Content uploaded by Christopher W. Dick
Author content
All content in this area was uploaded by Christopher W. Dick
Content may be subject to copyright.
doi: 10.1098/rstb.2004.1532
, 1611-1622359 2004 Phil. Trans. R. Soc. Lond. B
R. Toby Pennington and Christopher W. Dick
rainforest tree flora
The role of immigrants in the assembly of the South American
References
http://rstb.royalsocietypublishing.org/content/359/1450/1611#related-urls
Article cited in:
Email alerting service
hereright-hand corner of the article or click
Receive free email alerts when new articles cite this article - sign up in the box at the top
http://rstb.royalsocietypublishing.org/subscriptions go to: Phil. Trans. R. Soc. Lond. BTo subscribe to
This journal is © 2004 The Royal Society
on May 22, 2011rstb.royalsocietypublishing.orgDownloaded from
Published online 7 October 2004
The role of immigrants in the assembly of the South
American rainforest tree flora
R. Toby Pennington
1
and Christopher W. Dick
2
1
Royal Botanic Garden Edinburgh, 20a Inverleith Row, Edinburgh EH3 5LR, UK
2
Smithsonian Tropical Research Institute, Unit 0948 APO, AA 34002, USA
The Amazon lowland rainforest flora is conventionally viewed as comprising lineages that evolved in bio-
geographic isolation after the split of west Gondwana (ca. 100 Myr ago). Recent molecular phylogenies,
however, identify immigrant lineages that arrived in South America during its period of oceanic isolation
(ca. 100–3 Myr ago). Long-distance sweepstakes dispersal across oceans played an important and poss-
ibly predominant role. Stepping-stone migration from Africa and North America through hypothesized
Late Cretaceous and Tertiary island chains may have facilitated immigration. An analysis of inventory
plot data suggests that immigrant lineages comprise ca. 20% of both the species and individuals of an
Amazon tree community in Ecuador. This is more than an order of magnitude higher than previous esti-
mates. We also present data on the community-level similarity between South American and palaeo-
tropical rainforests, and suggest that most taxonomic similarity derives from trans-oceanic dispersal,
rather than a shared Gondwanan history.
Keywords: Amazon rainforests; molecular systematics; biotic interchange; long-distance dispersal
1. INTRODUCTION
The neotropics is species rich, both at the local and land-
scape scale. It contains an estimated 90 000 plant species,
more than any other continental area (Thomas 1999), and
the rainforests of the Amazon basin have the world’s high-
est alpha-diversity as measured by the number of species of
trees in a single hectare (1 hectare ðhaÞ¼10
4
m
2
)
(Valencia et al. 1994; Romoleroux et al. 1997; De Oliveira
& Mori 1999). The prevailing biogeographic model views
the lowland South American flora as a product of Gondwa-
nan vicariance, and assumes that the contribution of sub-
sequent immigrant lineages to South America is neglible
(Raven & Axelrod 1974; Gentry 1982). This idea has
recently been challenged through recognition of the impor-
tance of interplate dispersal routes for tropical plants
(Morley 2003) and fossil calibration of molecular phylo-
genies of major rainforest taxa. Our principal aim is to
examine the contribution to the contemporary Amazon
tree flora of lineages that have arrived in South America
since its split from Africa (ca. 100 Myr ago).
We first review the possible migration routes by which
plants may have arrived in South America since its split
from Africa. We then evaluate phylogenetic evidence indi-
cating that some tree groups assumed to be indigenous to
South America are in fact, immigrants from other con-
tinents. We examine two related ideas promoted by Alwyn
Gentry. First, that the contribution of immigrant taxa to
South American rainforest is negligible (Gentry 1982).
Second, that most floristic similarities between rainforests
in Africa and the neotropics derive from Gondwana vicar-
iance (Gentry 1993). Based upon evidence from molecular
phylogenies, we analyse the contribution of ‘immigrant
taxa’, in effect those that entered South America after its
split from western Gondwana, to an Amazon rainforest tree
community in Ecuador. Then we analyse the community
level floristic similarities between two neotropical forest
inventory plots (Yasunı
´
, Ecuador, and the BCI, Panama)
with a tropical forest plot in Cameroon, in light of the
immigration models reviewed here.
(a) Splendid isolation?
The conventional model of South American biotic his-
tory stems from palaeontological studies of mammals, and
is encapsulated in the evocative title of Gaylord Simpson’s
(1980) book about South American mammalian evolution,
‘Splendid isolation’. This model regards South America as a
biogeographically isolated landmass after its split from
Africa (ca. 100 Myr ago; Goldblatt 1993; Pitman et al.
1993) until the closure of the Isthmus of Panama (ca.3
Myr ago; Coates & Obando 1996), and thus explains the
evolution of many South American endemic mammal fam-
ilies. The island continent model is central to studies of
plant biogeography and features in Raven & Axelrod’s
(1974) seminal paper on angiosperm biogeography:
‘clearly, the history of South American biota has been one
of evolution in isolation of an initial West Gondwanaland
stock shared with Africa’ (p. 549). Despite refinements of
plate tectonic theory over the past 30 years, the model of in
situ South American floristic evolution remains the prevail-
ing view (e.g. Burnham & Graham 1999; Young et al.
2002).
Author for correspondence (t.pennington@rbge.org.uk).
One contribution of 16 to a Discussion Meeting Issue ‘Plant phylogeny
and the origin of major biomes’.
Phil. Trans. R. Soc. Lond. B (2004) 359, 1611–1622 1611
#
2004 The Royal Society
doi:10.1098/rstb.2004.1532
on May 22, 2011rstb.royalsocietypublishing.orgDownloaded from
Strictly speaking, the model is geologically accurate. By
85 Myr ago, it was estimated that an 800 km wide seaway
isolated Africa and South America (Burnham & Graham
1999). At the same time, wide seas divided South and
North America, a condition that had existed since the Early
Jurassic (Burnham & Graham 1999). There is no firm geo-
logical evidence for any continuous land connections
between South and North America until the closure of the
Isthmus of Panama at 3 Myr ago. Biotically, however, the
model of isolation may be too simplistic, as evidenced by
several recent molecular phylogenetic studies of plants,
which demonstrate arrivals in South America from the Late
Cretaceous and through the Tertiary (e.g. Chanderbali et
al. 2001; Renner et al. 2001). In many cases, the plant
immigrants have become important elements of the
neotropical flora, both in terms of species richness and
ecological dominance.
Oceanic dispersal must be implicated in all cases of
migration to South America during its long period of iso-
lation. However, it is not necessary to invoke long-distance
marine dispersal in every case. Although there is no
unequivocal evidence for continuous land connections
among the continents after their initial separations, it has
been hypothesized that migration was possible through
stepping stone migration along island chains: (i) between
Africa and South America until the Late Cretaceous
through islands of the Walvis Ridge – Rio Grande Rise and
Sierra Leone – Ceara
´
Rises (Morley 2000, 2003; Morley &
Dick 2003); (ii) from North America through islands of
the proto-Antilles (Iturralde-Vinent & MacPhee 1999;
Graham 2003) and along a Central American land
bridge in the Late Miocene (Bermingham & Martin 1998);
and (iii) along a southern route from Australia through
Antarctica (Morley 2003).
2. LAND CONNECTIONS TO SOUTH AMERICA
(a) African – South American connections
Based upon the simultaneous appearance of novel pollen
types in Africa and South America, Morley (2000, 2003)
suggested that transatlantic dispersal was facilitated until
the Maastrichtian (ca. 76 Myr ago) by putative island
chains of the Rio Grande Rise – Walvis Ridge and the
Ceara
´
– Sierra Leone Rises (figure 1). Among the pollen
taxa appearing simultaneously on either side of the south
Atlantic are ancestral Proteaceae and Palmae (Arecaceae),
Sapindaceae and Myrtales (Morley 2003). Parrish (1993)
reviewed evidence from the Deep Sea Drilling Programme
that shows shallow water and aerial volcanic features in
both the Walvis Ridge and Rio Grande Rises during the
Eocene. This corroborates subsidence estimations by
Theide (1977) suggesting that the Rio Grande Rise may
have been above sea level until the Oligocene. Thus it
seems plausible that island chains may have been more
extensive before these areas had subsided. Morley (2000,
2003) suggested that these island chains might have facili-
tated plant dispersal most effectively when sea levels were
relatively low, for example at the end of the Turonian
(ca. 88 Myr ago; Haq et al. 1988; Marcellari 1988). How-
ever, recent results from the Ocean Drilling Program
(Zachos et al. 2004) failed to find evidence for dry land in
the Late Cretaceous and Early Tertiary sediments along
Walvis Ridge.
(b) North American – South American connections
(i) Proto-Greater Antilles
Geological and tectonic models for the Caribbean have
been contentious, but Graham (2003) and Morley (2003)
indicate that consensus is developing around the tectonic
model of Pindell et al. (1988) and Pindell & Barrett (1990).
This model suggests that the proto-Greater Antilles origi-
nated in the Early Cretaceous in the eastern Pacific Ocean
adjacent to North and South America, which were separat-
ing. The Caribbean plate moved through the portal
between the continents, carrying the submerged proto-
Greater Antilles (figure 2). These only became islands after
the Caribbean plate collided with the Bahamas plate, with
the principal time of emergence in the Middle Eocene, ca.
50 Myr ago (Iturralde-Vinent & MacPhee 1999). In the
Middle to Late Eocene, the proto-Greater Antilles may
have provided an island corridor between North and South
boreotropics
Antarctica route
1
2
Figure 1. Early Tertiary, Early Eocene plate tectonic reconstruction and palaeogeography adapted, with permission, from Morley
(2003, fig. 5), indicating closed canopy rainforests (cross hatched). Dotted lines approximate present-day coastlines. Dotted
arrows indicate the approximate position of the Sierra Leone (1) and Walvis (2) ridges. Solid arrows indicate migration routes
through and from the boreotropics and Antarctica.
1612 R. T. Pennington and C. W. Dick Assembly of South American rainforest flora
Phil. Trans. R. Soc. Lond. B (2004)
on May 22, 2011rstb.royalsocietypublishing.orgDownloaded from
America. An arc of volcanic islands between North and
South America may have provided a similar pathway
somewhat earlier in the Late Cretaceous (Iturralde-Vinent
& MacPhee 1999). Gentry (1982) suggested that some
groups with strongly differentiated components in both
tropical North and South America, such as Cactaceae,
might reflect Late Cretaceous or Early Tertiary migration
by these routes, with subsequent isolation until the
closure of the Isthmus of Panama. Savage (1982) inter-
preted distributional patterns in frogs in the same way.
(ii) Greater Antilles and Aves Ridge
Iturralde-Vinent & MacPhee (1999) have proposed
another migrational pathway comprising the Greater Antil-
les and the currently submerged Aves ridge (figure 3).
GAARlandia may have existed for ca. 3 Myr at the
Eocene–Oligocene boundary (35–33 Myr ago), and was
not considered by earlier authors such as Raven & Axelrod
(1974). The GAARlandia hypothesis has been invoked to
explain the dispersal of Styrax (Fritsch 2001a), in addition
to animal groups such as sloths (see MacPhee et al. 2000).
GAARlandia provides a plausible link between North and
South America because at this time, the western portion of
the Greater Antilles was separated from Central America
only by two narrow straits.
(iii) Late Miocene Central American land bridge
There is ample evidence that the Isthmus of Panama was
completed as a terrestrial corridor ca. 3 Myr ago (Coates &
Obando 1996), and that this event allowed a continental
exchange of biota between North and South America. The
profound effects of this biotic exchange are encapsulated in
its usual title, the ‘Great American Interchange’ (Simpson
1980; Stehli & Webb 1985). However, geological evidence
indicates that the staggered and gradual rise of the Isthmus
influenced neotropical biogeography over a longer period
of ca. 10 Myr (Stehli & Webb 1985; Zeh et al. 2003). The
mammalian fossil record indicates some exchange of fauna
between North and South before 3 Myr ago. This is gener-
ally assumed to reflect a process of ‘island hopping’, but the
existence of lineages of freshwater fishes in Panama dated
at the end of the Miocene (4–7 Myr ago), derived from
South American groups, implies an earlier, entirely dry
land route, because these organisms cannot survive salt
water (Bermingham & Martin 1998). The ‘Bermingham–
Martin’ hypothesis has subsequently been used to explain
the historical biogeography of the trans-Isthmian harlequin
beetle-riding pseudoscorpion Cordylochernes scorpioides
(Zeh et al. 2003).
(c) The Antarctica route
In the Late Cretaceous and Palaeocene, the landmasses
now represented by South America, Antarctica, New Zealand,
Australia, the South Sandwich Islands and South Georgia
formed a continuous connection (Morley 2003; figure 1).
This facilitated dispersal for taxa such as Drimys, Protea-
ceae, Nothofagus and Cunoniaceae (Morley 2003). Because
of its near polar position, however, this is unlikely to have
been a significant land bridge for megathermal (frost intol-
erant) plants. However, climates may have been suffi-
ciently warm at the Late Palaeocene – Early Eocene
thermal maximum for this to have acted as a migration
route between South America and Australia for meso- and
macrothermal taxa such as Bombacaceae (Malvaceae ‘bao-
babs’), Sapindaceae and Polygonaceae (Morley 2003).
3. SOURCE AREAS
In the survey of connections to South America, we indi-
cated North America, Africa and Australasia as potential
sources for migrant floras. However, these landmasses
were themselves receiving immigrants from tropical south-
east Asia and elsewhere. To understand contemporary flor-
istic similarities between southeast Asia and South
America, for example, we must consider the floristic con-
nections that existed between Africa, North America and
Eurasia during the Late Cretaceous and Tertiary. In this
paper, we follow Morley (2003) and consider Australasia as
North America
accretion of
Nicaragua Rise
Farallon
Plate
Caribbean
Plate
Yucatan
Puerto
Rico
accretion of
W C Cordillera
passive margin
Cuban
Arc
Aves
Ridge
South America
proto-greater Antilles
shelf edge
proto-Caribbean
Ocean
seafloor
anomaly 34
Bahamas
Panama
Costa
Rica
Arc
Figure 2. Campanian reconstruction of Middle America modified, with permission, from Morley (2003, fig. 6), highlighting the
proto-Greater Antilles, which may have provided a filter-bridge in the Early Tertiary. Orange, thick continental crust; v symbols,
volcanic arcs.
Assembly of South American rainforest flora R. T. Pennington and C. W. Dick 1613
Phil. Trans. R. Soc. Lond. B (2004)
on May 22, 2011rstb.royalsocietypublishing.orgDownloaded from
historically isolated, and receiving few immigrants from
Southeast Asia, and do not discuss it further.
(a) Tropical Laurasia – boreotropical
and Madrean–Tethyan vegetation
Under the ‘boreotropics hypothesis’ (e.g. Lavin &
Luckow 1993) the proximity of the North American and
Eurasian tectonic plates until the Late Eocene or Early Oli-
gocene allowed plant dispersal over the North Atlantic,
either through direct land connection or limited water gaps.
It was also a time of thermal maxima, which would have
allowed the survival of megathermal (‘tropical’) organisms
at latitudes at which they do not occur today. The presence
of tropical biomes at high latitudes is evidenced by Eocene
fossil floras found in North America and Europe (reviewed
by Wolfe 1975; Tiffney 1985a,b), which contain taxa clo-
sely related to extant species in tropical southeast Asia,
Central America and the Antilles. The Beringian land
bridge between North America and Asia would also have
been a potential migration path for megathermal organisms
(Wolfe 1975). The boreotropics hypothesis suggests wide
migration of northern tropical elements between Eurasia,
North America and Africa. The extent and duration of the
boreotropical biome is unclear. Although there seems to
have been floristic uniformity between European and
North American mid- to high-latitude floras during the
Eocene, many boreotropical taxa did not appear in Asia
until the Miocene (Manchester 1999). This might reflect
the poor Asian fossil record (Manchester 1999), but it is
clear that boreotropical elements migrated at different
times, and by the differing routes of the Beringian land
bridge, the North Atlantic land bridge, and across Eurasia
along the Tethys seaway (figure 1).
The fossil record indicates that boreotropical vegetation
was a mixture without modern analogue. The modern rela-
tives of boreotropical taxa are found in habitats ranging
from northern deciduous hardwood forests to tropical and
paratropical rainforests (Tiffney 1985a,b). The climates
supporting this vegetation cannot have been strongly seaso-
nal, but there is evidence for a sclerophyllous ‘Madrean–
Tethyan’ flora characteristic of more seasonal climates that
ranged along the Tethyan coast from North America to
southeast Eurasia and North Africa (Axelrod 1975). Fossil
evidence shows sclerophyllous taxa entering the record in
the Early Eocene, and this Madrean–Tethyan sclerophyll
vegetation formed a broad and continuous belt across
southern North America and Eurasia (figure 4) until it was
disrupted by increased climatic cooling and drying at the
end of the Oligocene (Axelrod 1975).
The boreotropical and Madrean–Tethyan areas offer a
viable Early Tertiary migration route between the Asian,
African and North American tropics. Earlier reviews of the
boreotropics hypothesis (e.g. Lavin & Luckow 1993) envi-
saged the major arrival of boreotropical elements to South
America after the closure of the Isthmus of Panama. How-
ever, the migration routes through the Lesser Antilles and
the Late Miocene Central American land bridge offer ear-
lier alternatives for both the boreotropical and Madrean–
Tethyan floras.
4. INVESTIGATING THE HISTORICAL ASSEMBLY
OF THE NEOTROPICAL FLORA
In addressing this problem, many authors, following the
methods of Raven & Axelrod (1974), have attempted to
designate the origin of a taxonomic group (typically a fam-
ily) as being from west Gondwana or Laurasia (e.g. Gentry
GAARlandia
500 km
Figure 3. Caribbean tectonic reconstruction reproduced, with permission, following Pindell (1994, fig. 2.6L), highlighting the
GAARlandia migration route. White, dry land and volcanic areas; dark grey, shallow water; light grey, deep water.
1614 R. T. Pennington and C. W. Dick Assembly of South American rainforest flora
Phil. Trans. R. Soc. Lond. B (2004)
on May 22, 2011rstb.royalsocietypublishing.orgDownloaded from
1982; Wendt 1993; Burnham & Graham 1999). The infer-
ence is derived from several lines of evidence, including:
current distribution, taking into account areas of maximum
diversity and endemism; fossil distribution; degree of
specialization of a taxon in a given area; assessment of sys-
tematic relationships based upon classical morphological
taxonomic methods; and contemporary powers of disper-
sal. Given the publication of Raven and Axelrod’s influen-
tial work soon after the general acceptance of plate tectonic
theory, it is understandable that vicariance explanations
took precedence. For example, for a pantropical distri-
bution, which is both tempting and parsimonious to
explain by Gondwanan vicariance, conflicting data such as
a young fossil record and a systematic position implying
recent origin, were assigned lesser importance. Examples
are several families in asterid orders such as Lamiales,
Gentianales and Boraginales, none of which has a
pre-Eocene fossil record, but which were interpreted as
originating earlier by Raven & Axelrod (1974), perhaps
because of their wide distributions.
Several authors have used the Raven and Axelrod
approach to infer the history of the neotropical flora. For
example, Gentry (1982) used Raven & Axelrod (1974) to
assign 145 neotropical families a Laurasian or Gondwanan
origin. Considering the numbers of South American spe-
cies in these families, he concluded that the Laurasian con-
tribution to the South American flora was less than 10%,
and even more insignificant for the lowland flora alone.
This view of the South American lowland flora has been
influential and supports the still widely accepted view of
Gondwanan origins with little or no subsequent immi-
gration.
(a) Evidence from molecular phylogenies
Raven & Axelrod’s (1974) analysis of the relative age and
specialization of families was based upon concepts of
angiosperm relationships derived from morphology-based
evolutionary taxonomies (e.g. Thorne 1968; Takhtajan
1969) that have been superseded by robust phylogenetic
schemes from sequences of multiple chloroplast and
nuclear genes (e.g. APG 1998; APG II 2003). This new
source of information, in conjunction with the fossil record,
offers a far more powerful instrument for assessing the
age and origin of the lineages (stem groups) leading to
contemporary plant families. Part of this power stems from
that fact that a molecular phylogenetic tree specifies exact
sister group relationships, which increases the applicability
of the fossil record. Sister lineages diverged at the same
time, so the fossil record of one group can be used to date
the stem lineage of its sister group, even if this lineage lacks
fossils entirely (Magallo
´
n et al. 1999; Magallo
´
n & Sander-
son 2001).
Studies integrating new molecular phylogenies of angio-
sperms with palaeobotanical information (e.g. Magallo
´
n et
al. 1999), also offer some reassurance that the fossil record
is not as incomplete and misleading as many authors
assume, and that the phylogenies themselves are relatively
accurate. There is a remarkable congruence between the
fossil record and estimated phylogenetic relationships (e.g.
Magallo
´
n et al. 1999). Families belonging to basally
branching (i.e. inferred to be relatively ancient) lineages
generally have a longer fossil record, whereas more apically
placed (i.e. more recent) families generally have a more
recent fossil record. An accurate phylogeny and a relatively
complete fossil record explain this congruence between the
molecular phylogenetic estimates and the fossil record. If
the fossil record were badly incomplete, we would expect
far more instances where putatively primitive lineages had
only recent fossil records.
There is, however, some disagreement between the fossil
record and molecular clock estimates on exact ages for
some angiosperm lineages. Entire chloroplast genomes
(Chaw et al. 2004), and combined nuclear 18S rDNA and
chloroplast DNA rbcL and atpB genes (Wikstrom et al.
2001) provide somewhat older estimates than the fossil rec-
ord for core eudicot clades, which account for 75% of
extant angiosperms and most most rainforest trees. Our
discussion encompasses the range of age estimates found in
the literature.
Biogeographic interpretations of Raven & Axelrod
(1974) should now be re-examined in the light of this new
evidence because it highlights inaccuracies, especially in
families now established as having relatively recent origins.
For example, the order Lamiales (sensu Magallo
´
n et al.
1999) is determined as ca. 44 Myr ago (Magallo
´
n et al.
1999) to ca. 74 Myr ago (Wikstrom et al. 2001). This
undermines the Gondwanan vicariance explanation for
the distribution of the Lamiales families Bignoniaceae
Figure 4. Early Tertiary continental positions and areas potentially occupied by Madrean–Tethyan sclerophyllous vegetation
(stippled), modified, with permission, from Axelrod (1975, fig. 1).
Assembly of South American rainforest flora R. T. Pennington and C. W. Dick 1615
Phil. Trans. R. Soc. Lond. B (2004)
on May 22, 2011rstb.royalsocietypublishing.orgDownloaded from
and Gesneriaceae, first suggested by Raven & Axelrod
(1974), and followed by subsequent authors (e.g. Gentry
1982; Hilliard & Burtt 2002). Likewise, Gondwanan
explanations for the transatlantic distributions of Mela-
stomataceae sensu stricto (i.e. excluding Memecylaceae)
and Malpighiaceae have been refuted by molecular
phylogenetic studies coupled to fossil calibrated molecular
clock analyses (Renner et al. 2001; Davis et al. 2002).
(b) Hypothesis testing using molecular phylogenies
Hypotheses of geographical origin and migration contain
implicit assumptions about phylogenetic relationships. It is
possible to reconstruct the phylogeny for a group and
determine whether the phylogenetic estimate fits a biogeo-
graphic scenario. Furthermore, and most pertinent in the
context of this volume, different biogeographic scenarios
involve events of different ages, and calibrating phylogenies
with a time dimension adds a second level of biogeographic
testing. In this paper, we are interested in testing hypoth-
eses about source areas and migration routes which have
led to the assembly of the South American flora.
(i) Gondwanan origin
If South American taxa are descendants of west
Gondwanan lineages that have evolved in situ without
subsequent transatlantic dispersal, we might expect them
to be monophyletic, with monophyletic sister groups in
Africa and/or Madagascar. This pattern of reciprocal
monophyly (sensu Cunningham & Collins 1994, 1998) is
displayed by calamoid palms (Baker & Dransfield 2000;
Baker et al. 2000a,b), where the three endemic neotropical
genera Mauritia, Mauritiella and Lepidocaryum form a
monophyletic group, with the African Raphia as its sister
group (figure 5). Because of the Late Cretaceous – Early
Tertiary fossil pollen record attributable to this group in
both South America and Africa, Baker & Dransfield (2000)
suggested that the divergence of Raphia and its neotropical
relatives was linked to the opening of the Atlantic Ocean.
Corroborating evidence that calamoid palms were present
before the split of west Gondwana is suggested by their
basal position in a palm family phylogeny (Asmussen &
Chase 2001), and a date of origin for palms just older than
100 Myr ago (Bremer 2000).
The same pattern of reciprocal monophyly would be pre-
dicted if filter dispersal was possible until the Maastrichtian
through putative island chains along the Walvis Ridge – Rio
Grande Rise or Sierra Leone – Ceara
´
Rises. Because of the
relatively close timing of the break-up of west Gondwana
(ca. 96 Myr ago) and the availability of the possible island
dispersal routes (96–76 Myr ago; Morley 2003), distin-
guishing these scenarios by dating phylogenies is difficult.
For either, the divergence of the sister groups should
date at least to the Late Cretaceous (figure 5). Baker &
Dransfield (2000) did not attempt to date their phylogeny
of calamoid palms, but the high degree of ITS sequence
divergence between the neotropical and African taxa does
suggest antiquity.
A dated phylogeny of Lauraceae (Chanderbali et al.
2001) provides compelling evidence for an influence of the
break-up of west Gondwana on the distribution of this
family. The South American Chlorocardium–Mezilaurus
clade occupies a basal and isolated position in the family.
Its putative Gondwanan origin is supported by a molecular
rate analysis of ITS and rpl16 datasets. However, the lack
of continental monophyly in more derived Lauraceae
groups favours a role of Tertiary dispersal in explaining the
pantropical distribution. Such overlapping patterns of
diversification and dispersal may be common, meaning
that simple patterns of South American – African/Mada-
gascan reciprocal monophyly, such as that found in the
calamoid palms, may be rare.
(ii) Laurasian boreotropical or Madrean–Tethyan origin with
entry to South America through the proto-Greater Antilles,
GAARlandia, Late Miocene land bridge
and Isthmus of Panama
A good example of how to use phylogenetic hypotheses
in this biogeographic context was provided by Lavin &
Luckow (1993), who asked whether elements of the tropical
North American flora were derived from South American
Gondwanan stock, or from northern boreotropical
vegetation. These two biogeographic hypotheses imply
different area cladograms, which can be produced by
Mauritia
clade
Raphia
age older than 60 Myr
Figure 5. Gondwanan phylogenetic pattern of reciprocal monophyly, as exemplified by calamoid palms (Baker & Dransfield 2000).
1616 R. T. Pennington and C. W. Dick Assembly of South American rainforest flora
Phil. Trans. R. Soc. Lond. B (2004)
on May 22, 2011rstb.royalsocietypublishing.orgDownloaded from
reconstructing phylogenies for groups with species in the
requisite areas of South America, tropical North America,
and with tropical Old World sister groups. Subsidiary evi-
dence can be gathered from the fossil record; a Gondwanan
explanation would predict Cretaceous fossils in South
America or Africa (unfortunately, Cretaceous fossils are
scarce for South America; Burnham & Graham 1999;
Burnham & Johnson 2004); a boreotropical explanation
would predict Early Tertiary fossils in North America,
Europe or Asia. Calibrating phylogenies with a time
dimension now allows a further distinction of the hypoth-
eses. A Gondwana explanation would predict a date of ca.
90–76 Myr ago (the separation of Africa and South
America, plus putative island chain connections) for the
divergence of Old and New World groups, whereas a
boreotropical explanation would predict a date of ca.
40–50 Myr ago. These are the approximate dates for the
existence of a north Atlantic land bridge and temperatures
high enough for megathermal taxa to migrate through the
Beringian land bridge (Tiffney 1985a,b; Wolfe 1975).
Boreotropical groups may have ultimately arrived in
South America through the Proto-Antilles, GAARlandia,
the late Miocene Central American land bridge, or as part
of the Great American Interchange. These alternatives are
not mutually exclusive, and they all predict the same clado-
gram topology: South American groups derived from trop-
ical North American relatives with Old World sister groups
(Lavin & Luckow 1993; figure 6). Furthermore, confi-
dence limits in assigning dates to cladogram nodes are too
wide to confidently distinguish between the Late Miocene
land bridge (4–7 Myr ago) and the closure of the Panama
Isthmus (3 Myr ago). An arrival in the Eocene or Early
Oligocene boundary (ca. 50 33 Myr ago) might, however,
be indicative of a Proto-Antillean or GAARlandia route.
Confidently distinguishing these two Antillean scenarios by
assigning dates will also be difficult.
Laurasian origins of South American taxa have been sug-
gested for Annonaceae (Doyle & Le Thomas 1997;
Richardson et al. 2004), Lauraceae (Chanderbali et al.
2001), and some groups of Leguminosae (Fabaceae)
(Lavin & Luckow 1993). The example of Lauraceae is
particularly instructive in that there is little doubt from the
fossil record dating from the Mid-Cretaceous (Chanderbali
et al. 2001) and its basal phylogenetic position in the
angiosperms (e.g. APG 1998; APG II 2003), that this is a
relatively ancient angiosperm family. Chanderbali et al.
(2001) show that the South American Chlorocardium–
Mezilaurus clade is basally branching in the Lauraceae
phylogenetic tree, consistent with a primitive presence of
Lauraceae in South America (see above). However, a neo-
tropical diversification containing species-rich genera such
as Ocotea (200 species) and Nectandra (100 species) that
are abundant in lowland South American rainforests is
nested within less species-rich African, Madagascan and
Macaronesian lineages. Chanderbali et al. (2001) hypothe-
size that this transatlantic distribution cannot have been
achieved by west Gondwanan vicariance because it is far
too young (ca. 23 Myr ago), but rather was the result of
migration through Laurasia, probably in Madrean–
Tethyan scleophyll vegetation. Given the antiquity of
Lauraceae, the species in these neotropical genera might be
considered prime candidates to have accumulated by the
‘museum’ model, where high species diversity is attained
over long time-scales with low rates of extinction.
However, they are clearly the result of recent radiation.
This again emphasizes that ancient biogeographic events
such as Gondwanan vicariance should not necessarily be
sought to explain the distribution and diversification of
lineages and species within ancient families.
(iii) Antarctica route
This southern route does not appear to have been impor-
tant for lowland tropical plants. However, certain elements
of the Andean flora may have used it to enter South Amer-
ica during the Tertiary. An example is Cunoniaceae, of
which Weinmannia is an important genus in Andean mon-
tane forests. A phylogeny of Weinmannia (Bradford 2002)
is insufficiently resolved to determine whether the
neotropical species of this genus can be considered immi-
grants, but the extensive Palaeogene fossil record of
age ca. 40–50 Myr
Figure 6. One potential boreotropical phylogenetic pattern, modified from Lavin & Luckow (1993), showing a South American
group derived from tropical North American relatives, and with Old World sister groups.
Assembly of South American rainforest flora R. T. Pennington and C. W. Dick 1617
Phil. Trans. R. Soc. Lond. B (2004)
on May 22, 2011rstb.royalsocietypublishing.orgDownloaded from
Cunoniaceae, including Weinmannia, from southern Aus-
tralia, plus Late Cretaceous fossil wood from Antarctica
(reviewed by Bradford et al. 2004), implies that this family
probably migrated along this southern route.
(iv) Long-distance, trans-oceanic dispersal
The ‘island’ model of South American biotic isolation
until the closure of the Isthmus of Panama denies a major
role for long-distance, trans-oceanic dispersal in contribu-
ting to the neotropical flora through the Cenozoic. This
would surprise early biogeographers such as Darwin
(1859) and Darlington (1957) who emphasized the
primacy of dispersal explanations in explaining the global
distribution of organisms. Acceptance of continental drift
provided an alternative, potentially universal explanation
that did not need to invoke ad hoc dispersal scenarios for
each taxon. This viewpoint was popularized by the field
of cladistic or vicariance biogeography (e.g. Nelson &
Platnick 1981), which attempts to seek common patterns
in the distributions of unrelated organisms.
Evidence from molecular phylogenies does, however,
indicate that such ‘waif ’ or ‘sweepstakes’ dispersal across
very wide physical barriers as the Atlantic ocean has indeed
occurred in multiple taxa and explains disjunctions at spe-
cies, generic and higher taxonomic levels (Lavin et al. 2004;
Renner 2004). It is remarkable that some of these examples
are of plants that show little adaptation for over-water dis-
persal, such as the rainforest tree Symphonia globulifera
(Dick et al. 2003), whose large, recalcitrant seeds
cannot survive immersion in salt water. In all these cases,
trans-oceanic dispersal is inferred because the node on a
phylogenetic tree that marks the movement from one conti-
nental area to the other has a geological date that does not
correspond with the availability of any stepping stone
migration route. Renner (2004) presented 11 examples of
transatlantic distributions at various taxonomic levels (spe-
cies, genus, tribe) that are all dated at 11 Myr ago
or less, and therefore explicable only by long-distance dis-
persal. For legumes, Lavin et al. (2004) show that out of 59
trans-oceanic crown clades, only eight are older than 25
Myr ago. This strongly implies that most have developed
their distributions by long-distance dispersal.
We predict that as more studies are published, further
well-corroborated examples of recent transatlantic disper-
sal to and from the neotropics will appear. For example,
chloroplast DNA restriction site data (Pennington 2003)
and ITS sequences (Skema 2003) indicate that Andira
inermis reached Africa from the neotropics within the past
few million years. Renealmia (Zingiberaceae) is the only
genus of the ginger family indigenous to the neotropics,
where there are ca. 56 species. This genus is also dis-
tributed in Africa, where there are ca. 15 species. ITS
sequence data for a handful of neotropical and African
species (M. F. Newman, unpublished data) show that the
African and neotropical species are separated by remark-
ably short branch lengths, implying that the biogeographic
patterns in this genus have been attained by recent
trans-oceanic dispersal followed by rapid speciation.
Frequent long-distance trans-oceanic dispersal may
obliterate underlying biogeographic patterns caused by
continental vicariance or biogeographic bridges or filters.
Lavin et al. (2004) conclude that this is the case in Legumi-
nosae, where there are no clear repeated patterns of area
relationships. Schrire et al. (2004) suggested that structure
in legume phylogeny is more explicable by ecological
setting, and Lavin et al. (2004) further suggest that the
frequency of long-distance dispersal between habitats is the
principal determinant of legume phylogenetic patterns.
Whether the frequency of long-distance dispersal seen in
legumes is found to be typical of other groups can only be
verified by more empirical data. However, it is also invoked
to explain some distribution patterns in Annonaceae and
Rhamnaceae (Richardson et al. 2004). Conversely, such
dispersal must have been rare in calamoid palms. Legumes,
however, dominate neotropical rainforests, dry forests and
woody savannahs, and understanding that long-distance,
trans-oceanic dispersal has been instrumental in shaping
the distribution of this family is important to our under-
standing of the historical assembly of these biomes.
(c) The historical assembly of the Amazonian
rainforest flora: a re-evaluation
The examples above demonstrate that the South American
flora has received immigrant taxa throughout the Cenozoic
from Laurasia, Africa and Australia, by stepping stone
migration and long-distance dispersal. Clearly, the
contribution of immigrant taxa to the South American
rainforest flora also merits re-evaluation, as suggested by
the authors of many of the studies reviewed (e.g. Chanderbali
et al. 2001; Renner et al. 2001). One approach is to dupli-
cate the analysis presented by Gentry (1982), and re-assign
areas of origin to all Amazonian rainforest families in light
of new molecular and biogeographic evidence of their
geographical origins. This approach is not yet feasible,
however, because insufficient phylogenies are available.
Furthermore, the approach may be flawed, in that
subfamilial groups may have multiple continental origins,
as Chanderbali et al. (2001) have shown for the Lauraceae.
Given these uncertainties, we restrict ourselves to four
major families that are important elements of Amazonian
forests, and for which reasonable phylogenetic information
is available: Leguminosae (Schrire et al. 2004), Annonaceae
(Richardson et al. 2004), Lauraceae (Chanderbali et al.
2001) and Melastomataceae sensu stricto (Renner et al.
2001). For Lauraceae and Annonaceae, the neotropical
genera identified as arrivals from Laurasia are listed in
Appendix A. Legumes are the dominant tree family of
South American rainforests and their global fossil record
indicates a Late Cretaceous or Early Tertiary origin
(Herendeen et al. 1992; Magallo
´
n et al. 1999; Schrire et al.
2004), with the oldest molecular clock based estimate at
74–79 Myr ago (Wikstrom et al. 2001; Chaw et al. 2004).
Schrire et al. (2004) suggest that the seasonally dry margins
of the Tethys seaway are the ancestral area for legumes,
implying that all South American taxa must be considered
immigrants. The classification of all legumes as immigrants
would be disputed by some authors (e.g. Raven & Polhill
1981; Morley 2000, 2003), and the putative Madrean–
Tethyan origin hypothesized by Schrire et al. (2004) is not
inferred by a clear optimization of an ancestral area, but
from inference of historical ecological preferences. Renner
et al. (2001) indicated that all South American Mela-
stomataceae sensu stricto are derived from Laurasian boreo-
tropical ancestors. Morley & Dick (2003) considered this
family as possibly Gondwanan, but their preferred biogeo-
graphic scenario is an African origin for South American
1618 R. T. Pennington and C. W. Dick Assembly of South American rainforest flora
Phil. Trans. R. Soc. Lond. B (2004)
on May 22, 2011rstb.royalsocietypublishing.orgDownloaded from
species, which therefore does not alter their status as immi-
grants.
Some other less ecologically important families have a
clear Laurasian origin based upon published phylogenies or
their fossil record, and it seems reasonable to assume that
their South American species are immigrants. We consider
these families to be Ulmaceae (Ampelocera, Celtis, Trema),
Rhamnaceae (Colubrina, Rhamnidium, Zizyphus (see
Richardson et al. 2004)), Rosaceae (Prunus), Staphylea-
ceae, (Huertea, Turpinia), Magnoliaceae (Talauma) and
Styracaceae (Styrax (Fritsch 2001b)).
We assessed the percentage of species that can be con-
sidered immigrants to an Amazon rainforest tree com-
munity (1 cm or greater diameter at breast height)
represented in a 25 ha forest inventory plot in Yasuni
National Park, Ecuador (see Valencia et al. (2004) for a
description of this inventory plot). We adopted a conserva-
tive approach for the Lauraceae, by excluding the 30 unde-
termined morphospecies. Out of the staggering 1104 tree
species thus far identified in the Yasuni plot, 232 (21%)
belong to families and genera that we classify as immigrant.
Even if we do not consider legumes as immigrants, 11% of
the Yasuni species consist of Cenozoic immigrants. We can
also assess the immigrant contribution to the Yasuni flora
in terms of numbers of individuals (1 cm or greater in
diameter) found in the 25 ha plot. Out of 152 353 indivi-
duals (Valencia et al. 2004), 30 156 (20%) are accounted
for by the 232 immigrant species. These results stand in
stark contrast to the view of Gentry (1982, p. 577), who
claimed that less than 10% of the entire neotropical flora,
and ‘virtually none’ of the lowland rainforest flora, had
Laurasian (i.e., immigrant) origin.
Our calculations probably underestimate the immigrant
contribution to the Amazonian flora. As is the case for
Lauraceae, Annonaceae and Melastomataceae sensu stricto,
we predict that elements of families previously considered
‘Gondwanan’ will be identified as South American immi-
grants as more phylogenies become available. For example,
a phylogeny of Manilkara (Sapotaceae; Clayton 2003) indi-
cates that this genus arrived in the neotropics by a boreo-
tropical route. This calls into question the assumption that
all of Sapotaceae, another ecologically dominant
Amazonian family (54 species in the Yasuni plot), can
be considered essentially ‘Gondwanan’ (Gentry 1982).
Similarly, a phylogeny of Ficus (Moraceae; Weiblen 2000),
which sampled only three of ca. 150 neotropical species,
showed two to be sister species and basally divergent within
the genus, but one (F. pertusa) to be nested high within an
Asian clade, consistent with a recent neotropical arrival.
(d) How similar are neotropical and tropical African
rainforest communities?
There have been many reviews of biotic relationships
between the neotropics and tropical Africa (e.g. Meggers et
al. 1973; Goldblatt 1993). In many cases these have
emphasized floristic differences. For example, Thorne
(1973) listed 111 seed plant genera restricted to tropical
and subtropical Africa, Madagascar and the neotropics,
but he estimated that these are fewer than 2.5% of the 4500
genera found on either continent. Considering more wide-
ranging pantropical genera did not make the genera shared
between South America and Africa exceed 10% of the total
generic flora (Thorne 1973). However, Gentry (1993)
pointed out that the composition of amphi-Atlantic rain-
forest communities showed much higher levels of simi-
larity. He highlighted that the family-level composition of
rainforests on both continents was almost identical, and
that Leguminosae, Moraceae, Lauraceae and Annonaceae
are dominant rainforest tree families on both continents.
Gentry also pointed out strikingly high levels of generic
similarity; for example, an average of 31% of genera found
in a series of 0.1 ha sample plots in Africa are also present
in the neotropics, and he indicated that the existence of
additional shared genera was potentially obscured by
provincial nomenclature.
Given that the earliest fossils of many species-rich
eudicot families such as legumes that are important
components of tropical rainforests date only to the Late
Cretaceous (Magallo
´
n et al. 1999; Lavin et al. 2004), the
occurrence of shared genera between Africa and South
America is probably most often the result of oceanic disper-
sal. In a similar fashion to Gentry (1993), we have com-
pared species lists of the CTFS forest inventories in BCI,
Panama (50 ha), and Yasunı
´
, Ecuador (25 ha) with lists
from Iturı
´
, Democratic Republic of Congo (40 ha), and
Korup, Cameroon (50 ha). Out of 174 genera in
the BCI plots, 30 (17%) are found in the African plots,
whereas 43 out of 333 genera in Yasuni (13%) are also
found in Africa. If we expand the comparison to other
palaeotropical rainforests in Asia through reference to
distribution information in Gentry (1996) and Mabberley
(1993), we find that 105 (32%) of the Yasuni genera are
found in the palaeotropics.
Gentry (1993, p. 542) explained the community-level
similarities between South American and African rain-
forests in terms of Gondwana vicariance: ‘...the strong
floristic similarities seen when individual sites or communi-
ties are compared strongly suggest that today’s forests
result to a much greater extent from common origin than
from chance immigration’. By contrast, we hypothesize
that these similarities also reflect the relatively recent arrival
of immigrant taxa by both long-distance dispersal and the
migration routes we have reviewed. The analysis for Legu-
minosae by Lavin et al. (2004) suggests that long-distance
dispersal has been the predominant force shaping the dis-
tribution of this family. Whether this is the case for other
tree families that are ecologically important in Amazonia
will be determined only by the availability of thoroughly
sampled, dated phylogenies. These will enable us to
disentangle the relative contributions of continental
vicariance, filter bridge migration routes and long-distance
dispersal in the historical assembly of the rainforest floras.
Finally, we ask why does the model of ‘splendid
isolation’ apply more weakly to the South American flora
than to its vertebrate fauna? We think the answer lies in the
greater capacities of plants to disperse over long distances
and establish founder populations. The traits that dis-
tinguish them from vertebrates include seed dormancy,
asexual reproduction and the capacity to form their
own rafts and waifs. Because the fossil record of South
American plants provides a relatively poor documentation
of the floristic interchange (e.g. Burnham & Graham 1999;
Morley & Dick 2003), the new advances in molecular
systematics will provide strong inferences about the role of
immigration in the assembly of this species-rich biome.
Assembly of South American rainforest flora R. T. Pennington and C. W. Dick 1619
Phil. Trans. R. Soc. Lond. B (2004)
on May 22, 2011rstb.royalsocietypublishing.orgDownloaded from
We thank James Richardson and two anonymous reviewers for
constructive criticism of this manuscript. We thank Bill Baker,
Helen Hopkins, Bob Hill, Cindy Skema, Josh Clayton, Mark
Newman, David Harris and Michelle Hollingsworth for pro-
viding unpublished data or advice on examples presented in
this paper. We thank Hugo Navarrete, Elizabeth Losos and
Consuelo Hernandez of the CTFS and the Pontifica Uni-
versidad Cato
´
lica del Ecuador for allowing us to use unpub-
lished data from the CTFS Yasunı
´
forest inventories. We
gratefully acknowledge Bob Morley, James Pindell and
Maureen C. Warwick for providing figures, and John-Paull
Shirreffs for help with graphics files.
APPENDIX A. NEOTROPICAL GENERA OF
LAURACEAE AND ANNONACEAE THAT ARE NOT
PRIMITIVELY SOUTH AMERICAN
REFERENCES
APG 1998 An ordinal classification for the families of flower-
ing plants. Ann. Missouri Bot. Gard. 85, 531–553.
APG II 2003 An update of the Angiosperm Phylogeny Group
classification for the orders and families of flowering plants:
APG II. Bot. J. Linn. Soc. 141, 399–436.
Asmussen, C. & Chase, M. W. 2001 Coding and noncoding
plastid DNA in palm systematics. Am. J. Bot. 88,
1103–1117.
Axelrod, D. I. 1975 Evolution and biogeography of Madrean–
Tethyan sclerophyll vegetation. Ann. Missouri Bot. Gard. 62,
280–334.
Baker, W. J. & Dransfield, J. 2000 Towards a biogeographic
explanation of the calamoid palms. In Monocots: systematics
and evolution (ed. K. L. Wilson & D. A. Morrison), pp.
545–553. Melbourne, Australia: CSIRO.
Baker, W. J., Dransfield, J. & Hedderson, T. A. 2000a Phy-
logeny, character evolution, and a new classification of the
calamoid palms. Syst. Bot. 25, 297–322.
Baker, W. J., Hedderson, T. A. & Dransfield, J. 2000b Mol-
ecular phylogenetics of subfamily Calamoideae (Palmae)
based upon nrDNA ITS and cpDNA rps16 intron sequence
data. Mol. Phylogenet. Evol. 14, 195–217.
Bermingham, E. & Martin, A. P. 1998 Comparative mtDNA
phylogeography of neotropical freshwater fishes: testing
shared history to infer the evolutionary landscape of lower
Central America. Mol. Ecol. 7, 499–517.
Bradford, J. C. 2002 Molecular phylogenetics and morpho-
logical evolution in Cunonieae (Cunoniaceae). Ann. Mis-
souri Bot. Gard. 89, 491–503.
Bradford, J. C., Fortune-Hopkins, H. C. & Barnes, R. W. 2004
Cunoniaceae. In The families and genera of vascular plants (ed.
K. Kubitzki), pp. 91–111. Heidelberg, Germany: Springer.
Bremer, K. 2000 Early Cretaceous lineages of monocot
flowering plants. Proc. Natl Acad. Sci. USA 97, 4707–4711.
Burnham, R. J. & Graham, A. 1999 The history of neotropical
vegetation: new developments and status. Ann. Missouri
Bot. Gard. 86, 546–589.
Burnham, R. J. & Johnson, K. R. 2004 South American
palaeobotany and the origins of neotropical rainforests. Phil.
Trans. R. Soc. Lond.B359, 1595–1610. (doi:10.1098/rstb.
2004.1531)
Chanderbali, A. S., Van der Werff, H. & Renner, S. S. 2001
Phylogeny and historical biogeography of Lauraceae: evi-
dence from the chloroplast and nuclear genomes. Ann. Mis-
souri Bot. Gard. 88, 104–134.
Chaw, S. M., Chang, C. C., Chen, H. L. & Li, W. H. 2004
Dating the monocot-dicot divergence and the origin of core
eudicots using whole chloroplast genomes. J. Mol. Evol. 58,
424–441.
Clayton, J. 2003 Pantropical genera: systematics and biogeo-
graphy. A pilot study for the pantropical tree Manilkara.
MSc thesis, University of Edinburgh, UK.
Coates, A. G. & Obando, J. A. 1996 The geologic evolution of
the Central American Isthmus. In Evolution and environment
in tropical America (ed. J. B. C. Jackson, A. F. Budd & A. G.
Coates), pp. 21–56. University of Chicago Press.
Cunningham, C. W. & Collins, T. 1994 Developing model
systems for molecular biogeography: vicariance and inter-
change in marine invertebrates. In Molecular ecology and
evolution: approaches and applications (ed. B. Schierwater,
B. Streit, G. P. Wagner & R. DeSalle), pp. 405–433. Basel,
Switzerland: Birkha¨user.
Cunningham, C. W. & Collins, T. 1998 Beyond area relation-
ships: extinction and recolonization in molecular marine
biogeography. In Molecular approaches to ecology and evol-
ution (ed. R. DeSalle & B. Schierwater), pp. 297–321. Basel,
Switzerland: Birkha¨user.
Darlington, P. J. 1957 Zoogeography: the geographical distri-
bution of animals. New York: Wiley.
Darwin, C. 1859 On the origin of species by means of natural
selection, or the preservation of favoured races in the struggle for
life. London: John Murray.
Davis, C. C., Bell, C. D., Mathews, S. & Donoghue, M. J.
2002 Laurasian migration explains Gondwanan disjunc-
tions: evidence from Malpighiaceae. Proc. Natl Acad. Sci.
USA 99, 6833–6837.
De Oliveira, A. A. & Mori, S. 1999 A central Amazonian terra
firme forest. I. High tree species richness on poor soils.
Biodiv. Conserv. 8, 1219–1244.
Dick, C. W., Abdul-Salim, K. & Bermingham, E. 2003
Molecular systematics reveals cryptic Tertiary diversification
of a widespread tropical rainforest tree. Am. Nat. 162,
691–703.
Annonaceae Bocageopsis
Annonaceae Cremastosperma
Annonaceae Desmopsis
Annonaceae Ephedranthus
Annonaceae Klarobelia
Annonaceae Malmea
Annonaceae Mosannona
Annonaceae Onychopetalum
Annonaceae Oxandra
Annonaceae Polyalthia
Annonaceae Pseudephendranthus
Annonaceae Pseudomalmea
Annonaceae Pseudoxandra
Annonaceae Ruizodendron
Annonaceae Sapranthus
Annonaceae Stenanona
Annonaceae Tridimeris
Annonaceae Unonopsis
Lauraceae Aiouea
Lauraceae Aniba
Lauraceae Dicypellium
Lauraceae Endlicheria
Lauraceae Kubitzkia
Lauraceae Licaria
Lauraceae Nectandra
Lauraceae Ocotea
Lauraceae Paraa
Lauraceae Pleurothyrium
Lauraceae Rhodostemonodaphne
Lauraceae Umbellularia
Lauraceae Urbanodendron
1620 R. T. Pennington and C. W. Dick Assembly of South American rainforest flora
Phil. Trans. R. Soc. Lond. B (2004)
on May 22, 2011rstb.royalsocietypublishing.orgDownloaded from
Doyle, J. A. & Le Thomas, A. 1997 Phylogeny and geographic
history of Annonaceae. Ge
´
ogr. Phys. Quatern. 51, 353–361.
Fritsch, P. W. 2001a Multiple geographic origins of Antillean
Styrax. Syst. Bot. 28, 421–430.
Fritsch, P. W. 2001b Phylogeny and biogeography of the
flowering plant genus Styrax (Styracaceae) based upon
chloroplast DNA restriction sites and DNA sequences on
the internal transcribed spacer region. Mol. Phylogenet. Evol.
19, 387–408.
Gentry, A. H. 1982 Neotropical floristic diversity: phytogeo-
graphical connections between Central and South America:
Pleistocene climatic fluctuations or an accident of the
Andean orogeny? Ann. Missouri Bot. Gard. 69, 557–593.
Gentry, A. H. 1993 Diversity and floristic composition of low-
land tropical forest in Africa and South America. In Biologi-
cal relationships between Africa and South America (ed. P.
Goldblatt), pp. 500–547. New Haven, CT: Yale University
Press.
Gentry, A. H. 1996 A field guide to families and genera of woody
plants of northwest South America (Colombia, Ecuador, Peru).
University of Chicago Press.
Goldblatt, P. 1993 Biological relationships between Africa and
South America. New Haven, CT: Yale University Press.
Graham, A. 2003 Geohistory models and Cenozoic
palaeoenvironments of the Caribbean region. Syst. Bot. 28,
378–386.
Haq, B. U., Hardenbol, J., Vail, P. R. & Baum, G. R.
1988 Mesozoic and Cenozoic chorostratigraphy and
eustatic cycles. In Sea level change: an integrated approach.
Special publication 42 (ed. C. K. Wilgus, B. S. Hastings,
H. Posamentier & J. Van Wagoner), pp. 71–108. Tulsa,
OK: Society of Economic Palaeontologists and Miner-
alogists.
Herendeen, P. S., Crepet, W. L. & Dilcher, D. L. 1992 The
fossil history of the Leguminosae: phylogenetic and biogeo-
graphic implications. In Advances in legume systematics, part
4. The fossil record (ed. P. S. Herendeen & D. L. Dilcher),
pp. 303–316. Kew, UK: The Royal Botanic Gardens.
Hilliard, O. & Burtt, B. L. 2002 The genus Agalmyla (Gesner-
icaeae-Cyrtandroideae). Edinburgh J. Bot. 59, 1–210.
Iturralde-Vinent, M. A. & MacPhee, R. D. E. 1999 Paleogeo-
graphy of the Caribbean region: implications for Cenozoic
biogeography. Bull. Am. Mus. Nat. Hist. 238, 1–95.
Lavin, M. & Luckow, M. 1993 Origins and relationships of
tropical North America in the context of the boreotropics
hypothesis. Am. J. Bot. 80, 1–14.
Lavin, M., Schrire, B., Lewis, G., Pennington, T., Delgado-
Salinas, A., Thulin, M., Hughes, C., Beyra Matos, A. &
Wojciechowski, M. F. 2004 Metacommunity process rather
than continental tectonic history better explains geographi-
cally structured phylogenies in legumes. Phil. Trans. R. Soc.
Lond.B 359, 1509–1522. (doi:10.1098/rstb.2004.1536)
Mabberley, D. J. 1993 The plant-book: a portable dictionary of
the higher plants. Cambridge University Press.
MacPhee, R. D. E., White, J. L. & Woods, C. A. 2000 New
Megalonychid Sloths (Phyllophaga, Xenartha) from the
Quaternary of Hispaniola. Am. Mus. Novitates 3303, 1–32.
Magallo
´
n, S. & Sanderson, M. J. 2001 Absolute diversification
rates in angiosperm clades. Evolution 55 , 1762–1780.
Magallo
´
n, S., Crane, P. R. & Herendeen, P. S. 1999 Phylo-
genetic pattern, diversity and diversification of eudicots.
Ann. Missouri Bot. Gard. 86, 297–372.
Manchester, S. R. 1999 Biogeographical relationships of
North American Tertiary floras. Ann. Missouri Bot. Gard.
86, 472–522.
Marcellari, C. E. 1988 Cretaceous palaeogeography and
depositional cycles of western South America. J. South Am.
Geol. 1, 373–418.
Meggers, B. J., Ayensu, E. S. & Duckworth, W. D. 1973 Trop-
ical forests ecosystems in Africa and South America: a compara-
tive review. Washington, DC: Smithsonian Institution Press.
Morley, R. J. 2000 Origin and evolution of tropical rainforests.
Chichester, UK: Wiley.
Morley, R. J. 2003 Interplate dispersal routes for megathermal
angiosperms. Perspectives Pl. Ecol. Evol. Syst. 6, 5–20.
Morley, R. J. & Dick, C. W. 2003 Missing fossils, molecular
clocks, and the origin of the Melastomataceae. Am. J. Bot.
90, 1638–1644.
Nelson, G. & Platnick, N. 1981 Systematics and biogeography:
cladistics and vicariance. New York: Columbia University
Press.
Parrish, J. T. 1993 The paleogeography of the opening South
Atlantic. In The Africa–South America connection (ed. W.
George & R. Lavocat), pp. 8–41. Oxford: Clarendon.
Pennington, R. T. 2003 Monograph of Andira (Leguminosae–
Papilionoideae). Syst. Bot. Monogr. 64, 1–143.
Pindell, J. L. 1994 Evolution of the Gulf of Mexico and the
Caribbean. In Caribbean geology, an introduction (ed. S. K.
Donovan & T. A. Jackson), pp. 13–40. Jamaica: The Uni-
versity of the West Indies Publishers Association.
Pindell, J. L. & Barrett, S. F. 1990 Geological evolution of the
Caribbean region: a plate tectonic perspective. In The
geology of North America, vol. H. The Caribbean region (ed. G.
Dengo & J. E. Case), pp. 405–432. Boulder, CO: Geologi-
cal Society of America.
Pindell, J. L., Cande, S. C., Pitman, W. C., Rowley, D. B.,
Dewey, J. F., LaBrecque, J. & Haxby, W. 1988 A plate-
kinematic framework for models of Caribbean evolution.
Tectonophysics 155, 121–138.
Pitman III, W. C., Cande, S., LaBreque, J. & Pindell, J. 1993
Fragmentation of Gondwana: the separation of Africa from
South America. In Biological relationships between Africa and
South America (ed. P. Goldblatt), pp. 202–213. New Haven,
CT: Yale University Press.
Raven, P. H. & Axelrod, D. I. 1974 Angiosperm biogeography
and past continental movements. Ann. Missouri Bot. Gard.
61, 539–673.
Raven, P. H. & Polhill, R. M. 1981 Biogeography of the Legumi-
nosae. In Advances in legume systematics, part 1 (ed. R. M. Polhill
&P.H.Raven),pp.27–34.Kew,UK:RoyalBotanicGardens.
Renner, S. S. 2004 Plant dispersal across the tropical Atlantic by
wind and sea currents. Int. J. Pl. Sci. 165(Suppl. 4), 23–33.
Renner, S. S., Clausing, G. & Meyer, K. 2001 Historical biogeo-
graphy of Melastomataceae: the roles of Tertiary migration
and long-distance dispersal. Am. J. Bot. 88, 1290–1300.
Richardson, J. E., Chatrou, L. W., Mols, J. B., Erkens, R. H. J.
& Pirie, M. D. 2004 Historical biogeography of two cosmo-
politan families of flowering plants: Annonaceae and Rham-
naceae. Phil. Trans. R. Soc. Lond.B359, 1495–1508.
(doi:10.1098/rstb.2004.1537)
Romoleroux, K., Foster, R., Valencia, R., Condit, R., Balslev,
H. & Losos, E. 1997 Especies len
˜
osas (dap > 1 cm) encon-
tradas en dos hecta
´
reas de un bosque de la Amazonı
´
a ecua-
toriana. In Estudios sobre diversidad y ecologı
´
a de plantas (ed.
H. Balslev), pp. 189–215. Quito, Ecuador: Pontificia
Universidad Cato
´
lica del Ecuador.
Savage, J. M. 1982 The enigma of the Central American
Herpetofauna: dispersal or vicariance. Ann. Missouri Bot.
Gard. 69, 464–547.
Schrire, B. D., Lavin, M. & Lewis, G. P. 2004 Global distri-
bution patterns of the Leguminosae: insights from recent
phylogenies. In Plant diversity and complexity patterns: local,
regional and global dimensions (ed. I. Friis & H. Balslev). Bio-
logiske Skrifter. (In the press.)
Simpson, G. G. 1980 Splendid isolation: the curious history of South
American mammals. New Haven, CT: Yale University Press.
Assembly of South American rainforest flora R. T. Pennington and C. W. Dick 1621
Phil. Trans. R. Soc. Lond. B (2004)
on May 22, 2011rstb.royalsocietypublishing.orgDownloaded from
Skema, C. 2003 Molecular systematics and biogeography of
Andira. MSc thesis, University of Edinburgh, UK.
Stehli, F. G. & Webb, D. S. 1985 A kaleidoscope of plates,
faunal and floral dispersals, and sea level changes. In The
great American biotic interchange (ed. F. G. Stehli & D. S.
Webb), pp. 3–16. New York: Plenum.
Takhtajan, A. 1969 Flowering plants, origin and dispersal.
Washington, DC: Smithsonian Institution Press.
Theide, J. 1977 Subsidence of aseismic ridges: evidence from
sediments on Rio Grande Rise (southwestern Atlantic
Ocean). Am. Assoc. Petroleum Geol. Bull. 61, 929–940.
Thomas,W.W.1999Conservationandmonographicresearch
on the flora of tropical America. Biodiv. Conserv. 8, 1007–1015.
Thorne, R. F. 1968 Synopsis of a putatively phylogenetic
classification of flowering plants. Aliso 6, 57–66.
Thorne, R. F. 1973 Floristic relationships between tropical
Africa and tropical America. In Tropical forests ecosystems in
Africa and South America: a comparative review (ed. B. J.
Meggers, E. S. Ayensu & W. D. Duckworth), pp. 27–47.
Washington, DC: Smithsonian Institution Press.
Tiffney, B. 1985a Perspectives on the origin of the floristic
similarity between eastern Asia and eastern North America.
J. Arnold Arboretum 66, 73–94.
Tiffney, B. 1985b The Eocene North Atlantic land bridge: its
importance in Tertiary and modern phytogeography of the
northern hemisphere. J. Arnold Arboretum 66, 243–273.
Valencia, R., Balslev, H., Paz, G. & Min
˜
o, C. 1994 High tree alpha
diversity in Amazonian Ecuador. Biodiv. Conserv. 3, 21–28.
Valencia, R., Foster, R. B., Villa, G., Condit, R., Svenning,
J.-C., Herna
´
ndez, C., Romoleroux, K., Losos, E.,
Marga
˚
rd, E. & Balslev, H. 2004 Tree species distributions
and local habitat variation in the Amazon: large forest plot in
eastern Ecuador. J. Ecol. 92, 214–229.
Weiblen, G. D. 2000 Phylogenetic relationships of func-tion-
ally dioecious Ficus (Moraceae) based upon ribosomal
DNA sequences and morphology. Am. J. Bot. 87, 1342–
1357.
Wendt, T. 1993 Composition, floristic affinities and origins of
the Mexican Atlantic slope rainforests. In Biological diversity
of Mexico, origins and distribution (ed. T. P. Ramamoorthy,
R. Bye, A. Lot & J. Fa), pp. 595–680. Oxford University
Press.
Wikstro¨m, N., Savolainen, V. & Chase, M. W. 2001 Evolution
of the angiosperms: calibrating the family tree. Proc.
R. Soc. Lond. B 268, 2211–2220. (doi:10.1098/rspb.2001.
1782)
Wolfe, J. A. 1975 Some aspects of plant geography of the
Northern Hemisphere during the Late Cretaceous and Ter-
tiary. Ann. Missouri Bot. Gard. 62, 264–279.
Young, K. R., Ulloa, C. U., Luteyn, J. L. & Knapp, S. 2002
Plant evolution and endemism in Andean South America:
an introduction. Bot. Rev. 68, 4–21.
Zachos, J. C. (and 27 others) 2004 Early Cenozoic extreme
climates: the Walvis Ridge transect. Proc. Ocean Drilling
Initial Rep. 208, 1–112.
Zeh, J. A., Zeh, D. W. & Bonilla, M. M. 2003 Phylogeography
of the harlequin beetle-riding pseudoscorpion and the rise of
the Isthmus of Panama
´
. Mol. Ecol. 12, 2759–2769.
GLOSSARY
BCI: Barro Colorado Island
CTFS: Center for Tropical Forest Sciences
GAARlandia: Greater Antilles and Aves Ridge
ITS: internal transcribed spacer
1622 R. T. Pennington and C. W. Dick Assembly of South American rainforest flora
Phil. Trans. R. Soc. Lond. B (2004)
on May 22, 2011rstb.royalsocietypublishing.orgDownloaded from
