MINI REVIEW

Tempo and mode in evolution: phylogenetic inertia, adaptation

and comparative methods

S. P. BLOMBERG & T. GARLAND, J R

Department of Biology, University of California, Riverside, CA, USA

Keywords: Abstract
adaptation; Before the Evolutionary Synthesis, ‘phylogenetic inertia’ was associated with
comparative methods; theories of orthogenesis, which claimed that organisms possessed an endog-
constraint; enous perfecting principle. The concept in the modern literature dates to
independent contrasts; Simpson (1944), who used ‘evolutionary inertia’ as a description of pattern in
natural selection; the fossil record. Wilson (1975) used ‘phylogenetic inertia’ to describe
phylogenetic inertia; population-level or organismal properties that can affect the course of evolution
statistics. in response to selection. Many current authors now view phylogenetic inertia as
an alternative hypothesis to adaptation by natural selection when attempting to
explain interspecific variation, covariation or lack thereof in phenotypic traits.
Some phylogenetic comparative methods have been claimed to allow quanti-
fication and testing of phylogenetic inertia. Although some existing methods do
allow valid tests of whether related species tend to resemble each other, which
we term ‘phylogenetic signal’, this is simply pattern recognition and does not
imply any underlying process. Moreover, comparative data sets generally do not
include information that would allow rigorous inferences concerning causal
processes underlying such patterns. The concept of phylogenetic inertia needs
to be defined and studied with as much care as ‘adaptation’.

direction (e.g. during the evolution of the toes and teeth

Introduction
of horses). However, he argued that: (1) the patterns
We review the concept of ‘phylogenetic inertia’ and showed many irregularities; (2) no evidence of an
consider if and how it can be studied by modern endogenous mechanism had ever been found; and (3)
comparative methods. Before the Evolutionary Synthe- invocation of such a mechanism was unnecessary
sis, phylogenetic inertia was associated with theories of because patterns in the fossil record could be explained
orthogenesis, which, according to Mayr (1982, pp. 529– by natural selection (Simpson, 1944, pp. 161–163). Thus,
530), claimed that organisms possessed an endogenous for Simpson, phylogenetic inertia (Simpson uses the term
‘perfecting principle’. The term ‘inertia’ was used by ‘evolutionary inertia’) represented a pattern of ‘rectilin-
some proponents of orthogenesis as a direct analogy with ear’ evolution, by which he meant linear directional
inertia in physics: once organisms begin to evolve in a trends in the fossil record, generated by the process of
particular direction, they tend to keep evolving in the natural selection. Rectilinear trends were part of the
same direction (cf. Burt, 2001). In his refutation of dynamics (tempo) of evolution, and not associated with
orthogenetic arguments, Simpson (1944) acknowledged phyletic stasis. This is in stark contrast with more recent
that the fossil record did sometimes exhibit patterns that definitions of the concept.
suggested evolution was proceeding in a particular Since the Synthesis, the concept of phylogenetic
inertia has undergone considerable revision. Instead of
just describing patterns in the fossil record, the term has
Correspondence: Theodore Garland, Jr, Department of Biology, University
of California, Riverside, Riverside, CA 92521, USA. become associated with various factors, other than
Tel.: (909) 787-3524; fax: (909) 787-4286; natural and sexual selection, that may affect phenotypic
e-mail: tgarland@citrus.ucr.edu evolution. Ridley (1983) claims that the term was

J. EVOL. BIOL. 15 (2002) 899–910 ª 2002 BLACKWELL SCIENCE LTD 899

900 S. P. BLOMBERG AND T. GARLAND

introduced by Wilson (1975), who writes (p. 32): five possible causes for phylogenetic inertia of coopera-
‘Phylogenetic inertia…consists of the deeper properties tive breeding: 1. Limited genetic or phenotypic variation
of the population that determine the extent to which its (similar to Wilson’s 1., above). 2. Pleiotropy and genetic
evolution can be deflected in one direction or another, as covariance. If the same genes underlie two or more traits,
well as the amount by which its rate can be speeded or then evolution of one of the traits (e.g. cooperative
slowed’. Wilson lists four mechanisms that create phy- breeding) may be constrained by selection on the
logenetic inertia in the social behaviour of animals. other(s) (similar to Wilson’s 4., above). 3. High correlations
1. Genetic variability. Organisms can only respond to with metric traits. This seems to be a special case of 2. 4.
selection in proportion to the genetic component of Functional interdependency of components of a trait. If
phenotypic variability. 2. Antisocial factors. Wilson notes juvenile survival is linked to cooperative breeding, then
that various idiosyncratic, lineage-specific effects can alternative strategies may find it impossible to ‘invade’.
affect the direction of evolution. 3. The complexity of the This appears to be a special case of selection on the
social behaviour. The more complicated the behavioural components of cooperative breeding. 5. Behavioural
phenotype and the supporting physiological machinery, plasticity. If individuals can modify their behaviour, then
the greater the inertia. 4. The effect of the evolution on other they can limit the effect of selection on traits (see also
traits. Inertia of behavioural phenotypes is increased if Garland et al., 1990).
they are correlated with other traits that may affect A recent attempt to clarify the term phylogenetic
fitness. Ridley (1983) notes that Cain (1964) lists similar inertia (and related terms) has been undertaken by Burt
processes under the term ‘genetic inertia’, which appar- (2001). He argues that phylogenetic inertia should be
ently is attributable to Darlington & Mather (1949). In treated as a phenomenological pattern description of traits
any case, Wilson proposes various processes (mechan- among species (fossil or extant). For Burt, phylogenetic
isms) that can affect the course of evolution. All of these inertia is defined in analogy to inertia in physics, ‘…a
would appear to involve well-accepted (if sometimes character with an unchanged character state will remain
obscure) biological principles, not any endogenous ‘per- unchanged and a character experiencing consistent
fecting principle’. directional change will maintain that evolutionary pat-
Clearly, Wilson’s (1975) conception of phylogenetic tern between generations of a lineage unless an external
inertia (factors that tend to resist selection) is the virtual resultant force acts on it’. Burt points out that in most
opposite of Simpson’s (1944) (patterns that have resulted situations we are unable to observe each generation, and
from selection). However, Wilson (1975) also says that so offers an alternative operational definition, ‘…a
phylogenetic inertia includes pre-adaptation, the concept character with an unchanged character state will remain
of traits that evolve for one function and later get unchanged and a character experiencing consistent
co-opted as an adaptation for another, different function. directional change will maintain that evolutionary pat-
As noted by Gittleman et al. (1996a), after discussing tern between branches of a phylogenetic tree unless an
phylogenetic inertia on page 32, Wilson (1975) empha- external resultant force acts on it’.
sized adaptive explanations for behavioural traits, and Burt’s (2001) analogy with physics is historically
phylogenetic inertia is mentioned in only one other place interesting, in that at first sight he appears to retrace
in the entire book. the steps of the pre-Synthesis orthogeneticists in drawing
Ridley (1983) defines phylogenetic inertia simply as a a close parallel with physical inertia; moreover, he
character that is shared among related species, although ignores subsequent attempts in the biological literature
he admits that this definition is unclear. Berger (1988) to give phylogenetic inertia more biological meaning
offers a similar definition. However, many modern (e.g. Simpson, 1944; Wilson, 1975; Edwards & Naeem,
researchers have followed Wilson’s (1975) sentiment, at 1993). Burt (2001) did not discuss how the various terms
least in part, and now equate phylogenetic inertia with he defines might actually be studied with comparative
nonadaptive (or maladaptive) phenotypic stasis. Gould & data. We can ask, however, ‘What are the forces?’
Lewontin (1979), for example, recognize phylogenetic Clearly, they are not to be interpreted as physical forces,
inertia in the fact that humans are not optimally designed but simply the usual evolutionary processes of natural
for upright posture, because much of our Bauplan and sexual selection, mutation, genetic drift, and gene
evolved originally for quadrupedal locomotion. Edwards flow. Here the analogy with physics breaks down for
& Naeem (1993) provide a particularly good discussion in various reasons. Unless overridden by selection, random
the context of the evolution of cooperative breeding in mutation and genetic drift will cause genetic and hence
birds. They define phylogenetic inertia as occurring phenotypic evolution in any finite population. Therefore,
‘…when traits persist in lineages after the cessation of (absolute) stasis cannot generally be assumed as the null
selective forces thought to have produced or maintained hypothesis for trait evolution. For quantitative pheno-
them or through episodes of selectively important envi- typic traits, selection can produce no change in the
ronmental oscillations’. They say their definition implies population mean (stabilizing selection), directional
a tendency for traits to resist change, despite environ- change or even disruptive change (selection against the
mental perturbations. Edwards & Naeem (1993) provide mean), yet the process is fundamentally the same in all

J. EVOL. BIOL. 15 (2002) 899–910 ª 2002 BLACKWELL SCIENCE LTD

 Phylogenetic inertia 901

cases: differential survival and reproduction of individu-

Examples of empirical studies that invoke
als determined by their genetically heritable traits.
phylogenetic inertia
Modern empirical studies, especially of behaviour,
Phylogenetic ‘inertia’ and ‘constraint’
often invoke ‘phylogenetic inertia’. Shapiro (1981), for
The various meanings and applications of ‘constraint’ instance, analysed egg deposition in pierid butterflies. He
in evolutionary biology have been discussed in great suggested phylogenetic inertia as a reason why some
depth (e.g. Maynard Smith et al., 1985; Antonovics & butterflies disperse their eggs on their new hosts in an
van Tienderen, 1991; Janson, 1992; McKitrick, 1993; inappropriate manner (adaptive on previous hosts, now
Schwenk, 1995), yet the relationship between con- nonadaptive on the new, current hosts), resulting in a
straint and phylogenetic inertia is unclear. Our own shortage of laying sites. Peterson (1991) attributed the
view is that modern biologists often use the two terms occurrence of delayed maturation in the soft-part colour
in an interchangeable fashion, and often casually. We of some New World jays to phylogenetic inertia, as a
surveyed five popular textbooks of evolutionary bio- hypothesized strong relationship between delayed mat-
logy in order to gauge the ‘received view’ of con- uration and plural breeding (groups that have two or
straints as well as phylogenetic inertia (Price, 1996; more breeding pairs) was not observed. Peterson sugges-
Ridley, 1996; Futuyma, 1998; Strickberger, 2000; ted that a loss of plural breeding in some species may
Freeman & Herron, 2001). Surprisingly, none of the not have coincided with selection against delayed
examined texts list ‘phylogenetic inertia’ in the table of maturation.
contents, glossary or index, and inspection of each Bon et al. (1995) described the behaviour of a
failed to find mention of the topic. However, all had population of mouflon (wild sheep, Ovis gmelini) in
some discussion of the role of constraints in evolution. the absence of predators. Ewes with older lambs isolate
A consensus textbook definition of ‘constraint’ might themselves in safe ranges, even when predators are
be: ‘A property of a trait that, although possibly absent, an effect attributed to phylogenetic inertia. The
adaptive in the environment in which it originally behaviour remains although the selection (predation by
evolved, acts to place limits on the production of new foxes) no longer exists. Chu (1994) studied the
phenotypic variants’. Similarly, Derrickson & Ricklefs evolution of delayed plumage maturation in shorebirds
(1988, p. 418) defined ‘phylogenetic constraints and concluded that it is an incidental consequence of
broadly as differentiation of the evolutionary respon- the phylogenetic inertia (retention) of molts in this
siveness of the phenotype, which may result from group. Prey handling in Eumeces gilberti lizards was
intrinsic factors (the genetic covariation patterns) or studied by de Queiroz & de Queiroz (1987), who
extrinsic factors (the array of selective pressures concluded that headfirst ingestion of prey was ancestral
impinging on diverse members of a clade)’. Thus, in tetrapods. Thus, this behaviour could not necessarily
developmental and genetic processes can restrict the be considered adaptive in E. gilberti, and hence was
types of phenotypes that arise in the future and, given attributable to phylogenetic inertia. Strike-induced
that the environment (and hence selective regime) chemosensory searching (chemosensory searching fol-
changes over time, yesterday’s adaptation may become lowing an attempted predation event) in anguid lizards
tomorrow’s constraint. was attributed to phylogenetic inertia and not adapta-
Ridley (1996) provides a good discussion of develop- tion by Cooper (1995), who found that the phyloge-
mental and historical constraints; Futuyma (1998) des- netic evidence favoured an ancient origin for this
cribes several kinds of constraints (e.g. physical, genetic, stereotypic behaviour. Inflexibility in the social struc-
developmental), and discusses their consequences for ture of hamadryas baboons (compared with common
evolution. Among the consequences listed by Futuyma baboons) was attributed to phylogenetic inertia by
(1998) are the absence of adaptive characters (organisms Barton et al. (1996), who noted that the typical baboon
may be constrained from evolving adaptive traits social structure persists in captivity in hamadryas
because of a lack of variation in the required direction), baboons, whereas in common baboons other social
presence of directional trends (trends may be observed structures appear readily.
because developmental pathways make some variants Sih et al. (2000) studied antipredator behaviour in
more likely than others), and low rates of evolution salamanders (the closely related Ambystoma barbouri and
(because of limited genetic variation). Note that all three A. texanum). They concluded that the behaviour of
of these consequences have been used as descriptions of A. barbouri in the presence of predatory sunfish could
phylogenetic inertia, from various perspectives. For be attributed to phylogenetic inertia because A. barbouri
example, Simpson (1944) describes inertia as directional showed ineffective antipredator behaviour, similar to
change, and several other authors describe phylogenetic A. texanum. A. barbouri evolved from an A. texanum – like
inertia as little to no evolutionary change even when ancestor, and because A. texanum lives in fishless eph-
selection should be favouring change (e.g. Wilson, 1975; emeral ponds, A. barbouri probably inherited its anti-
Edwards & Naeem, 1993). predator behaviour from an ancestor that also lived in

J. EVOL. BIOL. 15 (2002) 899–910 ª 2002 BLACKWELL SCIENCE LTD

902 S. P. BLOMBERG AND T. GARLAND

fishless ponds. A. barbouri’s antipredator behaviour share traits that are adaptive for those regimes. In any
appears not to have been moulded by selection in the case, casting phylogenetic inertia and adaptation by
new predator-rich environment. selection as alternative hypotheses may be inappropriate.
The role of phylogenetic inertia in evolution has been Orzack & Sober (2001) argue that both phylogenetic
discussed in depth with reference to sexual dimorphism inertia and adaptation can contribute to trait values, and
in body size and canine size in primates. Cheverud et al. that they can be considered as orthogonal factors in an
(1985) and Lucas et al. (1986) argued for the role of evolutionary, statistical analysis of trait values (see also
phylogenetic inertia in the evolution of body size and Reeve & Sherman, 2001). Orzack and Sober define
canine size, respectively. The results of Cheverud et al. phylogenetic inertia as the influence of the initial state of
(1985) have been heavily criticized by Ely & Kurland a character on its end state, an interpretation that is at
(1989), and their implementation of the autocorrelation least compatible with Wilson (1975). This definition
comparative method (see below) apparently included a implies that phyletic stasis (absence of evolutionary
mathematical error (Rohlf, 2001). Various other studies change in a trait) is neither sufficient nor necessary
have also purported to find evidence for a phylogenetic evidence for phylogenetic inertia, in contrast to many
‘effect’ on dimorphism in body size (e.g. Smith & recent uses of the term in the empirical literature (see
Cheverud, 2002). However, other authors have claimed previous section; Reeve & Sherman, 2001). Lack of
no role for phylogenetic inertia in canine tooth size of change in a trait is not sufficient because the trait may be
primates (Plavcan & van Schaik, 1992). under stabilizing selection (Griffiths, 1996). It is also not
What these examples show is not whether phyloge- necessary because daughter species always inherit at least
netic inertia was confirmed or falsified in each case – some trait values from their ancestors. It then becomes a
however, defined by the researcher – but that question of how much of each trait is attributable to
phylogenetic inertia is a widely used explanation for adaptation to the current environment, and how much
the existence of some biological phenomena. Apparently, to ancestry (Edwards & Naeem, 1993).
the concept continues to be of value whether or not the
definition is clear or the criteria for establishing its
Phylogenetic inertia and analytical
existence are adequate.
methods for comparative data
Following publication of two seminal papers in 1985
Phylogenetic inertia and adaptation:
(Cheverud et al., 1985; Felsenstein, 1985), quantitative
alternative hypotheses?
and statistical aspects of ‘the comparative method’ (sensu
Many current researchers view phylogenetic inertia and Harvey & Pagel, 1991) have advanced tremendously.
adaptation by natural selection as alternative hypotheses These advances have followed from at least six ideas: (1)
for the presence (or absence) of a character in a taxon. adaptation by natural selection should not be inferred
Most often, phylogenetic inertia is viewed as a null casually from comparative data; (2) independent phylo-
hypothesis against which to test hypotheses of adaptation genetic information can greatly increase the types and
(e.g. Edwards & Naeem, 1993). However, this is not quality of inferences that can be drawn from comparative
always a clear distinction. Indeed, phylogenetic inertia data (e.g. estimation of character states for hypothetical
has been called a ‘last explanatory resort’ by Shapiro ancestors); (3) among-species data cannot be assumed to
(1981), and there is some concern that the two hypo- represent independent and identically distributed sam-
theses (adaptation and phylogenetic inertia) actually ples from a ‘population’ for purposes of statistical analy-
occupy different ‘levels of analysis’ (sensu Sherman, ses; (4) assumptions about the way characters have
1988; see Edwards & Naeem, 1993; Reeve & Sherman, evolved (such as by a process like Brownian motion or a
2001). Phylogenetic inertia may be an explanation of the more complicated model (e.g. see Garland et al., 1993;
origin of a trait, whereas if stabilizing selection is acting, Hansen, 1997; Orzack & Sober, 2001; Freckleton et al.,
adaptation by natural selection may be involved in the 2002; Martins et al., 2002) are required for statistical
maintenance of the trait. Thus, traits are a product both of inferences; (5) choice of species to include in a compar-
their evolutionary history and natural selection in the ative study should be guided by knowledge of their
recent and current environment. Alternatively, a trait phylogenetic relationships; (6) most comparative studies
may have evolved originally by natural selection, are purely correlational, so the ability to draw causal
experienced such strong selection that genetic variation inferences from them (e.g. about the importance of
was eliminated, and then persisted because of the lack of natural selection in shaping biological diversity) can be
genetic variation [point 1 of Wilson (1975) and Edwards greatly enhanced by additional types of information,
& Naeem (1993)], even in the face of altered environ- such as can be obtained from experimental studies of
mental conditions and hence a changed selective regime. selection acting within present-day populations or mech-
Moreover, as many workers have noted, if related species anistic studies of how organisms work (e.g. Coddington,
tend to share environmental characteristics, and hence 1988; Baum & Larson, 1991; Brooks & McLennan,
selective regimes, then we would expect them also to 1991; Harvey & Pagel, 1991; Lynch, 1991; Eggleton &

J. EVOL. BIOL. 15 (2002) 899–910 ª 2002 BLACKWELL SCIENCE LTD

 Phylogenetic inertia 903

Vane-Wright, 1994; Garland & Adolph, 1994; Garland & inertia did not account for the association between
Carter, 1994; Leroi et al., 1994; Doughty, 1996; Rose & fluctuating asymmetry and sexual selection [in pri-
Lauder, 1996; Autumn et al., 2002). mates]’. Similarly, Manning & Chamberlain (1994), after
These ideas have led to refinement of the analysis of conducting an analysis using phylogenetically indepen-
adaptation through comparative studies and the devel- dent contrasts, conclude ‘…it is unlikely that phylo-
opment of various phylogenetically based statistical genetic inertia can explain the relation between gametic
methods. For continuous-valued characters, four major redundancy and haploid chromosome number [in pri-
statistical methods have emerged (recent reviews in mates]’. Manning and Chamberlain therefore interpret
Garland et al., 1999; Rohlf, 2001): phylogenetically phylogenetic comparative analyses as falsifying the hy-
independent contrasts (Felsenstein, 1985; Garland et al., pothesis of phylogenetic inertia if phylogenetically ‘cor-
1992), phylogenetic autocorrelation (Cheverud & Dow, rect’ statistics show significant results, because in using
1985; Cheverud et al., 1985; Gittleman & Kot, 1990; these methods the likely lack of independence of trait
Gittleman & Luh, 1994), generalized least-squares values among related species has been eliminated (at
approaches (Grafen, 1989; Martins & Hansen, 1997; least in principle). Similarly, Hosken et al. (2001), after
Garland & Ives, 2000), and Monte Carlo simulations conducting an analysis using independent contrasts
(Martins & Garland, 1991; Garland et al., 1993). Some of conclude, ‘…after using independent contrasts to control
these approaches have been linked, directly or indirectly, for phylogenetic inertia in these data, baculum length [in
to the study of ‘phylogenetic inertia’. In this section, we bats] was not significantly associated with mating system,
briefly review these and some related methods in order testis mass or body mass’, and Iwaniuk et al. (1999)
to consider whether they can in fact provide measures of performed an independent contrasts analysis on brain
phylogenetic inertia in comparative data sets. A thorough size and forelimb dexterity in carnivores, ‘to account for
discussion of methods is beyond the scope of this paper, confounding effects of phylogenetic inertia…’
but some additional discussion can be found in Blomberg We argue that, in general, phylogenetically independ-
et al. (in press). ent contrasts are ill-suited to the study of phylogenetic
Biologists have long recognized that distinct evolu- inertia because the mathematical definition of independ-
tionary lineages (clades) may show quantitative differ- ent contrasts attempts to remove all effects of ancestry
ences in such traits as body size, brain size or metabolic from the calculation (see also Orzack & Sober, 2001).
rate (the latter two after correction for correlations with [The same arguments would apply to existing generalised
body size). These differences have often been referred to least squares (GLS) methods because independent con-
as ‘grade shifts’ (Huxley, 1958; Simpson, 1961). As trasts are a special case of them (Garland & Ives, 2000;
reviewed in Harvey & Pagel (1991), long before 1985 Rohlf, 2001).] A simple example will illustrate this fact.
nested A N O V A s and A N C O V A s were used to partition Suppose that an ancestral species A (with trait value a)
variance among taxonomic levels (implicitly assumed to undergoes speciation to form two daughter species, D1
represent clades), and differences that were identified and D2, with trait values d1 and d2, respectively. The
were sometimes discussed in terms of phylogenetic evolution of species A to D1 has caused the trait to evolve
inertia or constraint. More recently, it has been recog- from state a to state d1 + a (i.e. by an amount equal to
nized that the incorporation of phylogenetic information d1). Similarly, evolution of A to D2 has caused the trait
into clade comparisons greatly reduces the probability of value to evolve to d2 + a. The total amount of evolution
finding statistically significant differences (Garland et al., is therefore (d2 + a) ) (d1 + a) ¼ d2 ) d1, which is
1993). Moreover, even if clade differences are identified, independent of a, and hence independent of A (and all
attributing their origin to any particular clade-specific other ancestors). [The actual calculation of contrasts
feature is problematic because most clades exhibit many involves division of the differences between species by
synapomorphies, i.e. shared, derived features that are the square root of the sum of the branch lengths leading
unique to themselves. Thus, although modern phyloge- from their ancestor (see Felsenstein, 1985), which does
netically based statistical methods, such as Monte Carlo not affect the present argument.] As independent con-
simulations, can protect us from inflated type I error rates trasts attempt to make orthogonal all previous evolution
during clade comparisons, they do not allow us to infer in the calculation, they cannot be used to measure
phylogenetic inertia if we do find significant differences phylogenetic inertia in any simple way, because inertia is
among clades (see also Derrickson & Ricklefs, 1988). related in some way to previous evolutionary history.
Several authors have implied that the use of phylo- Nevertheless, Burt (1989) proposes a test based on
genetically independent contrasts (Felsenstein, 1985) independent contrasts for what he describes as phylo-
allows insight concerning ‘phylogenetic inertia’. For genetic inertia. Burt (1989) argues that if one trait
example, Manning & Chamberlain (1993), using an changes faster than another, then there should be a
independent contrast method, found significant associa- negative relationship between the degree of change
tions between fluctuating asymmetry and canine tooth within each contrast for the two traits (the within-
dimorphism, canine length, mass dimorphism and contrast ‘slope’, calculated as the contrast in the X
competition type. They conclude that ‘Phylogenetic variable divided by the contrast in the Y variable), and

J. EVOL. BIOL. 15 (2002) 899–910 ª 2002 BLACKWELL SCIENCE LTD

904 S. P. BLOMBERG AND T. GARLAND

divergence time. The reasoning behind this is that if components: y ¼ q W y + e, where y is the vector of
changes in trait X cause changes in the selective regime trait values, W is a weighting matrix that expresses the
experienced by trait Y, but trait Y has not responded in phylogenetic relationships of the species under study, e is
time to reach its optimum, then the faster X changes, the a vector representing values unique to each terminal
more Y will ‘lag’ behind its optimum, leading to an taxon, and q is a scalar autocorrelation coefficient. It was
association between the within-contrast ‘slope’ and time suggested that e could be used to test hypotheses about
since divergence. We are unaware of any applications of adaptation among terminal taxa. This has been criticized,
Burt’s method, aside from his own analysis of the data of however, because (subject to limitations of estimation), e
Sessions & Larson (1987). Deaner & Nunn (1999) use a is intended to consist only of values unique to the
similar method, except they examine the relationship of terminal taxa and does not contain information about
the residuals of the contrasts (of one trait regressed on higher taxa. It is not clear why an hypothesis of
the other) with divergence times. Under the ‘lag’ hypo- adaptation should be tested only with variation that has
thesis, small residuals will be associated with shorter arisen since the most recent furcations in a phylogeny
divergence times. (Harvey & Pagel, 1991; Garland et al., 1999).
The definition of phylogenetic inertia according to Burt Use of the autocorrelation method for testing hypo-
(1989) is the same as the ‘evolutionary lag’ of Deaner & theses of phylogenetic inertia centres on interpretation of
Nunn (1999) (although the latter authors do not cite the both the autocorrelation coefficient, q, and the ‘true R2’
former). Evolutionary lag occurs when changes in one [both of which may be affected by transformation of the
trait occur later in evolution than changes in a second W matrix (Gittleman & Kot, 1990; Garland et al., 1999)].
trait. Lag can thus be caused by different strengths of Positive and statistically significant q-values are thought
selection on each trait, or by phylogenetic inertia (sensu to indicate significant phylogenetic inertia, and one can
Wilson, 1975; Edwards & Naeem, 1993) owing to compare q-values among traits to establish which traits
insufficient genetic variation in the lagging trait. exhibit more or less phylogenetic inertia (e.g. Morales,
[According to Simpson (1944), the origin of the term 2000). Additionally, one can calculate the true R2, which
evolutionary lag is probably Darlington (1939).] The Burt provides a measure of the proportion of the total
and Deaner-Nunn methods for measuring evolutionary phenotypic variance explained by phylogeny and can
lag are conceptually simple, but they do not correspond be used to judge the fit of the model (Cheverud et al.,
to most definitions of phylogenetic inertia as usually 1985; Martins & Hansen, 1996). R2 has also be used as a
expressed. Lag can cause a pattern that may be recognized measure of the phylogenetic ‘effect’ in its own right
as phylogenetic inertia, but lag alone is not the same as (Gittleman & Kot, 1990; Gittleman et al., 1996a). How-
phylogenetic inertia. ever, it is clear from the statistical model that q only
In one sense, phylogenetic inertia may correspond to represents the degree of similarity among species, given
some type of lag: lag of a trait in tracking environmental their phylogeny. This is similar to Ridley’s 1983 defini-
optima set by a particular selection regime. To quote tion of phylogenetic inertia (shared traits among species),
Simpson (1944, p. 179), ‘Response to selection pressure is but it lacks any explanatory power. As with other
not instantaneous, and inertia, in the sense of lag in statistics that have been derived solely from comparative
following a shifting optimum, is an important element in data (and a phylogenetic tree), q does not provide us with
evolution’. This definition has added to the confusion any information on the cause of species resemblance.
surrounding the terminology of phylogenetic inertia. Similar criticisms can be made of related techniques that
Simpson’s final likening of inertia to lag, after earlier claim to measure phylogenetic inertia, including phylo-
defining evolutionary inertia as rectilinear evolution, genetic correlograms (Gittleman & Kot, 1990; Gittleman
illustrates one of the problems of the definition of et al., 1996a,b; Diniz-Filho, 2001) and phylogenetic
phylogenetic inertia. Some authors use it as a synonym eigenvector regression (Diniz-Filho et al., 1998). Rohlf
for phylogenetic or evolutionary lag. For example, Sih (2001) discusses the above methods and concludes that
et al. (2000) use phylogenetic inertia in the sense of phylogenetic autocorrelation cannot be made equivalent
Simpson (1944), when describing the lack of adaptive to independent contrasts or generalized least-squares
response of the behaviour of salamanders (Ambystoma methods under any known model of evolution; thus, it is
barbouri) to predation by sunfish (the selective regime). difficult to interpret biologically the results of an auto-
The phylogenetic autocorrelation method introduced correlation analysis. Rohlf (2001) also points out a
by Cheverud & Dow (1985) and Cheverud et al. (1985) mathematical error in all previous calculations using
would appear to show more promise for quantifying the autocorrelation method, which complicates inter-
phylogenetic inertia. Indeed, the latter paper is titled pretation of published results.
‘The quantitative assessment of phylogenetic con- An important phylogenetically based statistical method
straints in comparative analyses…’ The method works was introduced by Lynch (1991). By analogy with
by fitting a model which partitions trait values (mean methods from quantitative genetics, Lynch used a
values for a series of species or populations) into mixed-model formulation and maximum likelihood
(1) shared phylogenetic and (2) independent methods to estimate ‘phylogeny-wide’ mean values for

J. EVOL. BIOL. 15 (2002) 899–910 ª 2002 BLACKWELL SCIENCE LTD

 Phylogenetic inertia 905

characters, the variance–covariance structure of the

components of the taxon-specific means, and the esti-
mated mean phenotypes for ancestral characters. Meth-
ods to test hypotheses of correlations among characters
were also presented. Lynch (1991) argues that the
variance–covariance structure of phylogenetic effects in
his model can be used to describe phylogeny-wide
macroevolutionary patterns, whereas the variance–cova-
riance structure of the residual effects may be used to
describe microevolutionary patterns in the data, once
measurement error is taken into account. In large part
because of computational difficulties, the method has
rarely been applied (but see Christman et al., 1997).
However, recent work has provided usable algorithms for
estimating parameters in the statistical model (although
this will generally require large sample sizes) and begun
to clarify how it relates to other comparative methods
(Freckleton et al., 2002; Martins et al., 2002; E. A. Fig. 1 Hypothetical example illustrating the presence of phylo-
Housworth and M. Lynch, pers. comm.). It is important genetic signal in comparative data, and how adaptation in response
to natural selection can reduce phylogenetic signal [numbers
to note that Lynch (1991) presented his method as a way
adjacent to branches are branch lengths in units of expected variance
to describe patterns in the data and not to make causal of character evolution (see Felsenstein, 1985; Martins & Garland,
inferences, and he concludes with remarks that empirical 1991; Garland, 1992; Garland et al., 1992)]. For a hypothetical tree
work in ecology and genetics is necessary to shed light on with six species (a–f) and trait values of 3–8, significant phylogenetic
the causal factors that underlie patterns that occur in signal is detected (P ¼ 0.047, based on 1000 randomizations) by
comparative data. the randomization test proposed in Blomberg et al. (2001; in press).
Maddison & Slatkin (1991) proposed to test for If species c, d, and e experienced a change in selective regime and
phylogenetic inertia in discretely valued characters via evolved to have values as shown (1, 2, 3, respectively), then
ancestor reconstruction followed by counting the num- phylogenetic signal would be obscured (randomization P ¼ 0.294).
ber of evolutionary transitions on a tree. After ancestor Blomberg et al. (in press) propose a statistic, which they term K, that
can be used to indicate the amount of phylogenetic signal that is
reconstruction, the number of character changes in a
present relative to the amount expected for the given topology,
clade are tabulated, and fewer changes imply greater
branch lengths, and under a Brownian motion model of character
phylogenetic inertia. One can then compare multiple evolution. For these two examples, K is 1.38 and 0.52, respectively.
traits for a given tree. Maddison & Slatkin (1991), under For discussion of other statistics that can quantify the amount of
a heading titled ‘Test for Phylogenetic Inertia’, use phylogenetic signal, see Freckleton et al. (2002) and Blomberg et al.
permutation of character values on a given phylogeny (in press).
to establish the null distribution of the minimum number
of transitions required by the character. If the observed
number of transitions is relatively small, then it can be in spirit to recent usage in systematic biology (e.g. Hillis &
concluded that the character is evolving slowly enough Huelsenbeck, 1992), and is also similar to the ‘phylo-
to retain phylogenetic information (Maddison & Slatkin, genetic effect’ of Derrickson & Ricklefs (1988).] An
1991). Although this approach tests for phylogenetic important fact to note is that, on average, for any
structure in the data, like other purely comparative hierarchical tree, closely related species will tend to
approaches it offers no clues as to why such structure resemble each other under such simple evolutionary
might exist. Moreover, although Maddison & Slatkin models as Brownian motion. This resemblance consti-
(1991) equate phylogenetic inertia with low rates of tutes phylogenetic signal, and its presence, then, clearly
evolution (see also Garland, 1992; Garland & Ives, 2000), does not require the invocation of such processes as
this is not necessarily a property of phylogenetic inertia natural selection (indeed, adaptation via natural selec-
under any of the definitions discussed above. tion may often serve to obscure phylogenetic signal, as
We have devised a test similar to that of Maddison & shown in Fig. 1). That is, random genetic drift alone,
Slatkin (1991), but for continuous-valued characters occurring along a hierarchical phylogeny, will result in a
(Blomberg et al., 2001; in press). However, we do not general tendency for related species to resemble each
claim to be measuring ‘phylogenetic inertia’. Instead, we other. This is why we favour the term phylogenetic
simply hope to detect ‘phylogenetic signal’, which we signal: it carries no connotation of lack of genetic
define as a tendency for related species to resemble each variation, developmental constraint, character interac-
other more than they resemble species drawn at random tion, etc. Whether or not statistically significant phylo-
from the tree (see Fig. 1 for a hypothetical example). genetic signal is detected for any given trait on any given
[This meaning of the term ‘phylogenetic signal’ is similar tree will depend on the statistical power of the test as well

J. EVOL. BIOL. 15 (2002) 899–910 ª 2002 BLACKWELL SCIENCE LTD

906 S. P. BLOMBERG AND T. GARLAND

as the presence of various factors that may lower the test, except for continuous data, and was not originally
amount of signal, including measurement error in the developed with phylogenetic data in mind. Abouheif
trait data, errors in the phylogenetic topology, errors in incorporates phylogenetic information by conducting the
branch lengths, and adaptation or other factors that may test on the original data, then randomly ‘rotating’ nodes
cause evolution to deviate from simple Brownian of the tree and calculating the test statistic a large number
motion. In addition, most comparative studies do not of times. The mean of this distribution is compared with a
involve data that were gathered for species reared under null sampling distribution obtained by randomly shuf-
common environmental conditions, and the possibility of fling taxa on the tips of the tree, and calculating the test
genotype–environment interactions could have unpre- statistic again, repeating the process a large number of
dictable effects on the amount of phylogenetic signal. times. If the original value for the mean test statistic is
Our method can be implemented with phylogeneti- greater than (say) 95% of the ‘random’ values, then it
cally independent contrasts, as follows (Blomberg et al., can be concluded that significant ‘phylogenetic autocor-
2001; in press). First, standardized contrasts are calcula- relation’ exists in the data. For comparative studies, and
ted for the original data on the specified topology and following earlier suggestions by Gittleman and cowork-
branch lengths. Secondly, the variance of these contrasts ers, Abouheif (1999) recommends using a phylogeneti-
is calculated. Thirdly, the data are permuted equiproba- cally based statistical method if and only if such a
bly across the tips of the tree a large number of times, and diagnostic test rejects the assumption of phylogenetic
again the variance of the contrasts is recorded for each independence. Moreover, if a method such as independ-
permutation. Permuting the data should, on average, ent contrasts is applied, he recommends application of his
eliminate phylogenetic signal in the data. Finally, the test to the transformed data to verify that the transfor-
original variance of contrasts is compared with the mation has indeed rendered the data independent. A
distribution of variances resulting from the permutation disadvantage of Abouheif (1999) method is the ad hoc
procedure. If the data show phylogenetic signal, then way that phylogeny is incorporated into the analysis by
most (e.g. 95%) of the variances resulting from the the particular randomization method employed. The
permuted data should be larger than the original effects of phylogeny are ‘randomized out’ of the analysis
variance. The P-value is then calculated as the proportion instead of being incorporated in a fundamental way. This
of variances from the permuted data that are less than is a result of trying to shoe-horn hierarchical data into a
the variance of the original contrasts (see example in linear data structure, for which the test for serial
Fig. 1). [Calculations can be carried out with the Phen- independence was first designed. Also, the test does not
otypic Diversity Analysis Programs (PDAP), available use branch length information, so it is unclear how the
from T.G. on request.] Simulations under Brownian results would be affected by different evolutionary
motion character evolution indicate that our method has models. Finally, no analyses of the statistical power to
proper type I error rates (it does not falsely claim detect phylogenetic autocorrelation have been presented.
phylogenetic signal when none exists) and good statis- In any case, Abouheif (1999) method does not allow
tical power for trees with 20 or more species (Blomberg causal inferences about the source of any phylogenetic
et al., in press). signal that might be detected, and accordingly he does
Most recently, Pagel (1999), Freckleton et al. (2002), not use the terms phylogenetic ‘inertia’ or ‘constraint’.
and Blomberg et al. (in press) have emphasized that Attempting to move beyond a simple pattern defini-
transformations of branch lengths, such as originally tion, Orzack & Sober (2001) define phylogenetic inertia
proposed by Grafen (1989), can be used to compare the as an influence of trait values of ancestors on the trait
fit of a continuum of phylogenies, ranging from a star (no values of their descendants. They argue that one can, in
hierarchical structure) to a given candidate tree and even principle, examine the degree to which both phyloge-
to a tree that is more hierarchical than the candidate. netic inertia and natural (or sexual) selection contribute
Thus, the branch-length transformation parameter can to descendent trait values. They propose a test for
be used as an index of the amount of phylogenetic signal phylogenetic inertia, based on taking differences between
in traits. Such procedures again constitute pattern trait values for multiple pairs of species in a phylogeny.
recognition, but as transformations can be formulated The hypothesis that trait values are affected by ancestors
under various explicit models of character evolution, at a given depth in the phylogeny is tested by calculating
they may allow insight into processes. differences (‘controlled comparisons’) between species
In the context of diagnosing comparative data to which share a common ancestor at the phylogenetic level
determine whether or not to use phylogenetically based of interest. To test a phylogenetic inertia hypothesis
statistical methods, Abouheif (1999) introduced a test while controlling for effects of selection, one would need
that can detect whether traits are ‘significantly correlated to examine a large number of descendants that have
to phylogeny’ or possess ‘historical nonindependence’, been subject to the same selective regime (with respect to
which we prefer to call phylogenetic signal. Abouheif’s the character in question) and whose ancestors had
method involves a test for serial independence in various character states. However, because the character
comparative data. This test is an analogue of a ‘runs’ states of ancestors are not given by data for extant species

J. EVOL. BIOL. 15 (2002) 899–910 ª 2002 BLACKWELL SCIENCE LTD

 Phylogenetic inertia 907

(which is all that is typically available in a comparative responses to selection on other traits may in fact be
study), and because the validity of using parsimony to under some kind of ‘constraint’ which limits the evolu-
reconstruct ancestral character states is questionable, tionary options of the organism (Wake, 1991; Garland &
they suggest that the test be modified to employ multiple Carter, 1994). In any case, if phylogenetic inertia is to be
pairs of extant species (e.g. X1 and X2, Y1 and Y2) in invoked as an explanation for biological phenomena,
which the first member of each pair has been subject to then the term should be defined as carefully as ‘adapta-
the same selective regime and the second members tion’ has been in the recent literature.
exhibit different character states. The question then Moreover, we believe that study of the relative
becomes, does X1 resemble X2 more than X1 resembles evolutionary lability of traits is a legitimate use of
Y2, and does Y1 resemble Y2 more than Y1 resembles X2. comparative data. For example, Gittleman et al.
This method has not yet been applied to real data. We (1996a,b) used autocorrelation methods to compare
note, however, that Janson (1992) presents a method in behavioural, morphological, and life history traits.
which character-state transitions are modelled using a Although the autocorrelation methods may be viewed
Markov process. Under this process, current character as conceptually flawed (see above) and all previous
states depend only on the most recent ancestral character applications have contained a mathematical error (Rohlf,
state, so it appears that this approach is in the spirit of 2001), the basic aims of Gittleman et al.¢s. papers were
Orzack & Sober (2001), although it does require worthy (we note also that they did not use the term
inferences about ancestral states. To our knowledge, ‘phylogenetic inertia’). New analytical methods should
Janson’s method has not been applied other than in his allow statistically valid comparisons of different types of
original paper on seed dispersal syndromes. traits on a given tree (i.e. for a given set of species) as
well as across trees (e.g. Maddison & Slatkin, 1991;
Blomberg et al., 2001, in press; Freckleton et al., 2002).
Conclusions
For example, de Queiroz & Wimberger (1993) and
The concept of phylogenetic inertia has changed during Wimberger & de Queiroz (1996) compared different
the history of evolutionary biology, and many research- types of traits (categorical data) across trees and found
ers have applied it in an ad hoc way, often without clear that, contrary to some expectations, behavioural char-
definition, similar to the concept of ‘constraints’ (Anto- acters (at least those chosen by systematists) show no
novics & van Tienderen, 1991). This makes it difficult to more homoplasy than do morphological characters.
determine what phylogenetic inertia really is, and However, the behavioural characters chosen for phylo-
whether it exists in nature. ‘Pattern’ definitions of genetic analysis may be unusual in that systematists use
phylogenetic inertia, such as Simpson (1944) and (less prior judgement before including a particular trait in an
clearly) Burt (2001), can be subject to quantification by analysis. Traits chosen for study by behavioural ecolo-
use of phylogenetic comparative methods (e.g. Maddison gists may exhibit different properties with regard to
& Slatkin, 1991; Gittleman et al., 1996a,b; Abouheif, homoplasy. Nevertheless, such findings are important
1999; Blomberg et al., 2001, in press; Freckleton et al., because although homoplasy is perceived as problema-
2002). ‘Process’ definitions, on the other hand, are not tical in cladistic analyses, it has the potential to provide
well suited to study by comparative methods alone the best evidence for adaptations to common environ-
because comparative data sets typically do not contain ments (Brooks & McLennan, 1991; Brooks, 1996; but
information on the (past) genetic architecture of the see Wake, 1991). We urge further development of
traits or the selective regimes to which they have been comparative methods for the study of phylogenetic
subjected (see also Leroi et al., 1994; Hansen, 1997; inertia – or at least phylogenetic signal.
Wagner & Schwenk, 2000; Reeve & Sherman, 2001).
Still, comparative analyses may suggest directions for
Acknowledgments
future research that could test underlying causal
hypotheses about what drives or impedes evolution We thank F. Bashey, C. K. Ghalambor, A. R. Ives, C. L.
within populations. The kinds of studies that are needed Nunn, S. H. Orzack, D. N. Reznick, and E. R. Sober for
are ones that make the possible causes of phylogenetic discussions and ⁄ or comments on the manuscript. For
inertia a serious object of investigation, and not just an access to unpublished manuscripts, we thank E. A.
explanation of last resort (Shapiro, 1981). For example, Housworth and R. P. Freckleton. This work was suppor-
studies that focus on quantifying the selective regime, ted by NSF grant DEB-0196384 to TG and A. R. Ives, and
heritabilities of traits, and genetic correlations among a Michael Guyer Postdoctoral Fellowship to SPB.
traits may shed light on whether organisms evolve along
the most favourable genetic trajectories (Schluter, 1996)
References
in response to identified selective agents. Such studies
should also take into account morphological, physiologi- Abouheif, E. 1999. A method to test the assumption of
cal, biochemical, developmental, and genetic ‘design phylogenetic independence in comparative data. Evol. Ecol.
limitations’, because traits that exhibit correlated Res 1: 895–909.

J. EVOL. BIOL. 15 (2002) 899–910 ª 2002 BLACKWELL SCIENCE LTD

908 S. P. BLOMBERG AND T. GARLAND

Antonovics, J. & van Tienderen, P.H. 1991. Ontoecogenophy- Deaner, R.O. & Nunn, C.L. 1999. How quickly do brains catch up
loconstraints? The chaos of constraint terminology. Trends with bodies? A comparative method for detecting evolution-
Ecol. Evol. 6: 166–168. ary lag. Proc. Roy. Soc. Lond. B 266: 687–694.
Autumn, K., Ryan, M.J. & Wake, D.B. 2002. Integrating Derrickson, E.M. & Ricklefs, R.E. 1988. Taxon-dependent
historical and mechanistic biology enhances the study of diversification of life-history traits and the perception of
adaptation. Quart. Rev. Biol. 77 (in press). phylogenetic constraints. Funct. Ecol. 2: 417–423.
Barton, R.A., Byrne, R.W. & Whiten, A. 1996. Ecology, feeding Diniz-Filho, J.A.F. 2001. Phylogenetic autocorrelation under
competition and social structure in baboons. Behav. Ecol. distinct evolutionary processes. Evolution 55: 1104–1109.
Sociobiol. 38: 321–329. Diniz-Filho, J.A., de Sant’Ana, C.E.R.D. & Bini, L.M. 1998. An
Baum, D.A. & Larson, A. 1991. Adaptation reviewed: a eigenvector method for estimating phylogenetic inertia.
phylogenetic methodology for studying character macroevo- Evolution 52: 1247–1262.
lution. Syst. Zool. 40: 1–18. Doughty, P. 1996. Statistical analysis of natural experiments in
Berger, J. 1988. Social systems, resources, and phylogenetic evolutionary biology: comments on recent criticisms of the
inertia: an experimental test and its limitations. In: The Ecology use of comparative methods to study adaptation. Am. Nat. 148:
of Social Behavior (C. N. Slobodchikoff, ed.), pp. 157–186. 943–956.
Academic Press, San Diego, CA. Edwards, S.V. & Naeem, S. 1993. The phylogenetic component
Blomberg, S.P., Ives, A.R. & Garland, T. Jr 2001. Detecting of cooperative breeding in perching birds. Am. Nat. 141: 754–
phylogenetic signal in comparative data. Am. Zool. 41: 1395. 789.
Blomberg, S.P., Garland, T. Jr & Ives, A.R. (in press). Testing for Eggleton, P. & Vane-Wright, R.I. (eds) 1994. Phylogenetics and
phylogenetic signal in comparative data. Evolution. Ecology. Linnean Society Symposium Series. Academic Press,
Bon, R., Joachim, J. & Maublanc, M.L. 1995. Do lambs affect London.
feeding habitat use by lactating female mouflons in spring Ely, J. & Kurland, J.A. 1989. Spatial autocorrelation, phyloge-
areas fee of predators? J. Zool., Lond. 235: 43–51. netic constraints and the causes of sexual dimorphism in
Brooks, D.R. 1996. Explanations of homoplasy at different levels primates. Int. J. Primatol. 10: 157–171.
of biological organization. In: Homoplasy. The Recurrence of Felsenstein, J. 1985. Phylogenies and the comparative method.
Similarity in Evolution (M. J. Sanderson & L. Hufford, eds), pp. Am. Nat. 125: 1–15.
3–36. Academic Press, San Diego, CA. Freckleton, R.P., Harvey, P.H. & Pagel, M. 2002. Phylogenetic
Brooks, D.R. & McLennan, D.A. 1991. Phylogeny, Ecology, and analysis and comparative data: a test and review of evidence.
Behavior. The University of Chicago Press, Chicago, IL. Am. Nat. (in press).
Burt, A. 1989. Comparative methods using phylogenetically Freeman, S. & Herron, J.C. 2001. Evolutionary Analysis. 2nd edn,
independent contrasts. In: Oxford Surveys in Evolutionary Biology 704 pp. Prentice Hall, Upper Saddle River, NJ.
(P. H. Harvey & L. Partridge, eds), pp. 33–53. Oxford Futuyma, D.J. 1998. Evolutionary Biology, 2nd edn, 763 pp.
University Press, Oxford. Sinauer and Associates, Sunderland, MA.
Burt, B.B. 2001. Evolutionary stasis, constraint and other Garland, T. Jr 1992. Rate tests for phenotypic evolution using
terminology describing evolutionary patterns. Biol. J. Linn. phylogenetically independent contrasts. Am. Nat. 140: 509–
Soc. 72: 509–517. 519.
Cain, A.J. 1964. The perfection of animals. In: Viewpoints in Garland, T. Jr & Adolph, S.C. 1994. Why not to do two-species
biology (J. D. Carthy & C. L. Duddington, eds), pp. 36–63. comparative studies: limitations on inferring adaptation.
Butterworth, London. Physiol. Zool. 67: 797–828.
Cheverud, J.M. & Dow, M.M. 1985. An autocorrelation analysis Garland, T. Jr & Carter, P.A. 1994. Evolutionary physiology.
of genetic variation due to lineal fission in social groups of Annu. Rev. Physiol. 56: 579–621.
rhesus macaques. Am. J. Phys. Anthr. 67: 113–121. Garland, T. Jr & Ives, A.R. 2000. Using the past to predict the
Cheverud, J.M., Dow, M.M. & Leutenegger, W. 1985. The present: confidence intervals for regression equations in
quantitative assessment of phylogenetic constraints in com- phylogenetic comparative methods. Am. Nat. 155: 346–364.
parative analyses: sexual dimorphism in body weight among Garland, T. Jr, Bennett, A.F. & Daniels, C.B. 1990. Heritability of
primates. Evolution 39: 1335–1351. locomotor performance and its correlates in a natural
Christman, M.C., Jernigan, R.W. & Culver, D. 1997. A population of vertebrates. Experientia 46: 530–533.
comparison of two models for estimating phylogenetic effect Garland, T. Jr, Harvey, P.H. & Ives, A.R. 1992. Procedures for the
on trait variation. Evolution 51: 262–266. analysis of comparative data using phylogenetically indepen-
Chu, P.C. 1994. Historical examination of delayed plumage dent contrasts. Syst. Biol. 41: 18–32.
maturation in the shorebirds (Aves: Charadriiformes). Evolu- Garland, T. Jr, Dickerman, A.W., Janis, C.M. & Jones, J.A. 1993.
tion 48: 327–350. Phylogenetic analysis of covariance by computer simulation.
Coddington, J.A. 1988. Cladistic tests of adaptational hypothe- Syst. Biol. 42: 265–292.
ses. Cladistics 4: 3–22. Garland, T. Jr, Midford, P.E. & Ives, A.R. 1999. An introduction
Cooper, W.E. Jr 1995. Strike-induced chemosensory searching by to phylogenetically based statistical methods, with a new
the anguid lizard Elgaria coerulea. Amphib.-Reptilia 16: 147–156. method for confidence intervals on ancestral values. Am. Zool.
Darlington, C.D. 1939. The Evolution of Genetic Systems. Cambridge 39: 374–388.
University Press, Cambridge. Gittleman, J.L. & Kot, M. 1990. Adaptation: statistics and a null
Darlington, C.D. & Mather, K. 1949. Elements of Genetics. George model for estimating phylogenetic effects. Syst. Zool. 39: 227–
Allen and Unwin, London. 241.

J. EVOL. BIOL. 15 (2002) 899–910 ª 2002 BLACKWELL SCIENCE LTD

 Phylogenetic inertia 909

Gittleman, J.L. & Luh, H.-K. 1994. Phylogeny, evolutionary Martins, E.P. & Hansen, T.F. 1996. The statistical analysis of
models and comparative methods: a simulation study. In: interspecific data: a review and evaluation of phylogenetic
Phylogenetics and Ecology (P. Eggleton & D. Vane-Wright, eds), comparative methods. In: Phylogenies and the Comparative
pp. 103–122. Academic Press, London. Method in Animal Behaviour (E. P. Martins, ed.), pp. 22–75.
Gittleman, J.L., Anderson, C.G., Kot, M. & Luh, H.-K. 1996a. Oxford University Press, Oxford.
Phylogenetic lability and rates of evolution: a comparison of Martins, E.P. & Hansen, T.F. 1997. Phylogenies and the
behavioral, morphological and life history traits. In: Phylogenies comparative method: a general approach to incorporating
and the Comparative Method in Animal Behavior (E. P. Martins, phylogenetic information into the analysis of interspecific
ed.), pp. 166–205. Oxford University Press, Oxford. data. Am. Nat. 149: 646–667.
Gittleman, J.L., Anderson, C.G., Kot, M. & Luh, H.-K. 1996b. Martins, E.P., Diniz-Filho, J.A.F. & Housworth, E.A. 2002.
Comparative tests of evolutionary lability and rates using Adaptive constraints and the phylogenetic comparative
molecular phylogenies. In: New Uses for New Phylogenies (P. H. method: a computer simulation test. Evolution 56: 1–13.
Harvey & A. J. Leigh Brown, J. Maynard Smith & S. Nee, eds), Maynard Smith, J., Burian, R., Kauffman, S., Alberch, P.,
pp. 289–307. Oxford University Press, Oxford. Campbell, J., Goodwin, B., Lande, R., Raup, D. & Wolpert, L.
Gould, S.J. & Lewontin, R.C. 1979. The spandrels of San Marco 1985. Developmental constraints and evolution. Quart. Rev.
and the panglossian paradigm: a critique of the adaptationist Biol. 60: 265–287.
programme. Proc. Roy. Soc. Lond. B 205: 581–598. Mayr, E. 1982. The Growth of Biological Thought: Diversity,
Grafen, A. 1989. The phylogenetic regression. Proc. Roy. Soc. Evolution, and Inheritance. Belknap Press, Cambridge, MA.
Lond. B 326: 119–156. McKitrick, M.C. 1993. Phylogenetic constraint in evolutionary
Griffiths, P.E. 1996. The historical turn in the study of adapta- theory – has it any explanatory power? Ann. Rev. Ecol. Syst. 24:
tion. Br. J. Phil. Sci. 47: 511–532. 307–330.
Hansen, T.F. 1997. Stabilizing selection and the comparative Morales, E. 2000. Estimating phylogenetic inertia in Tithonia
analysis of adaptation. Evolution 51: 1341–1351. (Asteraceae): a comparative approach. Evolution 54: 475–484.
Harvey, P.H. & Pagel, M.D. 1991. The Comparative Method in Orzack, S.H. & Sober, E. 2001. Adaptation, phylogenetic inertia,
Evolutionary Biology: Oxford Series in Ecology and Evolution. and the method of controlled comparisons. In: Adaptationism
Oxford University Press, Oxford. and Optimality (S. H. Orzack & E. Sober, eds), pp. 45–63.
Hillis, D.M. & Huelsenbeck, J.P. 1992. Signal, noise, and Cambridge University Press, Cambridge.
reliability in molecular phylogenetic analyses. J. Heredity 83: Pagel, M. 1999. Inferring the historical patterns of biological
189–195. evolution. Nature 401: 877–884.
Hosken, D.J., Jones, K.E., Chipperfield, L. & Dixson, A. 2001. Is Peterson, A.T. 1991. Sociality and ontogeny of coloration in the
the bat os penis sexually selected? Behav. Ecol. Sociobiol. 50: blue-and-black jays. Wilson Bull. 103: 59–67.
450–460. Plavcan, J.M. & van Schaik, C.P. 1992. Intrasexual competition
Huxley, J.S. 1958. Evolutionary processes and taxonomy with and canine dimorphism in anthropoid primates. Am. J. Phys.
special reference to grades. Uppsala University Arssks. 6: 21–39. Anthropol. 87: 461–477.
Iwaniuk, A.N., Pellis, S.M. & Whishaw, I.Q. 1999. Brain size is Price, P.W. 1996. Biological Evolution, 429 pp. Saunders College
not correlated with forelimb dexterity in fissiped carnivores Publishing, New York, NY.
(Carnivora): a comparative test of the principle of proper mass. de Queiroz, A. & de Queiroz, K. 1987. Prey handling behavior of
Brain, Behav. Evol. 54: 167–180. Eumeces gilberti with comments on headfirst ingestion in
Janson, C.H. 1992. Measuring evolutionary constraints: a squamates. J. Herp. 21: 57–63.
Markov model for phylogenetic transitions among seed de Queiroz, A. & Wimberger, P.H. 1993. The usefulness of
dispersal syndromes. Evolution 46: 136–158. behavior for phylogeny estimation: levels of homoplasy in
Leroi, A.M., Rose, M.R. & Lauder, G.V. 1994. What does the behavioral and morphological characters. Evolution 47: 46–
comparative method reveal about adaptation? Am. Nat. 143: 60.
381–402. Reeve, H.K. & Sherman, P.W. 2001. Optimality and phylogeny.
Lucas, P.W., Corlett, R.T. & Luke, D.A. 1986. Sexual dimorphism A critique of current thought. In: Adaptationism and Optimality
and tooth size in anthropoids. Human Evol. 1: 23–39. (S. H. Orzack & E. Sober, eds), pp. 64–113. Cambridge
Lynch, M. 1991. Methods for the analysis of comparative data in University Press, Cambridge.
evolutionary biology. Evolution 45: 1065–1080. Ridley, M. 1983. The Explanation of Organic Diversity: the Compar-
Maddison, W.P. & Slatkin, M. 1991. Null models for the number ative Method and Adaptations for Mating. Oxford University
of evolutionary steps in a character on a phylogenetic tree. Press, Oxford.
Evolution 45: 1184–1197. Ridley, M. 1996. Evolution, 2nd edn. Blackwell Science, Cam-
Manning, J.T. & Chamberlain, A.T. 1993. Fluctuating asymme- bridge, MA.
try, sexual selection and canine teeth in primates. Proc. Roy. Rohlf, F.J. 2001. Comparative methods for the analysis of
Soc. Lond. B 251: 83–87. continuous variables: geometric interpretations. Evolution 55:
Manning, J.T. & Chamberlain, A.T. 1994. Sib competition and 2143–2160.
sperm competitiveness: an answer to ‘Why so many sperms?’ Rose, M.R. & Lauder, G.V. (eds) 1996. Adaptation, 511 pp.
and the recombination ⁄ sperm number correlation. Proc. Roy. Academic Press, San Diego, CA.
Soc. Lond. B 256: 177–182. Schluter, D. 1996. Adaptive radiation along genetic lines of least
Martins, E.P. & Garland, T Jr 1991. Phylogenetic analyses of the resistance. Evolution 50: 1766–1774.
correlated evolution of continuous characters: a simulation Schwenk, K. 1995. A utilitarian approach to constraint. Zoology
study. Evolution 45: 534–557. 98: 251–262.

J. EVOL. BIOL. 15 (2002) 899–910 ª 2002 BLACKWELL SCIENCE LTD

910 S. P. BLOMBERG AND T. GARLAND

Sessions, S.K. & Larson, A. 1987. Developmental correlates of Strickberger, M.W. 2000. Evolution, 721 pp. Jones and Bartlett
genome size in plethodontid salamanders and their implica- Publishers, Sudbury, MA.
tions for genome evolution. Evolution 41: 1239–1251. Wagner, G.P. & Schwenk, K. 2000. Evolutionarily stable
Shapiro, A.M. 1981. The pierid red–egg syndrome. Am. Nat. 117: configurations: functional integration and the evolution of
276–294. phenotypic stability. Evol. Biol. 31: 155–217.
Sherman, P.W. 1988. The levels of analysis. Anim. Behav. 36: Wake, D. 1991. Homoplasy – the result of natural selection, or
616–619. evidence of design limitations? Am. Nat. 138: 543–567.
Sih, A., Kats, L.B. & Maurer, E.F. 2000. Does phylogenetic Wilson, E.O. 1975. Sociobiology. The New Synthesis. Harvard
inertia explain the evolution of ineffective antipredator University Press, Cambridge.
behavior in a sunfish-salamander system? Behav. Ecol. Socio- Wimberger, P.H. & de Queiroz, A. 1996. Comparing behavioral
biol. 49: 48–56. and morphological characters as indicators of phylogeny. In:
Simpson, G.G. 1944. Tempo and Mode in Evolution. Columbia Phylogenies and the Comparative Method in Animal Behavior. (E. P.
University Press, New York. Martins, ed.), pp. 206–233. Oxford University Press, Oxford.
Simpson, G.G. 1961. Principles of Animal Taxonomy. Columbia
University Press, New York. Received 22 March 2002; accepted 10 July 2002
Smith, R.J. & Cheverud, J.M. 2002. Scaling of sexual size
dimorphism in body mass: a phylogenetic analysis of Rensch’s
rule in primates. Int. J. Primatol. 23: 1095–1135.

J. EVOL. BIOL. 15 (2002) 899–910 ª 2002 BLACKWELL SCIENCE LTD