Survey
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
Journal of Biogeography (J. Biogeogr.) (2004) 31, 943–957 ORIGINAL ARTICLE Holocene vegetation change and the mammal faunas of South America and Africa Mario de Vivo* and Ana Paula Carmignotto Museu de Zoologia, Universidade de São Paulo, São Paulo, Brazil ABSTRACT Aim Although sharing many similarities in their vegetation types, South America and Africa harbour very dissimilar recent mammal faunas, not only taxonomically but also in terms of several faunistic patterns. However late Pleistocene and mid-Holocene faunas, albeit taxonomically distinct, presented many convergent attributes. Here we propose that the effects of the Holocene climatic change on vegetation physiognomy has played a crucial role in shaping the extant mammalian faunistic patterns. Location South America and Africa from the late Pleistocene to the present. Methods Data presented here have been compiled from many distinct sources, including palaeontological and neontological mammalian studies, palaeoclimatology, palynology, and publications on vegetation ecology. Data on Pleistocene, Holocene and extant mammal faunas of South America and Africa allowed us to establish a number of similar and dissimilar faunistic patterns between the two continents across time. We then considered what changes in vegetation physiognomy would have occurred under the late Pleistocene last glacial maximum (LGM) and the Holocene climatic optimum (HCO) climatic regimes. We have ordained these proposed vegetation changes along rough physiognomic seral stages according to assumptions based on current botanical research. Finally, we have associated our hypothesized vegetation changes in South America and Africa with mammalian faunistic patterns, establishing a putative causal relationship between them. Results The extant mammal faunas of South America and Africa differ widely in taxonomical composition; the number of medium and large species they possess; behavioural and ecological characteristics related to herbivore herding, migration and predation; and biogeographical patterns. All such distinctions are mostly related to the open formation faunas, and have been completely established around the mid-Holocene. Considering that the mid-Holocene was a time of greater humidity than the late Pleistocene, vegetation cover in South America and Africa would have been dominated by forest or closed vegetation landscapes, at least for most of their lower altitude tropical regions. We attribute the loss of larger-sized mammal lineages in South America to the decrease of open vegetation area, and their survival in Africa to the existence of vast savannas in formerly steppic or desertic areas in subtropical Africa, north and south of the equator. Alternative explanations, mostly dealing with the disappearance of South American megamammals, are then reviewed and criticized. *Correspondence: Mario de Vivo, Museu de Zoologia, Universidade de São Paulo, Av. Nazaré, 481 Ipiranga, São Paulo, SP 04263-000, Brazil. E-mail: [email protected] Main conclusions The reduction of open formation areas during the HCO in South America and Africa explains most of the present distinct faunistic patterns between the two continents. While South America would have lost most of its open formations within the 30 latitudinal belt, Africa would have kept large areas ª 2004 Blackwell Publishing Ltd www.blackwellpublishing.com/jbi 943 M. de Vivo and A. P. Carmignotto suitable to the open formation mammalian fauna in areas presently occupied by desert and semi-arid vegetation. Thus, the same general climatic events that affected South America in the late Pleistocene and Holocene also affected Africa, leading to our present day faunistic dissimilarities by maintaining the African mammalian communities almost unchanged while dramatically altering those of South America. Keywords Mammals, South America, Africa, vegetation change, HCO, LGM. INTRODUCTION South America and Africa share many landscape similarities, including the presence of extensive rain forests, savannas, steppes and deserts, but their contemporary mammalian faunas are quite distinct (Keast, 1972; Vrba, 1993). Convergent patterns such as would be expected as a result of faunistic evolution over similarly structured landscapes (Bourlière, 1973) are actually not striking (McNaughton et al., 1993; Cristoffer & Perez, 2003). The faunas differ widely in taxonomical composition, they differ in the number of medium and large species each continent possess; behavioural and ecological characteristics related to herbivore herding, migration and predation; and biogeographical patterns. However, this has not always been the case. Both faunas have presented, at least from the Miocene onwards, convergent characteristics, with diversified forest and open-formation faunas, the latter inhabited by medium to large grazers and browsers, all preyed upon by specialized carnivores (Cooke, 1972; Patterson & Pascual, 1972). In the Quaternary, this scenario remained almost unaltered in Africa but changed dramatically in South America, where open-formation mammal fauna lost practically all similarities to that of Africa. Most authors believe that South America and Africa have evolved their quite differently structured extant mammalian faunas due entirely to distinct causes acting on each continent. Considerable attention has been paid to the role of humans (e.g. Martin, 1967, 1984), climatic change (e.g. Ochsenius, 1985; Cartelle, 1999) or the impact of North American immigrant mammals on the ecosystems (e.g. May, 1978; Marshall et al., 1982; Webb, 1985; Marshall & Cifelli, 1990). A few faunistic comparisons have also appeared, but they have been exclusively descriptive (Keast, 1972; Bourlière, 1973). Vrba (1993) argued that global climatic events should be employed in explaining faunistic patterns in South America and Africa, however, she primarily described how climate change affected the faunistic composition in South America as a result of the Great American Interchange of the late Pliocene and its after-effects during the Pleistocene. Additionally, all previous attempts to study the faunas have assumed that the late Pleistocene was the most significant period of change (Martin & Klein, 1984; Ochsenius, 1985; Marshall & Cifelli, 1990; Webb & Rancy, 1996; Cartelle, 1999), but recent effort at dating fossil mammals in South America revealed several extinct species as living well into the middle Holocene (Faure 944 et al., 1999; Baffa et al., 2000). Faure et al. (1999) dated a fossil assemblage from north-eastern Brazil including the camelid Palaeolama, the horse Equus and the giant armadillos, Glyptodon and Hoplophorus, as c. 8490 and 6890 yr bp. Baffa et al. (2000) dated a Toxodon from a karstic cave in southeastern Brazil to between 8000 and 5400 yr bp. This places these large mammals well into the middle Holocene, and considerably changes all previous views that the South American megafauna would have been extinct by that time. Besides, it indicates that analogously similar faunistic patterns between South America and Africa have persisted beyond the late Pleistocene. Here we will demonstrate how the same late Pleistocene and Holocene global events could have simultaneously affected the faunas and disrupted the pattern of shared analogous similarities between South America and Africa. The impact of climate change on vegetation at these times certainly affected floristic composition. However, physiognomy change was probably more significant for the mammalian faunas, with landscapes alternatively showing denser and sparser facies through time. Higher yearly average precipitation existing during the Holocene climatic optimum (HCO) would favour denser vegetation physiognomies, thus reducing the availability of savanna-like habitats for open formation faunas. However, lower averages would open existing physiognomies and open formation faunas would be benefited. Our continental level reconstruction of the vegetation changes in South America and Africa provide a model that indicates why the same climatic events led to a diversification of savanna mammal patterns in Africa while practically extirpating them in South America. IDENTIFYING FAUNISTIC PATTERNS Past and present faunistic patterns South America and Africa have always possessed taxonomically distinct mammal communities throughout the Tertiary and Quaternary (Patterson & Pascual, 1972; Maglio & Cooke, 1978; Wilson & Reeder, 1993). This is significant in the sense that any past or present similarities in ecological or behavioural patterns are analogous, not homologous. Throughout the Tertiary and Quaternary, South America harboured 20 orders of mammals and Africa 13 (only terrestrial forms; aquatic and volant mammals excluded). The recent fauna is much less diverse for South America, with only 12 extant Journal of Biogeography 31, 943–957, ª 2004 Blackwell Publishing Ltd Vegetation change and mammal faunas in South America and Africa orders and 11 in Africa (eight ordinal extinctions for South America and only two for Africa). Considering the extant faunas of the two continents, seven orders are shared, but have only two genera in common, the carnivores Panthera and Mustela, which are practically cosmopolitan taxa (Nowak, 1999). Below is a summary of ecological, behavioural and biogeographical patterns that involve past and recent mammal faunas in South America and Africa. Diversification of continental faunas across weight categories Figure 1 shows that Africa is richer in number of species for any category above 5 kg; only in the weight category below 5 kg is South America richer than Africa (622 and 587 species, respectively). There are no reliable weight estimates for extinct mammals in both continents but Anderson (1984) furnishes some adequate size estimates. If Pleistocene faunas could be included in our graph, diversity within the weight categories would have been quite similar between South America and Africa. Most of the extinct South American medium and largesized mammals were associated with open formations (MacFadden & Shockey, 1997; Cartelle, 1999; Rancy, 1999; Cristoffer & Peres, 2003) therefore attracting attention to this particular assemblage and its African counterpart. Presence of grazer and mixed grazer-browser terrestrial herbivores Number of species Among the largest African herbivores are the elephant, zebra, rhinoceros, giraffe and hippopotamus, but the bulk of its savanna diversity lies in the multitude of medium and largesized bovids (Bigalke, 1972; Peters, 1983). The extant South American counterparts are the comparatively much less impressive tapir, capybara, deer and the Andean camelids. All South American herbivores are either mixed grazers and browsers, or exclusive browsers, while Africa has a large number of grazers (McNaughton & Georgiadis, 1986). As – – – – Figure 1 Number of species per weight category; South America in black and Africa in grey. Weight data from several sources, mainly Nowak (1999). Africa presents more mammal species in all categories above 5 kg, while South America is richer in the category below 5 kg (see text). Journal of Biogeography 31, 943–957, ª 2004 Blackwell Publishing Ltd South America has extensive areas of open vegetation and did possess a diversified grazer herbivore fauna (Patterson & Pascual, 1972; Rancy, 1999; MacFadden, 2000), the loss of this guild strongly affects any contemporary comparison. In fact, bovids, which are responsible for much of the present day distinctiveness of Africa, failed to enter South America at the end of the Pliocene. This could lead us to think that their mere absence would promote the faunistic distinction just described, but South America did possess autochtonous herbivore orders, such as Litopterna and Notoungulata which never extended their distribution outside the Americas. Finally, even if all bovids are excluded from the analysis, Africa would still present an assemblage of large mammals for which no equivalents can be found in modern South America. Herding and migratorial behaviour Group size among South American herbivores is usually small, and solitary species are not infrequent, while Africa possesses many herding herbivores, with groups ranging from dozens to hundreds of individuals. These herds frequently migrate seasonally in search of better pastures, while no herbivore migration is known in the extant South American mammal fauna except for small altitudinal shifts in the Andean camelids (Kingdom, 1979; Nowak, 1999). The bulk of African mammal species exhibiting herding and migratorial behaviours require vast expanses of open vegetation to perform them (Owen-Smith, 1988; Kappelman et al., 1997). The fossil record shows that Plio-Pleistocene and present African mammal faunas were similarly structured (Vrba, 1993; Kappelman et al., 1997), but this is not clear for the Pleistocene mammals of South America. Cartelle & Bohórquez (1982) believe that the terrestrial sloth Eremotherium may have been gregarious, while Webb (1999) envisages ‘vast herds of herbivores’ for the continent. We suspect that South American gomphoteriid mastodons may have lived in groups, as well as the horses (Equus), as do phylogenetically close forms of today. The Xenarthra is a monophyletic group and all extant xenarthrans are solitary; the most parsimonious hypothesis is to suppose that giant armadillos and terrestrial sloths would also have been solitary, otherwise independent acquisitions of social behaviour would have to be accepted. This is, of course, possible, but less probable. Finally, we simply have no parallel for how to characterize the herding behaviour of animals belonging to extinct endemic orders of South American mammals, such as the Litopterna and Notoungulata. These animals could have formed ‘vast herds’, lived in small groups, lived solitarily, or any combination of the former. Specialized carnivore hunting and social behaviour: presence of carrion eaters All of the living South American predators hunt solitarily (with the exception of canid Speothos venaticus, a small forest dweller), while Africa has both solitary and group-organized forms. Group organized hunters are mostly open vegetation 945 M. de Vivo and A. P. Carmignotto forms, like lions (Panthera leo), wild hunting dogs (Lycaon) and hyenas (Hyaena, Crocuta). As prey size and availability are factors that strongly influence the predation patterns of carnivores (Vrba, 1980), group hunting has probably evolved as a response to herd herbivory and the complex behaviours involved can only be fully performed in open areas. South America presented, for instance, sabre tooth felids adapted to preying on large mammals and even specialized ‘giant’ vampire bats, but these predators disappeared with their prey (Simpson, 1980; Reig, 1981; Trajano & de Vivo, 1991). Finally, Africa has a group of socially organized carnivores (hyenas) that behave as opportunistic carrion eaters, which has no extant mammalian equivalent in South America. For carrion to be a major food source, it is probable that a large herbivore biomass must be available. presenting vicariant patterns for species and subspecies include giraffes (Giraffa), dik-diks (Madoqua), kudus (Tragelaphus), oryx (Oryx), antelopes of the genus Hippotragus, the lechwe (Kobus), reedbucks (Redunca), gazelles (Gazella), African buffalo (Syncerus), zebras (Equus), white rhinoceros (Ceratotherium), jackals (Canis), hyenas (Hyaena, Crocuta) and lions (P. leo). The patterns just described probably emerged due to the disappearance, in South America, of open vegetation large mammals and their survival in Africa. The fossil record reveals the presence of some vicariant species in South America represented by Patagonian and Brazilian provinces species (Cartelle, 1999). Nevertheless the fossil record is not accurate enough to support or reject the presence of vicariant taxa along the South American continent as a whole. RESULTS AND DISCUSSION Contemporary biogeographical patterns Today, South America has two distinct mammalian faunistic provinces, the Brazilian and the Patagonian (Hershkovitz, 1972). The Brazilian subregion actually extends from Central America to Colombia, west of the Andes, and to the northern half of Ecuador. To the east of the Andean cordillera, it encompasses almost the entire tropical and subtropical South America southwards to Bolivia and from there across Paraguay to southern Brazil and adjacent Uruguay. The Patagonian subregion includes the remaining parts of the continent: all of South America west of the Andes from central Ecuador southwards, and east of the Andes from south Bolivia to Argentina and Uruguay. The Andes have no parallel in Africa in continental extension and average altitude. Their presence distorts the expected zonal distribution of alternating latitudinal humid and drier life zones due to their powerful influence as a barrier to moisture carrying winds. To the west of the Andes, dry and very dry climates prevail in tropical latitudes even at sea level southwards from central Ecuador. On the contrary, the expected zonal humid temperate climates at latitudes around 60 appear only in Chile, and to the other side of the Andes, the Argentinean Patagonia is steppic in climate. The Patagonian province of South America includes many genera and even families that do not occur in the Brazilian subregion and vice versa. At least 27 genera of rodents occur exclusively in the southern South America or at the climatically equivalent high altitudes in the tropical Andes (e.g. Lagostomus, Dolichotis, Abrocoma, Andalgalomys, Phyllotis, Graomys and Eligmodontia), the camelids Llama and Vicugna, the cervids Pudu and Hippocamelus, the armadillos Chaetophractus, Chlamyphorus and Zaedyus, the carnivore Lyncodon, while entire families are tropical (the tree sloths Bradypodidae and Megalonychidae; the Myrmecophagidae anteaters; Cebidae and Atelidae primates; Echimyidae, Agoutidae and Dinomyidae histricognath rodents; Redford & Eisenberg, 1992; Eisenberg & Redford, 1999). In Africa, the most common pattern is the replacement of vicariant species and subspecies along the north–south axis of open formations (Haltenorth & Diller, 1977; Kingdom, 1977, 1979). Genera 946 Commentaries on faunistic patterns The patterns 1 to 4 described above are real: they underline the distinctiveness of the extant South American and African mammalian faunas. One question is if these patterns can in fact be associated with the open formations. Another question relates to the forest faunas: are they just as distinctive as the open formation ones? Our assumption is that large size, herding and migratory behaviour, and social hunting techniques are strongly related to open formations. This is not only intuitive, but also supported by several lines of research (Clutton-Brock & Harvey, 1983; Peters, 1983; Owen-Smith, 1988; Kappelman et al., 1997). The terrestrial medium to large herbivores of the African rain forest are either small-sized lineages (e.g. Cephalophinae bovids) or dwarfed vicariants to savanna taxa. The duikers (Cephalophinae) represent a bovid lineage that is clearly forest adapted. Their size is small (around 20 kg body mass), and they are either solitary or live in small groups, up to three individuals (Haltenorth & Diller, 1977; Peters, 1983). The forest buffalo (Syncerus caffer nanus) is a distinct taxon than the savanna forms, its body mass reaches a maximum of 300 kg, a weight comparable with that of the South American tapirs. The two savanna subspecies are much larger, reaching up to 800 kg body mass. The differences in group size are remarkable: the forest buffalo lives in groups of three to 12 individuals, while the savanna’s taxa range from 20 to 2000 (Haltenorth & Diller, 1977). The forest elephant (Loxodonta cyclotis, a distinct species according to Roca et al., 2001), is also quite smaller, with up to half the weight and size of the savanna form. Average group sizes are also quite distinct. The forest elephant averages 3.2 individuals per group, while savanna elephants average 10, but the latter can form loose aggregations of up to 1000 individuals in the rainy season (Haltenorth & Diller, 1977). It is revealing that forest elephants actually use mostly forest clearings for much of their activities; moving through the forest along paths to reach alternative clearings (Vanleeuwe & Gautier-Hion, 1998). The case of the forest buffalo and elephant is highly significant because it indicates that elephants and buffalos Journal of Biogeography 31, 943–957, ª 2004 Blackwell Publishing Ltd Vegetation change and mammal faunas in South America and Africa cannot be considered habitat generalists. The fact that forest and savanna present vicariant taxa means that speciation, or at least significant differentiation, has occurred. It is beyond the scope of this article to discuss the history of speciation in the African rain forests, but it is plausible that the large herbivores presently living in African rain forests represent populations once living in the open, that have been encircled by expanding forest sometime in the past. These populations would have become isolated from their savanna sisters and differentiated. Analogously, it can be assumed that, while a fraction of the large-sized savanna mammalian fauna of South America could have vicariously inhabited South American rain forests, it would still represent a minority of the taxa involved. However, there is no evidence for the existence of such a large-sized fauna in South American forests (Kay & Madden, 1997; MacFadden & Shockey, 1997; Rancy, 1999; Cristoffer & Perez, 2003). Cristoffer & Perez (2003) discuss the rain forest mammals of South America and Africa, and indicate that forest lineages in Africa have consistently evolved more terrestrial offshoots (in the case of the African primate Cercopithecus). To this example we could add the evolution of baboons (Papio) and our own (Homo) (Reed, 1996). If African rain forests repeatedly produced terrestrially adapted lineages, they are quite poor in specialized arboreal locomotory adaptations. Prehensile tails evolved independently at least six times in South America: twice in Primates (Cebus and the ateline genera, e.g. Brachyteles), rodents (Sphiggurus and Coendou), carnivores (Potos), xenarthrans (Tamandua and Cyclopes) and marsupials (all South American genera). In Africa, only pangolins (Manis) possess prehensile tails. Emmons & Gentry (1983) showed that African rain forests not only practically lack prehensile tailed mammals, but also are very poor on gliding forms (only three genera, Anomaluridae rodents), which abound in Southeastern Asia. Emmons & Gentry (1983) attributed the evolution of specialized locomotory adaptations in the world’s rain forests to selective pressures derived from their differently structured forest canopies and strata. It is not our aim to discuss the forest patterns, but it is intriguing that South America has so many independent acquisitions of prehensile tails, and Southeastern Asia possesses several gliding vertebrates, while Africa shows a remarkable poverty of both, and its forest lineages have frequently evolved terrestrial offshoots. Climate change and its influence on vegetation Climate change has been evoked as playing an important role in the megamammal Pleistocene extinctions (Graham & Lundelius, 1984; Guilday, 1984; Ochsenius, 1985; Cartelle, 1999), but besides pointing out general consequences such as drought, authors have refrained from building a model in which tropical vegetation types would function under distinct climatic regimes. For the purpose of evaluating the impact of late Pleistocene and Holocene climatic changes on the vegetation, we have made the following assumptions. Journal of Biogeography 31, 943–957, ª 2004 Blackwell Publishing Ltd The geographical distribution of vegetation types We have considered climatic changes affecting the present geographical configuration of major vegetation types in South America and Africa. There is a considerable body of research on botanical palaeocommunities which reveals that floristic composition has changed in the past 20 Ka (Coetzee, 1993; Houérou, 1997; Colinvaux et al., 2000). Despite the growing palynological knowledge, these reconstructions, however, remain local and would render impossible any analysis that considers South America and Africa simultaneously with all their distinct biomes. Besides, we know very little about past relationships between extinct mammal lineages and particular floristic elements beyond the general inferences about grazing and/or browsing habits. Thus, our focus is on physiognomic changes of vegetation rather than floristic (see below). How average precipitation levels affect vegetation physiognomy Our second assumption is generally applicable to tropical landscapes, in that plant formations tend to present denser concentrations of woody elements with greater water availability and vice versa (Fig. 2; Furley & Newey, 1983; Walter, 1984; Rizzini, 1997; Furley, 1999). The density of woody species directly reflects on the amount of sunlight reaching the soil and thus affects grass biomass. One way to demonstrate that density of woody elements in a vegetation varies with water availability through changing precipitation levels is to examine the literature on areas monitored for the effects of fire on vegetation. This is, of course, an indirect approach as the climate (precipitation level) has not actually changed in these areas. We assume that a severe dry period lasting for a large number of consecutive years, as would be the case at the beginning of a glacial period, enhances the openness of a landscape through fires, either natural or man made, while lesser deficits diminish the impact of natural fire, as attested by several sources (Walter, 1984; Miranda et al., 2002; Oliveira-Filho & Ratter, 2002). In Africa, a number of studies have focused on the vegetation changes due to the effects of seasonal fire and herbivory on the vegetation. Salvatori et al. (2001) and Mapaure & Campbell (2002) have shown that fire and herbivory play a significant role in the decrease of woodland area and its transformation in grasslands, but these authors have not been able to properly distinguish between the independent contributions of fire and herbivores. However, Dublin et al. (1990) showed that decrease in woodland cover was due primarily to fire, large herbivores playing a part mainly in the lack of regeneration of woodland after its eradication by fire. Swaine et al. (1992) were able to monitor controlled areas and demonstrated that trees become more abundant if savannas are protected from fire. These authors showed that in protected areas of savanna close to forest vegetation, the regeneration included an expansion of forest species, while in areas away from forests, the savanna trees increased their density. 947 M. de Vivo and A. P. Carmignotto Evergreen forest Semi-evergreen forest Woodland Tall Cerrado Savanna Dry forest Mixed grass and shrub Dry forest and shrub Grassland Steppe Steppe In South America, there are no large wild herbivores as in Africa, but the net results are essentially the same. Studies on the density of tree cover in the Venezuelan (San José & Farinãs, 1983, 1991) and Central Brazilian savannas (Moreira, 2000; Hoffmann & Moreira, 2002) have shown that protection from fire leads to denser physiognomies. McNaughton et al. (1993) provide comparisons between vegetation structures of arid and semi-arid regions of South America and Africa, concluding that great general similarities can be observed, including the response of open and dense savannas and woodland to fire. Our model assumes that vegetation is much less vulnerable to fire if precipitation levels are consistently higher, as would occur during the prevalence of wetter climates and vice versa. The role of soil in areas covered by savannas is, of course, essential, but not as important in determining physiognomies than it is for floristics. Following Oliveira-Filho & Ratter (2002), high fertility unflooded soils will always sustain some kind of forest, from semi-evergreen to dry. Low fertility unflooded soils will support savanna vegetation, but if water availability is high and seasonality not strongly marked, tall Cerrado (a kind of savanna forest) and dense savanna occur; even semi-evergreen forests can be found. Low fertility soils with strong water deficits will sustain several kinds of open savanna. The estimated effect of average water availability on vegetation physiognomy is summarized in Fig. 2. This figure presents two distinct paths of vegetation physiognomic change, both commencing at levels of good water availability and degrading to steppic or desertic conditions. The first series ranges from rain forests and the second from dense arboreal savanna. By keeping these two series separate we underline the fact that savannas are soil dependent (Furley & Newey, 1983; Sarmiento, 1984; Walter, 1984; Rizzini, 1997; Oliveira-Filho & Ratter, 2002). 948 Figure 2 Vegetation type physiognomies under increasingly drier and/or more markedly seasonal climates (from top to bottom). Elephants indicate those habitats suitable to medium to large savanna mammals. We believe that tall Cerrado, a kind of forested savanna of Central Brazil and known as ‘Cerradão’ by South American botanists, could not have supported large mammals because it is denser and have less grasses than typical African woodland. How general were physiognomic changes in vegetation Any vegetation map for South America and Africa is a simplified view of how the floras and physiognomies are spatially distributed. A vegetation such as the ‘Amazonian rain forest’ actually encloses what we understand as ‘typical’ rain forest and many other subtypes and enclaves. This makes our assumptions (1) and (2) not immediately applicable at regional or local scales, but adequate at continental level. Thus, when we generalize the alteration of a vegetation physiognomy from ‘evergreen forest’ to ‘semi-evergreen forest’ or ‘dry forest’, we are not stating that the entire biome has changed uniformly in these directions, but only that the indicated resulting physiognomy would be the predominant one. Our focus on the LGM and at the HCO makes distinctions between the two periods straightforward: at the LGM the prevalent climate over South America and Africa was drier, while at the HCO it was wetter. We assume that the distinct climatic regimes of the LGM and HCO prevailed over both continents, acting over their entire range of different kinds of vegetation cover. The intensity and kind of the climatic change We have considered a single variable in our model: the precipitation levels. The rationale is the same as that we have employed for the vegetation, i.e. we are aware that climate change involves alteration in a vast number of variables, such as temperature, aerial and marine circulation patterns, seasonality and others. However, not only are there no available climatic reconstructions at such detailed level for South America and Africa simultaneously, but also if such reconstructions were available, our data on the mammals would Journal of Biogeography 31, 943–957, ª 2004 Blackwell Publishing Ltd Vegetation change and mammal faunas in South America and Africa certainly lag behind. We find that the best climatic variable that can be reasonably studied with regard to its effect on vegetation physiognomy is water availability, translated as yearly average precipitation. As we cannot evaluate how much the average precipitation would drop or increase, nor if seasonal patterns would be maintained, we therefore have assumed that under wetter and drier climates, all regions would receive more or less rain by approximately the same amount, thus affecting vegetation physiognomy in one direction. Finally, as we cannot know how much more or less precipitation a continent received under a certain climatic regime, we cannot ascertain whether a vegetation change would be slight or else range across two or more of the categories depicted in Fig. 2. Below we briefly describe the present vegetation types of both continents and the effect of climate change in the savanna habitats and their mammalian faunas. Present vegetation physiognomy in South America and Africa In South America the main areas covered today by evergreen forest vegetation are the Amazonian rain forest (including, to the west of the Andes, Colombia and northern Ecuador) and the Atlantic rain forest of eastern Brazil (Fig. 3a). These forests are mainly tropical, but a significant temperate rain forest occurs in Chile, south of Valdivia. Between the Amazonian and Atlantic tropical rain forests lies a vast strip of more or less open vegetation types, extending across the continent. In north-eastern Brazil is the Caatinga, a type classified as dry forest and shrub (Eva et al., 2002). The central Brazilian Cerrado is generally described as a savanna, but actually includes savanna physiognomies and mesophytic forests as well. In Bolivia, Paraguay and northern Argentina there is also a dry forest and shrub, known as Chaco. An important area of savannas occur in Colombia and Venezuela (the Llanos), and two major enclaves of savannas can be found in northern Brazil and adjacent Guyana (Gran Sabana). To the south, Argentina has the extensive grasslands of the Pampas, which gradually changes into steppe as we move south into Patagonia. Just north of the Chilean temperate rain forest, climates become ever more drier and the vegetation changes accordingly: northwards from Valdivia, vegetation is gradually altered into deciduous forest, shrub forest, steppe and desert (Atacama). A large section of the Chaco is tropical to temperate, and the Pampas and the Patagonian steppe are entirely temperate. Thus, most of South America can be described as consisting of three major areas of evergreen forest – the Amazonian, the Atlantic, and the Chilean – with a succession of more or less open vegetation types extending from equatorial to temperate latitudes in between the first two and northwards from the third (Hueck, 1966; Hueck & Seibert, 1981; Rizzini, 1997; Eva et al., 2002). Africa presents a similar picture, albeit distinctively geographically organized (Fig. 3a). The main body of evergreen forest is equatorial, occurring from the Atlantic coast to central Journal of Biogeography 31, 943–957, ª 2004 Blackwell Publishing Ltd parts of Africa, at the Albertine Rift, with an eastern arc sweeping through Mozambique, Tanzania and southern Kenya. Savanna, and savanna and forest mosaics surround this forest. To the north, increasingly seasonal and drier climates predominate, constituting a transitional region between the savannas and the Sahara, known as the Sahel; the vegetation becomes ever more open till the desertic Saharan landscape. To the east and south of the equatorial rain forests the same general pattern of open formations can be verified, but more important areas of woodland and mixed savanna and woodland can be found, from central Angola to Tanzania and Mozambique. To the south of this vast ‘woodland’ area, climate becomes drier and vegetation changes accordingly into the xeric regions of the Namibian desert and the Kalahari. Forests or forest mosaics appear in the south-eastern coast of Africa, in Mozambique, mirroring the South American Atlantic forest. In summary, most of the African open formations lie surrounding evergreen forests, within regions dominated by tropical and subtropical climates. Africa does not possess sizeable land within temperate domains, but the moderately high areas of the Ethiopian and East African plateaus include altitudinal vegetation gradients analogous to those found in parts of the South American Andes (Cooke, 1972; Dickson, 1992; McMaster, 1992; Carroll, 2001). Vegetation physiognomies in South America and Africa in the late Pleistocene and Holocene Accepting the assumptions described above, we may discuss what changes would occur over the continents in terms of their vegetation cover. Considering that, world-wide, the LGM of the late Pleistocene was drier than the present (Flenley, 1979; Whitmore & Prance, 1987; Clapperton, 1993; Colinvaux et al., 2000; Whitlock et al., 2001), we associate our maximum vegetational openness scenario with this phase. Figure 3b shows our reconstruction of vegetation physiognomies for both continents during the LGM. As can be seen, subjected to lower precipitation averages, evergreen forests would degrade to semi-evergreen and could as well present more open patches of savanna-like physiognomies in a mosaic pattern. This kind of change would occur along vast areas of rain forest if precipitation dropped significantly. That this kind of change took place in South America can be verified by the fact that the Amazonia is mainly covered by rain forest, but include hundreds of isolated patches of savanna (Hueck & Seibert, 1981). Areas covered by semi-evergreen forests, dry forests, woodlands and savannas would generally open up as well. Forest would be mostly restricted to mountain ranges subjected to orographic rain and to other areas where rainfall would still be high enough. The retracting forests would leave space for drier seral stages in the forest succession (as seen in Fig. 2), and in areas with poor soils, savannas could be established. Vegetation types that are presently subjected to semi-arid and arid climates, such as deserts, South American grasslands, steppes and the south African areas of mixed grass 949 M. de Vivo and A. P. Carmignotto and shrub would become even drier. Mountain ranges would be affected somewhat differently, in that altitude and not only latitude would play a part. Altitudinal zonation would probably be found at correspondingly lower altitudes than those found today. As we enter the Holocene, climatic conditions changed. Figure 3c shows our reconstruction of the prevailing vegetation physiognomies in South America and Africa at the HCO. For Africa, there is a well-established pattern of ever more humid climate reaching the hypsithermal interval, or the HCO at the middle Holocene, followed by a return to drier conditions in the present (Coetzee, 1993; Houérou, 1997). During the HCO, African rain forests would be under ideal humidity conditions, probably expanding to tropical areas where soil could support it. Open woodlands would likewise present denser vegetation. Thus, open formations would be eliminated in certain areas or reduced in others, but would actually expand into regions previously too dry to support it, such as desertic areas. For South America, the HCO is less consensual. Some authors (Bradbury et al., 1981; Bigarella & de Andrade-Lima, 1982; Hare, 1992; Joly et al., 1999; de Oliveira et al., 1999) propose the same pattern as that just described for Africa, that is, the HCO occurring from early to middle Holocene. However, palynological studies throughout the continent resulted in somewhat conflicting results in the sense that some regions would have received more rainfall with others experiencing desiccation at the same moment, even within the tropics (Ledru et al., 1998; Salgado-Labouriau et al., 1998; Behling & Hooghiemstra, 2001). The same can be said of researches relying on different or broader data bases (Clapperton, 1993; Colinvaux et al., 2000). Generally, they sustain that most of the Holocene was drier than the present, gradually becoming ever more humid in the last 5000 years, approximately – although not uniformly – over the entire tropical South America. Irrespective of a middle or late Holocene HCO for South America, both Amazonian and Atlantic rain forests would be under their best climatic conditions, while dry forests such as those of Caatinga and Chaco would turn into semi-evergreen forests. Tropical savannas could maintain their floristic identity but its woody component would have become densely packed, with marked reduction of grassland areas. Grasslands and wetlands within tropical South America could probably still be found, but predominant physiognomies would be dense savanna and tall, forest savannas. Figure 3d presents a summary of our model, showing that areas within and without the 30 latitudinal belt would present complementary and opposite characteristics during the LGM and the HCO. A major distinction between South America and Africa is established in that, during the HCO, the latter continent would maintain large expanses of suitable habitats for medium- and large-sized open vegetation mammals, both to the north and to the south (the Sahara and the southern xeric regions of Namibia and the Kalahari), whereas South America would present its physiognomically suitable areas only at the colder, temperate lower latitudes of the southern part of the continent. 950 Mammal faunas and vegetation change Changing climate effects on vegetation and mammal faunas is a well-documented feature of the evolutionary history of mammals. The Miocene, with its world-wide expansion of grassland habitats is associated with the radiation of several herbivore lineages both in South America and Africa; at the end of the Miocene, when grasslands retreated, extinction of part of these lineages followed suit, at least in South America (Patterson & Pascual, 1972; Janis, 1993; Vrba, 1993; MacFadden, 2000). We would expect the same pattern occurring during the Pleistocene–Holocene climatic and vegetation change of the magnitude proposed here. In addition, we should expect the entire mammal fauna to be affected, not only its larger-sized lineages. Roughly, faunistic assemblages adapted to open vegetation environments should either become extinct or isolated within pockets if this vegetation is replaced by forest, and when open landscapes replace forests, the same would happen with the forest-adapted taxa. This is indeed the case, as we demonstrate below. Pardinãs et al. (2002) summarized several dramatic changes in the composition of small mammal communities in southern South America from the late Miocene to the Holocene. The ecological implications of these changes in the taxonomic structure of the communities is not entirely understood, but these authors attributed most of the small mammal distributional changes to climate and its effect on vegetation. Evidence that the Amazonian rain forest was widely penetrated by open formations is provided by some studies on the distribution of small mammals with pronounced habitat fidelity. The marsupial Lutreolina occurs from the Buenos Aires province in Argentina northwards until the Brazilian States of São Paulo, Goiás and Mato Grosso do Sul, always being associated with open vegetation habitats. Its distribution is then interrupted by the Amazonian rain forest, and the same species reappear in the open areas of Venezuela and Colombia (Nowak, 1999). Nunes (2001) studied small mammal communities within several isolated savanna enclaves within the Amazon basin and found that some of its rodents represented disjunct populations from species inhabiting the northern and central South American savannas, which implies that these rodents had a wider distribution in open habitats and survived in savanna pockets when the forests returned. We have already mentioned above that the presence of vicariant species such as the forest elephant (L. cyclotis) and the forest buffalo (S. caffer nanus) within the African rain forest may represent the same phenomenon of forest retraction followed by expansion and ‘capture’ of open enclaves for that continent, although the dates involved may be different. In the case of the South American small mammals, most of the species did not speciate, indicating that the disjunction events may be quite recent. The study of extinct Amazonian mammals reinforces the same scenario. Webb & Rancy (1996) and Rancy (1999) describe in detail the Amazonian Pleistocene open formation mammal fauna and concluded that there was a need for savanna habitats to sustain them. However, Colinvaux et al. Journal of Biogeography 31, 943–957, ª 2004 Blackwell Publishing Ltd Vegetation change and mammal faunas in South America and Africa 30 N (a) Desert Desert Woodland Savanna Savanna (b) Open savannas Open savannas Equator Evergreen forest Evergreen forest Dry forest Mosaic of open forests and savannas Mosaic of open forests and savannas Woodland Savanna Desert Open savannas Semi evergreen and evergreen forest Mixed grass and shrub Grassland Dry forest 30 S Evergreen forest Grassland / Steppe Desert Grassland / Steppe Desert / Cold Steppe (c) Savanna Suitable during HCO Unsuitable during LGM Dense savanna (d) Suitable during LGM Unsuitable during HCO Dense savanna Evergreen forest Evergreen forest Dense savanna Suitable during LGM Unsuitable during HCO Evergreen and semi evergreen forests Evergreen and semi evergreen forests Evergreen forest Savanna Temperate formations Suitable during HCO Unsuitable during LGM Suitable during HCO Unsuitable during LGM Figure 3 Present day major plant formations of South America and Africa (a) and geographical vegetation change for the two continents at the LGM (b) and the HCO (c). A summary of this change is depicted in (d), where darker grey are for areas of predominantly suitable megafauna habitat at the LGM and unsuitable habitats at the HCO; and lighter grey are for areas predominantly suitable habitats at the HCO and unsuitable at the LGM. White areas in the maps of South America mostly indicate complex Andean vegetation not considered here. (2000), working on western Amazonia Pleistocene palynological record, sustained that most of the Amazon lowlands remained as a forest throughout glacial cycles; proposing that these extinct mammals actually lived in forests, and that all extinct grazers would be restricted to river margins, where grasses could be found. We agree with most authors (Webb & Rancy, 1996; Rancy, 1999; Cristoffer & Perez, 2003) in that Amazonian forest as it is today could not support the megafauna. In our view the question of physiognomy vs. floristics is crucial here. Arboreal or grass pollen do not furnish an immediate picture of how densely or sparsely forested a region would be; instead they reveal floristic assemblages. Thus, for instance, arboreal pollen might be recovered from a site sustaining a woodland or dry forest (see Pennington et al., 2000), which is rather more open and more suitable to the Journal of Biogeography 31, 943–957, ª 2004 Blackwell Publishing Ltd megafauna than a rain forest. Spatial organization of plants is not immediately revealed by pollen lacustrine deposits, it is only suggsted by it, as structural reconstruction depends on the knowledge of our present-day floristic communities. If the HCO resulted in the spread of closed, forest environments over previously open landscapes, we should expect that forest-adapted mammals would leave signals of their presence in areas presently covered by open formations. Voss & Myers (1991), Emmons & Vucetich (1998) and Silva et al. (2003) documented that several species of small sigmodontine and histricognath rodents that are otherwise still living in forest areas elsewhere have been found as ‘Pleistocene’ fossils in the central Brazilian site of Lagoa Santa, in the State of Minas Gerais. This locality is today covered by savanna vegetation, and most of the rodents listed by those authors are 951 M. de Vivo and A. P. Carmignotto forest dwellers. Their disappearance from a savanna locality and survival in forests elsewhere indicates that ecological changes in vegetation distribution were probably responsible for their local extinction. Cartelle & Hartwig (1996) described two Pleistocene extinct folivorous monkeys from a fossil site within the present-day Caatinga (dry forest and shrub). This kind of vegetation cannot support highly herbivorous monkeys, and thus they must have inhabited considerably moister forests. de Vivo (1997) showed several cases of disjunction between species of arboreal mammals occurring in eastern Amazonian rain forest and north-eastern Brazilian Atlantic rain forest. Between these two rain forests, there is the Caatinga, which cannot presently support these species. These disjunctions appear to be quite recent because no differentiation can be detected for the populations involved. In Africa, there is evidence for the expansion of savanna mammal communities to areas covered today by desert. Savanna invaded what is today the Saharan desert at the Holocene HCO, as is beautifully attested by the rock art left by humans from 8000 to 1000 yr bp. Wild savanna mammals, such as giraffe, hippopotamus, elephant and several bovids were depicted in hunting scenes, and gradually were replaced in the paintings by cattle. Finally, at the end of the HCO even cattle are no longer depicted and the Sahara took its present arid configuration (Camps, 1974; Scarre, 1988; Houérou, 1997). Thus, there is considerable evidence that open and forestadapted mammal faunas have changed their distribution patterns in the past few thousand years, probably as a result of change in the vegetation physiognomy. If these vegetation changes have occurred as depicted in Fig. 3, the large-sized South American mammal lineages, mostly adapted to open formations, would have found reduced adequate landscapes in tropical and subtropical regions during the HCO, with its massive forest expansion. Most populations would become extinct and the few isolated ones would be much more vulnerable to accidental extinction or to human predation. In Africa, the savanna-adapted mammals would still find vast expanses of suitable habitats both in the subtropical north and south parts of the continent, and the populations would be under much less stress. As glacial and interglacial cycles occurred throughout the Quaternary, some of the faunistic patterns may have been established earlier. However, there is evidence that the last glacial and interglacial cycle was particularly intense and abrupt relative to previous cycles, at least relative to the last 100 Ka (Hambrey & Harland, 1981; Haffer, 1982; Bartlein & Prentice, 1989; Roy et al., 1996). All Quaternary glaciation cycles supposedly had the same general effect on vegetation, albeit with varying intensity. Thus, we think that most of the previous Pleistocene extinctions could be attributed to the alternating glacial and interglacial cycles, but the final coup would have been administered by the last one. It remains to be explained why the southern cone of South America, which was subjected to colder and drier conditions at the LGM, would not have supported most of the savanna 952 mammals at the Holocene HCO. The mammalian fossil record tells some of the story. Cartelle (1999) compiled data from several independent researchers of fossil Argentine mammals of the late Pleistocene (LGM) which show that this temperate fauna migrated to northern, more tropical localities at this period, separated by severe cold and steppic environments. (see also Ficcarelli et al., 1997; Coltorti et al., 1998; Núñez et al., 2001). During the HCO, expanding savannas could have become available again only in the temperate southern cone of South America. Animals adapted to tropical conditions would find it a harsh environment and thus only a fraction of the late Pleistocene tropical savanna fauna would have returned. Our hypothesis thus predicts that any Holocene assemblage of mammals from the temperate southern cone will be poorer than contemporary tropical ones. Most of the present day dissimilarities between Africa and South America would then be established at the Holocene. A critique of previous hypotheses on the extinction of the South American megafauna Alternative hypotheses have been proposed, namely (1) the action of human hunters entering the American continent and overexploiting the fauna (Martin, 1967, 1984); (2) the elimination of phyletic lineages due to a faunistic imbalance caused by the invasion of North American mammals in South America after the Great American Interchange of the late Pliocene (May, 1978; Marshall et al., 1982; Webb, 1985); and (3) the intense dry phase of the late Pleistocene (Ochsenius, 1985; Cartelle, 1999). Martin (1967, 1984) proposed that humans entering the Americas some 12 000 years ago overhunted a ‘naive’ mammal fauna that, having never faced such resourceful predators, provided easy and abundant meat. We reject it for several reasons discussed below. Martin (1984) proposed two possible features that would be discordant to his model: (1) the extinction of small mammals (not vulnerable to human impact) and (2) the extinction of large mammals before prehistoric human arrival. Both phenomena have been recorded for South America. The extinction of small forest mammals in areas presently covered by savannas has already been explored in the previous section. Ficcarelli et al. (1997) and Coltorti et al. (1998) described and dated sites in the Andes of Ecuador and in the northern Peruvian coast where mastodons and giant sloths became locally extinct before the arrival of man at the end of the Pleistocene (between 16 670 and 12 350 yr bp), while for areas in the Equatorial Andes more subjected to cold climates, the extirpation of the megafauna dated from c. 20 980 yr bp. Ficcarelli et al. (1997) cite other similar cases for the Argentine plains. Núñez et al. (2001) also describe and date a site in northern Chile where the arrival of humans found the megafauna already gone, and archaeological sites reveal remains of extant mammals (late Pleistocene, c. 10 800 yr bp). Ficcarelli et al. (1997), Coltorti et al. (1998) and Núñez et al. (2001) attributed the disappearance of large mammals Journal of Biogeography 31, 943–957, ª 2004 Blackwell Publishing Ltd Vegetation change and mammal faunas in South America and Africa from these sites to climatic change (excessive cold and dryness at the end of the Pleistocene). Additionally, Martin (1984, p. 360) argues that if the disappearance of large mammals through human agency was ‘truly swift and devastating’, few kill sites would have been found. Indeed, no kill sites have been found in South America, and the extended presence of large mammals in the midHolocene (described above) suggests that the absence of such sites is not favourable to Martin’s scenario. There are a few archaeological sites where evidence of some large mammals hunted by humans have been found (Taima-Taima, Venezuela, Gruhn & Bryan, 1984; southern Patagonia, Markgraf, 1985; north-eastern Brazil, Guérin et al., 1996; Tagua-Tagua and Monteverde, Chile, Núñez et al., 2001). The analysis of these publications clearly indicates that humans did hunt large mammals in South America, but the sites can hardly be described as ‘kill and processing sites’. The dating of midHolocene large mammals in South America (summarized earlier) far extends the presence of these animals in the continent simultaneously with the presence of humans, contradicting an extremely rapid rate of megamammal extirpation through excessive hunting. We believe that human predation of large South American mammals did occur, but that it may have played a marginal role, probably in the elimination of small isolated populations. Beck (1996) disagrees with Martin’s blitzkrieg hypothesis for North America, based on the geographical and temporal distribution of the megafaunal remains which do not fit the direction and timing of human migration from north-west to south-east North America, as should have happened under Martin’s assumptions. Beck (1996) emphasizes that his data does not preclude the possibility that changes in climate were a more important factor. Martin (1984) argues that the long coevolutionary history of humans and the megafauna in Africa and southern Asia would account for the survival of the animals in those regions. The animals would have learned to recognize humans as threatening and developed avoidance behaviours. As large mammals were present in the South American Holocene, the ‘overkill’ period would have to be extended from at least 12 000 to c. 6000 yr bp (some authors claim earlier dates for human entry in the American continent; see Rudgley, 1999). With such long period of co-habitation, South America would not have any ‘naive’ mammals left; they would all have learned to recognize and avoid humans, which is exactly what large mammals in Africa and Asia did. A second hypothesis dealing with (but not only) the South American mammal extinctions at the end of the Pleistocene is related to predictions contained in the Equilibrium Theory (MacArthur & Wilson, 1967), in which land masses are believed to reach relatively stable numbers of lineages given enough time. May (1978), Marshall et al. (1982), Webb (1985) and Marshall & Cifelli (1990), applying the Equilibrium Theory to the palaeontological history of New World mammal faunas, generally considered it as a viable alternative explanation for the South American extinctions. To them, the number Journal of Biogeography 31, 943–957, ª 2004 Blackwell Publishing Ltd of taxa that resulted from the sum of North American invaders plus the southern native mammals far exceeded the carrying capacity of the South American continent after the Great American Interchange at the end of the Pliocene. We reject this hypothesis. If the Equilibrium Theory explains the South American extinctions, we should expect a decrease in the number of lineages after the Great American Interchange, and that indeed occurred. However, it is not at all clear why these extinctions were concentrated on larger mammals, as demonstrated by Lessa et al. (1997). If an excessive number of lineages is the factor behind extinctions, we should expect that small, medium and large mammals would have been equally affected, but that is not what happened as we associate larger body sizes to the habitat preferences of the mammals (as discussed above, larger mammals tending to live in open formations), we believe that our hypothesis furnishes a better explanation for the known facts. Ochsenius (1985) and Cartelle (1999) proposed that the South American megafauna extinctions occurred in the late Pleistocene due to climatic change, relating it to the cold and dry conditions of the LGM. We agree with both authors that this indeed occurred, at least locally (Ficcarelli et al., 1997; Coltorti et al., 1998; Núñez et al., 2001). Our hypothesis, however, considers the presence in the mid-Holocene of some lineages of large mammals, which was ignored by both authors. The LGM climatic conditions cannot be assumed as the sole explanation for the extinctions, and a new perspective must be considered. The Holocene and its climatic events should be taken into account. CONCLUDING REMARKS All characteristics related to the special savanna attributes of herbivores and predators are largely dependent on the existence of vast extensions of open spaces with abundant grasses. In South America these spaces were severely reduced in tropical and subtropical areas during the HCO. The present savannas of Africa were probably affected in the same manner, but then Africa had ‘escape’ areas in its deserts, north and south. These deserts, under a more humid climate, would present savannalike physiognomies, thus sustaining entire savanna communities. The same climatic events could additionally explain the contemporary existence of vicariant patterns in the African savanna vs. distinct faunistic assemblages in the South American open formations (tropical and temperate). In Africa, the savannas invading former desertic regions, plus savanna isolates inside the tropical areas of the continent, could sustain isolated savanna mammal communities, which suffered differentiation resulting in the current vicariant pattern. Evidently, this could have occurred before, during the several glacial and interglacial cycles of the Quaternary, not only in the last one, thus reinforcing differentiation. However, South America would have lost most of its tropical and subtropical savannas. Grasslands in temperate South America, already somewhat depopulated in the late Pleistocene (see above), did not recover. What remained in the tropical and subtropical vs. the 953 M. de Vivo and A. P. Carmignotto temperate areas of the continent were the distinct faunistic complements found today. In conclusion, we believe that South America and Africa did share similar faunistic patterns until quite recently (middle Holocene) and that vegetation changes associated with fluctuating rainfall levels physiognomically altered habitats available for the open formation fauna. Present-day faunistic differences between these continents would arise not because the events affecting them were different, but because open formations survived in far greater extensions in Africa than in South America during the more humid phases of the Holocene. ACKNOWLEDGMENTS Erika Hingst-Zaher, Mario de Pinna, Eliana Marques Cancello, Alexandre Reis Percequillo, Daniel Jelin and two anonymous referees read earlier versions of this manuscript and offered valuable criticism and suggestions. This research was supported by FAPESP (grants 98/05075-7 and 00/06642-4). REFERENCES Anderson, E. (1984) Who’s who in the Pleistocene: a mammalian bestiary. Quaternary extinctions. A prehistoric revolution (ed. by P.S. Martin and R.G. Klein), pp. 40–89. The University of Arizona Press, Tucson, AZ. Baffa, O., Brunetti, A., Karmann, I. & Neto, C.M.D. (2000) ESR dating of a toxodon tooth from a Brazilian karstic cave. Applied Radiation and Isotopes, 52, 1345–1349. Bartlein, P.J. & Prentice, I.C. (1989) Orbital variations, climate, and palaeoecology. Trends in Ecology and Evolution, 4, 195– 199. Beck, M.W. (1996) On discerning the cause of late Pleistocene megafaunal extinctions. Paleobiology, 22, 91–103. Behling, H. & Hooghiemstra, H. (2001) Neotropical savannah environments in space and time: late Quaternary interhemispheric comparisons. Interhemispheric climate linkages (ed. by V. Markgraf), pp. 307–323. Academic Press, San Diego. Bigalke, R.C. (1972) The contemporary mammal fauna of Africa. Evolution, mammals, and southern continents (ed. by A. Keast, F.C. Erk and B. Glass), pp. 141–194. State University of New York Press, Albany. Bigarella, J.J. & de Andrade-Lima, D. (1982) Palaeoenvironmental changes in Brazil. Biological diversification in the tropics (ed. by G.T. Prance), pp. 27–40. Columbia University Press, New York. Bourlière, F. (1973) The comparative ecology of rain forest mammals in Africa and Tropical America: some introductory remarks. Tropical forest ecosystems in Africa and South America: a comparative review (ed. by B.J. Meggers, E.S. Ayensu and W.D. Duckworth), pp. 279–292. Smithsonian Institution Press, Washington DC. Bradbury, J.P., Leyden, B., Salgado-Labouriau, M., Lewis W. M. Jr, Schubert, C., Binford, M.W., Frey, D.G., Whitehead, D.R. & Weibezahn, F.H. (1981) Late Quaternary environ954 mental history of Lake Valencia, Venezuela. Science, 24, 1299–1305. Camps, G. (1974) Les Civilisations Préhistoriques de l’Afrique du Nord et du Sahara. Doin, Paris. Carroll, A. (2001) Africa’s natural realms. National Geographic Magazine, Map Supplement, 200 (3), September. Cartelle, C. (1999) Pleistocene mammals of the Cerrado and Caatinga of Brazil. Mammals of the Neotropics, the Central Neotropics, vol.3, Ecuador, Peru, Bolivia, Brazil (ed. by J.F. Eisenberg and K.H. Redford), pp. 27–46. The University of Chicago Press, Chicago and London. Cartelle, C. & Bohórquez, G.A. (1982) Eremotherium laurillardi (Lund, 1842). 1. Determinação especı́fica e dimorfismo sexual. Iheringia, Serie Geológica, 7, 45–63. Cartelle, C. & Hartwig, W.C. (1996) A new extinct primate from the Pleistocene megafauna of Bahia, Brazil. Proceedings of the National Academy of Sciences, 93, 6405–6409. Clapperton, C.M. (1993) Nature of environmental changes in South America at the Last Glacial Maximum. Palaeogeography, Palaeoclimatology, Palaeoecology, 101, 189–208. Clutton-Brock, T.H. & Harvey, P.H. (1983) The functional significance of variation in body size among mammals. Advances in the study of mammalian behaviour (ed. by J.F. Eisenberg and D.G. Kleiman), pp. 632–663. Special Publications no.7, The American Society of Mammalogists, Pittsburgh, Pennsylvania. Coetzee, J.A. (1993) African flora since terminal Jurassic. Biological relationships between Africa and South America (ed. by P. Goldblatt), pp. 37–61. Yale University Press, New Haven. Colinvaux, P.A., de Oliveira, P.A. & Bush, M.B. (2000) Amazonian and neotropical plant communities on glacial timescales: the failure of the aridity and refuge hypotheses. Quaternary Science Reviews, 19, 141–169. Coltorti, M., Ficcarelli, G., Jahren, H., Moreno Espinosa, M., Rook, L. & Torre, D. (1998) The last occurrence of Pleistocene megafauna in the Ecuadorian Andes. Journal of South American Earth Sciences, 11, 581–586. Cooke, H.B.S. (1972) The fossil mammal fauna of Africa. Evolution, mammals, and southern continents (ed. by A. Keast, F.C. Erk and B. Glass), pp. 89–139. State University of New York Press, Albany. Cristoffer, C. & Perez, C.A. (2003) Elephants versus butterflies: the ecological role of large herbivores in the evolutionary history of two tropical worlds. Journal of Biogeography, 30, 1357–1380. Dickson, K.B. (1992) The land. Africa, The New Encyclopaedia Britannica, 13, 45–51. Dublin, H.T., Sinclair, A.R.E. & McGlade, J. (1990) Elephants and fire as causes of multiple stable states in the SerengetiMara woodlands. Journal of Animal Ecology, 59, 1147–1164. Eisenberg, J.F. & Redford, K.H. (1999) Mammals of the Neotropics, the Central Neotropics, vol.3, Ecuador, Peru, Bolivia, Brazil. The University of Chicago Press, Chicago and London. Emmons, L.H. & Gentry, A.H. (1983) Tropical forest structure and the distribution of gliding and prehensile-tailed vertebrates. The American Naturalist, 121, 513–524. Journal of Biogeography 31, 943–957, ª 2004 Blackwell Publishing Ltd Vegetation change and mammal faunas in South America and Africa Emmons, L.H. & Vucetich, M.G. (1998) The identity of Winge’s Lasiuromys villosus and the description of a new genus of echimyid rodent (Rodentia: Echimyidae). American Museum Novitates, 3223, 1–12. Eva, H.D., de Miranda, E.E., Di Bella, C.M., Gond, V., Huber, O., Sgrenzaroli, M., Jones, S., Coutinho, A., Dorado, A., Guimarães, M., Elvidge, C., Achard, F., Belward, A.S., Bartholomé, E., Baraldi, A., de Grandi, G., Vogt, P., Fritz, S. & Hartley, A. (2002) A vegetation map of South America. Official Publications of the European Communities, Luxembourg. Faure, M., Guérin, C. & Parenti, F. (1999) The Holocene megafauna from the Toca do Serrote do Artur (São Raimundo Nonato archaeological area, Piauı́, Brazil). Comptes Rendus de l’Academie des Sciences, serie II fascicule A- Sciences de la Terre et des Planetes, 329, 443–448. Ficcarelli, G., Azzaroli, A., Bertini, A., Coltorti, M., Mazza, P., Mezzabotta, C., Moreno Espinosa, M., Rook, L. & Torre, D. (1997) Hypothesis on the cause of extinctions of the South American mastodons. Journal of South American Earth Sciences, 10, 29–38. Flenley, J.R. (1979) The equatorial rain forest: a geological history. Butterworths, London. Furley, P.A. (1999) The nature and diversity of neotropical savanna vegetation with particular reference to the Brazilian cerrados. Global Ecology and Biogeography, 8, 223–241. Furley, P.A. & Newey, W.W. (1983) Geography of the biosphere. An introduction to the nature, distribution and evolution of the world’s life zones. Butterworths, London. Graham, R.W. & Lundelius, E.L., Jr (1984) Coevolutionary disequilibrium and Pleistocene extinctions. Quaternary extinctions. A prehistoric revolution (ed. by P.S. Martin and R.G. Klein), pp. 223–249. The University of Arizona Press, Tucson, AZ. Gruhn, R. & Bryan, A.L. (1984) The record of Pleistocene megafaunal extinctions at Taima-taima, northern Venezuela. Quaternary extinctions. A prehistoric revolution (ed. by P.S. Martin and R.G. Klein), pp. 128–137. The University of Arizona Press, Tucson, AZ. Guérin, C., Curvello, M.A., Faure, M., Hugueney, M. & Mourer-Chauviré, M. (1996) The Pleistocene fauna of Piauı́ (northeastern Brazil): palaeoecological and biochronological implications. Revista da Fundação Museu do Homem Americano, 1, 55–103. Guilday, J.E. (1984) Pleistocene extinction and environmental change: case study of the Appalachians. Quaternary extinctions. A prehistoric revolution (ed. by P.S. Martin and R.G. Klein), pp. 250–258. The University of Arizona Press, Tucson, AZ. Haffer, J. (1982) General aspects of the refuge theory. Biological diversification in the tropics (ed. by G.T. Prance), pp. 6–24. Columbia University Press, New York. Haltenorth, T. & Diller, H. (1977) Säugetiere Afrikas und Madagaskars. BLV Verlagsgesellschaft mbH, München. Hambrey, M.J. & Harland, W.B. (1981) The evolution of climates. The evolving earth (ed. by L.R.M. Cocks), Journal of Biogeography 31, 943–957, ª 2004 Blackwell Publishing Ltd pp. 137–152. British Museum of Natural History and Cambridge University Press, London and Cambridge. Hare, F.K. (1992) Climate and weather. Climate change. The new encyclopaedia Britannica, Vol. 16, 15th edn (ed. by R. McHenry), pp. 484–497. Encyclopaedia Britannica, Inc., Chicago. Hershkovitz, P. (1972) The recent mammals of the Neotropical region: a zoogeographic and ecological review. Evolution, mammals, and southern continents (ed. by A. Keast, F.C. Erk and B. Glass), pp. 311–431. State University of New York Press, Albany. Hoffmann, W.A. & Moreira, A.G. (2002) The role of fire in population dynamics of woody plants. The Cerrados of Brazil (ed. by P.S. Oliveira and R.J. Marquis), pp. 159–177. Columbia University Press, New York. Houérou, H.N. (1997) Climate, flora and fauna changes in the Sahara over the past 500 million years. Journal of Arid Environments, 37, 619–647. Hueck, K. (1966) Die Waelder Südamerikas. Gustav Fischer Verlag, Stuttgart. Hueck, K. & Seibert, P. (1981) Vegetationskarte von Südamerika. Gustav Fischer Verlag, Stuttgart. Janis, C.M. (1993) Tertiary mammal evolution in the context of changing climates, vegetation, and tectonic events. Annual Review of Ecology and Systematics, 24, 467–500. Joly, C.A., Aidar, M.P.M., Klink, C.A., McGrath, D.G., Moreira, A.G., Moutinho, P., Nepstad, D.C., Oliveira, A.A., Pott, A., Rodal, M.J.N. & Sampaio, E.V.S.B. (1999) Evolution of the Brazilian phytogeography classification systems: implications for biodiversity conservation. Ciência e Cultura Journal of the Brazilian Association for the Advancement of Science, 51, 331–348. Kappelman, J., Plummer, T., Bishop, L., Duncan, A. & Appleton, S. (1997) Bovids as indicators of Plio-Pleistocene paleoenvironments in East Africa. Journal of Human Evolution, 32, 229–256. Kay, R.F. & Madden, R.H. (1997) Mammals and rainfall: paleoecology of the middle Miocene at La Venta (Colombia, South America). Journal of Human Evolution, 32, 161–199. Keast, A. (1972) Comparisons of contemporary mammal faunas of southern continents. Evolution, mammals, and southern continents (ed. by A. Keast, F.C. Erk and B. Glass), pp. 433–501. State University of New York Press, Albany. Kingdom, J. (1977) East African mammals. An atlas of evolution in Africa, vol. III part A (carnivores). The University of Chicago Press, Chicago. Kingdom, J. (1979) East African mammals. An atlas of evolution in Africa, vol. III part B (large mammals). The University of Chicago Press, Chicago. Ledru, M.-P., Salgado-Labouriau, M.L. & Lorscheitter, M.L. (1998) Vegetation dynamics in southern and central Brazil during the last 10,000 yr BP. Review of Palaeobotany and Palynology, 99, 131–142. Lessa, E.P., Van Valkenburgh, B. & Fariña, R.A. (1997) Testing hypotheses of differential mammalian extinctions subsequent to the Great American Biotic Interchange. 955 M. de Vivo and A. P. Carmignotto Palaeogeography, Palaeoclimatology, Palaeoecology, 135, 157–162. MacArthur, P.H. & Wilson, E.O. (1967) The theory of island biogeography. Princeton University Press, Princeton, New Jersey. MacFadden, B.J. (2000) Cenozoic mammalian herbivores from the Americas: reconstructing ancient diets and terrestrial communities. Annual Review of Ecology and Systematics, 31, 33–59. MacFadden, B.J. & Shockey, B.J. (1997) Ancient feeding ecology and niche differentiation of Pleistocene mammalian herbivores from Tarija, Bolivia: morphological and isotopic evidence. Paleobiology, 23, 77–100. McMaster, D.N. (1992) Animal life. Africa. The New Encyclopaedia Britannica, 13, 56–59. McNaughton, S.J. & Georgiadis, N.J. (1986) Ecology of African grazing and browsing mammals. Annual Review of Ecology and Systematics, 17, 39–65. McNaughton, S.J., Sala, O.E. & Oesterheld, M. (1993) Comparative ecology of African and South American arid to subhumid ecosystems. Biological relationships between Africa and South America (ed. by P. Goldblatt), pp. 548–567. Yale University Press, New Haven. Maglio, V.J. & Cooke, H.B.S. (1978) Evolution of African mammals. Harvard University Press, Cambridge. Mapaure, I.N. & Campbell, B.M. (2002) Changes in miombo woodland cover in and around Sengwa Wildlife Research Area, Zimbabwe, in relation to elephants and fire. African Journal of Ecology, 40, 212–219. Markgraf, V. (1985) Late Pleistocene faunal extinctions in southern Patagonia. Science, 228, 1110–1112. Marshall, L.G. & Cifelli, R.L. (1990) Analysis of changing diversity patterns in Cenozoic land mammal age faunas, South America. Palaeovertebrata, 19, 169–210. Marshall, L.G., Webb, S.D., Sepkoshi, J.J. & Raup, D.M. (1982) Mammalian evolution and the great American interchange. Science, 215, 1351–1357. Martin, P.S. (1967) Prehistoric overkill. Pleistocene extinctions: the search for a cause (ed. by P.S. Martin and H.E. Wright Jr), pp. 75–120. Yale University Press, New Haven. Martin, P.S. (1984) Prehistoric overkill: the global model. Quaternary extinctions. A prehistoric revolution (ed. by P.S. Martin and R.G. Klein), pp. 354–403. The University of Arizona Press, Tucson, AZ. Martin, P.S. & Klein, R.G. (1984) Quaternary extinctions. A prehistoric revolution. The University of Arizona Press, Tucson, AZ. May, R.M. (1978) The evolution of ecological systems. Scientific American, 239, 118–133. Miranda, H.S., Bustamante, M.M.C. & Miranda, A.C. (2002) The fire factor. The Cerrados of Brazil (ed. by P.S. Oliveira and R.J. Marquis), pp. 51–68. Columbia University Press, New York. Moreira, A.G. (2000) Effects of fire protection on savanna structure in Central Brazil. Journal of Biogeography, 27, 1021–1029. 956 Nowak, R.M. (1999) Walker’s mammals of the world, Vol. I/II. Johns Hopkins University Press, Baltimore and London. Nunes, A. (2001) Gradientes estruturais dos hábitats em savanas amazônicas: implicações sobre a distribuição e ocorrência das espécies de pequenos mamı́feros terrestres (Rodentia, Didelphimorphia). Doctorate Thesis, Universidade Federal do Rio de Janeiro, Rio de Janeiro. Núñez, L., Grosjean, M. & Cartajena, I. (2001) Human dimensions of late Pleistocene/Holocene arid events in southern South America. Interhemispheric climate linkages (ed. by V. Markgraf), pp. 105–117. Academic Press, San Diego. Ochsenius, C. (1985) Late Pleistocene aridity in the Neotropic as extinction cause of the South American Landmegafauna. Zentralblatt fuer Geologie und Palaeontologie Teil I Allgemeine Angewandte Regionale und Historische Geologie, H.11/12, 1691–1699. de Oliveira, P.E., Barreto, A.M.F. & Suguio, K. (1999) Late Pleistocene/Holocene climatic and vegetational history of the Brazilian Caatinga: the fossil dunes of the middle São Francisco River. Palaeogeography, Palaeoclimatology, Palaeoecology, 152, 319–337. Oliveira-Filho, A.T. & Ratter, J.A. (2002) Vegetation physiognomies and woody flora of the Cerrado biome. The Cerrados of Brazil (ed. by P.S. Oliveira and R.J. Marquis), pp. 91–120. Columbia University Press, New York. Owen-Smith, R.N. (1988) Megaherbivores – the influence of very large body size on ecology. Cambridge studies in ecology. Cambridge University Press, Cambridge. Pardinãs, U.F.J., D’Elı́a, G. & Ortiz, P.E. (2002) Sigmodontinos fósiles (Rodentia, Muroidea, Sigmodontinae) de América del Sur: estado actual de su conocimiento y prospectiva. Mastozoologia Neotropical, 9, 209–252. Patterson, B. & Pascual, R. (1972) The fossil mammal fauna of South America. Evolution, mammals, and southern continents (ed. by A. Keast, F.C. Erk and B. Glass), pp. 247–309. State University of New York Press, Albany. Pennington, R.T., Prado, D.E. & Pendry, C.A. (2000) Neotropical seasonally dry forests and Quaternary vegetation changes. Journal of Biogeography, 27, 261–273. Peters, R.H. (1983) The ecological implications of body size. Cambridge studies in ecology. Cambridge University Press, Cambridge. Rancy, A. (1999) Fossil mammals of the Amazon as a portrait of a Pleistocene environment. Mammals of the Neotropics, the Central Neotropics, vol. 3, Ecuador, Peru, Bolivia, Brazil (ed. by J.F. Eisenberg and K.H. Redford), pp. 20–26. The University of Chicago Press, Chicago and London. Redford, K.H. & Eisenberg, J.F. (1992) Mammals of the Neotropics, the Southern Cone, vol. 2, Chile, Argentina, Uruguay, Paraguay. The University of Chicago Press, Chicago and London. Reed, K.E. (1996) Early hominid evolution and ecological change through the African Plio-Pleistocene. Journal of Human Evolution, 32, 289–322. Journal of Biogeography 31, 943–957, ª 2004 Blackwell Publishing Ltd Vegetation change and mammal faunas in South America and Africa Reig, O.A. (1981) Teoria del origen y desarolo de la fauna de mamı́feros de America del Sur. Monographiae Naturae, 1, 1–162. Rizzini, C.T. (1997) Tratado de Fitogeografia do Brasil. Aspectos Ecológicos, Sociológicos e Florı́sticos. Ãmbito Cultural Edições, Rio de Janeiro. Roca, A.L., Georgiadis, N., Pecon Slattery, J. & O’Brien, J. (2001) Genetic evidence for two species of elephant in Africa. Science, 293, 1473–1477. Roy, K., Valentine, J.W., Jablonski, D. & Kidwell, S.M. (1996) Scales of climatic variability and time averaging in Pleistocene biotas: implications for ecology and evolution. Trends in Ecology and Evolution, 11, 458–463. Rudgley, R. (1999) The lost civilisations of the Stone Age. The Free Press, New York, NY. Salgado-Labouriau, M.L., Barberi, M., Ferraz-Vicentini, K.R. & Parizzi, M.G. (1998) A dry climatic event during the late Quaternary of tropical Brazil. Review of Palaeobotany and Palynology, 99, 115–129. Salvatori, V., Egunyu, F., Skidmore, A.K., de Leeuw, J. & van Gils, H.A.M. (2001) The effects of fire and grazing pressure on vegetation cover and small mammal populations in the Maasai Mara National Reserve. African Journal of Ecology, 39, 200–204. San José, J.J. & Farinãs, M.R. (1983) Changes in tree density and species composition in a protected Trachypogonsavanna, Venezuela. Ecology, 64, 447–453. San José, J.J. & Farinãs, M.R. (1991) Changes in tree density and species composition in a protected Trachypogon savanna protected for 25 years. Acta Oecologica, 12, 237–247. Sarmiento, G. (1984) The ecology of Neotropical savannas. Harvard University Press, Cambridge. Scarre, C. (1988) Past worlds. The Times atlas of archaeology. Times Books, London. Silva, C.R., Percequillo, A.R., Iack-Ximenes, G.E. & de Vivo, M. (2003) New distributional records of Blarinomys breviceps (Winge, 1888) (Sigmodontinae, Rodentia). Mammalia, 67, 147–152. Simpson, G.G. (1980) Splendid isolation. The curious history of South American mammals. Yale University Press, New Haven. Swaine, M.D., Hawthorne, W.D. & Orgle, T.K. (1992) The effects of fire exclusion on savanna vegetation at Kpong, Ghana. Biotropica, 24, 166–172. Trajano, E. & de Vivo, M. (1991) Desmodus draculae Morgan, Linares and Ray, 1988, reported for southeastern Brazil, with palaeoecological comments (Phyllostomidae, Desmodontinae). Mammalia, 55, 456–459. Vanleeuwe, H. & Gautier-Hion, A. (1998) Forest elephant paths and movements at the Odzala National Park, Congo: the role of clearings and Marantaceae forests. African Journal of Ecology, 36, 174–182. de Vivo, M. (1997) Mammalian evidence of historical ecological change in the Caatinga semiarid vegetation of northeastern Brazil. Journal of Comparative Biology, 2, 65–73. Journal of Biogeography 31, 943–957, ª 2004 Blackwell Publishing Ltd Voss, R.S. & Myers, P. (1991) Pseudoryzomys simplex (Rodentia: Muridae) and the significance of Lund’s collections from the caves of Lagoa Santa, Brazil. Bulletin of the American Museum of Natural History, 206, 414–432. Vrba, E.S. (1980) The significance of bovid remains as indicators of environment and predation patterns. Fossils in the making (ed. by A.K. Behrensmeyer and A.P. Hill), pp. 247– 271. University of Chicago Press, Chicago. Vrba, E.S. (1993) Mammal evolution in the African Neogene and a new look at the great American interchange. Biological relationships between Africa and South America (ed. by P. Goldblatt), pp. 393–432. Yale University Press, New Haven. Walter, H. (1984) Vegetation und Klimazonen – Grundriss der globalen Ökologie. Eugen Ulmer GmbH & Co., Stuttgart. Webb, S.D. (1985) Late Cenozoic dispersals between the Americas. The Great American Interchange (ed. by F.G. Stehli and S.D. Webb), pp. 357–386. Plenum, New York. Webb, S.D. (1999) Isolation and interchange: a deep history of South American mammals. Mammals of the Neotropics, the Central Neotropics, vol. 3, Ecuador, Peru, Bolivia, Brazil (ed. by J.F. Eisenberg and K.H. Redford), pp. 13–19. The University of Chicago Press, Chicago and London. Webb, S.D. & Rancy, A. (1996) Late Cenozoic evolution of the Neotropical mammal fauna. Evolution and environments in tropical America (ed. by J. Jackson, N. Budd and A. Coates), pp. 335–358. University of Chicago Press, Chicago. Whitlock, C., Bartlein, P.J., Markgraf, V. & Ashworth, A.C. (2001) The midlatitudes of North and South America during the Last Glacial Maximum and Early Holocene: similar paleoclimatic sequences despite differing large-scale controls. Interhemispheric climate linkages (ed. by V. Markgraf), pp. 391–416. Academic Press, San Diego. Whitmore, T.C. & Prance, G.T. (1987) Biogeography and Quaternary history in tropical America. Clarendon Press, Oxford. Wilson, D.E. & Reeder, D.M. (1993) Mammal species of the world. A taxonomic and geographic reference. Smithsonian Institution Press and American Association of Mammalogists, Washington and London. BIOSKETCHES Mario de Vivo is Chair of Vertebrates and Curator of Mammals and Birds at the Museu de Zoologia, Universidade de São Paulo. His main research interests are the systematics of South American mammals and their biogeography. Ana Paula Carmignotto is in the PhD Program of Zoology at the Instituto de Biociências, Universidade de São Paulo, currently developing her thesis research at the Museu de Zoologia, USP. Her interests are ecology and biogeography of open formation small mammals. 957