995) Biomechanics of the hip and birth in p.arly Homo. Am. J. Phys. Anthropol. 98:527-574.
RuffCB, and Runestad JA (1992) Primate limb bone structural adaptations. Ann. Rev. Anthrop. 21:407-433.
Ruff CB, Trinkaus E, and Holliday TW (1997) Body mass and encephalization in Pleistocene Homo. Nature
387: 173-176.
Ruff CB, Trinkaus E, Walker A, and Larsen CS (1993) Postcranial robusticity in Homo, I: Temporal trends and
mechanical interpretation. Am. J. Phys. Anthropol. 91:21-53.
Ruff CB, and Walker A (1993) Body size and body shape. In A Walker and R Leakey (eds.): The Nariokotome
Homo Erectus Skeleton. Cambridge: Harvard Univ. Press, pp. 234-265.
Schmid P (1983) Eine rekonstruktion des skelettes von A.L. 228-1 (Hadar) und deren konsequenzen. Folia Prima-
tol. 40:283-306.
Schmitt D, Stern JR, and Larson SG (1996) Compliant gait in humans: Implications for substrate reaction forces
during australopithecine bipedalism. Am. J. Phys. Anthropol. Suppl. 22:209.
Stern JT, and Susman RL (1983) The locomotor anatomy of Australopithecus afarensis. Am. J. Phys. Anthropol.
60:279--317.
Stern JT, and Susman RL (1991) "Total morphological pattern" versus the "magic trait": Conflicting approaches to
the study of early hominid bipedalism. In Y Coppens and B Senut (eds.): Origine(s) de la Bipedie chez les
Hominides, Cah. Paleoanthrop. Paris: Editions du CNRS, pp. 99--111.
Susman RL, Stern JT, Jr., and Jungers WL (1984) Arboreality and bipedality in the Hadar hominids. Folia Prima-
tol. 43: 113-156.
Suzman 1M (1980) A new estimate of body weight in South African australopithecines. In RE Leakey and BA
Ogot (eds.): Proceedings of the 8th Panafrican Congress of Prehistory and Quaternary Studies Nairobi, 5 to
10 September 1977. Nairobi: The International Louis Leakey Memorial Institute for African Prehistory, pp.
175-179.
Tague RG, and Lovejoy CO (1986) The obstetric pelvis of A.L. 228-1 (Lucy). J. Hum. Evol. 15:237-255.
Trinkaus E, Churchill SE, and RuffCB (1994) Postcranial robusticity in Homo, II: Humeral bilateral asymmetry
and bone plasticity. Am. J. Phys. Anthropol. 93: 1-34.
Walker A (1973) New Australopithecus femora from East Rudolf, Kenya. J. Hum. Evol. 2:545-555.
Washburn SL (1950) The analysis of primate evolution with particular reference to the origin of man. Cold Spring
Harbor Symp. Quant. BioI. 15:67-78.
Weidenreich F (1941) The extremity bones of Sinanthropus pekinensis. Paleont. Sinica (N.S. D.) 5D: 1-150.
White TD, and Suwa G (1987) Hominid footprints at Laetoli: Facts and interpretations. Am. J. Phys. Anthropol.
72:485-514.
WolpoffMH (1976) Fossil hominid femora. Nature 264:812-813.
Wood B (1976) Remains attributable to Homo in East Rudolf succession. In Y Coppens, FC Howell, GL Issac and
REF Leakey (eds.): Earliest Man and Environments in the Lake Rudolf Basin. Chicago: Univ. of Chicago
Press, pp. 490-506.
INDEX

A.L., 288, 225, 228, 233, 412, 450-466; see also A us- Aloualta, 6, 45--52,54-56,58, 145,344,367,370,
tralopithecus afarensis 408; see also Howling monkey(s)
Abduction, 37, 111-115, 117-119, 121-122, 124-128, Aloualta caraya, 55
135--136, 159, 163--164,344-345,347,363, Aloualta palliata, 21,45-46,49,408
399,402,439,450-453,462-463 Aloualta seniculus, 367
Acceleration, 167, 178, 250-251,278--279,282-283, Alphadon, 209; see also Marsupial(s)
285 Amazon Basin, 63
Acetabulocristal buttress, 462-463 Amebelodont gomphothere, 379
Acetabulum, 104, 112-122, 124-128,412,451 Amerindian, 460
Aclavicu1ate condition, 159 Amplitude of activity, 163
Activity budget(s), 19, 99, 397, 399,424 Analogues, 31, 426
Activity pattern, 19, 21,61,65 Analysis of Covariance, 181-182, 184, 187, 191,
Ad libitum, 399 195--196
Adapid,114 Anecdotal observations, 34
Adaptation, 16, 23, 41, 45-46, 55, 58, 76, 79, 81, 91, Anecdotal reports, 6
95- 97,114,127-128,131,145,154,158,160, Anesthetized animals, 119, 241
162,164,166-168,175,195,197,212,242, Angle of attack, 167, 194-196
249,252,256-258,261-264,271-272,287, Angled supports, 37, 39, 71,73
289,310,324,337-338,340-342,344-350, Anisotropic, 239
353-356,360,363--365,367,379-380, Ankles, 347
383-388,410-415,420-422,432,435, Anomalurid,379
438-441,444-445,450,465 ANOVA,20
Adaptive radiation, 96, 339-340, 350, 353, 357 Antelopes, 379
Adaptive significance, 45, 289 Anthropoids, 112-113,124,126,137,141,259,266,
Adduction, 36, 119, 125--126, 135-136,343,362,374 325,337,342-343,345,353,362,365--366,
Aegyptopithecus zeLcds, 367, 370, 372, 381 387,389; see also African apes; Asian apes;
Aepycerotine, 379 Catarrhines; Cercopithecoid(s); Gibbon(s);
Aerial phase, 17, 194 New World monkeys; Old World monkeys;
Africa, 95--96, 98, 337, 355, 378--379,404,421-424 P1atyrrhine(s)
African apes, 96-97, 103, 162,256,258,272,325, Anti1opine, 379
340,344,354,374,387-388,410,422,424, Ants, 287, 402, 404
431,435-440,443-445; see also Bonobo(s); Aotus, 64, 147
Chimpanzee; Gorilla; Pan Apes, 54, 76, 95--97, 101-102, 113, 160,272,
Afropithecus, 337-350, 384, 387 325--326,337-350,353-355,357,383-384,
Age-class, 61 387-389,397,409-410,413,421,424,427,
Age-sex composition (or groups), 15, 17,24 429,431,435,438,443,449,465
Aleut, 455 Apomorphies, 339
Allometry, 278--294, 297-302, 304-305, 309-311, Arboreality, 80, 96-97,102-103,128,340,348,355,
313,316-318,320-321,323--327,359,362, 359,363,386,402,405,407,413,431,435
381,426,458,460 Ardipithecus ramidus, 388, 409, 421,431

471
472 Index

Arm-hanging, 22, 259, 342, 399, 401-403, 410-411, Birds, 48, 161, 196,250,267
413-415 BMNH M. 16334,364--365,370
Arm-span, 63 Body length, 34, 63
Arm-swinging, 353 Body mass, 193-194,224,290,301,310-311,313,
Articular surfaces, liS, 125,224--225,370,443--444 316,318,322,324,326,362,380,386,409,
Artiodactyls,379 450,453-459,462,465-466
Asia, 337, 354 Body orientation, 33-34, 36, 38,41, 112
Asian apes, 258, 325, 435 Body size(s), 8, 31, 61-62, 64, 76, 89-90, 95-97, 99,
Ateles, 6,45-52,54--58,90, 145-146, 151,343,366, 102-103,145-146,161,168,248,279,288,
370; see also Spider monkey(s) 302,309-311,316-318,322-327,348-349,
Ateles geoffroyi, 21, 45-46, 49,139 354--355,360,380-381,383,386-388,405,
Atelines, 45, 52, 54, 56,145-146,148,151,154, 407,422,426
344--345,347,360,365 Body weight, 37, 56, 63-64, 10 I, 146, 154, 158, 166,
Australia, 151,255,273 169,179,188,242,251,323,354,360,
Austra1opithecine(s), 214, 223, 257,405,407, 381-382,388,398,403,409,411-414,
409-410,412-415,422,431,449-450,459, 442-443,451-453,458,464
461-463,465-466 Bone deformations, 237-238, 241, 243, 250-252
Australopithecus, 114,224-225,228,257,388,397, Bone strain, 237, 242, 249, 251
407-409,411-413,422,442-445,449-450, Bonobo(s), 97, 290-294, 300, 364--365, 374, 377,
465-466 381-382; see also Pan paniscus
Australopithecus afarensis, 114, 215, 225, 228, 257, Boselaphine, 379, 380
397,407,409-413,415,422,431,442-445, Bout(s), 10-15,20,24,34,36,65,381,401-402,404,
450,466 ' 445
Australopithecus africanus, 257, 407, 409, 422, 466 Bovids, 127,379
Australopithecus anamensis, 388, 409, 422, 431 Brachial index, 289, 410
A ustralopithecus boisei, 224 Brachiation, 79, 148,242,250,337,342,354
Australopithecus robustus, 408-409 Brachiator, 424
Avian morphology, 24 Braking, 179, 182, 190, 192
Aye-ayes, 271-272; see also Daubelltonia Branch, see also Substrate(s); Support
size, 46-47, 51-52, 53,64,71-72
Baboon, 161, 168, 181,184,311,313,317,426; see oblique, 47, 73-74
also Papio obliquely-angled, 71
Bai Hokou, 95, 97- 99, 101-103,440 obliquely-inclined, flexible, 57
Bantu, 454 Bridging, 32, 102, 112, 128. 159, 170,340-341,344,
Baragoi, Kenya, 356, 364 346,348,381
Bearded saki, 149,384; see also Chiropotes Buluk, 338-339
Behavioral categories, 9, 13, 15,32-34,40-41 Burramyids, 151
Behavioral data, 5, 9, 22, 133, 140
Behavioral reconstruction, 5 Cacajao, 32, 64, 370, 384,387; see also Uakaris
Behavioral repertoire(s), 16, 21, 40 Calcanei, 442-443
Bending, 176, 191,238-239,242-243,245-246, Calcaneum(us), 356, 358, 374, 376,411,439
248-249,251-252,313,315,318,323,327, Callicebus moloch, 64
347,384,450-453,458-460,462,464,466 Callicebus torquatus torquatus, 80
Biacetabular breadth, 412, 450-451, 458, 465-466 Callithrixjacchus,134-135
Biaxial strain, 241 Callitrichids, 38,132-134,137-139
Biceps brachii, 362,410-411 Callitrichine, 63, 73, 76, 147
Biceps femoris, 161 Caluromys, 139, 168
Bicipital groove, 366, 370, 410 Cameras, 177-178,211
Biomechanical principles, 5, 464 Canine(s), 75, 269, 349, 354, 357, 381, 383-384
Biomechanics, 5, 38, 40-41, 45,128,237-239,241, Canopy, 45, 64, 81, 83,89,405,407
249,256,261,266,288,290,310,316, Capitate, 342-343
326-327,449-451,456,458,462-464,466 Caprine,379-380
Biostratigraphy, 379 Caprotragoides potwaricus, 379-380
Biped, 257, 424, 426,428,430-431,465 Captive, 6, 23, 89-90, 147,438
Bipedalism, 169-170,290,354,388,397-405, Captivity, 9, 23, 32,115
411-415,419-432,437-441,443--444 Capuchin(s), 58, 145,409; see also Cebus
Bipedality, 256-257, 400, 402, 443-444, 449-450, Capybara, 168
465-466 Carcasses, 355, 387
Bird wing, 267-268; see also Avian morphology Carnivores, 117,119,132, 168,379
Index 473

Carpometacarpal joints, 344, 386 Chiropotes satanas, 32-33, 149

CaUUThines, 342-343,354-367,375,382,384-385, Choerolophodon,379
387-388 Clamberer, 101,424
Catarrhini, 298, 359 Cambering, 364, 381
Categorization, 6, 17, 23-24, 34, 41, 47 Clarke's t-test, 317
Cat(s), 125, 159--161, 164-166, 168, 196 Clawlike nails, 64, 73
Cebuella pygmaea, 64 Claws, 104, 131-133, 136-137, 139--140, 167-168
Cebupithecia, 151 Climacoceridae, 379
Cebus,45-52,54-56,58,64, 135, 145, 154,367,370, Climb(ing), 10, 12, 16-17, 19,22,32,62,68-70,
408; see also Capuchin(s) 74-75,79--80,82,84-85,90,95-98,100,102,
Cebus albi/rons, 64 111-113, 118, 122, 124, 126-128, 160, 162,
Cebus apella, 64, 135, 367 164,167-168,170,197,242,244,249,257,
Cebus capucinus, 21, 45-46, 49 260-261,288-291,324,337,340-342,
Center of gravity, 34, 37, 151, 166, 176, 181, 191, 344-348,350,354,362,364,378,386,400,
194,196,412-413,415,451,463-464 404,406,410-411,413-415,420-421,
Center of mass, 56, 115, 158, 167 424-426,428,432,439--440,444-445
Central African Republic, 95-97, 440 Clinging, 39, 68, 70, 111-112, 117, 126-127, 148,400
Centroidal neutral axis, 238 Coati,48
Cercocebus, 147,353,370; see also Mangabeys Colobinae, 356, 388
Cercocebus albigena, 147 Colobines, 148,3\0,313,315-318,320-323,
Cercopithecid(s), 79, 81-82, 87, 90, 243, 249, 343, 325-327,353,358-359,363,386,388,409
356,386 Colobus, 6, 19--21,62,73,80,82-83,85-89,91,
Cercopithecinae, 356 364-365,370,372,374-377
Cercopithecines, 310-311, 313, 315-318, 320-321, Colobus badius, 19--21,62,73,80,82,86-89,91,377
323,326-327,354,358,361,364-365,367, Colobus guereza, 21, 364-365, 370, 374-377
409 Colobus polykomos, 82-83, 85-86, 88-89, 91, 372
Cercopithecoid(s), 147-148, 151,310-311,313-316, Compression, 40, 47, 64, 170, 175-176, 195,216,
318-322,324-325,327,340,343-344,347, 238-239,241-243,245-246,248,256,315,
353-368,370,378,382-383,385-386,388, 318,320,322,325-327,398
438 Compressive forces, 40, 195
Cercopithecoidea, 386 Computer, 32,41, 113-119, 121-122, 127-129, 178,
Cercopithecoides williamsi, 359 205-207,211,213-214,217-220,241,
Cercopithecus aethiops, 353, 359, 363, 368, 370, 372, 243-245, 432
377 Computerized tomography, 207, 209
Cercopithecus ascanius, 368, 370 Conservatism, 279, 287, 289, 294, 298-299, 305
Cercopithecus cephus, 151, 153,368,370 Continuous recording, 12
Cercopithecus diana, 82, 84, 86-89, 91, 368, 370 Continuous sampling, 10
Cercopithecus Ihoesti, 147-148,368,370 Convergence(s), 23,257-258,261-262
Cercopithecus mitis, 368, 370, 372 Coracoid process, 411
Cercopithecus neglectus, 368, 370 Cortex, 243,245-246
Cercopithecus nictitans, 368, 370, 372 Cortical bone, 313, 325, 458
Chacma baboon, 311; see also Baboon; Papio Cortical control, 166, 169--170
Cheirogaleids, 111, 114 Corticospinal system, 165
Cheirogaleus, 115, 122, 124 Corticospinal tract, 165
Cheirogaleus major, 115 Costa Rica, 46, 58
Cheirogaleus medius, 115 Cost-benefit analysis, 419-420
Chewing, 399, 414 Counterbalance, 149, 151
Chi square, 15-16,24,34,63,65,70,85,114,154, Craniocaudal axis, 34, 36
176-177,179,227,249,310,400,413, Cranium, 214, 255, 258, 266, 358
440-441 Cross sections, 228, 243, 248, 252
Chicago Bulls, 12, 14, 18 Crown
Chicken, 248 diameter, 19
Chimp: see Pan periphery, 46-48, 52, 54, 56-58
Chimpanzee, 80, 158, 161-165,216,228,234,236, Cryptic, 167
338,377,399,400,410-411,413-415,419, Cuboid, 358, 374, 376
425-426,428,430-431,436-437,439-441, Cursors, 158-161
464; see also Pan
Chiropotes, 31-36, 38-40,149--151,349,370,384, Darajani baboon, 313, 317; see also Papio
387 Data acquisition, 207
 474 Index

Data collection, 6, 10, 12, 19,24-25,32,34,41,64, Epicondyle, 360-361, 367, 372, 378, 386
197,211,244,251 Epiphyses(is), 224, 362, 363
Daubentonia, 262, 271; see also Aye-ayes E~thnocebus,353,359,363,367,370,372,377;see
Daubentoniidae, 262, 271 also Patas
Day range, 61, 354, 419-420, 431-432 Eulemur, 10,"17, 19,21, 73, 148; see also Lemur(s)
Degree of habituation, 17 Eulemur fulvus, 10, 13, 19,21
Deltoid, 161-164,411 Eulemur macaco, 148
Dendropithecus macinnesi, 359, 367 Eulemurrubriventer, 10, 13, 19,21, 73
Diameter at breast height, 19, 64 Europe, 337
Didelphids, 151-152 Evolution, 5, 16, 22-23, 33, 40-41, 45-46, 73, 91,
Didelphis virginiana, 140 104,131-132,139-140,142,145-146,151,
Diet(s), 6, 8, 21, 23, 45-46, 48-50,61,63-64,66,73, 154,161,165-166,169-170,175,193,197,
75,96-98,102,145,255,260,349,354-355, 205,217,219,255-258,260-266,268,
378,380,383-388,407-410,413,420 271-273,277-279,284,287-288,302,
Digit(s), 34, 36-37, 73,104,132-137,139-140,142, 304-305,325,337-338,340,348-350,354,
268,291,299,438,444-445 386-389,398,400,419-425,428,430-432,
Dimorphism, 63, 76, 260,311,349,354,380-381, 440,445,465
383 Experimental approaches, 17,23-24, 104, 125, 133,
Distance traveled, 10-12,412 178,217,237,266-271,425,438
Distribution of foods, 98 Extension, 36, 39,47,57,64,118-119,122,124,
Diurnal, 114,354,429 126-128,135-136,151,159-160,168,176,
Diversity, 96, 119, 147,264,266-268,272,338,421 180,191,217,273,287,291,298,300,310,
Dogs, 159, 161, 196,241,248 324,342,344-345,347,357,359-361,367,
Dorsiflexion, 225, 228, 370, 444-445 378,410-411,420,438-439,465
Douc langurs, 378 Extinct, 5-6, 33, 40, 45, 58, 62, 104, Ill, 151, 205,
Drill, 367; see also Mandril/us 211,215-217,359,421,432
Dry season, 19,63,99, 102,440 Exudates, 45, 64, 66, 73; see also Gums; Saps
Dryopitheclls, 357, 367, 372, 380, 383, 386
Duration, 10, 12-13, 17,157, 165, 179,194,251,281, Feedback loops, 169
287 Feeding, 18-19,21,24,35-36,40-41,45-46,48-52,
54-58,63-75,81-82,88-89,96,98-100,102,
Eating, 103,261,397,405,408 126,139,145,149-150,154,256,287,289,
Echidna, 125 380,385,387,397-403,405-406,410,
Ecology, 45-46, 58,62-64, 73, 75-76,95-97,104, 413-415,424,426-432,440-441,445
132,255,260-262,271,287,338,340, Feet, 17,34,36-37,73,82,101,124,131,158,168,
349-350,354-355,388,397,407,415, 266,268,345-347,398,411,435-436,438,
419-420,424,430,432 440,442-445,465-466
Ecomorphology,6, 10,24,63,76 Femur, 39, 104, Ill, 115-122, 124-128,210,223,
Efficient, 13,41,45, 57, 131, 139,251,411,415,426, 311,313-316,318-323,326,362-363,378,
428,465-466 380-382,411-412,449-454,456-466
Elbow, 159, 168, 176-177, 179-181, 185, 187, Ferret, 125
191-192,195-197,243,249,252,256-257, Fibula, 224-225, 339, 344-345, 411
339-342,347-348,361-362,367,378,386 Ficus, 62
Electric resistance gauges, 239-240 Field observations, 17, 147, 149
Electromyograph( ic), 162, 360 Field study(ies), 6, 8, 18,31-32,41,62,79,89,91,95
Electron micrographs, 385 Fingers, 132, 135-136, 140-141, 168,267,339,411,
E~G, 104, 161-165, 170,205,439-440 413
Enamel, 349, 385, 407, 409, 413 Finite element method, 215-217, 219
Energetic(s), 23-24, 61, 76, 97,197,402,412, Fisher's Exact test, 65, 404
419-421,424-426,428-429,431-432 Flexion, 34, 36-37, 39-40,47,64, 104, 117-119,
Energy 121-122, 124, 126-128, 133, 135-136,
budget, 424 140-141,151,159-160,168,176,181,187,
conservation, 45 191,195-197,324,342,344-345,347,360,
costs, 168, 398, 425 362,365-366,378,411,438,441,445,
expenditure, 402, 412, 414, 425-430 465-466
Engineering strain, 238 Flowers, 47-49, 55, 98
Entocuneiform, 363, 370, 375, 378 Focal animal(s), 34, 46-47, 64, 98, 440
Environmental variables, 10,31,252,261-262 Folivores, 409
Eotragus,379 Folivory,95, 114,349,383,385,410
Index 475

Food, 19,32,45-47,55,57-58,61-65,68-69,71,73-74, Gait(s) (cont.)

82,88,96,9&-104,132-133,135-137,139-141, lateral sequence, 157, 165-166
147-148,164,166,169,175,197,349-350, pattern, 157,450,462,464-466
354-355,381,383,387,397-398,400,402-406, Galagids, 127
409-410,413-414,424,426,431,440,445 Galagines, 261, 271
Foot, 57, 127, 146, 151, 157, 159, 167, 169,224, Galago, 73,112,115, 118, 121-122, 124-128
256-257,339,344-347,363,411-412,422, Galagos, 89, III, 124, 127
435-439,441-443,445,450,465 Gallery forest, 63, 380, 388
Footfall(s), 158, 162, 166, 178,323 Gallop(ing), 157, 161, 167,244
Foraging, 19,46,4&-52,54-58,63-75,81- 82, 85-89, Garcinia, 403-404
91,97-98,101-103,139,141,146,165, Gastrocnemius, 161,445
16&-169,354-355,381,387,404-405,414,420, Gastropods, 380
424,428,431-432,437-438,440-441,444-445 Gazella, 379
Force Geladas, 353, 386; see also Theropithecus
absorption, 33 Gibbon(s), 54, 57, 169,242,249,289,325,342,354,
analysis, 104 377-378,383,435,437-438; see also Hylobates
application, 33 Giraffokeryx, 379
plate, 164, 166, 177-178 Glenohumeraljoint, 159, 161,347,360,365
transducers, 237, 249-250 Glenoid cavity, 366
vectors, 451, 453 Gluteus medius, 464
per-unit-mass ratio, 251 Golden-handed tamarins, 104, 133; see also Saguinus
Forearm, 135, 180, 190,243-244,24&-249,361,365, Gombe, 80, 39&-399, 411, 424
370,378 Goodness of fit test, 65
Forelimb(s), 34, 37,47, 56-58, 101,104,133,137,157- Gorilla, 95-96, 98,224, 353,365, 370, 372,375-378
170,175-182,187,191,193,195-197,242-243, Gorilla(s), 62, 89, 95-103, 224, 228, 230-232,
249,289,302,324,337-338,340-342,344, 234-236,287,325,341-342,354,377,383,
346-348,350,367,378,398,400-403,414,441 387,410,435-440,443-444
Forest, 13, 19,31,62,64-67,71,73-74,76,80-83, Gramnivores, 409
85-89,91,96,101,145,151,379-380,397, Grasping, 34, 36-37, 47, 57, 64, 6&-70, 74-75, 82,
403-404,415,419,423,431,441,444 101,104,126,131-137,139,142,146-147,
Fort Ternan, 356, 364, 37&-380 151,164-169,175,196-197,344-345,
Forward momentum, 39 347-348,361,378,386,414,438,444
Fossil, 6, 24, 32-33, 91, 95,104, III, 113-115, 121-122, Grazing, 380
124-128,150-151,205,209,211,213-214, Great apes, 31, 382; see also African apes; Asian apes
217-219,223-224,228,230,232-233,243,256, Greater trochanter, 115, 11&-119, 126
337-341,343-346,350,353,355-356,370,381, Greater tubercle, 357, 359-360,364-365,368,378
385-386,388,409-411,421,424,430,441,444, Grewia,403
449,453-454,460,462 Grip(ping), 131-133, 135-136, 140-141, 146,148,
Fovea capitis, 115, 117, 119 lSI, 154, 165, 168,400,410-411,413
Fox, 125 Guilds, 381
Frequency(ies), 12-13, 16-19,21-24,34,38,40,45, Gums, 64, 73; see also Exudates; Saps
58,62,65,69,71-76,80-81,86,90,97, Guri Lake, 34
99-100,140,163,167,196,244,250-251,
288,323-324,403,405-406,408-409,440 Habitat(s),6, 10, 15, 17-24,31,41,61-62,74-76,
Friction, 33, 37, 145, 168 79-81,86--91,95-99,101-103,165-169,175,
Frugivores, 349, 408-409, 413 177,251,327,354,378,380,386-388,397,
Frugivory,46, 54, 56, 58, 95-97, 145,261,349,383, 403,413,415,420,423,430-432,443,445
409-410 Habituated, 98, 398
Fruit(s), 45-50, 54-55, 57-58, 64-66, 73, 96--99, Habituation, 34
101-102,114,131-132,139-140,146,151, Hadar, 225,423, 442-446
260-261,266,349,355,381,383-387,39&-399, Hagenia, 10 I
402,404-405,408-411,413-415,441,444 Hallux, 36-37, 73, 139, 151-152,345,363,374,378,
422,437-439,442,444
G test(s), 16,22,24,83,98,320 Hamster, 125
Gait(s), 104,118, 157-159, 165-168, 170, 194-197, Hapalemur, 251,261
243,437,449-450,462,464-466 Harungana madagascarensis, 403
asymrnetric(al), 158, 167 Hee1(s), 169,435-445
diagonal sequence, 64, 157, 165-166 Hemiacodon, 114, 121-122, 124-127; see also
irregular, 32 Omomyidae
476 Index

Herbivory, 95, 97, 258 Infraspinatus, 161-164

Heterochrony, 277-280, 287, 302, 305 Inga, 66,75
High forest, 63, 80 Insect(s), 45, 47-48, 64, 66, 70-71, 73, 75,114,131,
Hind limb(s), 37, 39--40, 47, 56-58, 64, 68-69, 74-75, 133,139,196
104, III, 114, 126, 128, 146, 148-149, 151, Instantaneous sampling, 10,34,46-47,64,98-99,398
154,158,160-161,164,166-169,175-176, Intercarpal joints, 342
179, 193, 196,243,249,256,323,378, 398, Interspecific comparisons, 6, 9,16,18,47-48,103,
403,411-412,414-415,441,444,454 122,126,287,289,302,304,310
Hip(s), 104, 111-119, 121, 124-128, 160, 162, 194, Intraspecific variability, 61, 63, 97
211,256-257,347-348,411-415,438, Intraspecific variation, 6, 16-18,24, 228, 23 I, 360, 381
449--454,458,460,462-466 Invertebrate prey, 48, 50, 58,64, 132
Hippopotamids,379 Inverted bipedal posture, 46, 49, 5 I-52, 54-55, 57-58
Histological structure, 139 Iquitos, Peru, 63
Holding, 46,65, 134,428,440 Ischial spine, I 15, 363
Home range, 63-64, 66-67, 74 Ischial tuberosities, 44 I
Homeobox genes, 266-269 Ischium, 115-119,362
Hominid(s), 31,114,223--224,232,354-355,386, Isometry(ies), 278, 282, 284-285, 289, 291, 294, 301,
388,398,404,409,411-412,419--426, 304,309,313,315-316,318,321,323,325
430-432,435,437-438,441-446,449,454,
458-459,462,465-466 Japanese macaque, 161
Hominoid(s), 224,258,289,337-350,354-355,357, Joint loading, I 12
359-361,365-367,370,378,383,385-388, Joint mobility, 104, 11\,128,169,378
397,409,411,432,439,444
Homo, 112,224,357,366,376,422,43 1,442-445, Kalodirr,338-339
450,453-455,457-463,465-466 Kaloma,356
Homo ergaster, 422, 431 Karisoke, 95, 97-99, 101, 103
Homo habilis, 224, 422, 431, 443 Kelvin's principle, 239
Homorus, 380 Kenya, 338-339,350, 355,379,388-389,446,454,
Horizontal branch(es), 37, 47,74 458
Horizontal supports, 36 Kenyapitheclls, 339, 348-349, 355-358, 364-368,
Horn, 379 370,372,374-385,387-388
Horse, 168, 246 Kibale Forest, 19,80,398
Howling monkey(s), 46, 49, 5 I, 56-58, 169; see also Kinematic(s), 17,23-24, 104, 127, 141, 159, 176-ln
A/ouatta 180, 184, 196-197,248,252,288,327,364,
Humans, 95, 112, 114, 140-141, 175, 194, 196,224, 367
228,230,232,234-235,256-258,272,337, Kinetic, 158, 176,178-180,249,251
353-354,366,374,387-389,397,411-413, Kinkajou(s), 137, 140
415,421-422,424,426,428,431,435, Kipsaramon, 356
438-439,443-445,449--451,454,456, Kipsigicerus labidotus, 379, 380
458-466 Kisingiri,338
Humerofemoral indices, 288-289 Knee, 168, 194,256-257,345,347-348,445,
Humerus, 159--164, 168, 197,242-243,267,311, 451-453,465
313--323,325-327,340-341,347,3~6-360, KNM-FT 2751,364-365
364-368,370,372,378,382,386 KNM-MB 12044,357-358,364
Hylohates, 289, 294, 297-302, 305, 360, 366, 370, KNM-MB 13,356,363
372,377,383 KNM-MB 16,356
Hylobates lar, 289, 294, 297-302, 305, 372-377 KNM-MB 19, 356, 360
Hylobates syndactylus, 54, 289 KNM-MB 2, 356
Hylobatids, 45, 54, 342 KNM-MB 21206, 359, 370
Hypergeometric distribution, 65 KNM-MB 24729, 364-365, 370
Hypermorphosis, 279, 281, 283--285, 290, 304 KNM-MB 3, 356, 360
Hypomorphosis, 279, 281, 283, 285, 290, 304 KNM-MB 93, 356, 363
Hypsodontlls tanyceras, 379--380 KNM-RU 17376, 359, 370
KNM-WK 18395, 341
Ilia(um), 347, 412, 452-456, 462-464, 466 Knuckle-walker, 424
Iliotibial tract, 451,453 Knuckle-walking, 96, 99--100,158,162,370'
Incisor(s), 75, 268-269, 349, 357, 383--384,407, Komba,356
409--410,413 Koru,338
Indriid(s), III, 127,250-251,261-262,271 Kudu,388
Index 477

La Selva Biological Station, 46, 48-49, 58 Locomotor classifications, 5, 33, 40--41

Laetoli, 422, 441--442, 445--446 Locomotor events, 13, 16
Lagomorphs, 379 Locomotor groups, 5, 24
Lagothrix, 145,366; see also Woolly monkey Locomotor modes, 45--46, 99, Ill, 114, 124, 159,237,
Landing, 36, 38-40, 167, 194,237,250-252 243,248,348,386
Laser, 211-215, 217-219, 223-225, 232 Locomotor pattern, 6, 23, 103, 125, 127,355,386
Lateral condyle, 411 Locomotor profile, 81, 85--87, 89, 91, III
Latissimus dorsi, 161-162, 164-165,347 Lomako Forest, 80
Leaf-eating, 261, 381, 409 Long bones, 242-243, 246, 310-311, 313, 321,326-327
Leaping, 10, 12, 16-17, 19,21,33,36,38-40,61-65, Lope, Gabon, 96
70,75,79--80,82,85, 111-112, 114, 124, Lophocebus, 363, 368, 370, 372; see also Mangabeys
126-128, 193,237,243,250-251,260-261, Loris, 112, 196
263,288,310,324,326-327,348,353,381 Lorisidae, 261
Least squares regression, 311, 457- 459 Lorisids, Ill, 126-127,271
Leaves,6,45,47-49,54,65, 70,98-99,102,234, Lorisiformes, 258
260,355,408-410,413 Lorisines, 159, 167,261,271
Legs, 47, 49, 57,119,126,438 Lorisoids, 114
Legumes,64,66, 73, 75 LOWESS regressions, 317, 320, 322-323, 325
Lemur catta, 354 Lowland gorillas, 97-103
Lemurfulvus, 122, 161,354 Lowland tropical forests, 96
Lemurids, III Lunate, 112, 115--121, 124-125, 128,342-343,370
Lemuriformes, 258
Lemuroid, 114 M. pronator teres, 361
Lemur(s), 10-13, 19--21, 112, 124-125, 148--149,262, M. triceps brachii, 361
354 Maboko Island, Kenya, 340, 355--356, 359, 363-367,
Leontopithecus rosalia, 134, 137 375,378--381,385--387,389
Lepilemur, 111-112, 115, 121, 124-125, 127-128,261 Mabokopithecus, 356, 359, 370, 385
Lesser tubercle, 360, 366, 370 Macacafascicularis, 147-148, 182, 184,294,
Lever, 193,324,362,437 297-302,305,370,372
Liana(s), 64, 67, 73, 80, 151 Macaca irus, 147
Liberia, 81 Macaca mulatta, 147,367,372
Life span, 10, 206 Macaca nemestrilla, 353, 367, 370, 372
Lift-off, 159--160, 179 Macaque(s),57, 147, 181, 184,237,242-244,
Ligamentum teres, 118, 122, 126 248--250,252,353,388
Limb excursion, 23,104,114,116,119,128,159--161, Madagascar, 250, 354
167,169,196,443-444,465 Magnetic Resonance Imaging, 114-115, 129,207,211
Limnopithecus, 356 Maha1e, 80, 398--399, 403, 411,413
Lizard,161-163 Maizania, 380
Load(ing), 112, 114, 116-117, 119, 124-125, 128, Majiwa, 356
168,215--217,219,236- 239, 242-243, 246, Malagasy lemurs, 11
252,310,327,340,343,346-347,436--437, Malagasy primates: see Lemur(s); Strepsirhine(s)
459,462 Mammal(s), 45, III, 125, 131-133, 137-140, 145,
Locherangan, 338 149,151,157-160,165,167-168,175,191,
Locomotion, 12-13,21,31,39,46,48,61,63,65,68--70, 193,195--197,243,256,269--271,310,321,
74,76,79--82,86-91,97-98,100-102,104, Ill, 324,388,410,421
114,127,157-158,162,164-167,169--170,175, Mandible, 209, 242, 258, 266, 268--270, 349, 384
177,180,182,191,193-197,205,207,217,236, Mandril/us, 310-311, 353, 366-367, 370, 377,
242-244,246,248--249,252,256,260,277-279, 381-382,408
287,289,302,321,325,327,337,340-342, Mangabeys, 147,353,409; see also Cercocebus; Lo-
344-348,350,353-354,362,365,370,378, phocebus
380-381,386-387,398,401--402,413--415, Manipulatory skills, 169
425--426,428,435--437,438,443-445 Mann-Whitney U test, 137, 139,402
Locomotor activity(ies), 12,69,79--80,82,86-87,97, Marmosa, 169
148,244,260 Marsupial(s), 139, 151, 168,209
Locomotor behavior(s), 5, 8,10-13,15,17,19--21, Mastication, 217, 258, 262, 349, 407, 410
23-24,31,61,69,73,76,79--81,85--87, Maturation, 278, 282, 285
89--91, Ill, 148--149,277,287,294,302,304, Mechanical properties, 323, 327
310,322,337-339,400,413,421--424 Medial epicondyle, 360-361, 367, 372
Locomotor categories, 6, 31 Medial longitudinal arch, 437, 442
478 Index

Mediolateral angle, 179 Neandertalians, 445

Megadontia, 349 Neanderthal, 214
Megadonty, 431 Nectar, 64, 66, 73, 132
Afegapedetes,380 Neomorphosis, 283-286, 302
Meissner's corpuscle, 140 Neoteny, 278-279, 282-283, 285
Afesopithecus pentelici, 359 Neotropics, 81,151-152
Mesozoic, 209 Neural control, 161, 165
Metacarpal(s), 133-134, 137, 139-141, 180,246,287, Neural crest, 266, 269
343-344,346,358,370,375 Neurological control, 165
Metacarpophalangeal joints, 133,346, 370, 375, 378 Neuromuscular conservation hypothesis, 161-162
Metatarsals, 39, 339, 345--346,437,444 New World monkeys, 6, 55, 73,134,145,258,325,
Metatarsophalangeal joints, 436, 443--444 342-343,353,359; see also Platyrrhine(s)
Mfangano Island, 338 Niche, 8, 145,255,260--262,266,271,354
Mice, 48, 266, 288 Non,cursors, 158-161
Aficrocebus, 112, 115, 119, 121-122, 124-125, 128; Notharctines, 114
see also Cheirogaleids Notharctus, 114-115, 121-122, 124-126, 128
Microwear, 349, 380, 385,407,409--410,413 Nuclear Magnetic Resonance, 211
Middle Awash, 423 Nut,cracking, 408, 410, 413
Middle Eocene, 114 Nyakach, 356, 380
Midsupport, 179-182, 192, 195 Nyanzameryx, 379
Miocene, 127,337-340,343-344,346--348,350, Nycticebus coucallg, 196; see also Lorisids; Strep'
354-355,357,367,378-379,382,384, sirhine(s)
386--388,444
Afiopithecus talapoin, 311 Occlusal forces, 410
Mobility, 90, 111-115, 124-126, 128, 162, 164, 166, OH35, 224-225, 228, 230--231,233-235
169-170,181,340,343,348,359-360,386, OH8, 224-225, 230--231, 234
411 , 465--466 Old World higher primates, 353, 355, 358, 387; see
Molar(s), 97, 268, 349, 356--358, 381, 383-385, also Catarrhines
409--410,413 Old World monkeys, 112, 126, 147, 175, 177, 191,
Afonanthotaxis poggei, 403 195,258,318,321,323-325,341,344,347,
Monkeys, 13,46,49,53,56,58,73,76,80-81,83, 353-355,361,363,382-383,386,389;see
85--89,91,103-104,113-114,119,137,139, also Cercopithecid(s)
146,148,159-160,162,167,181,184,187, Old World primates, 63, 81
191, 195, 288, 294, 310, 324-325, 340--345, Olduvai Gorge, 223-224, 422, 431
347-348,362,378,386,388,438,449 Olecranon fossa, 367
Aforotopithecus, 338, 348 Olecranon, 342, 361
Morphometric tehniques, 104, 223, 233, 235, Olfactory bulbs, 131
255-257,260--263,269-271,454--458 Oligo, Miocene, 367
Moruorot, 338 Olive colobus. 310, 313: see also Procolobus
Mountain gorilla(s), 95--103, 287; see also Gorilla Omomyidae, 114-115, 126, 128
Moustached guenon, 151, 153; see also Cercopithecus Omomyines, 114, 127
cephus Omomys, 114, 120--122, 124, 126--127
Moustached monkeys, 288, 294; see also Cercopi, Ontogeny, 76, 104, 139, 142, 146, 169-170,206--207,
thecus cephus 217,255,266--273,277-279,281-291,294,
Moustached tamarins, 63-67, 69-72, 74-76; see also 298-302,304-305,327,365,376,378,386,
Saguinus mystax 420,462
Multi,male groups, 354 Opossum(s), 125, 159, 161-164
Muscle(s), 39, 45,104,119,145,158,161-166,169, Orangutan(s), 62,101, 126, 169,325--326,342,354,
176,207,242-243,248,251-252,258,324, 378,383,408--410,435,438; see also POllgo
342,345,347,349,362,410--412,437,439, Oreopithecid, 356
440,445,450,458,464 Orthograde, 38, 47, 112, 124, 148,354
Muscular effort, 46, 324, 465 Os coxae, 412
Musculoskeletal system, 46, 252, 421 Ostriche(s),379-380
Otolemur crassicaudatus, 73; see also Cheirogaleids
Nails, 65, 73,104,131-133,137,139,197
Napak, Uganda, 356 Padre Isla, 22, 63-69, 71-72, 74-75
Nasalis larvatus, 311, 377, 382; see also Proboscis Paedomorphosis, 277-279,282,284-286,290--291,
monkey 293-294,301
Navicular, 358 Paleobiogeography, 379
Index 479

Paleoecology, 378, 386 Pithecia aequatorialis, 64

Paleoenvironments, 378-379 Pithecia pithecia, 32-33
Paleotropics, 81 Pitheciini, 32
Palm swamps, 64 Pitheciins, 349
Pan, 55, 112,224,258,268,290-294,296,300-302, Pitting, 409, 413
305,349,353,364-365,370,372,374-378, P1antarflexion, 149,411,444
381-382, 408, 424, 444; see also Chimpanzees Plantigrady, 169,345,437-439,441,445
Pan paniscus, 290-291, 294, 296, 300, 305, 382; see Platyrrhine(s), 32, 56, 104, 132-133, 138, 145, 147,
also Bonobo(s) 149-151,154,325,360-361,367,370,387;
Pan troglodytes, 55, 224, 290-294, 296, 300, 305, see also New World monkeys
349,370,372,375-377,444 Pliocene, 419, 422, 431, 435, 442, 444
Papillae, 140 Pliopithecus vindobonensis, 367, 370, 372
Papio, 353, 359, 361, 363, 366-367,370,372, Plio-Pleistocene, 223, 355, 421-422, 430-431, 443,
375-377, 382; see also Baboon 445-446,459
Parametric tests, 15-16, 48, 315 Poisson's ratios, 238
Paranomalurus,380 Pollen, 380
Parasagittal plane(s), 119, 159 Pollex, 34,344,382
Parsimony, 386, 466 Polychromatic techniques, 207
Parturition, 412 Pongid, 439, 442; see also African apes; Asian Apes;
Patas, 119, 161-162, 164, 181, 184,353,359,363, Great apes
367,372,377; see also Erythrocebus Pongo,55,112,360,370,372,377,444
Patch size, 97, 102 Positional behavior(s), 5, 6, 9-10, 12, 15-16, 18-19,
Patellar groove, 380, 411 21-25,31-34,38,40-41,45-52,54-58,
Pecos Pueblo Amerindian, 456 61-65,68-71,73-76,79-80,90,91,95,
Pectoralis muscles, 161-165,411 97-99,102-103,124, 139, 148, 150-151,277,
Pedal arches, 443 279,290,294,310,353-355,386,397-398,
Pedal ball, 437 400-401,403,407,410,435,440-441,445
Pelvis, III, 115, 117-119, 124-125,214-215, Postcrania, 91
411-413,415,422,443,445,449-455, Postcranial anatomy, 33, 291, 294, 337-338, 348, 350,
462-466 378,386,413,422
Peramorphosis, 277-279, 282, 284, 286, 289-290, Postcranial skeleton, 40, 223, 286, 337, 350, 363, 379,
298-299,301 387
Periosteum, 243 Postcranium, 31,255,257-258,260,266,346,358,
Perodicticus, 112-113, 115, 121-122,124-125,128; 422
see also Lorisids; Potto; Strepsirhine(s) Postural behavior, 5, 63-64, 73,98, 149,400,440
Peru, 63, 66, 75,80,442 Postural events, 13
Petaurids, 151 Posture, 8, 12, 17,23, 31-32, 34, 36, 39,46,48-49,
Phalangeal flexure, 131 51-58,61-62,64,68-70,74-76,80,97,
Phalangerids, 151 99-102,104, III, 114-115, 117, 119,
Phalanges, 133, 139-141,287,344-347,356,363, 121-122,125-128,148-150,160,168-170,
377-378,411,443-444 176-177,191,193,195-197,256,310,322,
Phalanx, 339, 344, 356, 370, 375-376, 381-382 324,338,340-348,361-362,365,370,398,
Phenotypes, 277 400-403,414-415,435,438-439,445
Phenotypic variation, 76 Potos, 134, 137,152
Philopatry,382 Potto, 111-112, 114, 121, 124-126, 128; see also
Photoelastic studies, 216-217 Lorisids; Perodicticus; Strepsirhine(s)
Photon absorptiometry, 310 Power grips, 134
Phylogenetic inertia, 311, 317 Predator(s), 167, 191,382,387,407,414,424
Phylogeny, 97, 114,255,258-260,271,287,289,311, Prehensile hand, 131-132, 140
317,324,.339-340,354,357-358,367,379, Prehensile tail(s), 56,104,145-148,150-152,154
385,388,423-424 Prehension, 64, 74, 146,444
Physical stress, 46 Premolar, 209, 357
Physiological factors, 61 Presbytis, 6, 353, 359, 363, 367-368, 370, 372, 382,
Physiology, 61, 76, 104,431 388
Phytoliths, 380 Presbyfis entellus, 353, 359, 367-368, 370, 382, 388
Pine forests, 19 Presbyfis rubicundus, 363, 368, 370, 372
Pisiform, 343 Primary forest, 19, 63
Pith, 47, 408 Primitive, 139,161-162,337,339,342-343,350,356,
Pithecia, 31-40, 64,149-151,384,387 367,374,384-386,422
 480 Index

Proboscis monkey, 310, 313,377; see also Nasalis lar- Reconstruction of behavior, 45, 58, 62, 91,114-115,
vatus 117-118, 125,207,209,214,340,355,367,
Procolobus, 311, 361; see also Olive colobus 383,385-388,407,413,420,455-456,466
Proconsu4337-350,359,365,367,370,372, 374, Recruitment, 104, 158, 161-166, 169
380-381,383-384,386-387 Red colobus monkey(s), 19,82,377-378; see also
Proconsulafricanus,338 Colobus badius
Proconsul heseloni, 338-339, 347, 349 Reduced major axis (RMA), 311, 313-317, 319-321,
Proconsul major, 338 455,457-459,461
Proconsulnyanzae, 338,344,347-349,380,383 Repertoire, 22-24, 38, 62-63, 68-72, 74-76, 100-101,
Procyon, 134, 137 112, 114, 124, 127-128, 139,257,346,348,
Procyonids, 134, 151 354,363,413,431,444-445
Progenesis, 281 Resting, 18,24,35,40,98-99,126,397,424-427,
Pronation, 37, 162,340-343,347,361,363,370,378 429,431
Pronograde, 32,39,47,64, 112, 124,326,338,347, Retraction, 159-160, 181
353,365,387 Rhinopithecus, 353, 361
Propithecus, 10, 13, 17,19,21,73,148,251; see also Rio Blanco, 63-67, 69, 71-72, 74-75
Indriid(s) Rolling movements, 56
Proportions, 10, II, 12, 14-16,20-21,104,133,137, Rotation, 117-118, 125, 127, 159, 179,207,248,
140-141,168,257-259,263,278,284-285, 344-345, 384
287-290,294,300-302,363,383,386,411, Running, 13,64,82--83,85,112, 157-158, 167, 194,
445,451 225,241,244,261,327,348,360,363,437,
Propulsion, 37, 39, 64, 158, 164-165,249 445
Propulsive force( s), 40, 166, 181 Rusinga Island, 338, 367
Propulsive peak, 179, 182, 187, 190-192 Rwanda, 95, 97
Prosimians, 38, 63, 76,104,134-135,160,237,259,
262,271,325,327,389,438 Sacrum, 115, 454
Protohominids. 405, 414 Saddle-back tamarins, 63, 74, 76; see also Saguinus
Protraction, 104, 133, 159-161, 166-170, 181, 184, fuscicollis
187,193-195,197,302,360,365,378 Sagittal angle, 179
Pubic symphysis, 115, 454 Sagittal planes, 243
Pygmies, 455, 457 Sagittal resultant, 179
Saguinus, 6, 21, 38, 61, 63, 66, 68, 70, 72, 74-75, 80,
Quadriceps femoris, 161 131-137, 139-141,370; see also Tamarin(s)
Quadrumanous, 101,291 Saguinusfuscicollis, 63, 74-76
Quadrupedalism, 10, 17,19,22,32-34,36-37,39, Saguinus geofJroyi, 134
46-47,49,51-52,54,56,61-65,68-70,75, Saguinus granatensis, 141
79-80,82,85,90,99-100,102,104,111-112, Saguinus midas, 139-140
114,121,124,128,139,157-162,164-169, Saguinus mystax, 21-22, 61, 63, 66, 68, 70, 72,75-76,
175-177,184,191,193,196,256,290-291, 80
321,323-327,341,346-348,353-354,360, Saimiri, 64, 90, 133-135, 137-139, 147-148,370
362,365,367,374,386-387,398,402,425, Saimiri sciureus, 21, 64,133-135,137,148
427-431,438,440,449 Sampling, 9-10,12-15,18,24,34,46,64-65,82--83,
Quasidigitigrade, 443 99,244,326,440
Santa Rosa National Park, 46, 48-49, 58
Rabbits, 158, 166 Saps, 64; see also Gums; Exudates
Raccoons, 137, 140 Savanna,80,353,377,379-380,386-387,419,
Radius, 160,242-243,246,249,267,341,343,362, 423-424,445
367,370,454 Savannah Monitor lizard, 162
Raleighvallen-Voltzberg Nature Reserve, 80 Scaling, 227, 279, 281-289, 291, 294, 298-302,
Ramapithecus, 380, 384 304-305,310-311,313,315-316,318,
Rank order, 12-13, 16,21-22,73,317-318,322 320-321,323,326-327,457-458
Ranomafana National Park, Madagascar, 19 Scan sampling, 440
Rat(s), 125,266,269-270 Scaphoid, 342-343, 370
Rate(s), 46, 48-49,51-52,54-55,57-58,71,90,136, Scapula, 159, 161-162, 164, 168,287,291-292,294,
139,277-279,281-285,290,304,403,414, 296-302,305,347,366
426-427,429,441 Sciurus carolinellsis, 134, 139
Reaction forces, 104, 112, 124, 167-168, 176, Sclerocarp, 348, 387
190-193,196,249,252,436,445,463, Seed eating hypothesis, 397
465-466 Seeds, 45, 64,348-349,355,385-386,397-398,409
Index 481

Selective pressures, 31,400 Substrate(s), 6, 8,10,15,18,23-24,32,34,40-41,45,

Semi-brachiate, 31 61,63,65,71-75,79,83,88-89,96-103,
Semi-brachiator, 6 124--125,127,167-168,176-182,187-188,
Semi-captivity, 9 190--194,196-197,249,251-252,310,316,
Semi-terrestrial(ity), 177,353-355,357,360, 323-324,327,353-355,362-364,379,
362-363,365-366,378,386-388 386-388,402,413-414,426,438,442,444;
Sesamoid, 345, 347, 437 see a/so Branch; Supports
Shear strains, 238 Supination, 135, 149, 162,340--343,347,370,378
Shoots, 47 Support phase, 159-164, 166, 180--182, 185, 187-188,
Shoulder, 22, 57,104,159,161-165,168,177, 191-192,195-197,241-242,245-246,248,
179-182,187,191-192,195,244--245, 450,464
255-259,347-348,359-360,364,378 Supports, see a/so Branch, Substrate(s)
Sifaka, 148, 288; see a/so Propithecus deformable, 37, 39-40
Simio/lIs, 356, 367, 372, 385 discontinuous, 65, 164
Sitting, 10, 13,32-37,40,49,51-52,55-57,54,64, size, 36-37, 39, 90
68-70,74--75,99,402,414--415,438,440-441 use,33-34,50, 72, 79,81, 83,86-91
SivapitheclIs, 342, 356-357, 374, 386 oblique1y-oriented,71-72
Skeleton, 6,23,31,45,76, 127-128, 134, 142, 148-151, Supracondylar ridge, 410
205,213-217,219,223,242,266,277,288,304-- Supracoracoideus, 161-163
305,311,327,337-339,347,357-358,363-364, Supraspinatus, 161-163,359-360,364
388,410,422,450-451,453-455,462,465-466 Surinam, 80
Skull, 210, 212-213,224,258,338,383 Survival, 46, 397, 420
Slippage, 37 Suspension, 45-51,53-58,61--62,96-97,99,
Small branches, 46,51-52,56-58,62,70--72,99,165, 101-102,111-112,114,128, 145-146,
169,413-414,444 148-151,154,160,162,164,170,290--291,
Small supports, 50--51, 72 325-326,337-338,341-342,344,346-348,
Small-object postural feeding hypothesis, 398 350,366,381,386,401-403,413,415
Social environment, 76 Suspensory behavior
Social organization, 354, 381, 383 forelimb, 74,158,164,166-167,169
Social rank, 405 Suspensory feeding, 46, 49, 54--58
Social structure, 6 Swimming, 79
Social systems, 381 Swing phase, 161-163,245,438
Soft tissue, 119, 126,455 Sykes monkeys, 148
Sole, 401, 437-438, 442 Sympatric, 103, 149,354,385
Songhor, 338 Synapomorphy(ies), 337, 344, 350, 356
South America, 151
Spearman correlation, 13, 181, 317, 322 Tai Forest, 79-81, 87-88, 91-92, 103,324--325,441
Spider monkey(s), 49,53,55-58, 146, 148, 154, Tail, 47, 49,55,57,126,145-151,154,348,379-380
160--162, 164, 242, 250; see a/so A te/es Takeoff,36,38-40,111,237,250,251,252
Spring hare, 380 Talapoin, 288, 294, 310--311, 372; see a/so Miopi-
Spring-mass models, 193-194 thecus ta/apoin
Squatting, 99, 414, 435, 437-441, 443-445 Ta1ar joint, 225, 227
Squirrel monkey(s), 104, 133-136, 139, 142, 148, 160, Ta1ar trochlear surface, 225, 227-228, 345
325; see a/so Saimiri Ta1atakely, 19,21
Stability, 37, 56, 73, 76, 126-128, 162, 166-167, 197, Ta10crural joint(s), 224--228, 230, 232-233
378,411,414--415 Talus, 224--228, 230--232, 344--345, 445
Stance phase, 117, 164, 168,245,248,360,437,466 Tamarin(s), 61, 63-76, 103-104, 132-136, 139,
Standing, 35-37 141-142; see a/so Saguinus
Stature, 64, 67, 411, 454--456, 459 Tanzania, 80,223-224,399,411,441-442,445
Stercu1iaceae, 380 Tarsal bones, 39, 127, 149,438,443
Strain, 125,216-217,237-252 Tarsiers, 127,261,263,271-272
Strain gauge(s), 217, 237, 239-243, 249-250, 252, 384 Tarsiidae, 261
Strepsirhine(s), 111-114, 119, 121, 124, 126, 128, Tarsiids,271
259,262,266,268,271,354 Tarsiiformes, 258
Stress, 46, 125,216-217,311,410,412-413,423, Tarsipedids, 151
431,441 Tarsius,39, 126-127
Stride(s), 157, 161, 167-168, 196,205,323,402,414,465 Teeth, 139,223,258,268,270,294,302,338,340,
Students I-lest, 48, 65 348-350,358,384,407-409,413
Subchondral bone, 115-118, 121, 127 Temporalis, 349, 383
 482 Index

Tension, 37, 47, 57, 64, 175, 197,238,242-243, Ulna, 160, 162,237,242-243,246-249,252,267,339,
245-246,256,318,342,411 341-343,356,361,367,370,374,382,386,422
Tenninal branche(s), 39,47,57,75,82,97,139,145, Ungulates, 168
175,197,348,401-403,405-406,408,411,414
Tennites, 287 Varecia, 19, 73,148-151
Terrestriality, 95, 353, 355, 357, 431 Vastus lateral is, 161
Tetrapod(s), 161, 164,267 Vatoharanana, 19,21
Thennoregulation, 421, 431 VCL, III, 114, 127: see Vertical clingers and leaping
Theropithecus, 353, 359, 363, 367, 370, 372, 382; see Velocity,39, 178, 194
also Geladas Venezuela, 33
Thorax, 162, 164,257 Vertebra(e), 39, 145, 150,256-258,347,410-412,465
3-0,104,179,207-208,211,213-215,217-218 Vertebrate(s), 47-48, 73, 104,161,163,310
Three-dimensional data, 205, 223, 225; see also 3-D Vertical clingers and leaping, 38, 114,250-252
Three-dimensional modeling, 112, 114 Vertical supports, 39, 47,56,70,72, 197,410,441
Thumb, 135-136, 344, 411; see also Pollex Vervet, 148, 159, 164-165, 181, 184, 187, 191, 195,
Tibia, 216, 224-225, 227-228, 230-232, 241-243, 313,377-378
246,249,422,445,452 Vibrissae, 139
Tibialis anterior, 161 Victoriapithecids, 386
Tibiotarsus, 248 Victoriapithecus, 355-363, 367, 377-378, 381, 383,
Time 385-386
budget(s), 419-421, 424, 426, 428-429, 431-432 Victoriapithecu leakeyi, 356
duration, 12 Victoriapithecus macinnesi, 356, 359, 363-364, 370,
samples, 10-15,24 372,383
sampling method, 13 Videotape, 12, 17,32,34,41,133-134,136-137,158,
and space, 18 177-179,195,207,218,244-245,251
Tinderet, 338, 380 Vine, 47
Titi monkeys, 80; see also Callicebus moloch; Callice- Virungas, 95-96
bus torquatus Visoke, 438, 440
Toe-off, 250-251, 411, 442 Visual predation, 131
Toes,97,132,339,411,413,436-439,442,444 Volar pad, 145, 154
Torsion, 238, 246, 248 Volar skin, 139-140
Touchdo1Nn, 159-160, 166, 168-169, 179, 181-182, Volar surface, 135-136
184,195-196,251
Trabecular bone, 124, 209-210 Walking, 22, 32-34, 37, 64, 79, 82, 100, III,
Tragelaphus kyaloae, 388 157-165,167,176-177,181-182,184,
Transverse dorsal ridge, 370 191-197,241-244,246,248,342,344,354,
Transverse processes, 347 360,363,370,378,388,398,400,412-414,
Trapezia, 343-344 421,425-426,437-438,442-445,464-466
Traveling, 18-19,24,35,63,70,81-82,85-89,91,145, Weight-bearing, 34, 37, 341-342, 345, 347, 360, 450,
161,178,181,397-398,424-426,428-432 454,464
Tree cro1Nn, 46-48,54-58,61-62,65,71,73 West Turkana, 422
Tree height, 19 Western black and 1Nhite colobus, 82-83
Tree shre1N(s), 125, 131, 134, 139-141 Western I01Nland gorillas, 95-96
Triceps brachii, 324, 361, 367 Wet season, 19,63,66,99,440
Triceps surae, 437, 439-440, 445 Woodland,22,80, 380, 386-388,397,415,419,430
Triquetral, 342-343 Woolly monkeys, 161, 164; see also Lagothrix
Triquetrolunate joint, 343 Wrist(s), 37,160,179-180,243,256-257,339,
Trochlea,249,344-345,359-360, 367,378 342-343,347,370.386,410
Trochlear notches, 341
Trotting, 158, 161, 167 X-ray CT, 207, 209-210, 218-219
Tuberous roots, 355, 387 X-rays,117-119,12I,128,207
Tupaia, 134, 139-141; see also Tree shre1N(s)
Turkanapithecus, 348 Zaire, 80
T1Nigs,47,81-82,85-86,402,410,413-414,444 Zenkerella, 379
T1No-sample test of Manly, II, 20 Zhoukoudian, 460-461
Type I errors, 15 Zinjanthropus, 224
Zona conoidea, 340-341, 378
Uakaris, 387; see also Cacajao Z-scores, 65, 218
Uganda, 80, 356 Zygomatic, 357, 384