Neotropical Seasonally Dry Forests and Quaternary vegetation changes

Abstract

Seasonally dry tropical forests have been largely ignored in discussions of vegetation changes during the Quaternary.
We distinguish dry forests, which are essentially tree‐dominated ecosystems, from open savannas that have a xeromorphic fire‐tolerant, grass layer and grow on dystrophic, acid soils. Seasonally dry tropical forests grow on fertile soils, usually have a closed canopy, have woody floras dominated by the Leguminosae and Bignoniaceae and a sparse ground flora with few grasses. They occur in disjunct areas throughout the Neotropics. The Chaco forests of central South America experience regular annual frosts, and are considered a subtropical extension of temperate vegetation formations.
At least 104 plant species from a wide range of families are each found in two or more of the isolated areas of seasonally dry tropical forest scattered across the Neotropics, and these repeated patterns of distribution suggest a more widespread expanse of this vegetation, presumably in drier and cooler periods of the Pleistocene.
We propose a new vegetation model for some areas of the Ice‐Age Amazon: a type of seasonally dry tropical forest, with rain forest and montane taxa largely confined to gallery forest. This model is consistent with the distributions of contemporary seasonally dry tropical forest species in Amazonia and existing palynological data.
The hypothesis of vicariance of a wider historical area of seasonally dry tropical forests could be tested using a cladistic biogeographic approach focusing on plant genera that have species showing high levels of endemicity in the different areas of these forests.

Full text

Journal of Biogeography,
27
, 261–273
© 2000 Blackwell Science Ltd
Original Article
Blackwell Science, Ltd
Neotropical seasonally dry forests and
Quaternary vegetation changes
R. Toby Pennington*, Darién E. Prado† and Colin A. Pendry* *
Royal Botanic Garden, 20a
Inverleith Row, Edinburgh EH3 5LR, U.K. and
†
Cátedra de Botánica, Facultad de Ciencia
Agrarias, Universidad Nacional de Rosario (2123) Zavalla, Argentina
Abstract
Seasonally dry tropical forests have been largely ignored in discussions of vegetation
changes during the Quaternary.
We distinguish dry forests, which are essentially tree-dominated ecosystems, from open
savannas that have a xeromorphic fire-tolerant, grass layer and grow on dystrophic, acid
soils. Seasonally dry tropical forests grow on fertile soils, usually have a closed canopy,
have woody floras dominated by the Leguminosae and Bignoniaceae and a sparse
ground flora with few grasses. They occur in disjunct areas throughout the Neotropics.
The Chaco forests of central South America experience regular annual frosts, and are
considered a subtropical extension of temperate vegetation formations.
At least 104 plant species from a wide range of families are each found in two or more
of the isolated areas of seasonally dry tropical forest scattered across the Neotropics, and
these repeated patterns of distribution suggest a more widespread expanse of this vegeta-
tion, presumably in drier and cooler periods of the Pleistocene.
We propose a new vegetation model for some areas of the Ice-Age Amazon: a type of
seasonally dry tropical forest, with rain forest and montane taxa largely confined to gallery
forest. This model is consistent with the distributions of contemporary seasonally dry
tropical forest species in Amazonia and existing palynological data.
The hypothesis of vicariance of a wider historical area of seasonally dry tropical forests
could be tested using a cladistic biogeographic approach focusing on plant genera that
have species showing high levels of endemicity in the different areas of these forests.
Keywords
Seasonally dry tropical forests, neotropics, Pleistocene refugia, Amazon, plant distribu-
tion patterns, vicariance, gallery forests, chaco, cerrado.
INTRODUCTION
The vegetation of seasonally dry areas of the tropics has
received relatively little attention from conservationists and
ecologists relative to that given to rain forests (Janzen, 1988;
Mooney
et al.
, 1995) and the lack of knowledge about the
vegetation of these regions extends into the debate about the
historical biogeography of the Neotropics. In the Neotropics
there are floristically and ecologically distinct types of forest
and savanna vegetation in these dry areas, and these vegeta-
tion types must be considered separately in biogeographical
analyses because their component species react differently to
environmental changes.
This paper aims to clarify differences between these types
of vegetation, and draws special attention to the significance of
one dry forest type, which we term ‘seasonally dry tropical
forest’. We discuss the evidence of both drying and cooling
in the Neotropics during the Upper Pleistocene and present
a novel synthesis of biogeographic and palynological data
which incorporates both of these processes. We propose a
new model of the vegetation of areas of the Ice-Age lowland
Neotropics: the Pleistocene seasonally dry tropical forest.
NEOTROPICAL DRY FOREST AND SAVANNA
VEGETATION
In discussing vegetation in seasonal areas of the tropics it is
useful to distinguish dry forests and savannas, although their
relationships are notoriously complex (Furley
et al.,
1992;
see below under ‘cerrado’). Seasonally dry tropical forest
Correspondence: R. Toby Pennington, Royal Botanic Garden, 20a Inverleith
Row, Edinburgh EH3 5LR, U.K.
JBI397.fm Page 261 Thursday, July 27, 2000 5:51 PM

262 R. T. Pennington, D. E. Prado and C. A. Pendry
© Blackwell Science Ltd 2000,
Journal of Biogeography
,
27
, 261–273
occurs where the rainfall is less than 1600 mm/year, with at
least 5–6 months receiving less than 100 mm (Gentry, 1995;
Graham & Dilcher, 1995). The vegetation is mostly deciduous
during the dry season, and along a gradient deciduousness
increases as rainfall declines, though in the driest forests
there is a marked increase in evergreen and succulent species
(Mooney
et al.
, 1995). Savannas are found under similar or
slightly wetter climatic conditions but tend to be on poorer
soils (Sarmiento, 1992); savanna trees frequently have sclero-
phyllous, evergreen leaves (Ratter
et al.
, 1997). Tropical dry
forests are essentially tree-dominated ecosystems with a more
or less continuous canopy and in which grasses are a minor
element (Mooney
et al.
, 1995), whilst a xeromorphic, fire-
tolerant grass layer is an important component of savannas
.
Forests
Seasonally dry tropical forests
Seasonally dry tropical forests have a smaller stature and
lower basal area than tropical rain forests (Murphy & Lugo,
1986), and thorny species are often prominent. Ecological
processes are strongly seasonal, and Net Primary Productivity
is lower than in rain forests because growth only takes place
during the wet season. There is a build up of leaf litter during
the dry season because sunlight penetrates to the forest floor
and decomposition ceases in the low relative humidity. Flower-
ing and fruiting phenologies are strongly seasonal, and many
species flower synchronously at the transition between the
dry and wet seasons whilst the trees are still leafless (Bullock,
1995). Conspicuous flowers and wind-dispersed seeds are
frequent, in contrast to rain forests.
We follow Murphy & Lugo (1995) and define seasonally
dry tropical forest in a distinctly general manner. It includes
formations as diverse as tall forest on moister sites to cactus
scrub on the driest. Many different names are used for the
vegetation which we include under this definition (e.g. tropical
and subtropical dry forests, caatinga, mesotrophic, mesophil-
ous or mesophytic forest, semideciduous or deciduous forest,
bosque caducifolio, bosque espinoso; see Murphy & Lugo,
1995 for a fuller discussion), and this plethora of names has
probably served to confuse the links among the forests
of different regions, rather than emphasize their similar-
ities. The Leguminosae and Bignoniaceae dominate the
woody floras of these forests throughout their range with the
Anacardiaceae, Myrtaceae, Rubiaceae, Sapindaceae, Euphor-
biaceae, Flacourtiaceae and Capparidaceae also more or less
strongly represented (Gentry, 1995). The Cactaceae are
prominent in the understorey, particularly at the formation’s
latitudinal extremes, and are an important element in the
diversity of these forests (Gentry, 1995). Seasonally dry tropical
forests usually have a closed canopy, with a sparse ground
flora consisting of rather few grasses, with Bromeliaceae,
Compositae, Malvaceae and Marantaceae also represented.
The largest areas of seasonally dry tropical forests in South
America are found in north-eastern Brazil (the ‘caatingas’,
extending south to eastern Minas Gerais), in two areas
defined by Prado & Gibbs (1993) as the ‘Misiones’ and
‘Piedmont’ nuclei (Fig. 1) and on the Caribbean coasts of
Colombia and Venezuela. Other, smaller and more isolated
areas of seasonally dry tropical forests occur in dry valleys in
the Andes in Bolivia, Peru, Ecuador, and Colombia, coastal
Ecuador and northern Peru, the ‘Mato Grosso de Goiás’ in
Central Brazil and scattered throughout the Brazilian cerrado
biome on areas of fertile soils (Ratter
et al.
, 1978). In Central
America, seasonally dry forests are concentrated along the
Pacific coast from Guanacaste in northern Costa Rica, to just
north of the Tropic of Cancer in the Mexican state of Sonora.
Within all of these areas seasonally dry tropical forests occur
within a complex of vegetation types depending on local
climatic, soil and topographic conditions. Phytosociological
analysis of diverse woody stands from Argentina to the
Brazilian Amazon indicate that seasonally dry tropical forests,
including Brazilian caatinga and semideciduous forests of the
Paranense province (
sensu
Cabrera & Willink, 1980), form
a cohesive unit quite distinct from both chaco (see below)
cerrado and rain forest (Prado, 1991, in press).
Seasonally dry tropical forests occur on fertile soils with a
moderate to high pH and nutrient status and low levels of
aluminium. Such soils are favourable for agriculture (Ratter
et al.
, 1978), which has resulted in enormous destruction of
these forests in many areas (e.g. less than 2% of seasonally
dry forests on the Pacific coast of Mesoamerica are still
intact; Janzen, 1988), a problem exacerbated by the large
human populations in many Neotropical dry forest life
zones (Murphy & Lugo, 1995).
There is good evidence from the contemporary distribu-
tion of species in the disjunct areas of seasonally dry tropical
forests for historical links among all these areas. Prado &
Gibbs (1993) compared dot maps of distributions of individual
Figure 1 The distribution of seasonally dry vegetation in the
Neotropics. Seasonally dry forest; 1, Caatingas. 2, Misiones
Nucleus. 3, Bolivian Chiquitano region. 4, Piedmont Nucleus.
5, Bolivian and Peruvian InterAndean valleys. 6, Pacific coastal
Ecuador. 7, Colombian InterAndean valleys. 8, Caribbean coast
of Colombia and Venezuala. 9, Central America. 10, Antilles.
Savannas: Ce, Cerrado. Ll, Llanos. Ru, Rupununi. Ch, Chaco.
JBI397.fm Page 262 Thursday, July 27, 2000 5:51 PM

Neotropical seasonal forest biogeography 263
© Blackwell Science Ltd 2000,
Journal of Biogeography
,
27
, 261–273
species and documented forty phylogenetically unrelated
species that are distributed in up to ten disjunct areas of
seasonally dry tropical forest, but which are absent from
intervening moist and savanna vegetation. Examination of
monographs for this paper revealed another sixty-four such
species, and more will certainly be revealed. There are two
possible explanations for these coincident distribution
patterns: the separation (vicariance) of formerly continuous
distributions, or multiple independent dispersal events among
different areas. Because it is more parsimonious to assume
a single event of vicariance rather than many independent
dispersal hypotheses, Prado and Gibbs concluded that ‘these
fragmentary and mostly disjunct distributional patterns are
vestiges of a once extensive and largely contiguous seasonal
woodland formation, which may have reached its maximum
extension during a dry-cool period
≈
18,000–12,000
bp
,
coinciding with the contraction of the humid forest’.
Recent taxonomic publications which provide more
examples of distributions congruent with those outlined
by Prado & Gibbs (1993) include Lewis (1998; p. 22) which
states: ‘… the overall pattern of
Poincianella-Erythrostemon
species [a group within the genus
Caesalpinia
] fits the “pleis-
tocenic arc” of Prado & Gibbs (1993)’, and monographs by
Barneby (1991) and Barneby & Grimes (1997).
Chloroleucon
mangense
Britton & Rose (Fig. 2) is a fine example of a species
with a widespread distribution growing in disjunct areas
of ‘deciduous and semideciduous, more or less xeromorphic
woodland or chapparal at elevations mostly below 600 m,
exceptionally to 1200 m, widespread in locally differentiated
forms from the S margin of the Sonoran desert in NW Mexico
to Colombia, Venezuela and the Antilles, thence south along
the E foothills and intermontane valleys of the Andes into
NE Bolivia’ (Barneby & Grimes, 1997; p. 150). These vegeta-
tion types are all included in our concept of seasonally dry
tropical forest. Other similar examples are
Mimosa tenuiflora
Benth. (Fig. 3) and
Mimosa hexandra
M. Micheli (Fig. 4)
(Barneby, 1991), several species of
Piptadenia
(Lewis, 1991)
and the genus
Leucaena
(Hughes, 1998).
A recent study of the biogeography of the Cactaceae has
uncovered numerous examples of taxa which conform to
those demonstrated by Prado & Gibbs (1993; N.P. Taylor,
pers. comm.). For example,
Praecereus euchlorus
s.l
. (Weber)
N.P. Taylor ranges through seasonally dry tropical forests in
southern Brazil, Paraguay, Bolivia, Peru, Ecuador, Colombia
and Venezuela and
Cereus jamacaru
DC. is found in eastern
Brazil, northern South America and the Caribbean. Numerous
similar examples can be found (e.g.
Brasilicereus
,
Brasiliopunta
,
Harrisia
,
Lepismium
,
Pilosocereus
; N.P. Taylor, pers. comm.)
A wider appreciation of the former extent of seasonally
dry forests can influence taxonomic decisions. For example,
Figure 2 Distribution of Chloroleucon mangense (Jacquin) Britton
& Rose s.l. Redrawn from Barneby & Grimes (1996).
Figure 3 Distribution of Mimosa tenuiflora (Willdenow) Poiret.
Redrawn from Barneby (1991).
Figure 4 Distribution of Mimosa hexandra M. Micheli. Redrawn
from Barneby (1991).
JBI397.fm Page 263 Thursday, July 27, 2000 5:51 PM

264 R. T. Pennington, D. E. Prado and C. A. Pendry
© Blackwell Science Ltd 2000,
Journal of Biogeography
,
27
, 261–273
Barneby & Grimes (1997) maintained the distinction between
Pithecellobium diversifolium
Bentham and
P. excelsum
(Kunth)
Bentham, but noted that they were distinguished ‘as much
by the immense geographical discontinuity between them as
much as by any macromorphological divergence’.
P.
diversifo-
lium
is a caatinga species, found in north-eastern Brazil, and
P.
excelsum
grows in seasonally dry forests both in Andean
valleys in northern Peru and Southern Ecuador, and on the
Pacific coast of Ecuador and northern Peru (Fig. 5). The only
absolutely diagnostic character in their key is geographical
range since there is overlap in the two morphological characters
cited. Postulating a more extensive and now fragmented range
for dry forest vegetation removes the temptation to maintain
these as separate species simply because of their wide disjunction.
Therefore, because of their ecological, structural and
floristic similarities Neotropical seasonally dry tropical
forests should be considered together in biogeographic
analyses although they occur in widely disjunct areas.
Chaco
The term Chaco has been redefined by Prado (1993b) and
here is applied to the vegetation of the plains of northern
Argentina, western Paraguay and south-eastern Bolivia, and the
extreme western edge of Mato Grosso do Sul state in Brazil
(Prado, 1993a). This vegetation extends over 800,000 km
2
,
in one of the few areas in the world where the transition
between the tropics and the temperate belt does not occur in
the form of a desert but rather as semiarid forests and wood-
lands (Morello, 1967). The Chaco is very flat throughout,
and its soils are derived from the massive accumulation of
fine loess and alluvial sediments during the Quaternary.
Stones of any size are completely absent, resulting in the
development of compact soils with impeded drainage. The
effects of past oceanic intrusions through the Chaco-Pampean
plains are clear, with a predominance of saline soils, some-
times with highly alkaline horizons.
The Chaco climate is distinguished by its strong seasonality,
with summer maxima of up to 49
°
C, the highest tempera-
tures recorded in South America, and severe winter frosts.
The rainfall declines from over 1000 mm/year in the east to
less than 500 mm/year in the west, with a dry season in the
winter and spring and a rainy season in the summer; the dry
season is generally negligible at the Chaco’s eastern edge, and
increases in duration from east to west. Thus, the vegetation
of the Chaco is subjected to low soil moisture and freezing
in the dry season and waterlogging and extremely high air
temperatures during part of the rainy season.
The forests of the Chaco are dominated by trees of the
genus
Schinopsis
, together with
Aspidosperma quebracho-
blanco
D.F.K. Schldt.,
Tabebuia nodosa
(Griseb.) Griseb.,
and several species of
Acacia
and
Bulnesia
. There is a dis-
continuous shrub layer, mainly consisting of prickly mimosoid
species and a generally sparse herbaceous layer of Brome-
liaceae and Cactaceae with a few grasses.
The floristic composition of the Chaco forests is quite dif-
ferent from that of the seasonally dry tropical forests (Prado,
1991; Prado & Gibbs, 1993), and their floristic links are to
the dry, temperate Monte and Andean Prepuna formations
(Cabrera, 1976). This reflects the regular frosts received by
the Chaco vegetation, which is thus essentially a subtropical
extension of a temperate formation. Therefore, despite the
Chaco forests having a strongly seasonal climate with a dry
period lasting for several months, they are excluded from our
definition of seasonally dry tropical forests both floristically
and ecologically. Links between the Chaco and seasonally
dry tropical forests have been postulated erroneously (e.g.
Hueck & Siebert, 1981; Gentry, 1995) and appear to be
based on similarities in their overall appearance rather than
an understanding of their floristic composition (Prado, 1991;
1993a, b). This extends to vernacular terms since the low,
seasonally dry tropical forests around Corumbá, Mato Grosso
do Sul, are known locally as ‘chaco’ (Ratter
et al.
, 1988).
Savannas
Cerrado
The cerrado biome (cerrado
s.l
.) covers some two million
square kilometres of central Brazil. It contains several vegeta-
tion types, but principally savanna woodland or ‘cerrado’
(cerrado
sensu restrictu
); in this paper cerrado refers to cerrado
s.r.
This grows on dystrophic, acid soils, with low calcium
and magnesium availability, and often with high levels of
aluminium (Furley & Ratter, 1988; Ratter
et al.
, 1997).
Cerrado soils are always well-drained and cerrado vegeta-
tion is intolerant of waterlogging. It varies from dense grass-
land with a sparse covering of shrubs and small trees, to an
almost closed woodland with a canopy height of 12–15 m
known as cerradão (Ratter
et al.
, 1997). These differences in
vegetation structure have been related to soil fertility gradients
(e.g. Goodland & Pollard, 1973), but other data fail to show
this correlation (e.g. Ribeiro, 1983). Fire is undoubtedly an
important factor throughout the biome (Ratter
et al.
, 1997),
Figure 5 Distribution of Pithecellobium diversifolium Bentham and
P. excelsum (Kunth) Bentham. Redrawn from Barneby & Grimes
(1997).
JBI397.fm Page 264 Thursday, July 27, 2000 5:51 PM

Neotropical seasonal forest biogeography 265
© Blackwell Science Ltd 2000,
Journal of Biogeography
,
27
, 261–273
and the woody flora of the cerrado shows adaptations such
as thick, corky bark, xylopodia and the ability to sprout from
lateral buds if the growing apex is killed by fire. The most
important families of woody plants are the Leguminosae,
Malphigiaceae, Myrtaceae, Melastomataceae and Rubiaceae;
less diverse, but very characteristic of the cerrado is the
Vochysiaceae. Cerrado is floristically distinct from the other
types of seasonal vegetation, and whilst cerradão on meso-
trophic soils may appear physiognomically rather similar to
seasonally dry forest, even there the floristic distinction is
maintained through a broad ecotone (Ratter, 1992).
Other vegetation types occur within the cerrado biome.
Gallery forests occur along the rivers, and contain many
species which also occur within the rain forests of the Amazon
and the Atlantic coast of Brazil (Oliveira-Filho & Ratter, 1995).
Wet ‘campos’, lacking trees except the fan palm
Mauritia
flexuosa
L.f., frequently occur between cerrados and gallery
forest, where there is extreme fluctuation of the water table.
Waterlogging in the wet season excludes woody cerrado spe-
cies, and in the dry season the soils are too dry for gallery
forest species. On areas of fertile soil, which are often asso-
ciated with calcareous rocks, seasonally dry tropical forest
occurs (Ratter
et al.
, 1978).
Disjunct, isolated areas of cerrado-like vegetation also
occur within the Brazilian Amazon rain forest (Eiten, 1972),
both in northern sites (e.g. in Amapá State and Roraima
State) and southern sites closer to the cerrados of Central
Brazil (e.g. Humaitá in Amazonas State) and even close to
the Amazon river itself at Alter do Chão (Pará State). These
occur on poor, sandy soils (Solbrig, 1993) and with the excep-
tion of Alter do Chão are depauperate in numbers of species
compared to the cerrados of Central Brazil (Sanaiotti, 1996).
The southern sites and Alter do Chão have floristic affinities
with central Brazil, whilst the northern sites show affinities
to the hydrologic savannas of the Llanos and the Rupununi
savannas (Ratter
et al.
, 1996; Sanaiotti, 1996).
Other savanna areas
The other large areas of savanna in South America are the
Venezuelan Llanos, and the adjoining savannas in Guyana
and Brazil. Floristically, these are species poor cerrado
(Lenthall
et al.
, 1999), although they have been classified
separately from this vegetation type (Eiten, 1972). They also
frequently differ ecologically since many are hydrologic
savannas, with extreme seasonal fluctuations in water table
levels. The principal Central American savannas, in Belize
and adjacent Guatemala and Mexico, are also hydrologic
savannas. They have a distinct floristic composition, and are
identified as a separate savanna phytogeographic zone in
multivariate analyses of woody savanna species (Lenthall
et al.
, 1999).
PALAEOCLIMATOLOGY
The contemporary, fragmented distribution of seasonally dry
tropical forest in the Neotropics must be considered in rela-
tion to the climatic fluctuations of the Quaternary. The
ice-ages are associated with cooling and drying of the global
climate (CLIMAP, 1976; Tricart, 1985; Colinvaux, 1989;
Thomas & Thorp, 1992; Clapperton, 1993), but the magnitude
and relative importance of the two processes in tropical South
America have long been a matter for debate. On the one
hand the proponents of ice-age drying (Haffer, 1969; Prance,
1982; Clapperton, 1993; Van der Hammen & Absy, 1994)
contend that reduction in precipitation relative to modern
conditions was the primary factor controlling the distribution
of plants during most of the Pleistocene; on the other hand
Colinvaux (1989, 1996) and Colinvaux
et al
. (1996a) believes
the low temperatures of the ice-age to be the most important
factor influencing Amazonian plant distributions. The
evidence for both drying and cooling comes from diverse
sources, but the record for the Neotropics is fragmentary
and it is unsafe to make predictions about areas as large as
the Amazon Basin on the strength of such a limited data set
(Irion
et al.
, 1995; Colinvaux, 1997; Simpson, 1997).
Fluctuations in precipitation
Haffer (1969) plotted current rainfall across the Amazon,
and postulated that with reduced precipitation falling in a
similar pattern, large parts of the region would experience
climates which could not support rain forest. He postulated
a series of rain forest refugia which would have been cont-
inuously forested, surrounded by areas in which rain forest
was present during times of high rainfall such as the present
interglacial, and replaced by open, nonforest vegetation dur-
ing the drier periods. Evidence for drying in the Neotropics
has come from palynology (see Van der Hammen & Absy,
1994; Hooghiemstra, 1997) and geomorphology (reviewed
by Clapperton, 1993).
Palynology and sedimentology
A core from Katira, in Rondônia, Brazil shows clear evid-
ence of a replacement of forest cover by open grassy vegeta-
tion (Van der Hammen, 1972; Absy
et al.
, 1991). The
drier period here has been dated to between 22,000 years
bp
and 11,000 years
bp
(Van der Hammen & Absy, 1994)
which thus coincided with the final phase of the last glaci-
ation. This site is located towards the southern edge of the
Amazonian rain forest, and its vegetation would be expected
to be sensitive to any reduction in precipitation. Carbon
isotope studies of soil organic matter from several sites in
Rondônia also indicate changes in the distributions of forest
and open vegetation (Pessenda
et al.
, 1998). A basin at Serra
dos Carajás in the north Brazilian plateau has yielded evide-
nce from both sedimentology and palynology of drier
periods during which rain forest was absent (Soubies
et al.
,
1991). Radiocarbon dates of 23,800 years
bp
and 10,600 years
bp
bracket the most recent dry period. This site is located at
the south-eastern end of the corridor of low rainfall which
crosses Amazonia, and thus forests there would also be pre-
dicted to be particularly sensitive to ice-age drying of the
climate. However, Colinvaux
et al
. (1996a) interpreted this
pollen data as indicating an increase in marsh plants due
merely to a local reduction of precipitation on an already
rather dry and ecotonal plateau.
JBI397.fm Page 265 Thursday, July 27, 2000 5:51 PM

266 R. T. Pennington, D. E. Prado and C. A. Pendry
© Blackwell Science Ltd 2000,
Journal of Biogeography
,
27
, 261–273
Outside of the Amazon basin, in southern and Central
Brazil, data from eleven pollen records from areas now
covered by cerrado,
Araucaria
forest, littoral vegetation and
seasonally dry tropical forest show clear evidence for drier
conditions between 10,000 and 7000
bp
either from gaps
in sediment deposition or by indications of taxa from
xerophytic vegetation (Ledru
et al.
, 1998).
Stone lines
Buried beds of stones and gravels have been found in many
soil profiles throughout Amazonia and taken as indications
of drier climates than exist at present (Ab’Sáber, 1982).
Most theories of the origin of these features require the
removal of finer material downslope by surface flow, allow-
ing coarser, stony material to accumulate, and these deposits
then being re-covered by sandy clays under a regime with less
erosion (Thomas & Thorp, 1992). In some cases stone lines
appear to indicate a period of dry climate but their interpreta-
tion is problematic given the uncertainties in the dynamics of
ancient vegetation types, the dynamics of landscapes under
these vegetation types, the nature of palaeoclimates and the
effects of rapid changes in these climates at the Pleistocene—
Holocene boundary (Thomas & Thorp, 1992). More recent
work (Irion
et al.
, 1995) suggested that the stone lines are the
result of
in situ
weathering within the soil profile and may
date from 10 million years ago or even earlier.
Dune systems
Relict dune systems are present in large areas of western
Venezuela and eastern Colombia (Clapperton, 1993), the
Chaco (Putzer, 1962), the São Francisco valley of the caatinga
region of Brazil (Tricart, 1985), the Beni Lowlands of north-
east Bolivia (Clapperton, 1993), and Roraima (Santos, 1992;
Santos
et al.,
1993). These dunes are no longer mobile, and
are indicative of a drier climate in former times, at least in
these areas. It is logical to assume that cool dry conditions
permitting dune mobility were most likely during Pleis-
tocene glacial conditions (Clapperton, 1993), but data from
the São Francisco valley indicate that such dry conditions
have existed during the late Tertiary and even the Holocene
(to 1200
bp
) with no particular concentration of activity
during glacial periods (Colinvaux, pers. comm.).
The Amazon fan
The offshore sediments deposited at the mouth of the Amazon,
known as the Amazon Fan, contain a complete pollen record
of the vegetation growing throughout the Amazon basin
since the formation of the Andes during the Miocene. Colin-
vaux (1997) has argued against ice age aridity because no
peak of grass pollen has been found in any core, though
Hooghiemstra (1997) suggested that this evidence should
be treated with caution. Firstly, the Amazon basin is a huge
area with a great range of vegetation supplying pollen to the
fan; montane forest, lowland rain forest, savanna, aquatic
and semiaquatic vegetation are all present within the basin,
and this will tend to blur any signal. Secondly, grass pollen
is known to travel short distances, and a well-developed
gallery forest could mask the presence of grassy vegetation.
Thirdly, sea level fluctuations have caused major changes in
the erosion patterns of the lower part of the Amazon, and
sediments will have been eroded and redeposited in the fan,
leading to still further mixing of the pollen types.
Temperature fluctuations
Acceptance of the cooling of the South American lowlands
has been slower than the acceptance of ice age aridity, and
indeed Haffer (1969) even suggested that the periods of aridity
might coincide with warming of the lowlands. The CLIMAP
(1976) estimates of sea surface temperatures indicated that
cooling at tropical latitudes was of the order of only about
2
°
C, but this conflicted with the evidence from glaciologists
(reviewed by Clapperton, 1993) and palynologists (Van der
Hammen, 1974; Hooghiemstra, 1995) that montane glaciers
and tree lines had descended during the ice ages. There has
been some attempt to reconcile the apparent cooling of the
uplands with the thermal stability of the lowlands by post-
ulating that ice-age vegetation zones were compressed due
to the effects of cold, catabatic winds descending from the
ice fields and steeper lapse rates in drier air (Colinvaux,
1996). However, data from studies of isotope ratios in corals
from Barbados (Guilderson
et al.
, 1994) and noble gases
in Brazilian fossil ground waters (Stute
et al.
, 1995) and
recent modelling of the global climate (Webb
et al.
, 1997),
suggest cooling of the Neotropical lowlands of about 5
°
C,
in contravention of the earlier estimates.
Pollen data from the western edge of Amazonia at the
base of the Andes and from south-eastern Brazil indicate a
descent of populations of montane species and the north-
ward migration of cool-adapted lowland species during the
last glacial cycle (reviewed in Colinvaux
et al.
, 1996a).
Podocarpus
and
Alnus
pollen found in the cores from Mera
(Liu & Colinvaux, 1985) and San Juan Bosco (Bush
et al.
,
1990) in Ecuador suggest that montane species descended up
to 1500 m. Palynological studies of sediment cores located on
a transect in south-eastern Brazil from Santa Catarina to Minas
Gerais (reviewed by Behling, 1998) show clear evidence for
cooling of 5–7
°
C in the LGM from their indications of sub-
tropical grasslands (campos) in highland regions up to 750 km
from where they occur today.
Araucaria
pollen is present in late
Pleistocene and early Holocene sediments from Salitre de Minas
in Minas Gerais, in an area where the modern vegetation is
cerrado, but today the nearest forests with
Araucaria
are
found 600 km to the south in São Paulo state (Ledru, 1993).
The core from Lake Pata (Colinvaux
et al.
, 1996b) in
north-west Amazonia shows a strong
Podocarpus
signal and
no increase in grass pollen during the glacial period. This
was taken to indicate invasion of the lowland rain forest by
montane elements due to cooling and maintenance of wet
ice-age conditions in lowland Amazonia. Whilst Pata is in an
area predicted to be ice-age savanna by Clapperton (1993),
it must be pointed out that it is within the permanently
forested refugium delimited by Van der Hammen & Absy
(1994).
There is now general agreement that during the coolest period
of the last Ice Age Amazonia was at least some 4
°
±
2
°
C cooler
JBI397.fm Page 266 Thursday, July 27, 2000 5:51 PM

Neotropical seasonal forest biogeography 267
© Blackwell Science Ltd 2000,
Journal of Biogeography
,
27
, 261–273
than it is at present, but there is less agreement about how much
drier it was, and how rainfall was distributed. It is widely
accepted that the climate of large areas of western Amazonia
was always wet enough to support some form of rain forest,
but the size of the continuously forested area is contested. There
is good evidence that at least some areas on the periphery of
Amazonia were considerably drier during the ice ages, but other
areas show signs of a wetter climate in the same period. It is
therefore impossible to generalize for the whole region from
the meagre data which is currently available and there is a
particular paucity of data from central and eastern Amazonia.
The climatic history of Central America is even more
fragmentary than that of South America, but the limited
evidence available makes it clear that it has also been subject
to equally large fluctuations in both temperature and precipita-
tion (Metcalfe & Davies, 1998; Whitlock, 1998).
THE BIOGEOGRAPHIC IMPORTANCE OF
SEASONALLY DRY TROPICAL FORESTS
Plant communities are temporary associations which are the
result of the interaction between the response of their constitu-
ent species to environmental processes and the historical
biogeography of those species (Huntly & Webb, 1989), and
contemporary associations must differ to some degree from
those which existed during the last Ice Age. It is generally
accepted that at that time the Neotropical environment was
both cooler and drier than at present, and the vegetation
which existed then would have reflected both of these factors.
Proponents of the Refuge Theory have claimed that
during the drier conditions of the Ice Ages the rain forests
of Amazonia were fragmented by the spread of vegetation
adapted to drier conditions (Haffer, 1969; Prance 1973), and as
a corollary that vicariance may have driven speciation in the
disjunct patches of rain forest which were continuously present
in areas of higher rainfall (Haffer, 1982). The locations of rain
forest refuges were inferred both from contemporary pat-
terns of endemism and high rainfall, and many have been
postulated to be on higher ground. Since the Refuge Theory
concerned the mechanisms of speciation of rain forest taxa,
the debate tended to focus on these forests and much less
attention was paid to the type of vegetation which was
assumed to have replaced them under a drier climate. Neo-
tropical Refuge Theory was heavily influenced by African
studies which showed that Ice Age climates were very much
drier than now, and savannas increased in extent, and for
many there has been an implicit acceptance that cerrado was
the dominant vegetation of Ice Age Amazonia. However, the
pollen data from the Amazon Fan do not lend any support
for the spread of such an open, grassy vegetation.
Colinvaux (1989, 1997) has long denied the importance
of Ice Age drying for Amazonian vegetation, and has instead
stressed cooling as the dominant environmental factor
affecting vegetation. The absence of grass pollen in the core
from Lake Pata (Colinvaux
et al.
, 1996b) support his con-
tention that closed forest existed throughout the last glacial
period, and the presence of
Podocarpus
indicates cooler condi-
tions, but less certain is his claim that ‘tropical rain forest
covered the inselberg and the surrounding lowlands through-
out the time spanned by the section’. The pollen profile is
equally consistent with a rather different type of vegetation.
Inventories of seasonally dry tropical forests (Ratter
et al.
,
1988; Prado, 1991; Oliveira-Filho & Ratter, 1995; Ratter,
unpublished) include species from thirty-two of the forty
genera listed in Colinvaux
et al
.’s (1996b) ‘Table 2’. Of
the four genera and one family listed as ‘strongly suggestive
of a tropical rain forest’ (p. 87)
Cedrela
species are gener-
ally typical of seasonal forest throughout the Neotropics
(Pennington, 1981; moreover
Cedrela
pollen is impossible to
distinguish from other genera of Meliaceae [T.D. Pennington,
pers. comm.]).
Clusia
is found in gallery forests in seasonally
dry forest areas (for example Bridgewater, Pennington
& Reynel (in prep.) recently collected a
Clusia
in gallery
forest surrounded by seasonally dry tropical forest in the
Huancabamba region of Peru) and is a common strangler of
cerradão.
Didymopanax
(=
Schefflera
) is a habitat generalist
widespread in gallery forests of dry regions of central Brazil
(Oliveira-Filho & Ratter, 1995) and Bombacaceae are often
abundant in seasonally dry tropical forest (Ratter
et al.
,
1988; Bridgewater
et al.
, in prep). Taxa which are character-
istic of rain forest are certainly represented in the Lake Pata
profile (Cryosophila, Guarea and Macrolobium), but they
may have been less widespread under the drier conditions
of the Ice Age, and instead growing in gallery forests, along
with the other species which could not tolerate the dry con-
ditions away from water courses. Meave & Kellman (1994)
showed that present-day savanna gallery forests in Belize
have a floristic composition similar to that of continuous
rain forests growing in wetter climates in the region. Sim-
ilarly, two species of Podocarpus which are found in upper
montane forests in eastern Brazil also grow in gallery forests
within the cerrado biome (Oliveira-Filho & Ratter, 1995).
We propose that the vegetation of some of lowland
Amazonia for at least part of the last Ice Age was neither
cerrado nor a type of cool-adapted rain forest. A third option
is suggested by the distribution patterns of seasonally dry
tropical forest species, and furthermore does not conflict with
any of the evidence which has already been presented in the
long-running debate about the Amazonian palaeoenviron-
ment. In our model a hitherto unrecognized type of vegeta-
tion in Amazonia was a form of closed forest which included
the more drought-tolerant rain forest elements, and many
species which are now restricted to seasonally dry tropical
forests. Like contemporary seasonal forests this formation
had a closed canopy, and the understorey would have con-
tained little grass. Gallery forests would have been an import-
ant feature and would have consisted of rain forest species
growing with montane taxa such as Podocarpus which had
a wider distribution under the cool Ice Age climate. At the
Pleistocene/Holocene boundary the onset of pluvial condi-
tions allowed the rapid spread of rain forest species out-
wards from the gallery forests, dividing the ranges of the
species now scattered in the isolated patches. Coexistence
of montane taxa (e.g. Podocarpus, Drimys) with those of
seasonally dry tropical forests (e.g. Tecoma, Zanthoxylum
and Zygophyllum) within part of the contemporary cerrado
JBI397.fm Page 267 Thursday, July 27, 2000 5:51 PM

268 R. T. Pennington, D. E. Prado and C. A. Pendry
© Blackwell Science Ltd 2000, Journal of Biogeography, 27, 261– 273
biome between 17,000 and 13,000 years bp was shown by
the study from Salitre in southern Brazil (Ledru, 1993).
Although there is considerable evidence of Ice Age moisture
reductions from geomorphology and pollen cores from
the periphery of the Amazon basin, there is little support for
the spread of nonforest, cerrado vegetation. This is intriguing
because much of Amazonia (≈ 75%) is covered by dystrophic
soils of the oxisol-ultisol group that also underlie the cerrado
(Furley, 1990). However, no tree species of the cerrado vegeta-
tion are found in the Amazonian tall forest except in isolated
areas of savanna vegetation within Amazonia and in tran-
sitional forest at the cerrado–forest boundary (Ratter &
Bridgewater, unpublished data). The extreme paucity of
species in the majority of isolated Amazonian cerrados (Ratter
et al., 1996; Sanaiotti, 1996) and the very marked floristic
differences between the savannas to the north and south of
Amazonia (Lenthall et al., 1999) also suggest that these areas
were never part of a more widely distributed Pleistocene
cerrado (Sanaiotti, 1996).
In contrast, some species characteristic of seasonally dry
tropical forests are now widely, though sparsely, distributed
in Amazonia, occurring at low frequencies in areas of Amazo-
nian rain forest. Examples are Commiphora leptophloeos (Mart.)
Gill., Aspidosperma pyrifolium Mart., Aspidosperma discolor
A. DC., Albizia inundata (Mart.) Barneby & Grimes, Couepia
uiti (Mart. & Zucc.) Benth. (Prado & Gibbs, 1993). A com-
posite map (Fig. 6) of fifty-seven taxa (listed in Table 1)
found in the Brazilian caatingas and other areas of season-
ally dry tropical forest clearly demonstrates penetration of
Amazonia by these elements of the seasonally dry tropical
forest flora. Presumably, these species are growing in areas
of Amazonia where the soils are more fertile, and indeed
important contemporary locations for seasonally dry forest
species are found in Rondônia and western Pará in Brazil
(Fig. 6) where the parent materials are unusually mineral-
rich (Furley, 1990; in press). Other contemporary locations
for seasonally dry forest species are found along the major
rivers, and this is also probably related to soil fertility. Suit-
able soils may have been present in some areas which are
currently under várzea, a type of seasonally flooded forest
inundated by white waters laden with mineral-rich sediments
from Andean erosion. During the Pleistocene glaciations sea
levels were up to 100 m lower than they are at present, and
consequently the Amazon and its tributaries excavated
deeper channels through the soft rocks of central Amazonia
to reach their equilibrium profile. This is evident at Obidos,
700 km from the mouth of the Amazon, where the bottom of
the river bed is 80 m below contemporary sea levels (Tricart,
1985). Erosional features such as these have now been
drowned by higher sea levels for many hundreds of km from
the sea (Tricart, 1985). At times of low sea levels the level of
the Amazon would also have been lower, and therefore there
would have been less flooding of the areas currently under
várzea, and a lowering of the water table, exposing the higher
areas for colonization by seasonally dry forest species. Várzea
areas account for almost 10% of Amazonian soils (Furley,
1990), and their dendritic distribution along river courses means
that they could act as important migration routes for season-
ally dry tropical forest species throughout Amazonia. It is
notable that many of the records of such species in Amazonia
(Fig. 6) are concentrated along major rivers. Investigation
of whether other records of these species are in areas of
fertile soil would reward future investigation. Such soils are
certainly available in the Guianas (e.g. Guyana; Pennington,
1993), where there is also an abundance of records (Fig. 6).
In epochs of less rainfall, it is also plausible that species of
seasonally dry tropical forest might extend their ranges in
areas of the cerrado biome (and other areas). Xeric condi-
tions might lead to an enrichment of the surface horizons of
soils by capillarity where mineral rich parent material is
close to the surface, and thus the areas with soils high in pH
and calcium would increase (Ratter et al., 1988). The wetter
climate of an interglacial would certainly cause greater
leaching and acidification producing the infertile soils which
are more suitable for savanna species and mesotrophic soils
(and the associated seasonally dry forest species) would only
be found in the small, and often tiny areas where they could
be continuously replenished by the weathering of base rich
rock. The contemporary distribution patterns of terrestrial
Cactaceae in areas of seasonally dry tropical forest separated
by areas of cerrado provide further evidence for more
widespread Pleistocene seasonally dry tropical forest within
the cerrado biome. These cacti are unable to tolerate the
regular fires of cerrado, and are thus unlikely to migrate
Figure 6 Composite map of the distribution of fifty-seven
seasonally dry forest species found in the Caatinga and in at least
one other area. See Table 1 for a list of the species.
JBI397.fm Page 268 Thursday, July 27, 2000 5:51 PM

Neotropical seasonal forest biogeography 269
© Blackwell Science Ltd 2000, Journal of Biogeography, 27, 261–273
Table 1 The nonendemic woody species
of the Brazilian Caatingas whose
distribution patterns have been used to
create the map in Fig. 6.
Family Species
Anacardiaceae Astronium concinnum Schott
Astronium fraxinifolium Schott
Astronium urundeuva (Fr. All.) Engl.
Cyrtocarpa velutinifolia (Cowan) Mitchell & Daly
Loxopterygium gardneri Engler
Loxopterygium grisebachii Hier. & Lor.
Loxopterygium huasango Spruce
Loxopterygium sagotti Hook. f.
Schinopsis brasiliensis Engler
Schinopsis peruviana Engler
Apocynaceae Aspidosperma cuspa (Kunth) Blake
Aspidosperma discolor A. DC.
Aspidosperma polyneuron Müll. Arg.
Aspidosperma pyrifolium Mart.
Aspidosperma riedelii Müll. Arg
Bignoniaceae Tabebuia aurea (Manso) Benth. & Hook.
Tabebuia impetiginosa (Mart.) Standley
Boraginaceae Cordia alliodora (R. & P.) Oken
Patagonula americana L.
Patagonula bahiensis Moricand
Burseraceae Commiphora leptophloeos (Mart.) Gill.
Capparidaceae Crateva tapia L.
Caricaceae Carica quercifolia (St.-Hil.) Hier.
Chrysobalanaceae Couepia uiti (Mart. & Zucc.) Benth.
Combretaceae Combretum leprosum Mart.
Convulvulaceae Ipomoea carnea Jacq. ssp. fistulosa (Mart. ex Choisy) Austin
Leguminosae (Caesalpinoideae) Hymenaea martiana Hayne
Hymenaea velutina Ducke
Peltophorum dubium (Spr.) Taub.
Poeppigia procera Presl.
Pterogyne nitens Tulasne
Senna spectabilis (DC.) Ir. & B.
Leguminosae (Mimosoideae) Albizia polyantha (Spr.f.) Lewis
Anadenanthera colubrina (Vell.) Bren.
Enterolobium contortisiliquum (Vell.) Mor.
Mimosa caesalpinifolia Benth.
Mimosa exalbescens Barneby
Piptadenia viridifolia (Kunth) Benth.
Leguminosae (Papilionoideae.) Amburana cearensis (Fr. All.) Smith
Geoffroea spinosa Jacq.
Machaerium acutifolium Vog.
Myroxylon balsamum (L.) Harms
Platypodium elegans Vog.
Phytolaccaceae Phytolacca dioica L.
Polygonaceae Ruprechtia laxiflora Meissn.
Rhamnaceae Zizyphus joazeiro Mart.
Rubiaceae Alseis floribunda Schott.
Coutarea hexandra (Jacq.) Schum.
Rutaceae Balfourodendron riedelianum (Engl.) Engl.
Sapotaceae Pouteria gardneriana (DC.) Radlk.
Solanaceae Brunfelsia uniflora (Pohl) D. Don
Solanum granuloso-leprosum Dunal
Sterculiaceae Sterculia striata St.-Hil. & Naud.
Ulmaceae Celtis pubescens (Kunth) Spr.
Phyllostylon brasiliensis Capanema
Phyllostylon orthopterum Hallier
Phyllostylon rhamnoides (Poiss.) Taub.
JBI397.fm Page 269 Thursday, July 27, 2000 5:51 PM

270 R. T. Pennington, D. E. Prado and C. A. Pendry
© Blackwell Science Ltd 2000, Journal of Biogeography, 27, 261– 273
across wide expanses of this vegetation. Only three species
are found in cerrado, two Discocactus and a Cereus, and
all possess unusual morphological adaptations which allow
them to survive fires (N.P. Taylor, pers. comm.).
These ideas may not however, be applicable to the whole
cerrado biome. Pollen data from the edge of the cerrado
biome at Lago do Pires, Minas Gerais, in south-eastern
Brazil, shows semideciduous forest to have replaced cerrado
within the last millennium (Behling, 1998). It is thus possible
that the rather short disjunction between the caatingas and
Misiones nuclei of seasonally dry tropical forests (Fig. 1)
was formerly much larger. More floristic, soil and palaeo-
ecological studies in such areas of vegetation transition in
south-eastern Brazil are clearly necessary.
Climatic cooling would also encourage the spread of spe-
cies of seasonally dry tropical forests into the Amazon Basin
just as it would promote the spread of montane taxa. Some
areas of seasonally dry forests currently experience a climate
far cooler than that of the Amazon Basin—for example, they
occur south of the Tropic of Capricorn in Argentina, and North
of the Tropic of Cancer in Mexico, and in the Ecuadorian
Andes seasonally dry tropical forest grows at ≈ 1000 m.
The hypothesis that a form of seasonally dry tropical forest
adapted to the cool conditions existed in the Neotropics
during parts of the Pleistocene is attractive because it fits all
the evidence for both cooling and drying which has been
presented in the debate on Neotropical palaeoclimatology, the
pollen record and the biogeographic analysis presented here.
A PHYLOGENETIC APPROACH TO TESTING
THE HYPOTHESIS OF A WIDER EXPANSE
OF SEASONALLY DRY TROPICAL FOREST
IN THE PLEISTOCENE
The biogeographic data from plant distributions presented
in this paper and by Prado & Gibbs (1993) are distribution
maps of widespread species that occur in more than one of
the isolated areas of Neotropical seasonally dry forests. In
contrast, species of other genera show high levels of endemicity
in the different areas of these forests (Table 2). Distribution
maps of two such examples, Loxopterygium (Anacardiaceae)
and Pereskia (Cactaceae) are presented in Fig. 7 and Fig. 8.
The hypothesis of vicariance of seasonally dry Neotropical
forests provides a tempting explanation for the origin of
these endemic species—by allopatric speciation caused by
the fragmentation of the wider historical area of this forest.
A cladistic biogeographic approach would provide a test
of this hypothesis of speciation driven by vicariance and thus
a test of the hypothesis of a wider area of seasonally dry
tropical forest during the cool-dry phases of the Pleistocene.
Morphological and molecular characters, analysed in a
cladistic framework, can be used to infer phylogenies of the
species in the genera with endemic species in the disjunct
areas of seasonally dry tropical forests. For each genus,
determining the relationships of its component species
enables inferences to be made of the historical relationships
between the areas of seasonally dry tropical forest in which
Genus Family No. of spp. Areas present
Astronium s.l. Anacardiaceae 12 1,2,3,4,8,9
Cyrtocarpa Anacardiaceae 4 1,8,9
Loxopterygium Anacardiaceae 5 1,4,6
Schinopsis Anacardiaceae 7 1,3,4,6
Pereskia Cactaceae 16 1,2,3,4,5,6,7,8,9,10
Isocarpha Compositae 5 1,5,6,9
Caesalpinia (Poincianella-
Erythrostemon group)
Leguminosae 56 1,2,3,4,5,9,10
Coursetia Leguminosae 38 1,4,5,8,10
Chloroleucon Leguminosae 10 1,2,3,4,5,6,7,8,9,10
Pithecellobium Leguminosae 13 1,6,7,8,9,10
Gossypium Malvaceae ≈ 19 1,5,9
Ruprechtia Polygonaceae 17 1,2,3,4,5,6,8,9
Basistemon Scrophulariaceae 8 4,5,8
Table 2 List of the genera which contain
endemic species disjunctly distributed
among different areas of seasonally dry
tropical forest. The numbers of the areas
correspond to the numbers in Fig. 1.
Figure 7 Distribution of Loxopterygium. 1, L. gardnerii Engl.;
2, L. grisebachii Hieron. & Lorentz; 3, L. huasango Spruce;
4, L. sagotti Hook. f. Redrawn from Prado & Gibbs, 1993.
JBI397.fm Page 270 Thursday, July 27, 2000 5:51 PM

Neotropical seasonal forest biogeography 271
© Blackwell Science Ltd 2000, Journal of Biogeography, 27, 261–273
they occur. If the patterns suggested by different, remotely
related genera are similar, the implication is that the same
process must have driven speciation in each genus. This
common process would be the vicariance of a wider expanse
of seasonally dry tropical forest.
This type of phylogenetic approach to assessing hypotheses
about biological diversification in Neotropical rain forests
was advocated by Bates et al. (1998). Because the areas of
endemism are more easily defined for seasonally dry Neo-
tropical forests (e.g. see Fig. 1), we believe that this approach
might be more successfully applied to this vegetation type.
Lavin (1998) confirmed this by converting phylogenies of
two sections of the genus Coursetia DC. (Leguminosae) into
a single, fully resolved, area cladogram for South American
areas of seasonally dry tropical forests. He concluded that:
‘…the high degree of endemicity of the South American species
of Coursetia, and the high degree of resolution in the taxon-
derived cladogram strongly suggests that the biogeography
of these species was strongly influenced by Quaternary frag-
mentation of the South American dry forests’ (Lavin, 1998
p. 141). Whether this represents a general area cladogram
for the areas (listed in Table 2) in which Coursetia occurs can
only be tested by its congruence with other taxon-derived
area cladograms. We intend to generate phylogenies for some
of the genera listed in Table 2 in order to further explore this
cladistic biogeographic approach to testing the hypothesis of
vicariance of Neotropical seasonally dry forests.
ACKNOWLEDGMENTS
This project is supported by a Leverhulme Trust grant to the
Royal Botanic Gardens Edinburgh. Darién Prado would like
to thank the British Council (Buenos Aires), who provided
financial support to enable him to visit Edinburgh. We thank
Sam Bridgewater, Paul Colinvaux, Peter Furley, Colin Hughes,
Matt Lavin, Gwil Lewis, Jim Ratter, Nigel Taylor and an
anonymous referee for their advice and comments.
REFERENCES
Ab’Sáber, A. N. (1982) The paleoclimate and paleoecology of
Brazilian Amazonia. Biological diversification in the tropics (ed.
by G. Prance), pp. 41–59. Columbia University Press, New York.
Absy, M. L., Cleef, A. F., Fournier, M., Martin, L., Servant, M.,
Sfeddine, A., Ferreira da Silva, M. F., Soubies, F., Suguio, K.,
Turcq, B. & Van der Hammen, T. (1991) Mise en évidence de
quatre phases d’ouverture de la forêt dense de sud-est de
l’Amazonie au cours des 60.000 dernières anneés. Comptes
Rendus Academie des Sciences, Paris, Series II, 312, 673– 678.
Barneby, R. C. (1991) Sensitivae Censitae. A description of the
genus Mimosa Linnaeus (Mimosaceae) in the New World.
Memoirs of the New York Botanical Gardens, 65, 1–835.
Barneby, R. C. & Grimes, J. W. (1996) Silk Tree, Guanacaste, Mon-
key’s Earring. A generic system for the synandrous Mimosaceae
of the Americas. Part I. Abarema, Albizia and Allies. 74(1), 1–292.
Barneby, R. C. & Grimes, J. W. (1997) Silk Tree, Guanacaste, Mon-
key’s Earring. A generic system for the synandrous Mimosaceae
of the Americas. Part II. Pithecellobium, Cojoba and Zygia.
Memoirs of the New York Botanical Gardens, 74(2), 1–149.
Bates, J. M., Hackett, S. J. & Cracraft, J. (1998) Area-relationships in
the Neotropical lowlands: an hypothesis based on raw distribu-
tions of Passerine birds. Journal of Biogeography, 25, 783–793.
Behling, H. (1998) Late Quaternary vegetational and climatic
changes in Brazil. Review of Palaeobotany and Palynology,
99, 143–156.
Bullock, S. H. (1995) Plant reproduction in Neotropical dry
forests. Seasonally dry tropical forests (ed. by S. H. Bullock,
H. A. Mooney and E. Medina), pp. 277–303. Cambridge
University Press, Cambridge.
Bush, M. B., Colinvaux, P. A., Weinmann, M. C., Piperno, D. R.
& Liu, K.-B. (1990) Late Pleistocene temperature depression
and vegetation change in Ecuadorian Amazon. Quaternary
Research, 34, 330–345.
Cabrera, A. L. (1976) Regiones Fitogeográficas Argentinas, 2nd
edn. Enciclopedia Argentina de Agricultura y Jardinería.
ACME S.A.C.I., Buenos Aires.
Cabrera, A. L. & Willink, A. (1980) Biogeografía de América
Latina 2nd Ed. Serie de Biología. Secretaría General de la
Oraganización de los Estados Americanos, Washington D.C.
Clapperton, C. M. (1993) Nature of environmental changes in
South America at the Last Glacial Maximum. Palaeogeography
Palaeoclimatology, Palaeoecology, 101, 189–208.
CLIMAP. (1976) The surface of the Ice-Age earth. Science, 191,
1131–1138.
Colinvaux, P. A. (1989) Ice-Age Amazon revisited. Nature, 340,
188–189.
Colinvaux, P. A. (1996) Quaternary environmental history and
forest diversity in the Neotropics. Evolution and environment
Figure 8 Distribution of Pereskia. The dashed line indicates the
limits of P. aculeata Miller, found in the Caribbean region and
eastern South America. The other species are distributed as follows:
1, P. bahiaensis Gürke, P. aureifolia Ritter, P. stenantha Ritter;
2, P. grandifolia Haworth; 3, P. nemorosa Rojas Acosta;
4, P. sacharosa Griseb.; 5, P. diaz-romeroana Cárdenas;
6, P. weberiana K. Schumann; 7, P. humboldtii Britton & Rose;
8, P. bleo (Kunth) DC.; 9, P. guamacho F.A.C. Weber;
10, P. lychnidiflora DC.; 11, P. zinniiflora DC.;
12, P. portulacifolia (L.) DC., P. quisqueyana Liogier.
Redrawn from Leuenberger (1986).
JBI397.fm Page 271 Thursday, July 27, 2000 5:51 PM

272 R. T. Pennington, D. E. Prado and C. A. Pendry
© Blackwell Science Ltd 2000, Journal of Biogeography, 27, 261– 273
in tropical America (ed. by J. B. C. Jackson, A. F. Budd and
A. G. Coates), pp. 359–405. University of Chicago Press, Chicago.
Colinvaux, P. A. (1997) The Ice-Age Amazon and the problem
of diversity, pp. 7–30. NWO/Huygenslezing 1997, The
Hague, NRC Handelsbad, Rotterdam.
Colinvaux, P. A., Liu, K.-B., Oliveira, P., Bush, M. B., Miller, M. C.
& Steinitz Kannan, M. (1996a) Temperature depression in the
lowland tropics in glacial times. Climatic Change, 32, 19–33.
Colinvaux, P. A., de Oliveira, P. E., Moreno, J. E., Miller, M. C. &
Bush, M. B. (1996b) A long pollen record from lowland
Amazonia: forest and cooling in glacial times. Science, 274, 85– 87.
Eiten, G. (1972) The cerrado vegetation of Brazil. Botanical
Review, 38, 201–341.
Furley, P. A. (1990) The nature and sustainability of Brazilian
Amazon soils. The future of Amazonia (ed. by D. Goodman
A. Hall), pp. 309–359. Macmillan, London.
Furley, P. A. (in press) The fragility and resilience of Amazonian
soils. Human impacts on Amazonia: the role of traditional
ecological knowledge in conservation and development (ed.
by D. A. Posey). Colombia University Press, New York.
Furley, P. A. & Ratter, J. A. (1988) Soil resources and plant
communities of the central Brazilian cerrado and their
development. Journal of Biogeography, 15, 97–108.
Furley, P. A., Proctor, J. & Ratter, J. A. (eds) (1992) Nature and
dynamics of forest-savanna boundaries. Chapman & Hall, London.
Gentry, A. H. (1995) Diversity and floristic composition of
neotropical dry forests. Seasonally dry tropical forests (ed. by
S. H. Bullock, H. A. Mooney and E. Medina), pp. 146–194.
Cambridge University Press, Cambridge.
Graham, A. & Dilcher, D. (1995) The Cenozoic record of tropical dry
forest in northern Latin America and the southern United States. Sea-
sonally dry tropical forests (ed. by S. H. Bullock, H. A. Mooney and
E. Medina), pp. 124–145. Cambridge University Press, Cambridge.
Goodland, R. J. & Pollard, R. (1973) The Brazilian cerrado vege-
tation: a fertility gradient. Journal of Ecology, 61, 219– 224.
Guilderson, T. P., Fairbanks, R. G. & Rubenstone, J. L. (1994)
Tropical temperature variations since 20,000 years ago: modul-
ating interhemispheric climate change. Science, 263, 663–665.
Haffer, J. (1969) Speciation in Amazonian forest birds. Science,
165, 131–137.
Haffer, J. (1982) General aspects of the Refuge Theory. Biological
diversification in the tropics (ed. by G. T. Prance), pp. 6–24.
Colombia University Press, New York.
Hooghiemstra, H. (1995) Environmental and palaeoclimatic
evolution in Late Pliocene-Quaternary Colombia. Palaeocli-
mate and evolution, with emphasis on human origins (ed. by
E. S. Vrba, G. H. Denton, T. C. Partridge and L. H. Burckle),
pp. 249–261. Yale University Press, New Haven.
Hooghiemstra, H. (1997) Tropical rain forest versus savanna;
two sides of a precious medal, pp. 32–43. NWO/Huygenslez-
ing (1997), The Hague, NRC Handelsbad, Rotterdam.
Hueck, K. & Siebert, P. (1981) Vegetationskarte Von Südamerika,
2nd edn. Fischer-Verlag, Stuttgart.
Hughes, C. E. (1998) Monograph of Leucaena (Leguminosae-
Mimosoideae) Systematic Botany Monographs, 55, 1–244.
Huntly, B. & Webb, T. III (1989) Migration: species’ response
to climatic variations caused by changes in the earth’s orbit.
Journal of Biogeography, 16, 5–20.
Irion, G., Müller, J., Nunes de Mello, J., Junk, W. J. (1995) Quater-
nary geology of the Amazonian lowland. Geo-Marine Letters
15, 172–178.
Janzen, D. H. (1988) Tropical dry forests. The most endangered
major tropical ecosystem. Biodiversity (ed. by E. O. Wilson),
pp. 130–137. National Academy Press, Washington D.C.
Lavin, M. (1998) Origin, diversity and biogeography of Neo-
tropical Fabaceae. Monographs in systematic botany from the
Missouri Botanical Garden, 68, 131–145.
Ledru, M.-P. (1993) Late Quaternary environmental and climatic
changes in central Brazil. Quaternary Research, 39, 90–98.
Ledru, M.-P., Salgado-Labouriau, M. L. & Lorscheitter, M. L.
(1998) Vegetation dynamics in southern and central Brazil
during the last 10,000 yr B.P. Reviews of Palaeobotany and
Palynology, 99, 131–142.
Lenthall, J. C., Bridgewater, S., Furley, P. A. & Ratter, J. A. (1999)
A phytogeographical analysis of the woody elements of the New
World savannas. Edinburgh Journal of Botany, 56, 293 – 305.
Leuenberger, B. E. (1986) Pereskia (Cactaceae). Memoirs of the
New York Botanical Garden, 41, 1–141.
Lewis, G. P. (1991) Five new taxa of Piptadenia (Leguminosae:
Mimosoideae) from Brazil. Kew Bulletin, 46, 159–168.
Lewis, G. P. (1998) Caesalpinia. A revision of the Poincianella-
Erythrostemon group. Royal Botanic Gardens, Kew.
Liu, K.-B. & Colinvaux, P. A. (1985) Forest changes in the Amazon
during the Last Glacial Maximum. Nature, 318, 556– 557.
Meave, J. & Kellman, M. (1994) Maintenance of rain forest
diversity in riparian forests of tropical savannas: implications
for species conservation during Pleistocene drought. Journal
of Biogeography, 21, 121–135.
Metcalfe, S. & Davies, D. (1998) Lacustrine records of Mid-Holocene
climates from Mexico, Central America and the Caribbean. Con-
ference Abstracts; PEP 1. Pole-Equator-Pole, Merida, Venezuela.
Mooney, H. A., Bullock, S. H. & Medina, E. (1995) Introduction.
Seasonally dry tropical forests (ed. by S. H. Bullock, H. A. Mooney
and E. Medina), pp. 1– 8. Cambridge University Press, Cambridge.
Morello, J. (1967) Bases para el estudio fitoecologicó de los grandes
espacios (el Chaco argentino). Cienca e Investigacion, 23, 252–267.
Murphy, P. & Lugo, A. E. (1986) Ecology of tropical dry forest.
Annual Review of Ecology and Systematics, 17, 67–88.
Murphy, P. & Lugo, A. E. (1995) Dry forests of Central Amer-
ica and the Caribbean. Seasonally dry tropical forests (ed. by
S. H. Bullock, H. A. Mooney and E. Medina), pp. 146–194.
Cambridge University Press, Cambridge.
Oliveira-Filho, A. T. & Ratter, J. A. (1995) A study of the origin
of central Brazilian forests by the analysis of the plant species
distributions. Edinburgh Journal of Botany, 52, 141–194.
Pennington, R. T. (1993) The Commonwealth and Government
of Guyana rain forest programme, phase I, site resource survey,
botanical studies. Natural Resources Institute, Chatham, UK.
Pennington, T. D. (1981) A monograph of Neotropical Meliaceae.
Flora Neotropica, 28, 1–470.
Pessenda, L. C. R., Gomes, B. M., Aravena, R., Ribeiro, A. S.,
Boulet, R. & Gouveia, S. E. M. (1998) The carbon isotope
record in soils along a forest-cerrado ecosystem transect: implica-
tions for vegetation changes in the Rondônia state, south-
western Brazilian Amazon region. Holocene, 8, 599–603.
Prado, D. E. (1991) A critical evaluation of the floristic links
between chaco and caatingas vegetation in South America.
PhD Thesis. University of St. Andrews, Scotland.
Prado, D. E. (1993a) What is the Gran Chaco vegetation in South
America?. I. A review. Contribution to the study of flora and
vegetation of the Chaco. V. Candollea, 48, 145–172.
Prado, D. E. (1993b) What is the Gran Chaco vegetation in South
JBI397.fm Page 272 Thursday, July 27, 2000 5:51 PM

Neotropical seasonal forest biogeography 273
© Blackwell Science Ltd 2000, Journal of Biogeography, 27, 261–273
America?. II. A redefinition. Contribution to the study of flora
and vegetation of the Chaco. VII. Candollea, 48, 615–629.
Prado, D. E. (in press) Seasonally dry forests of tropical South
America: from forgotten ecosystems to a new phytogeographic
unit. Edinburgh Journal of Botany, in press.
Prado, D. E. & Gibbs, P. E. (1993) Patterns of species distribu-
tions in the dry seasonal forests of South America. Annals of
the Missouri Botanical Garden, 80, 902– 927.
Prance, G. T. (1973) Phytogeographic support for the theory of
Pleistocene forest refuges in the Amazon Basin, based upon
evidence from distribution pattern in Caryocaraceae, Chryso-
balanaceae, Dichapetalaceae and Lecythidaceae. Acta Amazon,
3, 5–28.
Prance, G. T. (1982) Forest refuges: evidence from woody
angiosperms. Biological diversification in the tropics (ed. by
G. T. Prance), pp. 41–59. Columbia University Press, New York.
Putzer, H. (1962) Geologie von Paraguay. Beiträge Zur Regionalen
Geologie der Erde (ed. by A. Bentz). Gebrüder Borntraeger, Berlin.
Ratter, J. A., Askew, G. P., Montgomery, R. & Gifford, D. R.
(1978) Observations on forests of some mesotrophic soils in
central Brazil. Revista Brasileira de Botanica, 1, 47–58.
Ratter, J. A., Pott, A., Pott, V. J. & da Cunha, C. N. & Hari-
dasan, M. (1988) Observations on the woody vegetation
types in the Pantanal and at Corumbá, Brazil. Notes of the
Royal Botanical Garden Edinburgh, 45, 503–525.
Ratter, J. A. (1992) Transitions between cerrado and forest veg-
etation in Brazil. Nature and dynamics of forest-savanna
boundaries (ed. by P. A. Furley, J. Proctor and J. A. Ratter),
pp. 417–430. Chapman & Hall. London.
Ratter, J. A., Bridgewater, S., Atkinson, R. & Ribeiro, J. F.
(1996) Analysis of the floristic composition of the Brazilian
cerrado vegetation II: comparison of the woody vegetation of
98 areas. Edinburgh Journal of Botany, 53, 153–180.
Ratter, J. A., Ribeiro, J. F. & Bridgewater, S. (1997) The Brazilian
cerrado vegetation and threats to its biodiversity. Annals of
Botany, 80, 223–230.
Ribeiro, J. F. (1983) Comparação da concentração de nutrientes
na vegetação arbórea de um cerrado e um cerradão no Distrito
Federal, Brasil. MSc Thesis, University of Brasilia, Brazil.
Sanaiotti, T. M. (1996) The woody flora and soils of seven Brazil-
ian Amazonian dry savanna areas. PhD Thesis, University of
Stirling, Scotland.
Santos, J. O. S. (1992) O pantanal setentrional e os campos de dunas
da Amazonia ocidental. Simposio Internacional do Quaternario
de Amazonia. FUA/IQUA/Unesco, Resumos, Manaus,
Santos, J. O. S., Nelson, B. W. & Giovannini, C. A. (1993) Corpos
de areia sob leitos abandonados de grandes rios. Ciência Hoje,
93, Suplemento Technologia, Paleoclimas da Amazônia, 22 –l25.
Sarmiento, G. (1992) A conceptual model relating environmental
factors and vegetation formations in the lowlands of tropical
South America. Nature and dynamics of forest-savanna bounda-
ries (ed. by P. A. Furley, J. Proctor and J. A. Ratter), pp. 583 –
601. Chapman & Hall, London.
Simpson, B. B. (1997) Reply from B. B. Simpson. Trends in Ecology
and Evolution, 12, 319.
Solbrig, O. T. (1993) Ecological constraints to savanna land use.
The world’s savannas. Economic driving forces, ecological
constraints and policy options for sustainable land use (ed. by
M. D. Young and O. T. Solbrig), pp. 21– 47. MAB series 12.
Parthenon Publications, London.
Soubies, F., Suguio, K., Martin, L., Leprun, J., Servant, M., Turcq, B.,
Fournier, M., Delaune, M. & Sfeddine, A. (1991) The Quater-
nary deposits of the Serra dos Carajas (State of Pará, Brazil) –
ages and other preliminary results. Proceedings Global Changes
in South America during the Quaternary. Boletim Instituto de
Geosciencias—Universidade de São Paulo. Publicacao Especial,
8, 223–243.
Stute, M., Forster, M., Frischkorn, H., Serejo, A., Clark, J. F.,
Schlosser, P., Broeker, W. S. & Bonani, G. (1995) Cooling of
tropical Brazil (5 oC) during the Last Glacial Maximum. Science,
263, 663–665.
Thomas, M. F. & Thorp, M. B. (1992) Landscape dynamics and
surface deposits arising from Late Quaternary fluctuations in
the forest–savanna boundary. Nature and dynamics of forest-
savanna boundaries (ed. by P. A. Furley, J. Proctor and J. A.
Ratter), pp. 215–253. Chapman & Hall, London.
Tricart, J. (1985) Evidence of Upper Pleistocene dry climates in
northern South America. Environmental change and tropical
geomorphology (ed. by I. Douglas and T. Spencer), pp. 197–
218. Allen & Unwin, London.
Van der Hammen, T. (1972) Changes in vegetation and climate
in the Amazon basin and surrounding areas during the Pleis-
tocene. Geologie en Mijnbouw, 51, 641–643.
Van der Hammen, T. (1974) The Pleistocene changes of vegetation
and climate in tropical South America. Journal of Biogeography,
1, 3–26.
Van der Hammen, T. & Absy, M. L. (1994) Amazonia during the
last glacial. Palaeogeography Palaeoclimatology and Palaeoecol-
ogy, 109, 247–261.
Webb, R. S., Rind, D. H., Lehmann, S. J., Healy, R. J. &
Sigman, D. (1997) Influence of ocean heat transport on the
climate of the Last Glacial Maximum. Nature, 385, 695–699.
Whitlock, C. (1998) Features of the full-glacial vegetation of the
Americas. Conference Abstracts; PEP 1. Pole-Equator-Pole,
Merida, Venezuela.
BIOSKETCHES
Toby Pennington and Colin Pendry are Senior Scientific
Officers in the Tropical Biology Group at the Royal
Botanic Garden Edinburgh. Toby Pennington’s research
focuses on the taxonomy of Neotropical Leguminosae,
with particular emphasis on phylogenetic reconstruction,
molecular systematics and biogeography. He has carried
out fieldwork in seven Neotropical countries.
Colin Pendry’s research in ecology and taxonomy has
included studies of nutrient cycling and altitudinal
zonation in rain forests and floristic accounts of SE Asian
flowering plants, particularly the Polygalaceae.
Darién Prado is Head Professor of Botany at the National
University of Rosario and affiliated to CONICET,
Argentina. His principal research interest is biogeography,
though he has published on phytosociology, Capparaceae
taxonomy, and plant reproductive biology. He has worked
on the dry forest vegetation of central South America for
the last 20 years, studying the distribution patterns of its
woody species from both herbarium material and extensive
fieldwork throughout the Neotropics.
JBI397.fm Page 273 Thursday, July 27, 2000 5:51 PM