Review of Palaeobotany and Palynology 120 (2002) 73^90

www.elsevier.com/locate/revpalbo

Holocene climate change and hydrarch succession in

lowland Amazonian Ecuador
Chengyu Weng a; , Mark B. Bush a , J. Stephen Athens b
a
Department of Biological Sciences, Florida Institute of Technology, College of Science and Liberal Arts,
150 W. University Blvd., Melbourne, FL 32901, USA
b
International Archaeological Research Institute Inc., 2081 Young St., Honolulu, HI 96826, USA
Received 1 February 2001; received in revised form 30 October 2001; accepted 12 November 2001

Abstract

Two sediment cores (Maxus 4 and Maxus 1) were obtained from lowland wetlands situated in the highly diverse
rain forests of the western Amazon basin, Ecuador. The cores, separated by ca. 57 km, were analyzed for fossil pollen,
charcoal and loss-on-ignition. The Maxus 4 core encompasses the last 9500 years, while the Maxus 1 core provides a
record for approximately the last 2000 years. As neither core registered evidence of human impacts or land use, an
unobscured image of Holocene vegetation and climate change was obtained. Alternating wet and drier periods on a
millennial scale characterized much of the Holocene. The unusually high values of Cecropia in the pollen sum between
8700 and 5800 cal yr BP suggest frequent disturbance or the presence of early successional habitat. After 5800 cal yr
BP a series of taxa rising and falling in prominence reveals a hydrarch succession that culminated in a relatively dry
period between ca. 4900 and 3700 cal yr BP. Seasonal flooding of the site is evident as local water tables rose between
3700 and ca. 1000 cal yr BP. After 1000 cal yr BP the modern wooded wetland was established. The second site
Maxus 1 reveals a detailed history of approximately last 2000 years. The principal feature of the record is a
disturbance event at about 1000 cal yr BP, followed by a period of hydrarch succession. 8 2002 Elsevier Science
B.V. All rights reserved.

Keywords: Amazonia; rain forest; fossil pollen; charcoal; climate change; hydrarch succession; Holocene

1. Introduction logical gradients and taxa with narrowly de¢ned

niches combine to produce very high K- and
Amazonia supports the largest tropical rain for- L-diversity (Gentry, 1988). The region of highest
est system in the world. Within this vast area K-diversity yet described is the Yasuni rain forests
there exists on all spatial scales an extraordinary of eastern Ecuador (Pitman et al., 1999). Consid-
diversity of plants and animals. Fine-grained eco- eration of the origin of this diversity led to the
refugial hypothesis (Ha¡er, 1969) and, explicitly,
the identi¢cation of the Napo refugium. However,
Colinvaux et al. (2000) and Colinvaux (1996) re-
ject the refugial mechanism as an explanation of
* Corresponding author. the richness of species and endemics in the Napo
E-mail address: cweng@¢t.edu (C. Weng). and other Amazonian regions. The immense bio-

0034-6667 / 02 / $ ^ see front matter 8 2002 Elsevier Science B.V. All rights reserved.
PII: S 0 0 3 4 - 6 6 6 7 ( 0 1 ) 0 0 1 4 8 - 8

PALBO 2448 23-7-02

74 C. Weng et al. / Review of Palaeobotany and Palynology 120 (2002) 73^90

diversity of the upper Napo drainage has made and Becker, 1996; Behling and Hooghiemstra,
this area a top priority for permanent protection 1999).
(Dinerstein et al., 1995). In this paper we present two paleoecological
The Amazon basin has long been home to records that provide the ¢rst complete Holocene
Amerindian peoples (Roosevelt et al., 1996), and fossil pollen and charcoal evidence from the Ya-
scattered paleoecological data provide some local suni region of Ecuador. We also use these data to
insights into land use. However, at the larger evaluate the impact of human activity and climate
scale, evidence is lacking regarding their in£uence change on Yasuni ecosystems over the last 9000
on the forests (e.g. Bush and Colinvaux, 1988; years.
Behling, 1996; Behling and Da Costa, 2000). If
other areas of lowland Ecuador are a guide, ripar- 1.1. Study area
ian forests could have been in£uenced by human
use and disturbance for much of the Holocene. The study area of the present project is roughly
Human occupation of lakeside sites and the culti- the northwestern arm of Yasuni National Park
vation of maize have a long (6000 years) history (Fig. 1). The Rio Napo and the Rio Yasuni ap-
in eastern Ecuador and other regions of Amazo- proximate the northern and southern boundaries,
nia (Athens and Ward, 1999; Bush and Colin- respectively, of this park subregion (Yasuni Na-
vaux, 1988; Bush et al., 1989; Piperno and Holst, tional Park extends east to the present Peruvian
1998; Piperno, 1990; Piperno and Becker, 1996). border and south to the Rio Curaray; the western
Human use of this region, however, abruptly de- border is irregular ^ see Fig. 1). The elevations of
clined 400^450 years ago with depopulation (Por- Yasuni National Park are between 175 and 400 m
ro, 1994). Similar abandonment is also inferred above sea level.
from records from Panama between 400 and 350 Our principal study site, Maxus 4 (0‡27PS,
yr BP (Bush and Colinvaux, 1994; Bush et al., 76‡37PW), lies near the Rio Yasuni, while Maxus
1992; Piperno et al., 1991). The coincidence of 1 is near the Rio Napo, the latter being one of the
these changes with the arrival of Conquistadors major rivers of the eastern Ecuadorian lowlands
leaves little doubt that the introduction of Old (Fig. 1). The Maxus 4 core was collected from a
World diseases was the primary factor (Denevan, marsh in the £oodplain of the Yasuni River ap-
1976; Denevan, 1992). Given this historical rec- proximately 60 km southeast of the community of
ord, it is obvious that the common perception of Pompeya on the Napo River (Fig. 1). The Maxus
the Amazon rain forest as primeval and largely 1 core was raised from what seems to be an in-
untouched by humans is misleading (Bale¤e, ¢lled abandoned meander (perhaps of the nearby
1988). However, the extent to which anthropogen- Rio Indillana), and is located about 3 km south-
ic in£uence has been a factor shaping Amazon east of the Napo River. This site is covered with
forests, including those of the Yasuni area, is an evergreen tropical rain forest. The two sites are
important issue in Amazonian ecology and con- separated by about 57 km, and the elevation of
servation. both sites is about 220 m above sea level.
Similarly, it is important to establish whether
there have been signi¢cant climatic or environ- 1.2. Climate
mental changes during the Holocene. Servant et
al. (1981) suggested a profound dry phase in Bo- The climate of the Yasuni area is generally wet
livia that purportedly caused an expansion of sa- and warm. Average monthly temperatures are be-
vanna into rainforest areas in the mid-Holocene. tween 24‡C and 27‡C for all months and average
To date there has been no independent evidence annual precipitation is over 3000 mm. Rainfall
to support this contention, although there have occurs in all months, but April^May and Octo-
been many reports of more modest cycles of low- ber^November are the wettest, and December^
ered lake levels from Holocene records across February and August are the driest. As in other
Amazonia (e.g. Liu and Colinvaux, 1988; Piperno areas, the largest current inter-annual variation in

PALBO 2448 23-7-02

 C. Weng et al. / Review of Palaeobotany and Palynology 120 (2002) 73^90 75

Fig. 1. Sketch map showing the location of the coring in relation to rivers and the Yasuni National Park boundary. Maxus 4
and Maxus 1 are study sites in this paper. Maxus 5 is the study site for Athens and Ward, 1999, and mentioned in this paper.

precipitation is attributable to the El Nin‹o South- classi¢ed into terra ¢rme forests, seasonally
ern Oscillation. Strong El Nin‹o events cause the £ooded forests, and swamp forests according to
dry seasons to be more marked, which in turn their hydroperiod.
causes river levels to drop. Isolated climatic events Terra ¢rme forests lie above the mean high riv-
such as windstorms sometimes cause blowdowns er level and seldom, if ever, £ood. The most im-
that open relatively large ( s 2 ha) canopy gaps portant woody families of the terra ¢rme are
(Foster and Terborgh, 1998; Nelson et al., 1994). Annonaceae, Arecaceae, Fabaceae, Lauraceae,
Over the last several decades natural ¢res are not Melastomataceae, Meliaceae, Moraceae, Rubia-
known from the Yasuni area (Pitman, unpub- ceae and Sapotaceae. Romoleroux et al. (1997)
lished data). estimated a total of 3114 tree and shrub species
for the Yasuni National Park and the adjacent
1.3. Vegetation ethnic reserve. In investigations of 58 1-ha tree
plots, between 114 and 307 species were identi¢ed
The Yasuni area is covered by a diverse, tall, per plot (Gentry, 1988 ; Pitman, unpublished
closed-canopy, moist evergreen forest; it is one of data).
the most species-rich forests yet described. Soil The seasonally inundated forests may be
moisture plays a determining role in vegetation £ooded for up to 90 days each year. The com-
distribution and local forest types are broadly monest tree in both the seasonally inundated

PALBO 2448 23-7-02

76 C. Weng et al. / Review of Palaeobotany and Palynology 120 (2002) 73^90

Table 1
Sediment description of the study cores
Layer Depth Sediment Average accumulation rate
(cm)
Core 4 I 0^18 dark gray silty clay, ¢ne rootlet and plant remains :
>
>
II 18^64 dark brown peat with silt, plant remains become ¢ner with 0.41 mm/yr
depth <
>
>
III 64^146 brown clay with a small amount ¢brous material 8
IV 146^252 black peat, fair coarse with abundant macro-plant remains 0.81 mm/yr
V 252^386 dark gray clay
1.1 mm/yr
VI 386^417 silty clay, abundant woody pieces 
VII 417^505 dark gray clay, grayish brown in lower part 0.98 mm/yr
VIII 505^641 dark gray clay 0.65 mm/yr
IX 641^800 silty clay with sand N/A
Core 1 I 0^100 dark brown humic silt :
>
>
>
II 100^224 dark gray humic silt with a small amount of ¢ne sand >
III 224^266 peat, no sand 4.7 mm/yr
<
>
IV 266^330 same as layer II with gradual increase in clay near base >
>
>
V 330^469 grayish brown silty clay with ¢ne sand and plant remains 8
VI 469^680 brownish gray silty clay with ¢ne sand :
>
>
VII 680^704 gray silty clay
N/A
VII 704^730 gray silt with ¢ne sand <
>
>
VIII 730^736 brown silty clay with some ¢ne sand 8

and terra ¢rme forests is the palm, Iriartea deltoi- and Iversen, 1989) and a spike of exotic Lycopo-
dea. The diversity of these £ooded forests is dium clavatum spores was added to facilitate
slightly less than that of the terra ¢rme forest quantitative analyses (Stockmarr, 1971). Identi¢-
(Pitman, unpublished data). cations were based on published atlases and keys
The swamp forests are the wettest and least (Absy, 1979; Bartlett and Barghoorn, 1973; Co-
diverse of the forest types (Pitman, unpublished linvaux et al., 1999; Hooghiemstra, 1984; Roubik
data). Often they are dominated by Mauritia, a and Moreno, 1991) and the pollen reference col-
large palm. The presence of Mauritia indicates a lection at the Florida Institute of Technology
consistently high water table, with standing water ( s 3000 species). At least 300 terrestrial pollen
often present 6 months of the year. grains were counted for each sample. Pollen per-
centages are calculated based on the total of ter-
restrial pollen grains.
2. Methods Organic content was determined by LOI at
105‡C, 550‡C and 1000‡C (Dean, 1974). Samples
We present data from sediment cores, Maxus 4 selected for charcoal quanti¢cation and identi¢ca-
and Maxus 1, two of a series of ¢ve cores raised tion of 0.5 cm3 were sieved through a 180-Wm-
with a Colinvaux^Vonhout coring device from mesh screen after treatment with hot 10% KOH
wetlands alongside a petroleum access road in and 0.1 M sodium pyrophosphate solution. Char-
eastern Ecuador in September, 1994 (Athens and coal fragments were scanned under a dissecting
Ward, 1999). Upon opening the cores in the lab- microscope at a magni¢cation of U20.
oratory, the sediments were described and sub- AMS radiocarbon dates were obtained from
sampled for pollen, loss-on-ignition (LOI) and ac- the Institute of Arctic and Alpine Research at
celerator mass spectrometry (AMS) radiocarbon the University of Colorado and Beta Analytic.
dating. Samples of 0.5 cm3 were taken at 10^30- The corresponding calendar ages were obtained
cm depth intervals. Pollen samples were treated using the CALIB 4.2 computer program (Stuiver
with standard palynological procedures (Faegri and Reimer, 1993).

PALBO 2448 23-7-02

 C. Weng et al. / Review of Palaeobotany and Palynology 120 (2002) 73^90 77

Fig. 2. Age^depth curves for (top) Maxus 4 and (bottom) Maxus 1. Also shown are stratigraphical lithology and organic content
curves. The bottom date for Maxus 1 is rejected because it is apparently too young. Note the scale di¡erence of the axes of age
(horizontal axes) for the two cores.

PALBO 2448 23-7-02

78 C. Weng et al. / Review of Palaeobotany and Palynology 120 (2002) 73^90

Table 2
AMS 14 C dates from Maxus cores
14
Depth Laboratory No. Materials C age Calendar years
(cm) (yr BP) mean (range)
Core 4 60^64 NSRL-11814 leaves 1120 Q 30 1051, 1030, 992 (970^1059)
149^150 NSRL-11185 wood 3430 Q 35 3686, 3663, 3650 (3630^3696)
244^246 NSRL-11186 wood 4350 Q 35 4868 (4861^4969)
405^406 NSRL-11187 wood 5450 Q 40 6280 (6201^6292)
509^510 NSRL-11188 sediment 6480 Q 45 7339, 7339, 7381 (7286^7388)
626^627 NSRL-11189 wood 8180 Q 45 9042, 9071, 9088, 9132, 9147, 9176, 9183 (8992^9212)
Core 1 147^148 NSRL-11684 wood 140 Q 30 265, 218, 142, 24, 2 (1^274)
225^226 NSRL-11685 sediment 250 Q 30 298 (288^308)
465^469 Beta-80900 plant macroremains 1000 Q 60 929 (798^960)
702^704 NSRL-11486 wood 160 Q 35 271, 209, 198, 193, 146, 14, 3 (1^281)

Log-transformed percentile data for 30 most ble 2, Fig. 2) for Maxus 4. The dates are inter-
abundant pollen taxa in the Maxus 4 core were nally consistent with no inversions. An extrapo-
ordinated using the PC-Ord 4.0 version of de- lated age for the bottom of the core is about 9850
trended correspondence analysis (McCune and cal yr BP, suggesting that it covers almost the
Me¡ord, 1999). entire Holocene (Fig. 2). Average accumulation
rates were between 0.4 and 1.1 mm/yr. The high-
est rate of deposition was in the middle of the
3. Results core (252^505 cm, around 1 mm/yr; Table 1).
The sediments and their accumulation rates sug-
3.1. Stratigraphy and LOI gest that the accumulation was continuous.
Four AMS dates were obtained for the Maxus
Maxus 4 and Maxus 1 are 8.00 m and 7.36 m 1 (Table 2, Fig. 2). The oldest date was 1000 Q 60
long, respectively. Sediments of these cores consist yr BP. The basal date was apparently too young
primarily of clay and peat, with silt and sand (probably from tree root materials), and is re-
becoming more important constituents in some jected (Fig. 2). The top two dates are consistent
layers (Table 1, Fig. 2). A band of ¢ne white with a relatively rapidly depositional environ-
volcanic ash was noted in Maxus 4 at 494^496 ment. The extrapolated age of the bottom of the
cm, and ash was also present in more dispersed core is 6 2000 years old (Fig. 2). We consider it
form between 641 and 798 cm. likely that the core represents only the late Holo-
In both cores, sediments fall into three basic cene period. The average sediment accumulation
categories: those with V5% organic content, a rate for the top 470 cm is about 4.7 mm/yr, which
value consistent with local riverine sediment, is much higher than Maxus 4.
those with V15% organic content, consistent
with other closed basin lakes in the Amazonia 3.3. Pollen analysis
£oodplain, and sediments with s 75% organic
material, consistent with shallow, permanently sa- 3.3.1. Maxus 4
turated swamps (Kalliola et al., 1993 ; Bush, un- The sandy sediments at the base of the core
published data). lack pollen, but all other layers are rich in pollen.
Based on a CONISS analysis (Grimm, 1987) and
3.2. Chronology and sediment accumulation rate major changes of abundant pollen taxa, the pollen
record is divided into ¢ve zones (Fig. 3).
Six AMS radiocarbon dates were obtained (Ta- Zone 1 ( s 570 cm; s 8300 cal yr BP): Pollen

PALBO 2448 23-7-02

 C. Weng et al. / Review of Palaeobotany and Palynology 120 (2002) 73^90
PALBO 2448 23-7-02

Fig. 3. Pollen percentage diagram of selected taxa from Maxus 4. AMS radiocarbon dates, corresponding calendar year ages, and lithology are also shown on the
left of the diagram. Organic content in the sediment is shown on the right of the diagram for comparison.

79
80 C. Weng et al. / Review of Palaeobotany and Palynology 120 (2002) 73^90

Fig. 4. Pollen concentration diagram of selected taxa from Maxus 4 (U1000 grains/cm3 ).

PALBO 2448 23-7-02

 C. Weng et al. / Review of Palaeobotany and Palynology 120 (2002) 73^90 81

concentrations are 6 100 000 grains/cm3 (Fig. 4), mataceae, Acalypha, Alchornea, Poaceae and As-
organic content of sediments in this zone is low teraceae are important.
(V10%). The pollen assemblage is diverse and is
not dominated by any single type. Medium high 3.3.2. Maxus 1
abundance of Malpighiaceae, Fabaceae, Myristi- Pollen concentration in this core is generally
caceae, Combretaceae/Melastomataceae and Ficus lower than in Maxus 4 (all less than 65 000
characterizes the pollen assemblages. Cecropia grains/cm3 , except for levels of 243 cm and 461
pollen is scarce ( 6 20%). Polypodium spores are cm, which are around 400 000 grains/cm3 ). The
very common (25^100% total pollen) (Fig. 3). low pollen concentrations are consistent with rel-
Zone 2 (570^350 cm, V8300^5700 cal yr BP): atively high sediment accumulation rate. The rec-
Pollen concentrations are very high ( s 100 000 ord is divided into four zones based on a CONISS
grains/cm3 , with the highest reaching 400 000 analysis (Grimm, 1987) of the percentile data
grains/cm3 ), and organic content is still low (Fig. 5).
(V10%) (Figs. 2 and 4). Extremely high Cecropia Zone 1 (736^475 cm, s V1000 cal yr BP):
pollen percentages (40^80%) (except at depth of Pollen concentration is 6 25 000 grains/cm3 . Or-
410 cm) characterize this zone. Concentrations ganic content is very low ( 6 5%) (Figs. 2 and 6).
of Cecropia pollen are generally between 50 000 Pollen of Urticaceae/Moraceae, Combretaceae/
and 350 000 grains/cm3 . Triporate Urticaceae/ Melastomataceae, Iriartea and Piper is common
Moraceae and Myristicaceae pollen are also abun- but none rises to dominate the assemblage. Cecro-
dant. pia is generally between 15% and 20%, but stead-
Zone 3 (350^150 cm, 5700^3500 cal yr BP): ily increases its abundance in the upper portion of
Pollen concentrations drop to between 50 000 the zone (to V35%). Fern spores, especially tri-
and 100 000 grains/cm3 . Organic content is very lete spores, are abundant at the beginning of the
high (up to 70%). Cecropia pollen percentages de- zone, reaching their maximum representation in
crease to very low levels ( 6 5%). Fabaceae, Mal- the core (V10%), but decline markedly in abun-
pighiaceae, Myristicaceae, Zanthoxylum, Lamia- dance near the top (Fig. 5).
ceae, Ficus and Sloanea sequentially dominated Zone 2 (475^400 cm, 800^1000 cal yr BP): Pol-
the pollen assemblages. len concentration is extremely high at 450 cm due
Zone 4 (150^70 cm, 3500-V1000 years ago): to high Cecropia concentration (nearly 400 000
Total pollen concentrations are very high again grains/cm3 ), but then decreases to 6 50 000
due to high concentrations of Mauritia and Sloa- grains/cm3 (Figs. 5 and 6). Organic content is be-
nea (with as many as 250 000 grains/cm3 ), but tween 15 and 20%. High Cecropia pollen percent-
dropped near the top of the zone to 6 50 000 age characterizes this zone ( s 30%, and reaching
grains/cm3 . Organic content is persistently low V90% at 450 cm). Ficus and Inga pollen are
( 6 20%). Mauritia and Sloanea dominate the pol- abundant at the top of this zone. Other pollen
len percentages ( s 20% to nearly 90%, and V10^ types have low percentage representation due to
75%, respectively). Iriartea is also an important the abundance of Cecropia.
component (V5%). Selaginella and other trilete Zone 3 (400^225 cm, V800^300 cal yr BP):
spores (mostly from Cyathaceae) reach maximal Pollen concentration is generally between 40 000
representation in this zone. and 80 000 grains/cm3 , but at level of 243 cm, it
Zone 5 (70^0 cm, 6 1000 years ago): Pollen reaches 480 000 grains/cm3 . Organic content is
concentration is low at the beginning of the low at the base of the zone (10^15%), but very
zone ( 6 100 000 grains/cm3 ), but increases to high at the top (230^270 cm, as high as 80%).
very high levels in the upper portion of the zone Iriartea pollen is the most abundant pollen type
( s 200 000 grains/cm3 ). Organic contents are very in this zone (V30%). Zanthoxylum, Tapirira,
high (60^70%, matching the levels in zone 3). Acalypha and Combretaceae/Melastomataceae
Myristicaceae pollen is an abundant component are also important components of the pollen spec-
of this uppermost zone. Combretaceae/Melasto- trum. Polypodium spores are consistently present

PALBO 2448 23-7-02

 82
C. Weng et al. / Review of Palaeobotany and Palynology 120 (2002) 73^90
PALBO 2448 23-7-02

Fig. 5. Pollen percentage diagram of selected taxa from Maxus 1. AMS radiocarbon dates and lithology are also shown on the left of the diagram. Organic content
in the sediment is shown on the right of the diagram for comparison.
C. Weng et al. / Review of Palaeobotany and Palynology 120 (2002) 73^90 83

Fig. 6. Pollen concentration diagram of selected taxa from Maxus 1 (U1000 grains/cm3 ).

PALBO 2448 23-7-02

84 C. Weng et al. / Review of Palaeobotany and Palynology 120 (2002) 73^90

Fig. 7. DCA ordination result of the ¢rst two axes for log-transformed pollen samples from Maxus 4. Rare species are down-
weighted. Eigenvalues for axis 1 and axis 2 are 0.275 and 0.069, respectively. Samples from each of the ¢ve pollen zones are dis-
tributed in a relatively close area (circled in the diagram), and samples in di¡erent zones are generally distributed in separate
areas, except for samples of zone 1 and zone 3, indicating the similarity of these two zones. Wet-adapted taxa (Mauritia, Ilex,
Iriartea and Sloanea) have the highest scores on axis 1. Thus, axis 1 expresses a dry to wet gradient. Zones are ordered from old
to young (from bottom to top), and the label of each sample means its level of depth (e.g. L284).

at s 10% of total pollen. The sediment for the 3.5. Detrended correspondence analysis (DCA)
top part of the zone is peat, di¡erent from its result
neighbors, which are silt with ¢ne sands. A peak
of Poaceae pollen (V30%) is found in this layer. The DCA ordination of the Maxus 4 data (Fig.
Zone 4 (225^0 cm, 6 300 cal yr BP): Pollen 7) produces a ¢rst axis that polarizes Mauritia-
concentration is low (30 000^60 000 grains/cm3 ). dominated samples from other samples. This
Organic content increases from 6 10% at the axis is of increasing £ood duration. The second
base to high ( s 60%) on the top. Cecropia and axis polarizes lakeside £oras into those that are
Iriartea are moderately abundant. Alchornea, Cel- dominated by Cecropia (negative extreme) from
tis, Trema and Urticaceae/Moraceae are also im- those that are not. Although zones 1 and 3 are
portant components. very similar in their composition, there is little
overlap of samples between contiguous zones.
3.4. Charcoal From this result, we know that samples from
the same pollen zones are relatively similar to
Almost no charcoal fragments were found in each other, and that changes between zones are
either core. At a few levels, one or two fragments relatively great. This suggests that the environ-
of charcoal ( s 65 Wm) were seen, but they are too mental conditions and vegetation components
rare to be meaningful. were stable during the period represented by

PALBO 2448 23-7-02

 C. Weng et al. / Review of Palaeobotany and Palynology 120 (2002) 73^90 85

each pollen zone, and environmental changes ever, at Maxus 4, some factor, or combination of
were substantial and relatively rapid between factors, maintained a large population of Cecro-
zones. pia close to the coring site for three millennia.
One explanation of a prolonged dominance of
Cecropia in the pollen rain is that humans were
4. Interpretation and discussion disturbing the landscape. At Ayauchi , Ecuador,
human disturbance was inferred from the pres-
4.1. Environmental history ence of charcoal, high abundances of Poaceae
(15^20% of non-Cecropia pollen), Cecropia (40^
The base of Maxus 4 is composed of clays with 90%) and Zea mays (Bush and Colinvaux,
sands. No pollen was preserved in this sediment, 1988). A charcoal peak is also documented in
indicating a prolonged oxidation prior to the for- Maxus 5 that seems to be roughly contemporane-
mation of the modern wetland. This sediment is ous with the Cecropia spike in Maxus 4, though it
the common matrix found beneath wetlands in was not accompanied by high Poaceae values or
western Amazonia and probably represents an an- Zea mays (Athens and Ward, 1999). Although the
cient soil surface. About 9300 cal yr BP, clays increase in Cecropia in Maxus 4 might be re-
were deposited that are rich in pollen indicating garded as a signal indicative of major human dis-
that the site had £ooded. The origin of the wet- turbance, the near or complete absence of char-
land is not known, but it could have been caused coal, Poaceae pollen, or Zea pollen in this core
by river migration or by tectonic activity. As no renders such an interpretation uncertain.
coarse sands underlie this zone it is unlikely that In the modern ecosystems, the principal place
this site was a main river channel, but it could in which large populations of Cecropia occur in
have been a white water (varzea) or blackwater natural settings is where river migration results in
(igapo) £ood system. The spectrum and concen- the accretion of a sandy beach. On such beaches
tration of pollen deposition at this time is consis- there is a predictable succession of Tessaria (As-
tent both with that of modern lowland rain forest teraceae) and Salix (Salicaceae) followed by Ce-
and with that of modern riverine muds. cropia. In these settings, Cecropia can form large,
Between 8700 and 5700 yr BP, Cecropia be- almost monotypic stands that are gradually in-
comes the most abundant pollen type, dominating vaded by larger taxa such as Ficus and Guarea
the pollen spectrum. At lakes in western Amazo- (Foster, 1990; Foster et al., 1986). The Cecropia
nia, Cecropia typically forms 15^20% of the pol- phase probably lasts less than 40 years (Foster,
len count (Bush and Colinvaux, 1988; Liu and 1990) at any one place and its population e¡ec-
Colinvaux, 1988; Lyons-Weiler, 1992), but at tively undergoes local migration at the pace of
Maxus 4 Cecropia accounts for 40^80% of the river migration, occupying new sandbars and
pollen count for 3000 years. beaches as they form and stabilize.
As Cecropia are short-lived pioneer trees that At Cocha Cashu, Peru, a 160-year history of
thrive in disturbed settings, their abundance in successional changes was tracked in the pollen
the early to mid-Holocene requires explanation. rain following abandonment of an oxbow. Cecro-
In the absence of human disturbance, Cecropia pia pollen was typically between 15 and 35% of
are fugitive species that invade tree-fall gaps and the pollen count. The highest representation of
primary successional habitats on the accreting Cecropia pollen, ca. 50%, came about 100 years
beaches associated with meandering rivers (Fos- after abandonment. After 160 years of isolation,
ter, 1990). The sudden £are in abundance of Ce- Cecropia representation had fallen to about 25%
cropia in Maxus 1 may be associated with a single (Bush, unpublished data). The Cocha Cashu rec-
natural disturbance event (Fig. 5). Such a signa- ord, unfortunately, does not provide evidence of
ture is consistent with an event that allows Cecro- an analog mechanism that could maintain unusu-
pia to establish a population for one generation ally high Cecropia representation for thousands of
prior to being snu¡ed out by competition. How- years.

PALBO 2448 23-7-02

86 C. Weng et al. / Review of Palaeobotany and Palynology 120 (2002) 73^90

At Maxus 4, if bankside populations are the cropia near to the site. Overlying samples contain
origin of the mid-Holocene peak of Cecropia, it higher concentrations of forest taxa suggesting
is evident that there was a higher energy £uvial that there may be a separate explanation for the
system causing a greater proportion of the £ood- occurrence of high concentrations of Cecropia in
plain to remain in an early successional state. these levels.
Otherwise the river must have been very close to Studies in semi-deciduous forests on Barro Col-
the coring site (the Rio Yasuni is presently about orado Island, Panama, showed that Cecropia is
1 km distant from Maxus 4). Ra«sa«nen (personal strongly over-represented in local pollen spectra
communication), however, has not seen evidence by a factor of about 17 relative to its basal area
of increased erosive activity in Ecuadorian rivers in the adjacent forest (Bush and Rivera, 1998,
at this time. 2001). In mature forests of the Yasuni, Cecropia
For Cecropia pollen to be found at 40^80% contributes ca. 6 1% basal area in £oodplain for-
suggests a very close source area, i.e. less than ests and terra ¢rme forests (Pitman, personal
0.5 km distant. Yet, Maxus 4 has a slow deposi- communication). Therefore, a background repre-
tional rate (compared with Maxus 1), suggesting sentation of Cecropia in the pollen record of ca.
that £ooding from nearby rivers was rare. Evi- 10^15% might be expected. Additional representa-
dence for a lack of regular £ooding also comes tion is likely as Cecropia take advantage of light
from the pollen spectrum. River muds do not availability and gap formation at lake or river
contain such high percentages of Cecropia pollen margins. This additional source of Cecropia e¡ec-
and contain a lower concentration of pollen than tively doubles its representation in the pollen rec-
is found at this site (Haberle, 1997 ; our unpub- ord.
lished data). In Maxus 4, a band of silty clay with At Maxus 4, after the initial successional phase
a markedly di¡erent pollen signature and low pol- at the base of the Cecropia zone, the proportion
len concentration occurs at 4 m and it is probable of Cecropia pollen is around 60%, approximately
that this represents either a phase of £ooding or a two to three times the amount normally found.
single £ood event. Thus, if a signi¢cant propor- Given that the forest still contains many large
tion of the pollen was derived from £oodwater, trees (as indicated by other components of the
the pollen composition of Maxus 4 would be ex- pollen spectrum), it appears probable that canopy
pected to resemble that of rivers, rather than this turnover was more rapid. Increased gap forma-
Cecropia-rich spectrum. It seems improbable that tion would favor increased Cecropia abundance.
a river with enough energy to maintain extensive One possible explanation is that periodic drought,
areas of primary succession lies so close to Maxus such as that experienced in 1983 on Barro Colo-
4 that its marginal vegetation dominates the pol- rado Island could cause mortality and increased
len spectrum, but does not regularly £ood into canopy turnover. In that drought, the most
this adjacent wetland. strongly a¡ected individuals were large canopy
The pollen concentration data (Fig. 4) reveal trees ( s 20 cm dbh), and their mortality increased
more detail of the Cecropia-dominated zone. from a ‘normal’ rate of 2% per annum to almost
The initial rise of Cecropia provides a peak in 7% per annum during the three years following
pollen concentration. The only other pollen taxon the drought (Condit et al., 1995, 1996). It will
to show a strong response at this time is Combre- be noted that this single drought induced a
taceae/Melastomataceae. Both of these families three-fold increase of large tree mortality, and
occur in almost all habitats in the modern forests, by inference, a substantial increase in canopy
and both could contribute to early successional gap formation.
phases, although Miconia spp. (Melastomataceae) The amount of canopy gap formation needed
are especially abundant in early successional envi- to explain the increase in Cecropia pollen in the
ronments. Given the paucity of other forest ele- Maxus 4 record would be about a doubling of gap
ments, this Cecropia peak is consistent with the area. Hence, the probable gap formation caused
formation of a riverine primary succession of Ce- by a severe drought is about the right scale of

PALBO 2448 23-7-02

 C. Weng et al. / Review of Palaeobotany and Palynology 120 (2002) 73^90 87

event to account for the Cecropia abundance. As invaded by swamp taxa. By 4900 cal yr BP,
there is no evidence to suggest that the moist for- when the pool had almost ¢lled with sediments,
est of this region was replaced by an overall drier peat started to form. A renewed £ooding event at
assemblage, we infer that the droughts were epi- ca. 3700 cal yr BP is suggested by a resumption of
sodic with su⁄cient return periodicity to maintain clay deposition and the dominance of Mauritia.
relatively high rates of canopy turnover. If this This transition has been found in several sites in
hypothesis is correct, the cycle of droughts would Amazonia (Behling and Hooghiemstra, 1999;
have started approximately 8700 cal yr BP and Bush et al., 2000; Frost, 1988; Lyons-Weiler,
ended approximately 5800 cal yr BP. Interest- 1992; Mayle et al., 2000) and probably represents
ingly, Athens and Ward (1999) suggest drier con- a regional increase in precipitation.
ditions at this same time in Maxus 5, though they At about 1000 cal yr BP, the hydrarch succes-
infer a climate regime characterized by more in- sion resumes and Mauritia is replaced by wetland
tense seasonality than at present. Distinguishing elements such as Myristicaceae (cf. Iryanthera)
between cyclical droughts, as we believe for and Combretaceae/Melastomataceae. For the
Maxus 4, and seasonally dry conditions as sug- last 1000 years, the hydrological and vegetational
gested for Maxus 5, however, will be a consider- conditions have been relatively stable, supporting
able challenge for paleoecology. an essentially modern vegetation.
The Yasuni inference of drought conditions co- This hydrarch succession in the last 1000 years
incides with the marked lowering of lake level at can be seen most clearly in Maxus 1. The basal
Lake Titicaca, which receives most of its water sediments of this core are similar to those at the
from the winds carrying Amazonian moisture base of Maxus 4, except pollen is preserved in
into the Andes (Hastenrath, 1985). Coinciden- Maxus 1, although the concentrations are very
tally, this same period has been suggested to be low (Fig. 6). The initiation of the Maxus 1 wet-
a time of reduced El Nin‹o activity (Keefer et al., land is probably similar to that of Maxus 4. This
1998; Sandweiss et al., 1996). However, the root lake was formed during the wet period docu-
cause of these drought events remains enigmatic. mented in Maxus 4 (3700^1000 cal yr BP). Sands
The onset of wetter conditions may have caused in the basal sediments are consistent with a river-
the drop in Cecropia abundance at 5800 cal yr BP. ine origin and are overlain by sandy clays typical
After 5800 cal yr BP, the pollen spectrum is more of an abandoned oxbow lake that is still season-
typical of undisturbed lowland forest. A sequence ally £ooded. A Cecropia peak occurred at about
of species rising and falling in abundance suggests 1000 cal yr BP when the sediment changed from
a hydrological succession. Very similar patterns of sandy clay to silty clay. This sedimentary transi-
succession have been noted in late Holocene tion may indicate the complete abandonment of
Ecuadorian sequences from An‹angucocha (Frost, the lake and a cessation of periodic £ooding by
1988) and Zancudococha (Lyons-Weiler, 1992). In the river. The AMS date indicates that this peak
the Colombian savannas of the Llanos Orientales, follows soon after the apparent onset of drier
Behling and Hooghiemstra (1999) suggest the conditions at Maxus 4. The Cecropia trees may
presence of a wet phase from ca. 6350 cal yr represent colonization of exposed shoreline, or a
BP. This further corresponds remarkably closely severe drought that opened canopy gaps. Unlike
to rising levels documented in two central Ama- the early Holocene record of Maxus 4, this Cecro-
zonian lakes at ca. 6500 cal yr BP (Bush et al., pia peak is quickly followed by peaks of Ficus,
2000), and suggests the widespread onset of wetter Cedrela and other trees in a typical succession
climatic conditions in the Holocene at this time. (Foster et al., 1986; Foster, 1990). By about 300
The short-lived peaks of pollen types evident in cal yr BP, sediments almost ¢lled the lake. Rela-
the Maxus 4 record could re£ect the successional tively high abundances of Poaceae and ferns sug-
phases as the basin ¢lls with sediment. However, gest the formation of a £oating mat, which has
it is equally likely to be a relatively stochastic se- been steadily invaded to provide the present
ries of events in which a small pool is serially wooded swamp. Evidence for human land use is

PALBO 2448 23-7-02

88 C. Weng et al. / Review of Palaeobotany and Palynology 120 (2002) 73^90

surprisingly sparse in Maxus 1 despite the pres- may have been important in helping to maintain
ence of nearby archeological sites (Netherly, species diversity through continual low levels of
1997). ecosystem disturbance (Connell, 1978) that pre-
vented competitive exclusion.

5. Conclusion
Acknowledgements
Our data from two cores in the eastern Ecua-
dorean lowlands document an almost complete We thank the government and people of Ecua-
Holocene sequence that re£ects vegetational and dor for allowing us to work in their country. This
climatic change. Unusually, these cores seem to study was supported by a grant from the National
lack evidence for human landscape disturbance, Scienti¢c Foundation DEB-9732951. Drs. H. Beh-
which fortuitously allows a clearer vision of nat- ling and H. Hooghiemstra reviewed the manu-
ural changes in the record than has been obtained script and provided valuable suggestions to im-
from other Ecuadorian sites. This ¢nding also prove it.
suggests a rather low or negligible impact on the
landscape by prehistoric Amerindians, presum-
ably indicative of quite low population densities
in the inter£uvial areas. References
A long-lasting plateau of Cecropia pollen
Absy, M.L., 1979. A Palynological Study of Holocene Sedi-
(8700^5800 cal yr BP) is tentatively interpreted ments in the Amazon Basin. Ph.D. Dissertation, University
to indicate a period in which increased tree mor- of Amsterdam, Amsterdam.
tality led to more gap formation and a relatively Athens, J.S., Ward, J.V., 1999. The late Quaternary of the
high population of Cecropia. One explanation to western Amazon: climate, vegetation and humans. Antiq-
uity 73, 287^302.
account for this mortality is sequences of severe
Bale¤e, W., 1988. Indigenous adaptation to Amazonian palm
droughts occurring on a decadal timescale. The forests. Principes 32, 47^52.
timing of this drought-prone period corresponds Bartlett, A.S., Barghoorn, E.S., 1973. Phytogeographic history
well with other early Holocene records from of the Isthmus of Panama during the past 12, 000 years: A
Amazonia, adjacent savannas, and the Andes. history of vegetation, climate, and sea-level change. In: Veg-
After 5800 cal yr BP, more uniform conditions etation and Vegetational History of Northern Latin Amer-
ica. Elsevier, Amsterdam, pp. 203^299.
allowed the development of mature forests, and a Behling, H., 1996. First report on new evidence for the occur-
hydrarch succession that gradually ¢lled the lake rence of Podocarpus and possible human presence at the
with sediment. The last 5800 years have seen al- mouth of the Amazon during the late-glacial. Vegetat.
ternating wet and dry millennial-scale events. The Hist. Archaeobot. 5, 241^246.
Behling, H., Da Costa, M.L., 2000. Holocene environmental
e¡ects of these oscillations of water level are evi-
changes from the Rio Curua¤ record in the Caxiuana‹ region,
dent in the hydrology of the site and in the com- eastern Amazon Basin. Quat. Res. 53, 369^377.
munities of wetland plants, rather than in any Behling, H., Hooghiemstra, H., 1999. Environmental history
regional change in vegetation types. We infer of the Colombian savannas of the Llanos Orientales since
that climatic conditions between 5800 and 4900 the Last Glacial Maximum from lake records El Pinal and
cal yr BP were relatively wet, from 4900 to 3700 Carimagua. J. Paleolimnol. 21, 461^476.
Bush, M.B., Colinvaux, P.A., 1988. A 7000-year pollen record
cal yr BP they were relatively dry, from 3700 to from the Amazon lowlands, Ecuador. Vegetatio 76, 141^
1000 cal yr BP they were relatively wet, and from 154.
1000 cal yr BP to present they were (and are) Bush, M.B., Colinvaux, P.A., 1994. A paleoecological perspec-
relatively dry. tive of tropical forest disturbance: Records from Darien,
As might be expected in such a wet region, the Panama. Ecology 75, 1761^1768.
Bush, M.B., Miller, M.C., De Oliveira, P.E., Colinvaux, P.A.,
landscape appears to have supported moist trop- 2000. Two histories of environmental change and human
ical rain forest throughout the Holocene. How- disturbance in eastern lowland Amazonia. Holocene 10,
ever, the climatic oscillations evident in this core 543^554.

PALBO 2448 23-7-02

 C. Weng et al. / Review of Palaeobotany and Palynology 120 (2002) 73^90 89

Bush, M.B., Piperno, D.R., Colinvaux, P.A., 1989. A 6000 Seed Dispersal. Dr W. Junk Publishers, Dordrecht, pp. 357^
year history of Amazonian maize cultivation. Nature 340, 370.
303^305. Frost, I., 1988. A Holocene sedimentary record from An‹angu-
Bush, M.B., Piperno, D.R., Colinvaux, P.A., Krissek, L., De cocha in the Ecuadorian Amazon. Ecology 69, 66^73.
Oliveira, P.E., Miller, M.C., Rowe, W., 1992. A 14,300 year Gentry, A.H., 1988. Changes in plant community diversity and
paleoecological pro¢le of a lowland tropical lake in Panama. £oristic composition on environmental and geographical
Ecol. Monogr. 62, 251^276. gradients. Ann. Missouri Bot. Gard. 75, 1^34.
Bush, M.B., Rivera, R.S., 1998. Pollen dispersal and represen- Grimm, E.C., 1987. CONISS: A Fortran 77 program for
tation in a neotropical forest. Global Ecol. Biogeogr. Lett. 7, stratigraphically constrained cluster analysis by the method
78^92. of the incremental sum squares. Comput. Geosci. 13, 13^35.
Bush, M.B., Rivera, R., 2001. Reproductive ecology and pol- Haberle, S.G., 1997. Late Quaternary vegetation and climate
len representation among neotropical trees. Global Ecol. history of the Amazon Basin: Correlating marine and ter-
Biogeogr. 10, 359^367. restrial pollen records. In: Flood, R.D., Piper, D.J. W.,
Colinvaux, P.A., 1996. Quaternary environmental history and Klaus, A., Peterson, L.C. (Eds.), Proc. ODP Sci. Results,
forest diversity in the Neotropics. In: Jackson, J.B.C., Budd, pp. 381^396.
A.F., Coates, A.G. (Eds.), Evolution and Environment Ha¡er, J., 1969. Speciation in Amazonian forest birds. Science
in Tropical America. University of Chicago Press, Chicago, 165, 131^137.
IL. Hastenrath, S., Kutzbach, J., 1985. Late Pleistocene climate
Colinvaux, P.A., De Oliveira, P.E., Bush, M.B., 2000. Ama- and water budget of the South American Altiplano. Quat.
zonian and neotropical plant communities on glacial time- Res. 24, 249^256.
scales: the failure of the aridity and refuge hypothesis. Quat. Hooghiemstra, H., 1984. Vegetational and Climatic History of
Sci. Rev. 19, 141^169. the High Plain of Bogota', Colombia: A Continuous Record
Colinvaux, P.A., De Oliveira, P.E., Moreno, J.E., 1999. Ama- of the last 3.5 Million Years. Dissertation Botanicae, 79,
zon Pollen Manual and Atlas. Harwood Academic Press, Vaduz.
New York. Kalliola, R., Puhakka, M., Danjoy, W., 1993. Amazonia Pe-
Condit, R., Hubbell, S.P., Foster, R., 1995. Mortality rates of ruana: Vegetacio¤n Hu¤meda Tropical en el Llano Subandina.
205 Neotropical tree and shrub species and their responses Gummerus Printing, Jyva«skyla«.
to severe drought. Ecol. Monogr. 65, 419^439. Keefer, D.K., deFrance, S.D., Moseley, M.E., Richardson,
Condit, R., Hubbell, S.P., Foster, R., 1996. Changes in tree J.B.I., Satterlee, D.R., Day-Lewis, A., 1998. Early maritime
species abundance in a Neotropical forest over eight years: economy and El Nin‹o events at Quebrada Tacahuay, Peru.
Impact of climate change. J. Trop. Ecol. 12, 231^256. Science 281, 1833^1835.
Connell, J.H., 1978. Diversity in tropical rain forests and coral Liu, K.-b., Colinvaux, P.A., 1988. A 5200-year history of Am-
reefs. Science 199, 1302^1310. azon rain forest. J. Biogeogr. 15, 231^248.
Dean, W.E.Jr., 1974. Determination of carbonate and organic Lyons-Weiler, J., 1992. Palynological Analysis of a Holocene
matter in calcareous sediments and sedimentary rocks by Sedimentary Sequence from the Western Amazon basin.
loss on ignition: comparison with other methods. J. Sedi- M.S. Thesis, Ohio State University, Columbus, OH.
ment. Petrol. 44, 242^248. Mayle, F.E., Burbridge, R., Killeen, T.J., 2000. Millennial-
Denevan, W.M., 1976. The aboriginal population of Amazo- scale dynamics of southern Amazonian rain forests. Science
nia. In: The native population of the Americas in 1492. 290, 2291^2294.
University of Wisconsin Press, Madison, WI. McCune, B., Me¡ord, M.J., 1999. PC_ORD. Multivariate
Denevan, W.M., 1992. The pristine myth: the landscape of the Analysis of ECological Data. MJM Software Design, Gle-
Americas in 1492. Ann. Assoc. Am. Geogr. 82, 369^385. neden Beach, OR.
Dinerstein, E., Olson, D.M., Graham, D.J., Webster, A.L., Nelson, B.W., Kapos, V., Adams, J.B., Oliveira, W.J., Braun,
Primm, S.A., Bookbinder, M.P., Ledec, G., 1995. A Con- O.P.G., do Amaral, I.L., 1994. Forest disturbance by large
servation Assessment of the Terrestrial Ecoregions of Latin blowdowns in the Brazilian Amazon. Ecology 75, 853^858.
America and the Caribbean. World Bank, Washington, DC. Netherly, P.J., 1997. Loma y ribera: patrones de asentamiento
Faegri, K., Iversen, J., 1989. Textbook of Pollen Analysis. prehistoricos en la Amazonia Ecuatoriana. Front. Invest.
Wiley, Chichester. (Quito) 11, 33^54.
Foster, M.S., Terborgh, J., 1998. Impact of a rare storm event Piperno, D., Holst, I., 1998. The presence of starch grains on
on an Amazonian forest. Biotropica 30, 470^474. prehistoric stone tools from the humid neotropics: Indica-
Foster, R.B., 1990. Long-term change in the successional for- tions of early tuber use and agriculture in Panama. J. Ar-
est community of the Rio Manu £oodplain. In: Gentry, chaeol. Sci. 25, 765^776.
A.H. (Ed.), Four Neotropical Rainforests. Yale University Piperno, D.R., 1990. Aboriginal agriculture and land usage in
Press, New Haven, CT, pp. 565^572. the Amazon Basin, Ecuador. J. Archaeol. Sci. 17, 665^677.
Foster, R.B., Arce, J.B., Wachter, T.S., 1986. Dispersal and Piperno, D.R., Becker, P., 1996. Vegetational history of a site
the sequential plant communities in Amazonian Peru £ood- in the Central Amazon basin derived from phytolith and
plain. In: Estrada, A., Fleming, T.H. (Eds.), Frugivores and charcoal records from natural soils. Quat. Res. 45, 202^209.

PALBO 2448 23-7-02

90 C. Weng et al. / Review of Palaeobotany and Palynology 120 (2002) 73^90

Piperno, D.R., Bush, M.B., Colinvaux, P.A., 1991. Paleoeco- nett, W., Imazio Da Silveira, M., Henderson, A., Sliva, J.,
logical perspectives on human adaptation in Panama. II: Cherno¡, B., Reese, D.S., Holman, J.A., Toth, N., Schick,
The Holocene. Geoarchaeology 6, 227^250. K., 1996. Paleoindian cave dwellers in the Amazon: the
Pitman, N.C.A., Terborgh, J., Silman, M.R., Nun‹ez, P.V., peopling of the Americas. Science 272, 373^384.
1999. Tree species distributions in an upper Amazonian for- Roubik, D.W., Moreno, E., 1991. Pollen and Spores of Barro
est. Ecology 80, 2651^2661. Colorado Island. Monographs in Systematic Botany, vol.
Porro, A., 1994. Social organization and political power in the 36, Missouri Botanical Garden, St. Louis, MO.
Amazon £oodplain: the ethnohistorical sources. In: Roose- Sandweiss, D.H., Richardson, J.B.I., Reitz, E.J., Rollins, H.B.,
velt, A. (Ed.), Amazon Indians from Prehistory to the Maasch, K.A., 1996. Geoarchaeological evidence from Peru
Present: Anthropological Perspectives. University of Arizo- for a 5000 years B.P. onset of El Nin‹o. Science 273, 1531^
na Press, Tucson, AZ, pp. 79^94. 1533.
Romoleroux, K., Foster, R., Valencia, R., Condit, R., Balslev, Servant, M., Fontes, J.-C., Rieu, M., Salie'ge, X., 1981. Phases
H., Losos, E., 1997. Espcies len‹osas (dap, 1 cm) encontracas climatiques arides holoce'nes dans le sud-ouest de l’Amazo-
en dos hectareas de un bosque de la Amazon|¤a ecuatoriana. nie (Bolivie). C.R. Acad. Sci. Paris Ser. II 292, 1295^
In: Valencia, R., Balslev, H. (Eds.), Estudios sobre Diversi- 1297.
dad y Ecolog|¤a de Plantas: Memoria del II Congreso Ecua- Stockmarr, J., 1971. Tablets with spores used in absolute pol-
toriano de Bota¤nica. Ponti¢cia Universidad Cato¤lica del len analysis. Pollen Spores 13, 615^621.
Ecuador, Quito, pp. 189^216. Stuiver, M., Reimer, P.J., 1993. Extended 14 C database and
Roosevelt, A.C., Lima Da Costa, M., Lopes Machado, C., revised CALIB radiocarbon calibration program. Radiocar-
Michab, M., Mercier, N., Valladas, H., Feathers, J., Bar- bon 35, 215^230.

PALBO 2448 23-7-02