Alternative models of vertebrate speciation inAmazonia: an overview

JUÈ RGEN HAFFERTommesweg 60, D-45149 Essen, Germany

Received 15 December 1995; revised and accepted 30 April 1996

The main hypotheses proposed to explain barrier formation separating populations and causing the

di�erentiation of vertebrate species in Amazonia are based on di�erent (mostly historical) factors, as

follows. (1) Changes in the distribution of land and sea or in the landscape due to tectonic move-

ments or sea-level ¯uctuations (Paleogeography hypothesis). (2) The barrier e�ect of Amazonian

rivers (River hypothesis). (3) A combination of the barrier e�ect of broad rivers and vegetational

changes in Northern and Southern Amazonia (River-refuge hypothesis). (4) The isolation of forest

blocks near areas of surface relief in the periphery of Amazonia during dry climatic periods of the

Tertiary and Quaternary (Refuge theory). (5) Competitive species interactions and local species

isolations in peripheral regions of Amazonia due to invasion and counterinvasion during cold/warm

periods of the Pleistocene (Disturbance-vicariance hypothesis). (6) Parapatric speciation across steep

environmental gradients without separation of the representative populations (Gradient hypothesis).

Several of these hypotheses are probably relevant to a di�erent degree for the speciation processes in

di�erent faunal groups or during di�erent geological periods. The paleogeography hypothesis refers

mainly to faunal di�erentiation during the Tertiary and in combination with the Refuge hypothesis;

Milankovitch cycles leading to global climatic-vegetational changes a�ected the biomes of the world

not only during the Pleistocene but also during the Tertiary and earlier geological periods. New

geoscienti®c evidence for the e�ect of dry climatic periods in Amazonia supports the predictions of

the Refuge theory.

Keywords: Amazonia; speciation; forest refugia; alternative models.

Introduction

The species-rich vertebrate faunas of the Amazon rainforest region inhabit vast levelplains from the Eastern base of the Andes to the Atlantic coast at the mouth of theAmazon River. Biologists interested in the origin of the innumerable species of theseforests suggested various historical or ecological factors to account for the spatial sepa-ration of populations, permitting their di�erentiation in allopatry (according to the modelof geographic speciation; Mayr, 1942, 1963). These factors include tectonic uplift orsubsidence of certain areas, the development or presence of river barriers, climatic ¯uc-tuations leading to changes in the vegetation cover in parts of Amazonia, or a combinationof these factors. Various hypotheses of speciation have been proposed based on the as-sumption that one of these modes of barrier formation was more e�ective than most or allothers. Below I summarize brie¯y the main characteristics of each of these hypotheses anddiscuss some of their relative merits: (1) Palaeogeography hypothesis, (2) River hypothesis,(3) River-refuge hypothesis, (4) Refuge hypothesis, (5) Disturbance-vicariance hypothesis,(6) Gradient hypothesis (Table 1). With the exception of the Refuge hypothesis, all of theother hypotheses have been introduced into the literature as casual suggestions by the

0960-3115 Ó 1997 Chapman & Hall

Biodiversity and Conservation 6, 451±476 (1997)

Ta

ble

1.

Hy

po

thes

eso

fg

eog

rap

hic

al

spec

iati

on

inA

ma

zon

iad

uri

ng

the

Ter

tiary

an

dQ

uate

rnary

per

iod

s(C

eno

zoic

),as

pro

po

sed

by

vari

ou

s

au

tho

rs

1.

Pa

leo

-

geo

gra

ph

y

hy

po

thes

is

2.

Riv

er

hy

po

thes

is

3.

Riv

er-r

efu

ge

hy

po

thes

is

4.

Ref

uge

hyp

oth

esis

5.

Dis

turb

an

ce±

vic

ari

an

ce

hyp

oth

esis

6.

Gra

die

nt

hyp

oth

esis

Fo

rest

red

uct

ion

du

rin

gd

rycl

ima

tic

per

iod

so

fth

e

Cen

ozo

ic

No

t

con

sid

ered

(irr

elev

an

t)

No

t

con

sid

ered

(irr

elev

an

t)

Wea

k;

on

lyp

eri-

ph

era

lp

ort

ion

s

inN

ort

her

nan

d

So

uth

ern

Am

azo

nia

a�

ecte

d

Str

on

g;

per

iph

eral

regio

ns

an

d

cen

tral

Am

azo

nia

a�

ecte

d

Wea

k;

on

lyp

eri-

ph

eral

po

rtio

ns

in

No

rth

ern

an

d

So

uth

ern

Am

azo

nia

a�

ecte

d

No

tco

nsi

der

ed

(irr

elev

an

t)

Ba

rrie

rs

sep

ara

tin

g

po

pu

lati

on

s

Co

nti

nen

tal

sea

s,p

late

au

s,

¯o

od

edp

lain

s

Riv

ers

(an

dth

eir

¯o

od

pla

ins)

Bro

ad

riv

ers

in

cen

tra

lA

ma

zon

ia

an

du

nfo

rest

ed

are

as

inth

e

hea

dw

ate

rre

gio

ns

Op

enfo

rest

san

d

no

nfo

rest

regio

ns;

river

slo

call

y

Eco

logic

all

y

un

suit

ab

le

fore

sts

Ste

epen

vir

on

-

men

tal

gra

die

nts

Ca

use

of

ba

rrie

r

Tec

ton

ic

mo

vem

ents

an

d/o

r

Dev

elo

pm

ent

of

riv

ers

or

ÿC

lim

ati

c¯

uct

uati

on

sd

uri

ng

the

Cen

ozo

icÿ!

ÿH

um

id/d

ryp

erio

ds

ÿ!C

old

/warm

per

iod

s

Su

ture

szo

nes

acr

oss

eco

ton

es

form

ati

on

sea

-lev

eld

isp

ersa

lo

fF

air

lyw

eak

Str

on

g[A

lso

¯u

ctu

ati

on

sfo

un

der

sa

cro

ssra

insh

ad

ow

e�ec

t

pre

-ex

isti

ng

riv

erb

arr

iers

du

eto

tect

on

ic

up

lift

an

d/o

r

sub

sid

ence

]

Au

tho

rsE

msl

ey(1

96

5)

for

bu

tter

¯ie

s

Sic

k(1

96

7),

Ca

pp

are

lla

(19

88

),

Her

shk

ov

itz

(19

77

)

Ay

res

(19

86),

Ay

res

eta

l.,

(1992),

Ca

pp

are

lla

(1991)

Ha�

er(1

969,

1974),

Van

zoli

ni

etal.

,

(1970),

Vrb

a(1

993)

Co

lin

vau

x(1

993)

En

dle

r(1

982)

452 Ha�er

respective authors without detailed discussions of patterns of di�erentiation and distri-bution of selected species groups over all of Amazonia. For this reason more data areavailable for review regarding the Refuge model than for any of the other hypotheses.Moreover, I emphasize the Refuge model in this article because it focuses on geographicalpatterns of paleoecological stability which is the unifying theme of this issue of Biodiversityand Conservation.

I am here concerned exclusively with the problem of the multiplication of species inAmazonia, i.e. the historical origin of tropical species richness. This problem needs to bedistinguished from ecological aspects, i.e. the problem of maintenance of tropical speciesrichness. In this respect ecologists study species packing and niche reduction, competition,predation and pest pressure. Other ecological factors include small-scale habitat mosaicsdue to the structure of the forest interior, gap phase dynamics, local topography (e.g. creekbeds in the forest and the intervening forest ¯oor) and ¯uvial dynamics (e.g. vegetationbelts along river beds that change their position rapidly). The analysis of these ecologicalphenomena of local patch dynamics of the environment and of habitat preferences ofanimal species contribute to an understanding in which way the multitude of sympatricspecies coexist in the same rainforest region. However, in view of the insu�cient spatialseparation of populations, these ecological factors do not contribute to an understandingof the historical problems of the origin of vertebrate species in Amazonia. These twodistinct and largely independent sets of problems referring, respectively, to the origin andcoexistence of rainforest species, are still today occasionally confused and mingled byecologists in their discussions of the `cause' of tropical species richness.

Alternative models of speciation

1. PALEOGEOGRAPHY HYPOTHESIS

Under this hypothesis, extant species and subspecies as well as their distribution patternsare thought to have originated when populations were repeatedly separated (and laterreconnected) due to paleogeographic changes in the distribution of forested land areas andcontinental seas in tropical South America. These changes were caused by fairly gentle(epeirogenic) movements, i.e. tectonic uplift or subsidence, over large regions or by morelocalized (orogenic) mountain building during the last 60 million years (Cenozoic era;Tertiary and Quaternary periods) and before; or these changes were caused by ¯uctuationsof world sea-level leading to the repeated ¯ooding (and subsequent falling dry) of low lyingareas in Amazonia and on the continental shelves. Tectonic movements and the formationof the surface relief in low lying and upland areas of South America (with or without sea-level changes) are considered as necessary and su�cient factors to explain the geographicalisolation of populations and resulting faunal di�erentiation (e.g. Croizat, 1976). Argu-ments of the Refuge hypothesis are used if climatic-vegetational changes are assumed inassociation with tectonic movements (see below).

The regional distribution and di�erentiation of the bird and butter¯y faunas inNorthwestern South America were interpreted on the basis of the geological history of theNorthern Andes, i.e. their early appearance during the Cretaceous as low-lying islandswhich were progressively uplifted during the Tertiary, thereby separating lowland faunasto the West and East of the Andes (e.g. Chapman, 1917, p 89; Emsley, 1965). It appearscertain that the early development and uplift of the Northern Andes in¯uenced decisivelythe di�erentiation of the ancestral faunas of the lowlands and the mountains in North-

Vertebrate speciation in Amazonia 453

western South America during the Tertiary. However, it remains di�cult to be morespeci®c, because repeated faunal immigrations probably again and again reached thetrans-Andean region from the Amazonian lowlands around the Northern end of theAndes during later geological periods (Ha�er, 1967, 1981). The origin of a particular trans-Andean endemic species cannot be ascribed to the isolation of an ancestral population inthe Paci®c lowlands due to the Miocene and Pliocene uplift of the Northern Andeswithout additional reasoning (as provided, e.g. by phylogenetic analyses of entire speciesgroups; Cracraft and Prum, 1988; Prum, 1988), because many trans-Andean endemicspecies and subspecies very probably originated from upper Amazonian immigrant pop-ulations that reached the Choco region of Paci®c Colombia and/or Middle America duringthe Quaternary, i.e. after the uplift and barrier formation of the Andes during the LateTertiary. In these latter cases, the Andes served as a pre-existing barrier and their earlieruplift is only indirectly linked to the origin of these younger (Quaternary) endemic trans-Andean taxa. Sea-level ¯uctuations and changing climatic-vegetational conditions inNorthern Colombia and Panama during the last 2 million years probably determined therepeated temporary continuity and discontinuity of the cis- and trans-Andean faunasthrough these regions. For this reason, there are many trans-Andean animal populationswhich are barely di�erentiated at the level of subspecies from their Amazonian repre-sentatives.

A portion of the present lower Amazon Valley, i.e. the region between Manaus andObidos, was apparently dry land during long periods of the Tertiary, when this areapermitted a direct faunal exchange between the land areas of the Guianan Shield to thenorth and the Brazilian Shield to the south (Fig. 1). Until the late Miocene, this `bridge'was a gentle divide between the broad upper Amazonian (SolimoÄes) basin to the West andthe comparatively small sedimentary basin in the lowermost Amazon Valley including theMarajo trough to the east (Mosmann et al., 1986). This low drainage divide disappearedduring the late Miocene tectonic movements leading to the continuity of the AmazonValley from the upper SolimoÄes region to the Atlantic Ocean.

During the Tertiary, no (or very little) e�ect is seen of the Iquitos Arch (Fig. 1) thatcrossed the upper Amazon basin from South to North during the previous Mesozoic andPaleozoic eras. It remains unknown whether the area over this arch has been above sea-level during any period of the geological past. A slight thinning of the Upper TertiaryPebas Formation over the former Iquitos Arch (Petri and FuÂlfaro, 1983, Fig. V-5) appearsinsu�cient to suggest a connection between the Guianan and Brazilian Shields in this areaduring the late Miocene. However, the paleogeographic situation is still very poorlyknown. Occasional marine incursions from the Paci®c Ocean and the Caribbean Seareached Western Amazonia during the Tertiary, when this region probably was coveredwith huge lakes, swamps and rivers. This area was closed o� from the Paci®c Ocean duringthe Middle Miocene due to the continued uplift of the Andes mountains which led to thereversal of the drainage pattern from a previous Western and Northwestern direction to an

Figure 1. Thickness maps of the Tertiary SolimoÄes Formation (above) and the Cretaceous Alter do

ChaÄo Formation (below) in Amazonia. Notice the drainage divide in central Amazonia (below

Manaus) during the Tertiary and the Iquitos Arch separating the Amazon basin from the sub-

Andean Peruvian-Acre basin during the Cretaceous; this arch was no longer active during the

Tertiary (from Mosmann et al., 1986).

454 Ha�er

Vertebrate speciation in Amazonia 455

Eastern direction. Complete continuity of the Amazon Valley to the Atlantic Ocean de-veloped during the Late Miocene, when the Northern Andes were strongly uplifted(Katzer, 1903; Nuttall, 1990; Hoorn, 1994; Hoorn et al., 1995; RaÈsaÈnen et al., 1995).

This rather simple paleogeographic setting of the greater Amazon region during theTertiary period (of ca 60 million years duration) does not seem to provide a su�cientlycomplex and changing geological background to have caused the intensive faunal evolu-tion and speciation that certainly took place on the stable land areas of the Guianan shieldto the north and the Brazilian shield to the South of the Amazon basin as well as inportions of the Amazon basin itself during the Tertiary. In any case, the geological dataare not su�ciently ®negrained to permit the details of physiographic changes in centralSouth America during this time interval to be analysed. Nevertheless, it appears likely thatother factors than the paleogeographic development of these regions determined thespeciation patterns of the Tertiary faunas in South America more e�ectively, such aspaleoclimatic ¯uctuations (caused by Croll-Milankovitch cycles) leading to repeatedecological vicariance events and the formation of Tertiary forest refugia (Refuge theory;see below).

The hypothesis of a huge lake supposedly covering most of Amazonia during the LatePleistocene±Early Holocene (Frailey et al., 1988) may also be mentioned as a paleogeo-graphic hypothesis. Disjunct pockets of rainforest located along the irregular margins of`Lago Amazonas' have been hypothesized (in a very general manner) as centres ofspeciation separated by ¯ooded lowland regions. It remains totally unknown what barrieror dam near the mouth of the present Amazon River should have prevented the waters ofthis lake, if it existed, from emptying into the Atlantic Ocean. No other author seems tohave adopted this model (unless we equate this lake with the less extensive interglacialembayments which probably covered portions of central Amazonia repeatedly duringQuaternary periods of raised sea-level; see Fig. 2C). Tuomisto et al. (1992) critically re-viewed the idea of `Lago Amazonas' labelling it as unsupported by geological ®eld data.

Without discussing any details or giving examples, Salo et al. (1986) and RaÈsaÈnen et al.(1990, 1992) suggested that the Late Tertiary to Quaternary development of the surfacerelief in upper Amazonia due to Sub-Andean foreland deformation may have causedspecies di�erentiation in the rainforest faunas when populations were isolated on gently

Figure 2. Three major speciation models for Amazonia (schematic representation).

(A) Riverine barrier hypothesis. Populations are separated by broad rivers; however, no geo-

graphical isolation exists in the headwater regions, where rivers cease to be barriers and dispersal is

fairly uninhibited (open dashed arrows). Stippled line follows outer limit of rainforest.

(B) River-refuge hypothesis. Rainforest is assumed to contract on broad latitudinal fronts during dry

climatic periods of the Cenozoic, presumably isolating forest animals in `semi-refugia' between the

broad lower courses of rivers (di�erent shading); the barrier e�ect of rivers during dry periods

probably was much reduced because of their narrower width.

(C) Refuge hypothesis (adapted from Ha�er, 1969, 1982). During dry climatic periods, isolated forest

refugia existed adjacent to pronounced surface relief in peripheral regions of Amazonia (Andes in the

West, Tepui mountains (T) and Guiana mountains (G) in the North and ParecõÂs mountains (P) in

RondoÃnia). Hatched area schematically indicates interglacial Amazonian embayment (sea-level

raised by several metres) which, however, was not contemporaneous with Pleistocene refuge for-

mation. Regions between forest refugia presumably were covered with variously extensive wooded

savannas, open liana forests and humid gallery forests.

456 Ha�er

Vertebrate speciation in Amazonia 457

uplifted arches (e.g. Serra do Moa Arch, Fitzcarrald Arch) separated by temporarily¯ooded basins (Pastaza-MaranÄoÂn Basin, Ucayali Basin, Acre Basin). While the occasional¯ooding of the low-lying basinal areas in eastern Peru may have contributed at times tothe separation of populations inhabiting the non-¯ooded areas, it is di�cult to visualize acomplete isolation of populations on these various gently uplifted arches leading to fullspeciation. On the other hand, the partial ¯ooding of the central Amazon Valley and ofportions of the coastal lowlands during interglacial high sea-level stands did probablycause or enhance the separation and di�erentiation of many animal populations innorthern and southern Amazonia, respectively.

2. RIVER HYPOTHESIS (OR RIVERINE BARRIER HYPOTHESIS)

Under this model, widespread and uniform ancestral populations of Amazonian animalsare assumed to have been divided into subpopulations and e�ectively isolated, when thenetwork of wide Amazonian rivers developed during the Late Tertiary and Early Qua-ternary periods (Fig. 2A). As Sick (1967, p. 501) stated, numerous ancestral Amazonianspecies supposedly had a vast and more or less uninterrupted distribution in a continuousforest region that was not (yet) traversed by large rivers. When these developed, manypopulations, especially those inhabiting the forest interior, became e�ectively separatedand deviated to the level of subspecies and species on opposite river banks (or on oppositesides of the ¯oodplains). Capparella (1988) similarly suggested that the development of theriver system in Amazonia dissected once-continuous lowland forest, leading to populationfragmentation and high diversity of avian species, especially in birds of the forest un-derstory. This basic River model of speciation appears to be ¯awed, because the devel-opment of the great forest and of its large rivers was probably one interrelated process.The forest cannot be envisioned without wide rivers and vice versa. Variants of thisoriginal River theory either refer only to the most recent readjustment of the river coursesor assume the active or passive dispersal of founders (individuals or groups of individuals)across the river barriers at any time after the formation of the rivers, interpreting thedisjunction of populations as secondary rather than primary.

The River hypothesis in its original form has been invoked on many occasions toexplain certain situations, but has never been discussed in any detail or tested on the basisof the distribution patterns of a large faunal group over the entire Amazon region. Fieldnaturalists observed since the last century that the Amazon River and some of its tribu-taries separate, at least for some distance, the ranges of many forest interior species andsubspecies of animals (Wallace, 1853; Bates, 1863 [for various groups of animals];Hellmayr, 1910, 1912; Snethlage, 1913; Mayr, 1942, p. 228; Sick, 1967 [for birds]; Hers-hkovitz, 1977; Ayres, 1986; Ayres and Clutton-Brock, 1992 [for primates]).

There is no question that the broad lower portions of many Amazonian rivers, often inconjunction with their much wider ¯oodplains, e�ectively separate populations of manyforest animals, thus causing the development or enhancement of genetic di�erences of theseparated though still conspeci®c populations. This has been demonstrated recently byCapparella (1988) on the basis of his study of allozymic di�erences among river-separatedforest bird populations. However, these genetic di�erences possibly disappear clinallytoward the headwater region of the respective rivers where the latter cease to be e�ectivebarriers and more or less uninhibited gene ¯ow connects the broadly intergrading popu-lations which, on the other hand, are indeed e�ectively separated by the wide lower

458 Ha�er

stretches of the same rivers and their ¯oodplains (Fig. 2A). Alternatively, when a wideriver valley separates two closely related taxa, these often are in direct contact in theheadwater region, where one of them crossed the narrower river course resulting in avariously extensive overlap zone (`good' species), a hybrid zone (subspecies) or in geo-graphical exclusion of their ranges along an abrupt parapatric contact zone without hy-bridization (paraspecies). Such contact zones (indicating lack of geographical isolation)prevent the direct application of the River theory. The question is still unsolved whetherthe barrier e�ect of Amazonian rivers during the geological past has often been su�cientlye�ective to have caused species di�erentiation to occur in animals, or whether this isolatione�ect in many or most cases was rather localized, leading to the development of intra-speci®c di�erences of certain populations on opposite sides of the lower stretches ofAmazonian river valleys, initiating the process of speciation in these cases.

Many authors invoking the River hypothesis of speciation overlooked and left undis-cussed the problems associated with the lack of spatial separation of populations in theheadwater regions. The question is not whether wide rivers are barriers to gene¯ow forunderstory birds and whether genetic di�erences have evolved in populations separated bywide river courses in central Amazonia. This appears obvious (as in many populationsseparated by other kinds of barriers on continents). Rather, the question is whether thebarrier e�ect of rivers in Amazonia has been su�ciently e�ective to have led routinely tofull speciation in understory birds and other animal groups (despite the lack of isolation inthe headwater regions of the rivers).

Future genetic studies of river-separated taxa may permit the distinction of certainalternatives:

(a) Both taxa are each other's closest relatives in a population phylogeny; in this case theymay be connected through continuous populations in the headwater region (River theory)or through a narrow hybrid zone (River theory, River-refuge theory or Refuge theory). Incase of a distributional gap in the headwater region one taxon may be derived from theother by jump dispersal across the river or through di�erentiation under the River-refugemodel.(b) Both taxa are only distantly related; this situation would favour an interpretationbased upon the Refuge model and reject most alternatives under (a) above

I list below a number of di�culties with the River theory of speciation which are usuallynot considered by authors favouring this hypothesis;

(1) The frequent transfer of an extensive portion of land to the opposite side of a river eachtime a meander loop is cut o� or a new river course carved out within the ¯oodplain. Inthis way, even poorly dispersing animals of the ¯oodplains are routinely `transported'passively across most small rivers and even large streams such as the SolimoÄes and theAmazon rivers.(2) The lack of geographical separation of populations in the forested headwater regionswhere the rivers cease to be barriers.(3) A number of representative taxa of the forest interior whose ranges are separated byrivers occupy (as uniform and phenotypically undi�erentiated populations) extensive areaswhich are traversed by larger rivers than those that separate the respective ranges of theserepresentatives. Future analyses will demonstrate whether or not many of such river-separated and uniform populations exhibit strong genetic di�erences.

Vertebrate speciation in Amazonia 459

(4) The occurrence of numerous secondary contact zones between Amazonian subspeciesand species of birds and other animals in continuous terra ®rme forest regions (theirlocations, in many cases, being unrelated to large rivers which these contact zones cross atright angles).

The problem of speciation in birds inhabiting river-created vegetation zones and instrong-¯ying canopy birds and other animals that readily cross broad rivers is left unex-plained by the River hypothesis whose application many authors restricted to birds of theforest understory. Among canopy birds are the large Ramphastos toucans that undertakeextensive movements in Amazonia during certain years of environmental stress. Theconspicuously reduced barrier e�ect of large rivers during the glacial (dry) periods oflowered sea-level of the Pleistocene also needs to be taken into consideration. During theseperiods the rivers were much narrower than today and ¯owed in the central deep portionof their current beds.

The above aspects prevent an application of the River theory as a likely model for theorigin of several ecological groups of birds including many, although not all, birds of theforest understory. Capparella (1988) and Silva et al. (1995) suggested that, for this groupas a whole, the River theory is a feasible model to explain the current patterns ofdi�erentiation and distribution without, however, discussing any examples from a set ofunderstory birds studied across the entire Amazon region. Capparella (l.c.) referred to anumber of bird populations of upper Amazonia and in the one case that Silva et al.(l.c.) mentioned (Hylexetastes), one taxon surrounds the wide Rio TapajoÂs in theheadwater region and is separated from its eastern representative by a 450 km-wide gapwith no records (where both may meet). Understory birds also form contact zoneswhose locations in certain regions of Amazonia are unrelated to river courses (Ha�er,1997).

FjeldsaÊ (1994, p. 217) dismissed the River theory concluding that `probably most of thedi�erentiation happened by isolation in habitat pockets along the Amazon Basin pe-riphery. As the species later dispersed, the rivers became natural `sutures' between some ofthe sister taxa'. In certain arboreal spiny rats, haplotype sharing across the Jurua River inwestern Amazonia was greater at its mouth than in the headwaters; moreover, twohaplotype clades uncovered correspond to headwaters versus mouth areas, not to oppositesides of the river, as would be expected by the River hypothesis (Patton et al., 1994).However, the meandering Rio Jurua that hardly represents a barrier for forest birds, is nota good example to test this theory.

I do not mean to say that Amazonian rivers were unimportant, but they appear to havebeen overrated as barriers. The broad lower courses of only a few large rivers represente�ective barriers to a portion of the vertebrate fauna, especially species of the forestinterior (many of which, however, surround the river barriers in the headwater regions).Generally speaking, the numbers of species whose ranges are delimited by rivers increasewith the width of Amazonian rivers. The broad Amazon River itself, of course, is a strongbarrier, although the e�ect seen in extant faunas may be in part the result of the separationof North and South Amazonian regions by the broad interglacial (brackish) Atlanticembayment that occupied central Amazonia when world sealevel was somewhat higherthan at present. Among a sample of 360 forest bird species, the upper Amazon (RioMaranoÂn) delimits the ranges of fewer than 20 species (Ha�er, 1978, 1993a). Goingdownstream this number increases along the middle Amazon (Rio SolimoÄes) to more than

460 Ha�er

50 species and reaches a total of over 150 species along the lower Amazon River below themouth of the Rio Negro. The barrier e�ect of the Amazon River decreases again near itsmouth where large islands like MarajoÂ, Mexiana and others facilitated increased avifaunalexchange from North to South (and vice versa) that was probably further enhanced by therepeated glacial lowering of world sealevel. The broad lower courses of the Amazoniantributary streams Rio Negro, Rio Madeira, Rio TapajoÂs and Rio Tocantins delimit theranges of 20 to 70 species in the above sample. The study of Ayres (1986) and Ayres andClutton-Brock (1992) demonstrated that the barrier e�ect of Amazonian rivers for pri-mates is comparable to that for birds. The decrease of the barrier e�ect of the AmazonRiver going upstream toward the Rio MaranÄoÂn is conspicuous. On the other hand, thebarrier e�ect for primates sharply increases below the mouth of the Rio Negro but de-creases again near the mouth of the Amazon.

3. RIVER-REFUGE HYPOTHESIS

The model that I designated as River-refuge hypothesis (Ha�er, 1993a,b) combines aspectsof the River hypothesis and of the Refuge hypothesis of faunal di�erentiation. Animalpopulations have been presumabaly isolated in `semi-refugia' separated by a combinationof the broad lower courses of several Amazonian rivers (plus their ¯oodplains) and byextensive, ecologically unsuitable terrain in the headwater regions of Northern andSouthern Amazonia that were more or less unforested during dry glacial climatic periodswhen the zone of tropical forests supposedly was contracted on broad latitudinal fronts(Ayres 1986; Ayres and Clutton-Brock 1992; Capparella 1991). This hypothesis (Fig. 2B)should not be included under the same designation as the River hypothesis, because thee�ect of repeated climatic-vegetational changes is not required under the latter hypothesis,whereas the e�ect of such changes is an essential part of the River-refuge hypothesis (andof the Refuge hypothesis; see below). The Amazon forest region contracted in a north-south direction under the River-refuge hypothesis but did not fragment. According to theRefuge theory, climatic-vegetational changes also a�ected central Amazonia, leadingto fragmentation of the Amazon forest. The designation `River-forest contractionhypothesis' would also be feasible for the former model. However, I prefer the label`River-refuge hypothesis', because this model is somewhat intermediate between the Rivermodel and the Refuge model and draws arguments from both. The River-refuge hy-pothesis was proposed on the basis of the patterns of distribution and di�erentiation ofAmazonian primates (Ayres, 1986; Ayres and Clutton-Brock, 1992) and a study of certainbird species of the rainforest understory (Capparella, 1991). On the other hand, Cerqueira(1982) had earlier interpreted the distribution patterns of Amazonian mammals on thebasis of the Refuge theory.

Assuming, as a ®rst approximation, a more or less uniform reduction of humidity andrainfall over Amazonia during dry climatic periods probably would not only lead to acontraction of the Amazon forest from the North and South, but would also cause theseparation of upper Amazonian forest blocks from lower Amazonian forest blocks alongthe conspicuous dry transverse belts that cross Southwestern and central Amazonia fromSoutheast to Northwest (Van der Hammen and Absy, 1994) (see Fig. 3). The coarsegeological strata underlying the rainforest near Pitinga (250 km north of Manaus) and inthe TapajoÂs region (see below) appear to corroborate this assumption thus favoring theRefuge model over the River-refuge model.

Vertebrate speciation in Amazonia 461

Fig

ure

3.

Ma

po

fa

nn

ua

lra

infa

llin

cen

tra

lS

ou

thA

mer

ica

at

pre

sen

t(l

eft)

an

dare

as

cover

edw

ith

hu

mid

rain

fore

std

uri

ng

ad

rycl

imati

cp

erio

do

f

the

pa

sta

ssu

min

ga

gen

era

l1

00

0m

m(4

0%

)re

du

ctio

no

fa

nn

ua

lra

infa

ll(r

ight)

;fr

om

Van

der

Ham

men

etal.

(1994).

462 Ha�er

4. REFUGE HYPOTHESIS

The Refuge model holds that forest and non-forest biomes changed continuously in dis-tribution during the geological past, breaking up into isolated blocks and again expandingand coalescing under the varyingly dry to humid climatic conditions of the Cenozoic(Tertiary and Quaternary periods). It is postulated that there have been many di�erentperiods of the formation of forest refugia and non-forest refugia during the peaks of dryand wet climatic periods, respectively, over the last 60 million years. During dry climaticperiods, extensive humid forests probably existed in fairly large regions of the Amazonianlowlands, where enough surface relief was present to create rainfall gradients, e.g. near therising Andes, around the mountains of Southern Venezuela and the Guianas that werebeing uplifted and eroded during the Tertiary, as well as in RondoÃnia (to the North of theParecõÂs mountains in central Brazil) and in the hilly areas of Eastern Para (Ha�er, 1969,1974, 1982; Vanzolini and Williams, 1970; Vanzolini, 1973; Prance, 1982). Open forestsand gallery forests probably existed in the regions between the postulated forest refugia,where variously extensive wooded savannas may, at times, have dominated the landscape(Fig. 2C). Gallery forests, of course, also served as refugia for some tropical forest biotasin open regions during dry climatic phases (Meave et al., 1991, 1994). Their signi®cance ascentres of di�erentiation and speciation, however, remains open. It also remains unknownwhether the above mentioned forest refugia, during the height of any of the dry periods ofthe Cenozoic, were reduced spatially to such an extent that they were represented essen-tially by gallery forests along rivers in otherwise more or less deforested areas (even in thelowlands near pronounced surface relief in peripheral regions of Amazonia). Theoreti-cally, this would represent an extreme situation under the Refuge hypothesis.

The origin of the complex hilly or mountainous surface relief in peripheral regions ofAmazonia due to tectonic uplift and erosional activity (Andes, tepui mountains insouthern Venezuela, table mountains in central Brazil, etc.) is seen as a precondition forthe formation of forest refugia. The geological movements leading to the formation of themountains around Amazonia by themselves (without the e�ect of climatic-vegetationaloscillations) and the e�ect of climatic ¯uctuations by themselves (without the geomor-phological structures) are each considered insu�cient to explain the geographical isolationand speciation of Amazonian vertebrate populations. Rather, the combination of a(developing) complex geological structure and surface relief in peripheral regions ofAmazonia (creating geographical rainfall gradients) together with the e�ect of globalclimatic ¯uctuations during the Cenozoic are the basic factors under the Refuge hypothesisthat caused ecological vicariance of vertebrate populations.

In addition to global climatic oscillations also more localized tectonic movements led tothe formation of ecological refugia (Ha�er, 1990): in di�erent parts of the world (Andes,Brazil, Himalayas, etc), rising mountain ranges caused climatic changes in the lowlands atthe leeward side of the mountains which led to extensive vegetational changes in theseareas (rain shadow e�ect). Refugia originated for di�erent geological and paleoclimaticalreasons during many periods of the history of the earth. Regarding the interaction of thedegree of topographic diversity of a landscape, climatic ¯uctuations and speciation see alsothe discussions by FjeldsaÊ (1992), Vrba (1993) and FjeldsaÊ and Lovett (1997).

Plant and animal populations isolated in the more or less restricted forest refugia eitherbecame extinct, survived without much change or di�erentiated to the taxonomic level ofsubspecies or species before the survivors came into secondary contact with their

Vertebrate speciation in Amazonia 463

representative populations of other refugia during a following favourable expansive phase.If at this time a refuge population of an ancestral species had reached sexual and ecologicalisolation from its neighbouring allies, it could disperse widely in the now-continuoushabitat before its extensive range was fragmented again during the next adverse climaticphase. Some species remained more or less restricted to the area of their origin withoutexpanding their ranges widely in the continuous forest region (in this way characterizingtoday a number of `areas of endemism' in Amazonia).

General reviews of this macroevolutionary `habitat theory' were also published by Vrba(1992, 1993). It remains unknown whether extinction of species or generation of specieswas prevalent during particular periods of refuge formation, leading to a reduction orincrease of overall regional species diversity, respectively.

Investigators in the tropics analysed data sets from two independent sources ± ®rst,from geoscienti®c studies (palynology, geomorphology, paleoclimatology, rainfall distri-bution in relation to the surface relief ) and, second, from biogeographic studies (e.g. themapping of areas of high biotic endemism and regions of secondary contact zones amongclosely related taxa). The coincidence between conclusions derived from these two inde-pendent sets of data led to the formulation of the general hypothesis (Ha�er, 1969;Vanzolini and Williams, 1970; Simpson and Ha�er, 1978; Ha�er, 1982; Cerqueira, 1982;Whitmore and Prance, 1987). It is incorrect when recent authors state that the Refugehypothesis was initially based on biogeographical data alone.

Rainfall over Amazonia is not evenly distributed but heavily in¯uenced by the surfacerelief and other factors. Assuming, as a ®rst approximation, an even reduction of rainfallby about 1000 mm during certain dry climatic periods, this would lead to the disruption ofthe presently continuous forest region into several isolated forest blocks, as predicted bythe Refuge hypothesis (Fig. 3).

A number of problems with the Refuge theory that need to be taken into considerationmay be summarized, as follows. (1) Large gallery forests could promote gene ¯ow betweenrefugia and reduce the amount of di�erentiation. (2) Humid forests between refugiamay, at times, have been replaced by dry forest (rather than nonforest vegetation). (3)Geoscienti®c evidence only suggests that there were climatic-vegetational changes inAmazonia, not that these changes caused speciation. (4) Only ecologically fairly narrowlyadapted forest species were a�ected, whereas populations of ecologically more ¯exiblespecies may not have been e�ectively isolated by the refugia. The fact that relatively youngspecies predominate in areas with a marked topography, while there is relatively littledi�erentiation seen in the ¯uvial basins (FjeldsaÊ, 1994, Fig. 3), appears not as a problemfor the Refuge model, because the postulated refugia were mainly associated with thetopographic relief in peripheral regions of Amazonia (base of the Andes, Tepui mountainsand mountains of central Brazil).

The vicariance analysis of 13 clades of Amazonian birds from four families supportedonly two patterns for the historical interrelationships of forest areas of endemism, i.e.primary upper/lower Amazonian vicariance and primary Guianan vicariance. As Prum(1988) stated, climatic-vegetational ¯uctuations are a likely cause of these series ofvicariance events.

The Refuge hypothesis proposed a plausible mechanism of faunal di�erentiation in thetropics and in higher latitudes through the formation of ecological refugia due to Croll-Milankovitch cycles during the Tertiary and Quaternary and superimposed on a globalcooling trend (see Discussion). This is a predictably reversible driving mechanism in the

464 Ha�er

order of tens to hundreds of thousands of years, consistent with the rate of biologicaldiversi®cation (Terborgh, 1992). The refugia presumably served as `species traps' and`species pumps'. This historical model appears capable of explaining the origin of the hightropical species diversity as compared to the less diverse faunas of the arctic and borealbiomes. The main factors are (a) the larger geographical area of tropical continentscompared to continental areas in the Temperate Zones and (b) the much longer time spansince the `accumulation' and diversi®cation of species began in the tropics during EarlyTertiary times (and before) compared to the Temperate Zones (which became temperateonly during the Late Tertiary global cooling trend; Shacklton et al., 1990). As a conse-quence, global climatic-vegetational ¯uctuations generated more isolated habitat frag-ments (refugia) during each of the many climatic reversals in the tropics than in higherlatitudes. By generating a larger number of isolated populations (and thus, on average,more new species) per unit of geological time, evolution proceeded `faster' in the tropics,although the speciation process itself, of course, has always proceeded in an identicalmanner in tropical and extratropical faunas of the world.

The basic assumptions under the Refuge hypothesis are (1) the fragmentation of theGuianan-Amazon forests into (few to many) more or less separated forest blocks ofdi�erent sizes during dry climatic phases of the Tertiary and Quaternary, and (2) thedi�erentiation of animal populations isolated in these refugia. The ®rst assumption is,strictly speaking, a geoscienti®c problem that will eventually be clari®ed by geoscienti®c(not biological) data. The second assumption will have to be evaluated in detail onceenough geoscienti®c data document the presumed fragmentation of the forests.

New evidence for the e�ect of dry climatic periods in Amazonia during the Quaternary

Although the data base documenting dry climatic periods and corresponding vegetationalchanges in Amazonia is still scarce, a number of interesting reports have been publishedin recent years (Fig. 4). Additional geological and geomorphological evidence fromAmazonia was reviewed earlier by Ab'Saber (1982) and Ha�er (1987). These data,together with the new results summarized below, may be considered as a geoscienti®c testof the Refuge hypothesis. It is remarkable that all geoscientists who published the resultsof their ®eld studies in recent years (whether they are geologists, geomorphologists, ge-ographers, palynologists, geochemists or paleontologists) agree on the basic premise of theRefuge hypothesis, i.e. a strong e�ect of vegetational shifts in Amazonia caused by periodsof dry climates during the geological past.

Large fossil dune ®elds have been discovered in northcentral Amazonia (Rio AracaÂ-Rio Branco region; Fig. 4, no.

9; Santos et al., 1993). The rainforest at Pitinga, ca 250 km north of Manaus (Fig. 4, no. 4) is underlain by coarse

and extremely poorly sorted sediments which have been deposited under semiarid climatic conditions and in the

absence of dense rainforest from most of this region (Veiga et al., 1988). The same interpretation applies to other

portions of the Brazilian Amazon region (XinguÂ, Teles Pires-Juruena, middle TapajoÂs and Northern RondoÃnia;

Bettencourt et al., 1988; Veiga et al., 1988). Similarly, Bibus (1983) had reported the widespread occurrence in the

middle Rio TapajoÂs region (Fig. 4, no. 3) and in the lowlands around the Serra do Cachimbo (Fig. 4, no. 2) of

coarse debris in surface depressions accumulated during a period of strong erosion when the Late Quaternary

climate was semi-arid and rainforest vegetation had largely disappeared from these regions. Geomorphological

observations in RondoÃnia (regions of Porto Velho and Humaita; Fig. 4, no. 5) by Emmerich (1988) also indicate a

semiarid climate and open vegetation in this portion of southern Amazonia during the Late Tertiary and dry

climatic phases of the Pleistocene. During the Late Pleistocene (last glacial stage), savanna vegetation was

widespread in the area to the southeast of Porto Velho, RondoÃnia, where Van der Hammen et al. (1994) studied

the pollen content of dated sections. Throughout the Acre Subbasin (region of the upper Rio PuruÂs and lower Rio

Acre; Fig. 4, no. 6), gypsum and aragonite precipitates associated with ®ne-grained sur®cial sediments indicate the

Vertebrate speciation in Amazonia 465

desiccation of an extensive ¯uvial±lacustrine system due to dry climatic conditions about 53 000 years ago, i.e.

during the last glacial cycle (Kronberg et al., 1991). Moreover, paleontological studies of fossil mammals led

Rancy (1991, 1993) to conclude that a vegetation consisting of wooded savannas and gallery forests was wide-

spread in upper Amazonia during some periods of the Pleistocene before dense rainforests again covered this area

completely (Fig. 4, no. 8).

The relatively small plateaus of the Serra dos CarajaÂs, State of Para (Fig. 4, no. 1) are covered today with open

canga vegetation and are surrounded on all sides by dense rainforest covering the slopes and intervening low-

lands. Geological and paleopollen analyses of a core collected in a swamp on one of the plateaus revealed four

periods of rainforest regression from this general region during the last 60 000 years (Absy et al., 1991).

In upper Amazonia, coarse gravels occur within terrace sediments along the Rio Caqueta in Southeastern

Colombia (Fig, 4, no. 7) and indicate temporary aggradational conditions markedly more torrential than those

currently prevailing along this river (Eden et al., 1982). These gravels may be interpreted to indicate periodic

phases of dry, or at least strongly seasonal, conditions in the Andean headwater region. On the basis of pollen

data from the middle Rio Caqueta region, Van der Hammen et al. (1994) concluded that, during drier intervals of

the middle Pleniglacial, savanna-caatinga-type vegetation could develop or extend somewhat locally. Paleocli-

matic data from two ice cores taken in the Andes of Peru led Thompson et al. (1995) to conclude that the Amazon

basin forest cover probably was markedly reduced (more patchy) during the Last Glacial Stage than it is today.

Figure 4. Location map of areas in Amazonia where additional paleoecological evidence for dry

climatic periods and associated vegetational changes of the Late Pleistocene has been discovered

in recent years. Further data from Amazonia are mentioned in the text. A rich data base for

Quaternary climatic-vegetational shifts is also available from northern South America and from

various portions of Brazil outside Amazonia (not indicated on this map). 1: Serra dos CarajaÂs; 2:

Serra do Cachimbo region; 3: Lower Rio TapajoÂs region; 4: Pitinga region; 5: Porto Velho and

Humaita region; 6: Rio Acre region; 7: middle Rio Caqueta region; 8: upper Rio Jurua region; 9: Rio

AracaÂ-Rio Branco region. Dotted line follows the approximate outer limit of the Guiana-Amazon

forest region prior to recent deforestation. See text for further details.

466 Ha�er

The results reviewed above indicate strong vegetational shifts in Amazonia due toalternating dry/humid climatic phases; in other words, the Refuge hypothesis so farwithstood the test by geoscientists (who failed to falsify its geoscienti®c premises). How-ever, presently available geoscienti®c data are still insu�cient to allow for the precisemapping of changes in distributions of forest and nonforest vegetation during the variousclimatic periods and, in particular, for the tracing of the size and location of areas ofremnant forests and savannas that presumably served as refugia for the Amazonian faunaand ¯ora during adverse climatic periods. Moreover, detailed knowledge on the shrinkingand expansion of the forest, by itself, does not falsify alternative theories of speciation. Asdiscussed below, other models of speciation are probably also relevant for the process offaunal diversi®cation in Amazonia.

Recent discussions of the Refuge theory

In view of much supporting geoscienti®c and biogeographic evidence, numerous authorsdiscussed the Refuge hypothesis favorably in interpretations of their own results of sys-tematic and biogeographic studies in the Neotropics, e.g. the contributors to several multi-author volumes (Prance, 1982; Prance and Lovejoy, 1985; Whitmore and Prance, 1987), aswell as Terborgh (1992) and Vanzolini (1992). Similarly, FjeldsaÊ (1994, p. 219) concludedthat Neotropical species probably `di�erentiate by isolation of relict populations in placeswhich remained ecologically stable, and within geologically well structured regions at theperiphery of an old biome' (such as Amazonia). Other authors directed critical commentsat distorted caricatures of this hypothesis stating, e.g., that the Refuge hypothesis refers toa single, fairly recent vicariance event, or that it claims the diversi®cation of the entireNeotropical biota is the result of habitat fragmentation exclusively during the Quaternary.Rather, Refuge hypothesis refers to the postulated origin of species in ecological refugiairrespective of the time period (Quaternary and Tertiary, as well as before). Other authorsbased certain 'tests' of the Refuge hypothesis on totally unrealistic criteria none of whichhad ever been stipulated as de®ning properties of this model (see Ha�er, 1993b, for adiscussion of some of these criticisms). As mentioned above, broad gallery forests andlocally humid conditions probably existed along major river valleys of those regions inAmazonia that were a�ected by generally dry climatic conditions. Therefore humid con-ditions, e.g., along the middle Amazon River during the last dry glacial stage, cannot beconsidered as evidence against the reduction of rainforest cover over Amazonia as a whole(contra MuÈller et al., 1995).

5. DISTURBANCE±VICARIANCE HYPOTHESIS

According to this hypothesis, `the prime environmental forcing of tropical forests in ice-age America was cooling rather than aridity ... The forests of the central Amazon wereprobably not markedly fragmented, though savanna regions at the periphery were prob-ably more extensive than now' (Colinvaux 1993, pp. 473, 485). During the glacial periodsof the Pleistocene, lowland biota are assumed to have inhabited the Amazonian bot-tomland (0 to 300 m above modern sea-level), where temperatures were ca 5°C, at times7°C, lower than at present. The peripheral portions of Amazonia that include the areas offaunal endemism are seen as dynamic borderlands between the uplands and the bottom-lands. In these regions species distribution and abundance oscillated throughout thePleistocene due to invasion and counterinvasion. These intense competitive species

Vertebrate speciation in Amazonia 467

interactions are assumed to have favored species isolations, thereby supposedly explaininghow these regions became centres of speciation and endemism.

Under this hypothesis, the peripheral areas of Amazonia are expected to be especiallyrich in endemic taxa because of the supposed environmental instability of these regions(rather than the implied ecological stability in the postulated remnant forests under theRefuge model). It is di�cult to follow Colinvaux's reasoning, because he did not take intoconsideration any of the geoscienti®c data that suggest dry climatic conditions in pe-ripheral and certain portions of central Amazonia (as reviewed above). Moreover, thishypothesis (based exclusively on pronounced temperature oscillations) refers only to theQuaternary and, in contrast to the Refuge hypothesis, is not applicable to the intensivefaunal di�erentiation during the generally warm Tertiary period.

In his wideranging review, Bush (1994) also stated that climatic cooling, rather thanaridity, was the factor driving a Pleistocene re-assortment of vegetation in Amazonia,although he did accept climatic drying (by about 20%) over Amazonia during glacialperiods, which led to the expansion of dry-adapted vegetation types into the transverseclimatic belts crossing lower Amazonia from Southeast to Northwest in the Manaus-SantareÂm region, as well as crossing Southwestern Amazonia along the border region ofPeru and Brazil. In this way, Bush (l.c.) accepted a separation of humid rainforest blocksin the Guianas and at the mouth of the Amazon River from the upper Amazonian forests(as discussed under the Refuge hypothesis). Moreover, he speaks of species that `onlysurvived in areas that were optimal'. Apparently, Bush (1994, p. 13) had species-speci®crefugia in mind when he stated:

`If the cooling and drying stressed individual species to the point where they went extinct over parts of their range

and only survived in areas that were optimal, a mechanism for allopatric speciation emerges. Each time a

population was stressed by climatic change, and this may occur several times for each Quaternary glaciation, the

chance of it becoming fragmented increases, especially where it is in competition with species that are better

adapted to the prevailing conditions. Species that contained considerable environmentally-related genotypic

variation may have contracted into optimal locations for each genotype. ... The presence of invading, cold-

adapted, or dry adapted taxa could have resulted in local competitive exclusion of some lowland taxa, or

genotypes, further increasing the possibility of isolation and allopatric speciation.'

It is evident that this generalized interpretation uses arguments derived from both theRefuge hypothesis (dry/humid cycles) and the Disturbance-vicariance hypothesis (cold/warm cycles). As in the model of Colinvaux (1993), the question arises under Bush'sinterpretation as to the speciation mechanism during the Tertiary, when primarily dry/humid cycles (but basically no cold/warm cycles) occurred.

6. GRADIENT HYPOTHESIS

This model predicts parapatric speciation across steep environmental gradients (bound-aries) without the separation of the representative populations (Endler, 1982). Centres ofdiversity correspond with zones of relative environmental uniformity, and zones of changecorrespond with zones of environmental change.

Mayr and O'Hara (1986) refuted this hypothesis with respect to the fauna of tropicalAfrica. Similarly, a biogeographical data set from South America does not support theGradient hypothesis (Cracraft et al., 1988; Prum, 1988), whereas Mallet (1993) found `it isnot easy to exclude parapatric di�erentiation' as a mode of speciation in AmazonianHeliconius butter¯ies (see, however, Futuyama and Shapiro, 1995; Brower, 1996).

468 Ha�er

Discussion

The di�erent historical hypotheses proposed by various authors to account for the processof vertebrate speciation in Amazonia emphasize the biogeographic e�ect of tectonicmovements and mountain building, the barrier e�ect of rivers, the changing compositionand distribution of animal and plant communities due to climatic-vegetational ¯uctuationsduring the Cenozoic, the e�ect of environmental gradients, or a combination of thesefactors, that resulted in the geographical isolation and speciation of animal populations.Only the Gradient hypothesis is based on the model of parapatric speciation; all otherhypotheses are based on the generally accepted theory of geographic (allopatric) speciation.Each of the models of allopatric speciation are probably relevant to a di�erent degree forthe speciation process in di�erent faunal groups or during di�erent geological periods.

Most biogeographers probably agree that the paleogeographical changes in the dis-tribution of land and sea, as well as the uplift of the Andes, Tepui mountains and otherranges, in¯uenced the early evolution of the faunas during the Tertiary. However, theorigin of many or most extant species and their distribution patterns probably cannot beunderstood solely on the basis of such geological processes. Simultaneous ecologicalvicariance events through global climatic-vegetational ¯uctuations of the Cenozoic leadingto the repeated separation of ecologically more or less stable `refugia' within or near areasof complex surface relief (Andes and peripheral areas of Amazonia) probably explain theorigin of most species and species groups as we see them in the Neotropical region today.The rainshadow e�ect during more localized processes of mountain building probably alsoled to the formation of refugia in some regions, thus contributing to the faunal di�eren-tiation.

Recently, FjeldsaÊ (1994) presented a similar analysis emphasizing the indirect signi®-cance of the geomorphological processes of mountain building and the direct e�ect ofpaleoclimatic ¯uctuations (`radiations ... during the last 6 MY period with increasingclimatic ¯uctuations of the Croll-Milankovitch type'; FjeldsaÊ 1994, pp. 210±211). Spe-ciation in the Andes took place in populations which were widely separated in ecologicallystable mountainous refugia (not along ecological gradients of mountain slopes followingthe Gradient hypothesis; Patton et al., 1992). The pattern under present-day (humid)climatic conditions, with long linear distributions along the Amazonian (eastern) slope ofthe Andes and dense altitudinal replacements of closely related species, is a secondary state(FjeldsaÊ 1992, 1994, p. 219). In other words, the geomorphological processes alone cannotexplain the separation of populations and the speciation process in lowland or montanefaunas of the tropics that directly depend on ecological vicariance due to climatic-veg-etational changes.

Tests of the various hypotheses are not yet fully conclusive. However, biogeographicalstudies of Afrotropical birds, vicariance analyses of Amazon forest birds and biochemicalstudies of Andean mice falsi®ed parapatric speciation (as assumed under the Gradienthypothesis) to be a feasible speciation model (Mayr and O'Hara, 1986; Prum, 1988; Pattonet al., 1992). The new geoscienti®c evidence for dry climatic periods collected in variousportions of upper and central Amazonia con®rmed predictions of the Refuge hypothesis(as discussed above). The interpretation of speciation pulses based on the e�ect ofMilankovitch cycles predict similar ages of di�erent `generations' of species across di�erentgroups of animals. Such analyses may be capable of testing the Refuge hypothesis when asu�cient number of phylogenetic studies become available.

Vertebrate speciation in Amazonia 469

Contact zones between sharply di�erentiated and geographically representative taxa ofbirds inhabiting the forest interior and the forest canopy reveal striking faunal disconti-nuities in a continuous rainforest environment (Ha�er, 1974, 1993a, 1997). An historicalinterpretation of such zones of geographical replacement as zones of secondary contactbetween the respective taxa is probable. This interpretation implies large-scale separationof the respective bird populations during one or more periods of geographic isolationduring the geological past. Many of these contact zones are independent of the rivercourses in peripheral and central portions of Amazonia and, therefore, also support theRefuge hypothesis (ca 20 such contact zones between 40 well di�erentiated taxa mapped insouthern Amazonia so far; Ha�er, 1997). It is true that, in general, range limits of manyAmazonian species are probably ecologically determined. However, the above argumentrefers to coinciding range limits of members of numerous pairs of closely related taxa thatexclude each other over great distances along contact zones that cross ecologically varyingterrain. An ecological explanation for the origin and location of such contact zones ap-pears highly unlikely.

I recommend distinguishing the various speciation models under separate designations,especially the Paleogeography hypothesis, the Refuge hypothesis, the River-refuge hy-pothesis and the River hypothesis (Table 1) to minimize the potential of misunderstandingin discussions of the historical biogeography of Amazonia.

A hierarchy of timeless environmental disturbance cycles (`Time's cycle') characterizethe nature of Amazonia, from short-term treefall cycles and ¯uvial cycles to long-termclimatic and paleoclimatic cycles (Fig. 5). The latter cycles, in conjunction with the

Figure 5. Spatial and temporal interrelations of environmental disturbance and change, biotic re-

sponse and vegetation patterns (after Delcourt et al. in Di Castri and Hadley, 1988).

470 Ha�er

directional processes of mountain building and erosion responsible for the development ofpronounced surface relief in peripheral regions of Amazonia, permit an understanding ofthe multiplication of species and the origin of Amazonian species diversity (an example of`Time's arrow', Fig. 6). Most of the hypotheses proposed depend on conspicuous envi-ronmental instability over Amazonia as the driving mechanism of speciation, except forthe original form of the River hypothesis and for the Gradient hypothesis (both of which,however, do not appear to be applicable to a major portion of the Amazonian vertebratefauna).

Recent geological studies in various parts of the world have established the fact thatastronomical Milankovitch cycles causing global climatic-vegetational ¯uctuations in¯u-enced the continuously changing distribution of forest and nonforest vegetation on earthnot only during the Ice Ages of the last 2 million years (Quaternary) but also during theentire Tertiary period and before. These cycles caused sea-level oscillations, rhythmicfacies changes of Mesozoic and Tertiary sedimentary strata, and climatic-vegetationalshifts on the continents (Herbert and Fisher, 1986; Olsen, 1986; Bartlein and Prentice,1989; Berger et al., 1989; Bennett, 1990). These geologically rather short-term (high-fre-quency) oscillations were superimposed on a gradual cooling trend of the earth's climatesince the beginning of the Cenozoic. The latitudinal thermal gradient steepened during the

Figure 6. Time's cycle and Time's arrow in the history of Amazonia. Schematic representation.

(Left) Hierarchic disturbance cycles from treefall cycles (1st order) and ¯uvial cycles (2nd order) to

climatic and paleoclimatic cycles (3rd order) illustrate Time's cycle. These disturbance cycles,

through generating habitat heterogeneity, contribute to the maintenance of the high tropical species

richness. (Right) Time's arrow of genealogy along contingent evolutionary pathways. New species

lineages are generated by paleoclimatic-vegetational cycles (symbolized by concentric stippled lines)

which explain the origin of the high tropical species richness. See text for further details.

Vertebrate speciation in Amazonia 471

course of the Late Tertiary when annual average temperatures increased in the tropics andthe summer temperature decreased in the high latitudes. It may be mentioned that, at leastfor the Andes (Hooghiemstra et al., 1994), evidence exists that the impact of the 100 000-year cycles became prominent only in the Pleistocene.

The cyclic formation and disappearance of forest and nonforest refugia due to the e�ectof global Milankovitch cycles during the Tertiary and Quaternary probably underlies, asa predictably reversible speciation mechanism, much of the organic evolution on thecontinents (Terborgh, 1992). This statement is not intended to diminish in any way thegeneral biogeographic signi®cance of paleogeographic changes in the distribution of landand sea, as well as geomorphological changes in South America and other regions due totectonic movements during the course of the geological history or of the far-reaching e�ectof continental rifting and subsequent continental drift. An understanding of the signi®-cance of these vicariance processes as well as of jump dispersal between islands in theoceans and between ecological `islands' on the continents is needed for a complete analysisof the zoogeographical history of the various groups of animals, in addition to that of thebiogeographic e�ect of climatic-vegetational ¯uctuations.

Acknowledgments

I thank F. FjeldsaÊ (Copenhagen) for his invitation to attend the workshop on `Biodiversityand Stability' where I presented this contribution. I am also grateful to Else and JonFjeldsaÊ for their hospitality during the workshop and to the Danish Research Council fora travel grant. J. FjeldsaÊ and an anonymous referee kindly reviewed the manuscript andmade a number of useful suggestions for improvement.

References

Ab'Saber, A.N. (1982) The paleoclimate and paleoecology of Brazilian Amazonia. In Biological

Diversi®cation in the Tropics (G.T. Prance, ed.) pp. 41±9. New York: Columbia University

Press.

Absy, M.L., Cleef, A., Fournier, M., Martin, L., Servant, M., Si�edine, A., Ferreira da Silva, M.,

Soubies, F., Suguio, K., Turcq, B. and Van der Hammen, T. (1991) Mise en eÂvidence de quatre

phases d'ouverture de la foret dense dans le sud-est de l'Amazonie au cours des 60 000 dernieÁres

anneÂes. PremieÁre comparaison avec d'autres reÂgions tropicales. C.R. Acad. Sci. Paris 312, seÂr.

II: 673±8.

Ayres, J.M.C. (1986) Uakaris and Amazonian ¯ooded forest. PhD dissertation, University of

Cambridge, Cambridge, England.

Ayres, J.M.C. and Clutton-Brock, T.H. (1992) River boundaries and species range size in Ama-

zonian primates. Am. Naturalist 140, 531±7.

Bartlein, P.J. and Prentice, I.C. (1989) Orbital variations, climate and paleoecology. Trends Ecol.

Evol. 4, 195±9.

Bates, H.W. (1863) The Naturalist on the River Amazon. London: Murray.

Bennett, K.D. (1990) Milankovitch cycles and their e�ects on species in ecological and evolutionary

time. Paleobiology 16, 11±21.

Berger, A., Loutre, M.F. and Dehant, V. (1989) Pre-Quaternary Milankovitch frequencies. Nature

342, 133.

Bettencourt, J.S., Muzzolon, R., Payolla, B.L., Dall'Igna, L.G. and Pinho, O.G. de (1988) DepoÂsitos

estanõÂferos secundaÂrios de regiaÄo central de RondoÃnia in Principais DepoÂsitos Minerais do

Brasil, vol. 3, Brasilia.

472 Ha�er

Bibus, E. (1983) Die klimamorphologische Bedeutung von stone-lines und Decksedimenten in

mehrgliedrigen Bodenpro®len Brasiliens. Zeitschr. Geomorph. N.F., Suppl.-Band 48, 79±98.

Brower, A.V.Z (1996) Parallel race formation and the evolution of mimicry in Heliconius butter¯ies:

a phylogenetic hypothesis from mitochondrial DNA sequences. Evolution 50, 195±221.

Bush, M.B. (1994) Amazonian speciation: a necessarily complex model. J. Biogeogr. 21, 5±17.

Capparella, A. (1988) Genetic variation in Neotropical birds: implication for the speciation process.

Acta XIX Congr. Intern. Ornith. (Ottawa 1986), vol. 2, 1658±64.

Capparella, A. (1991) Neotropical avian diversity and riverine barriers. Acta XX Congr. Intern.

Ornith. (Aukland 1990), vol. 1, 307±16.

Cerqueira, R. (1982) South American landscapes and their mammals. In Mammalian Biology in

South America (M.A. Mares and H.H. Genoways, eds) pp. 53±75. Spec. Publ. 6, Pymatuning

Laboratory of Ecology, University of Pittsburgh.

Chapman, F.M. (1917) The distribution of bird-life in Colombia; a contribution to a biological

survey of South America. Bull. Am. Mus. Nat. Hist. 36, 729 pp.

Colinvaux, P. (1993) Pleistocene biogeography and diversity in tropical forests of South America. In

Biological Relationships between Africa and South America (P. Goldblatt, ed.) pp. 473±99. New

Haven: Yale University Press.

Cracraft, J. and Prum, R.O. (1988) Patterns and processes of diversi®cation: speciation and his-

torical congruence in some Neotropical birds. Evolution 42, 603±20.

Croizat, L. (1976) BiogeografõÂa analõÂtica y sinteÂtica (`PanbiogeografõÂa') de las AmeÂricas; tomo 1±2.

Biblioteca de la Academia de Ciencias FõÂsicas, MatemaÂticas y Naturales, vol. 15. Caracas.

Di Castri and Hadley, M. (1988) Enhancing the credibility of ecology: interacting along and across

hierarchical scales. GeoJournal 17, 5±35

Eden, M.J., McGregor, D.F.M. and Morelo, J.A. (1982) Geomorphology of the middle CaquetaÂ

basin of eastern Colombia. Zeitschr. Geomorph. N.F. 26, 343±64.

Emmerich, K.H. (1988) Relief, BoÈden und Vegetation in Zentral- und Nordwest-Brasilien unter

besonderer BeruÈcksichtigung der KaÈnozoischen Landschaftsentwicklung. Frankfurter Geowiss.

Arbeiten, Serie D (Phys. Geogr.) 8, 218 pp.

Emsley, M.G. (1965) Speciation in Heliconius (Lep., Nymphalidae): Morphology and geographic

distribution. Zoologica (N.Y.) 50, 191±254.

Endler, J. (1982) Pleistocene forest refuges: fact or fancy? In Biological Diversi®cation in the Tropics

(G.T. Prance, ed.) pp. 179±200. New York: Columbia University Press.

FjeldsaÊ, J. (1992) Biogeographic patterns and evolution of the avifauna of relict high-altitude

woodlands of the Andes. Steenstrupia 18, 9±62.

FjeldsaÊ, J. (1994) Geographical patterns for relict and young species of birds in Africa and South

America and implications for conservation priorities. Biodiv. Conserv. 3, 207±26.

FjeldsaÊ, J. and Lovett, J.C. (1997) geographical patterns of old and young species in African forest

biota: the signi®cance of speci®c montane areas as evolutionary centres. Biodiv. Conserv. 6, 325±

46.

Frailey, C.D., Lavina, E.L., Rancy, A. and Pereira de Souza, J. (1988) A proposed Pleistocene/

Holocene lake in the Amazon Basin and its signi®cance to Amazonian geology and biogeog-

raphy. Acta AmazoÃnica 18, 119±43.

Futuyama, D.J. and Shapiro, L.H. (1995) Hybrid zones. Evolution 49, 222±6.

Ha�er, J. (1967) Speciation of Colombian forest birds west of the Andes. Am. Mus. Novitates 2294,

57 pp.

Ha�er, J. (1969) Speciation in Amazonian forest birds. Science 165, 131±7.

Ha�er, J. (1974) Avian speciation in tropical South America. Publ. Nuttall Ornith. Club 14,

390 pp.

Ha�er, J. (1978) Distribution of Amazon forest birds. Bonner zool. BeitraÈge 29, 38±78.

Ha�er, J. (1981) Aspects of Neotropical bird speciation during the Cenozoic. In Vicariance Bioge-

ography (G. Nelson and D.E. Rosen, eds) pp. 371±94. New York: Columbia University Press.

Vertebrate speciation in Amazonia 473

Ha�er, J. (1982) General aspects of the refuge theory. In Biological Diversi®cation in the Tropics

(G.T. Prance, ed.) pp. 6±24. New York: Columbia University Press.

Ha�er, J. (1987) Quaternary history of tropical America. In Biogeography and Quaternary History in

Tropical America (T.C. Whitmore and G.T. Prance, eds) pp. 1±18. Oxford: Clarendon and

Oxford University Press.

Ha�er, J. (1990) Geoscienti®c aspects of allopatric speciation. In Vertebrates in the Tropics (G. Peters

& R. Hutterer, eds) pp. 45±60. Museum A. Koenig, Bonn, Germany.

Ha�er, J. (1993a) On the `river e�ect' in some forest birds of southern Amazonia. Boletim do Museu

Paraense E. Goeldi (Brasil), ZoologõÂa 8, 217±45 (`1992').

Ha�er, J. (1993b) Time's cycle and Time's arrow in the history of Amazonia. Biogeographica 69, 15±

45 (C.R. de seÂances de la Societe de BiogeÂographie, Paris).

Ha�er, J. (1997) Contact zones between birds of southern Amazonia. Ornith. Monogr. (in press).

Hellmayr, C.E. (1910) The birds of the Rio Madeira. Novitates Zool. 17, 257±428.

Hellmayr, C.E. (1912) Zoologische Ergebnisse einer Reise in das MuÈndungsgebiet des Amazonas

(L. MuÈller). II. VoÈgel. Abhandl. KoÈnigl. Bayer. Akad. Wiss., Math.-Phys. Klasse 26, 1±142.

Herbert, T.D. and Fischer, A.G. (1986) Milankovitch climatic origin of mid-Cretaceous black shale

rhythms in central Italy. Nature 321, 739±43.

Hershkovitz, P. (1977) Living New World Monkeys (Platyrrhini), vol. 1. Chicago: University of

Chicago Press.

Hooghiemstra, H., Melica, J.L., Berger, A. and Shacklton, N.J. (1993) Frequency spectra and

paleoclimatic variability of the high-resolution 30-1450 ka Funza I pollen record (Eastern

Cordillera, Colombia). Quat. Sci. Rev. 12, 141±56.

Hoorn, C. (1994) Fluvial palaeoenvironments in the intracratonic Amazonas Basin (Early Miocene-

early Middle Miocene, Colombia). Palaeogeogr. Palaeoclimatol. Palaeoecol. 109, 1±54.

Hoorn, C., Guerrero, J., Sarmiento, G.A. and Lorente, M.A. (1995) Andean tectonics as a cause for

changing drainage patterns in Miocene northern South America. Geology 23, 237±40.

Katzer, F. (1903) GrundzuÈge der Geologie des unteren Amazonasgebietes. Leipzig: M. Weg.

Kronberg, B.I., Benchimol, R.E. and Bird, M.I. (1991) Geochemistry of Acre Subbasin sediments:

window on Ice-Age Amazonia. IntercieÃncia 16, 138±41.

Mallet, J. (1993) Speciation, raciation, and color pattern evolution in Heliconius butter¯ies: evidence

from hybrid zones. In Hybrid Zones and the Evolutionary Process (R.G. Harrison, ed.) pp. 226±

60. Oxford: Oxford University Press.

Mayr, E. (1942) Systematics and the Origin of Species from the Viewpoint of a Zoologist. New York:

Columbia University Press.

Mayr, E. (1963) Animal Species and Evolution. Cambridge, MA: Harvard University Press.

Mayr, E and O'Hara, R.J. (1986) The biogeographic evidence supporting the Pleistocene forest

refuge hypothesis. Evolution 40, 55±67.

Meave, J. and Kellman, M. (1994) Maintenance of rainforest diversity in riparian forests of tropical

savannas: implications for species conservation during Pleistocene drought. J. Biogeogr. 21,

121±35.

Meave, J., Kellman, M., MacDougall, A. and Rosales, J. (1991) Riparian habitats as tropical forest

refugia. Global Ecol. Biogeogr. Lett. 1, 69±76.

Mosmann, R., Falkenhain, F.U.H., Goncales, A. and Nepomuceno, F. F. (1986) Oil and gas po-

tential of the Amazon Paleozoic basins. In Future Petroleum Provinces of the World

(M.T. Halbouty, ed.). Amer. Assoc. Petrol. Geol., Memoir 40, 207±41.

MuÈller, J., Irion, G., Nunes de Mello, J. and Junk, W. (1995) Hydrological changes of the Amazon

during the last glacial±interglacial cycle in central Amazonia. Naturwissenschaften 82, 232±5.

Nuttall, C.P. (1990) A review of the Tertiary non-marine molluscan faunas of the Pebasian and other

inland basins of north-western South America. Bull. Brit. Mus. Nat. Hist. (Geol.) 45, 165±371.

Olsen, P.E. (1986) A 40-million-year lake record of Early Mesozoic orbital climatic forcing. Science

234, 842±8.

474 Ha�er

Patton, J.L. and Smith, M.F. (1992) mtDNA phylogeny of Andean mice: a test of diversi®cation

across ecological gradients. Evolution 46, 174±83.

Patton, J.L., Silva, M.N. and Malcolm, J.R. (1994) Gene genealogy and di�erentiation among

arboreal spiny rats (Rodentia: Echimyidae) of the Amazon basin: a test of the riverine barrier

hypothesis. Evolution 48, 1314±23.

Petri, S. and FuÂlfaro, V.J. (1983) GeologõÂa do Brasil. SaÄo Paulo: Universidade de SaÄo Paulo.

Prance, G.T. (ed.) (1982) Biological Diversi®cation in the Tropics. New York: Columbia University

Press.

Prance, G.T. and Lovejoy, T.E. (eds) (1985) Amazonia. Key Environments. Oxford: Pergamon

Press.

Prum, R.O. (1988) Historical relationships among avian forest areas of endemism in the Neotropics.

Acta XIX Congr. Intern. Ornith. 2, 2562±72.

Rancy, A. (1991) Pleistocene mammals and paleoecology of the western Amazon. PhD dissertation,

University of Florida; Ann Arbor; University Micro®lm International.

Rancy, A. (1993) A paleofauna da Amazonia indica aÂreas de pastagem com pouca cobertura vegetal.

CieÃncia Hoà je (SaÄo Paulo) 16, no. 93, 48±51.

RaÈsaÈnen, M.E., Salo, J.S., Jungnert, H. and Pittman, L.R. (1990) Evolution of the western Amazon

lowland relief: impact of Andean foreland dynamics. Terra Nova 2, 320±32.

RaÈsaÈnen, M., Neller, R., Salo, J. and Jungner, H. (1992) Recent and ancient ¯uvial deposition

systems in the Amazonian foreland basin, Peru. Geol. Mag. 129, 293±306.

RaÈsaÈnen, M.E., Linna, A.M., Santos, J.C.R. and Negri, F.R. (1995) Late Miocene tidal deposits in

the Amazonian foreland basin. Science 269, 386±90.

Salo, J., Kalliola, R., Hakkinen, I., Makinen, Y., Niemela, P., Puhakka, M. and Coley, P.D. (1986)

River dynamics and the diversity of Amazon lowland forest. Nature 322, 254±58.

Santos, J.O.S., Nelson, B.W. and Giovannini, C.A. (1993) Corpos de areia sob leitos abandonados

de grandes rios. CieÃncia Hoà je (SaÄ o Paulo) 16, no. 93, 22±5.

Shacklton, N.J., Berger, A. and Peltier, W.R. (1990) An alternative astronomical calibration of the

lower Pleistocene time scale based on EDP Site 677. Trans. Roy. Soc. Edin. Earth Sci. 81, 251±

61.

Sick, H. (1967) Rios e enchentes na AmazoÃnia como obsta culo para a avifauna. Atas Simp. SoÃbre a

Biota AmazoÃnica, vol. 5 (Zoologia), 495±520.

Silva, J.M.C. de, Novaes, F.C. and Oren, D.C. (1995) A new species of the genus Hylexetastes

(Dendrocolaptidae) from eastern Amazonia. Bull. Brit. Ornith. Club 115, 200±6.

Simpson, B.B. and Ha�er, J. (1978) Speciation patterns in the Amazonian forest biota. Ann Rev.

Ecol. Syst. 9, 497±518.

Snethlage, E. (1913) UÈ ber die Verbreitung der Vogelarten in Unteramazonien. J. Ornith. 61, 469±

539.

Terborgh, J. (1992) Diversity and the Tropical Rain Forest. New York: Freeman.

Thompson, L.G., Mosley-Thompson, E., Davis, M.E., Lin, P.-N., Henderson, K.A., Cole-Dai, J.,

Bolzan, J.F. and Liu, K.-b. (1995) Late Glacial Stage and Holocene tropical ice core records

from HuascaraÂn. Peru, Science 269, 46±50.

Tuomisto, H., Ruokolainen, K. and Salo, J. (1992) Lago Amazonas: Fact or fancy? Acta AmazoÃnica

33, 353±61.

Van der Hammen, T. and Absy, M.L. (1994) Amazonia during the last glacial. Palaeogeog. Pa-

leoclimatol. Palaeoecol. 109, 247±61.

Vanzolini, P.E. (1973) Paleoclimates, relief, and species multiplication in equatorial forests. In

Tropical Forest Ecosystems in Africa and South America: a Comparative Review (B.J. Meggers,

E.S. Ayensu and W.D. Duckworth, eds) pp. 255±8. Washington: Smithsonian Press.

Vanzolini, P.E. (1992) Paleoclimas e especiacËaÄo em animais da AmeÂrica do Sul tropical. Estudos

AvancËados (SaÄo Paulo) 6, no. 15, 41±65.

Vertebrate speciation in Amazonia 475

Vanzolini, P.E. and Williams, E.E. (1970) South American anoles: geographic di�erentiation and

evolution of the Anolis chrysolepis species group (Sauria, Iguanidae). Arquivos de ZoologõÂa (SaÄo

Paulo) 19, 1±298.

Veiga, A.T.C., Dardenne, M.A. and Salomao, E.P. (1988) Geologia dos aluvioes aurõÂferos e est-

anõÂferos da AmazoÃnia. Anais XXXV Congr. Brasil. Geol. (BeleÂm, ParaÂ) 1, 164±77.

Vrba, E.S. (1992) Mammals as a key to evolutionary theory. J. Mammal. 73, 1±28.

Vrba, E.S. (1993) Mammal evolution in the African Neogene and a new look at the Great American

interchange. In Biological Relationships between Africa and South America (P. Goldblatt, ed.)

pp. 393±432. New Haven: Yale University Press.

Wallace, A.R. (1853) A Narrative of Travels on the Amazon and Rio Negro. London: Reeve.

Whitmore, R.T. and Prance, G.T. (eds) (1987) Biogeography and Quaternary History in Tropical

America. Oxford Monogr. on Biogeogr. 3. Oxford: Clarendon Press.

476 Ha�er