The historical bridge between the Amazon and the Atlantic Forest of Brazil: a study of molecular phylogeography with small mammals

Journal of Biogeography, 30, 71 86 The historical bridge between the Amazon and the Atlantic Forest of Brazil: a study of molecular phylogeography with small mammals Leonora P. Costa Museum of Vertebrate Zoology, University of California, Berkeley, Berkeley, CA, USA Abstract Aim To examine how the genetic diversity of selected taxa of forest-dwelling small mammals is distributed between and within the major rain forest domains of Amazonia and Atlantic Forest and the intervening interior forests of Brazil, as inferred by the relationships between gene genealogies and geography. I also addressed the historical importance of the central Brazilian forests in connecting Amazon and Atlantic Forest populations of rodents and marsupials. Methods I evaluated variation in the mitochondrial cytochrome b gene to estimate the levels of sequence divergence between those taxa occurring throughout the Amazon, Atlantic Forest, and forests in the Cerrado and Caatinga regions. I inferred the hierarchical relationships between haplotypes, populations and formal taxa using the cladistic approach of maximum parsimony. I compared areas and the clades identified by superimposing cladograms on the geographical distribution of samples. The degree of concordance both in phylogeny and the depth of the nodes in these phylogenies, in addition to patterns of geographical distribution of clades, permitted me to make inferences on how, when and where the taxa differentiated. Results Sequence similarity is often greater between samples from the Atlantic Forest and either Amazon or central Brazilian forests than it is within each of the two rain forest domains. The Atlantic Forest clades are either not reciprocally monophyletic or are the sister group to all the other clades. There is some indication of northern and southern components in the Atlantic Forest. Given the geographical distribution of clades and the relatively deep levels of divergence, the central Brazilian area does not behave as a separate region but is complementary to either Amazon or Atlantic Forest. Patterns of area relationships differ across taxa, suggesting that different processes and or historic events affected the diversification within each lineage. Main conclusions The Amazon and the Atlantic forests are not exclusive in terms of their small mammal faunas; both overlap broadly with taxa occurring in gallery forests and dry forests in central Brazil. Central Brazilian forests are an integral part of the evolutionary scenario of lowland small mammals, playing an important role as present and past habitats for rain forest species. Therefore, representatives from this area should always be included in analyses of the evolutionary history of lowland rain forest faunas. The incongruence of branching patterns among areas is in agreement with recent results presented for Neotropical passerine birds and indicates that a single hypothesis of Neotropical area relationships is unlikely. These findings reinforce the idea that speciation in the Neotropics will not be explained by any single model of vicariance or climatic changes. Correspondence: Departamento de Zoologia, Instituto de Ciências Biológicas, Universidade Federal de Minas Gerais, Av. Antônio Carlos, 6627, 31260-901 Belo Horizonte, MG, Brazil. E-mail: leonora@icb.ufmg.br Ó 2003 Blackwell Publishing Ltd

72 L. P. Costa Keywords Neotropical small mammals, Atlantic Forest, Amazon, Cerrado, Caatinga, phylogeography, evolutionary history of rain forest mammals. INTRODUCTION The aim of this paper is to investigate the biogeography and evolutionary relationships of the non-volant, small-mammal fauna distributed across the Amazon, Atlantic Forest and the central Brazilian forests. The Amazon and the Atlantic forests morphoclimatic domains of South America (Ab Saber, 1977) encompass the most diverse tropical forests in the world. Between these two forests lies a belt of more open vegetation, including the Argentinean and Paraguayan Chaco, the Caatinga in north-eastern Brazil, and the central Brazilian Cerrado, the latter being the second largest domain in Brazil extending over 2 million km 2. This dry corridor of open vegetation, also called ÔThe major South American disjunctionõ (Brieger, 1969), has been considered as an important restraint to species migration between the two rain forest regions (Moojen, 1948; Raven & Axelrod, 1974; Rizzini, 1979; Mori et al., 1981; Por, 1992). But if at first glance, the Amazon and Atlantic forests appear widely isolated, maps of present-day vegetation clearly shows that the ubiquitous gallery forests and series of deciduous and semi-deciduous forest patches constitute a network of interconnected forests through otherwise open landscape (Oliveira-Filho & Ratter, 1995; Vivo, 1997). Other particularities of the Cerrado and Caatinga regions provide evidence of past connections as well. Although dominated by a semi-deciduous arboreal savanna, the Cerrado is composed of a mosaic of subunits that vary from grasslands to dry forests. Along the rivers that transect this mosaic, there are strips of mesic forests, the gallery forests. Comprising <10% of the total area of the Cerrado, gallery forests occur throughout the region. Oliveira-Filho & Ratter (1995) provide many instances of floristic links of gallery forest in the Cerrado region to either Amazonian or Atlantic rain forests. Redford & Fonseca (1986), based on a literature review, showed that the greatest proportion of the Cerrado mammalian fauna is composed of species characteristic of mesic domains, and that the composition of mammal species in gallery forests is very similar to that of both the Amazon and Atlantic rain forests. Among the 100 species of mammals known to occur in the Cerrado, only eleven are endemic to this domain, with fifty-five being found also in the Amazon and sixty-four in the Atlantic rain forests. The majority of the non-endemic mammal species occur in the Cerrado mainly as a consequence of the presence of gallery forests (Redford & Fonseca, 1986). A similar pattern of endemism and habitat use was found by Silva (1995) for birds of the Cerrado, but he pointed out that dry forests, additionally to gallery forests, are keystone elements in maintaining avian diversity in the Cerrado region. Therefore, dry forests should also be investigated as important elements in the biogeographical history of the region. The Caatinga is a xerophytic vegetation type characterized by spiny deciduous trees and shrubs in association with succulent plants, cacti and bromeliads (Kuhlmann, 1977; De Oliveira et al., 1999). This near-desert vegetation surrounds inland ÔbrejoÕ forests, enclaves of moist forest caused by precipitation on the slopes of various plateaus (Mori, 1989). Botanical studies of these mesic enclaves and the northeastern coastal forests revealed floristic disjunction with either the Amazon or the Atlantic Forest (Rizzini, 1963; Bigarella et al., 1975; Coimbra-Filho & Câmara, 1996). Species of small mammals typical of rain forests [such as the slender mouse opossum Marmosops incanus (Lund) and the murid rodent Rhipidomys mastacalis (Lund)] are currently found in ÔbrejoÕ forests in the states of Ceará and Bahia (Y. Leite and L. P. Costa, unpublished data). Based on mammalian distributional patterns, the fossil record and ecological physiology, Vivo (1997) suggested that a mesic forest capable of supporting arboreal rain forest mammals existed in the area presently occupied by the Caatinga, connecting the eastern Amazonian and the Atlantic rain forests. Therefore, forests are far from a negligible component of central Brazilian vegetation, and their role as migration routes for forest species must be recognized, especially when evidence exists for a previously wider forest cover (Oliveira- Filho & Ratter, 1995; Vivo, 1997). The Amazon and Atlantic forests were probably once continuous in the past, becoming separated as increasing aridity in the Tertiary formed the belt of xeromorphic formations between them (Bigarella et al., 1975). The palynological record of the Quaternary showed that between 33,000 and 25,000 BP, the central Brazilian region was moister than today and was covered by rain forests (Ledru, 1993), and during the last glacial maximum (18,000 12,000 yr BP), the present-day corridor of xeric vegetation was covered by extensive woodland (Prado & Gibbs, 1993). These findings indicate the predominance of seasonal arboreal vegetation during most of the Pleistocene. A palynological profile from the latest Pleistocene (10,990 10,540 yr BP) from the Caatinga region revealed pollen of taxa found in the present Amazonian and Atlantic forests; the pollen concentration is high, probably reflecting a large and well-drained watershed under a climate conducive for a dense forest cover (De Oliveira et al., 1999). The extent to which these climatic fluctuations and associated vegetation changes affected the patterns of distribution and diversification of the small mammal fauna remains a central question in understanding the evolution of lowland forest communities. Recent developments in the field of molecular biology provide the tools to further investigate phylogenetic

Phylogeography of Neotropical forest small mammals 73 Figure 1 Diagram showing the distribution of the Amazon (Am) and the Atlantic Forest (AF), the Cerrado (hatched) and the Caatinga (stippled). In order to examine the evolutionary distinctiveness of the rain forest biomes, a cladistic approach was employed to determine if the sister-group relationships of small mammals were found within (left) or between (right) the Amazon and the Atlantic forests. relationships among organisms. When associated with geographical distributions, this information enables us to understand processes of diversification, to construct patterns of phylogenetic relationships and to test explicit hypotheses of biogeographical history (e.g. Smith & Patton, 1993; Patton et al., 1994). The study of the principles and processes governing the geographical distributions of genealogical lineages constitutes the new field of phylogeography, as defined by Avise et al. (1987). Previous work on phylogeography of small mammals has addressed the relationships among taxa distributed across Amazonian and Atlantic forests. In a preliminary analysis, Patton et al. (1997) utilized Cracaft s (Cracraft, 1985, 1988; Cracraft & Prum, 1988) framework for small mammals, and found the samples from the Atlantic Forest to form a monophyletic group that was also the most divergent phylogenetically. A high degree of genetic divergence (above 12%) between Atlantic Forest and Amazonian clades of three marsupial taxa (Micoureus Lesson, Philander Tiedemann, and Didelphis Linnaeus) and one rodent [Oryzomys capito (Olfers)] suggested that these two domains have been isolated for a long period of time. Mustrangi & Patton (1997) examined alternative hypotheses of relationships between Atlantic Forest and Amazonian species of Marmosops Matschie, and in this case the monophyly of the Atlantic Forest species could not be confirmed. The considerations above raise several questions regarding the evolutionary history of the Neotropical rain forest mammal fauna. How much diversification has occurred exclusively within the Amazon or the Atlantic Forest and how much diversification has been shared among these two rain forest domains (Fig. 1)? Have the central Brazilian forests played a historical role as extensions of the rain forest domains for small mammal species, enabling gene exchange through two regions otherwise isolated? Is there a common pattern of diversification across different mammal lineages, such as marsupials and murid rodents? One way to answer these questions is to integrate molecular data under a phylogenetic framework for multiple taxa with present-day geographical distributions. The work I report here is an extension of that previously carried out by Patton et al. (1997), and the specific topics covered in this paper are: (1) How phylogenetically distinct are the small mammal taxa of the Amazon with respect to their Atlantic Forest relatives, and vice-versa? (2) Do the central Brazilian gallery and dry forests play a role in largescale patterns of geographical distribution of the genetic diversity and differentiation of lowland rain forest small mammals, acting as connections between the Amazon and the Atlantic Forest? If so, what are the phylogenetic affinities of the small mammals living in these interior forests with respect to the mammals living in the two rain forest domains? (3) Given the geographical distribution of clades, is there a consistent pattern of area-relationship across taxa? METHODS I focused my analysis on geographical clades of eleven monophyletic groups of small mammals, including representatives of five genera (Caluromys J. A. Allen, Philander, Micoureus, Marmosops, Gracilinanus Gardner and Creighton) and two widespread species of marsupials [Metachirus nudicaudatus (Desmarest) and Marmosa murina (Linnaeus)], two groups of Oryzomyini rodents (Oryzomys megacephalus group and O. macconnelli group), and two arboreal sigmodontine rodent genera (Rhipidomys Tschudi and Oecomys Thomas). Twenty-two collection sites within gallery forests, semi-deciduous forests and dry forests in central and north-eastern areas in Brazil were sampled. Non-volant small mammals were trapped using live- and snap-traps, and prepared according to standardized museum procedures. Liver tissue samples were collected in the field and preserved in alcohol or liquid nitrogen. DNA was extracted using the ChelexÒ (Bio-Rad, Hercules, CA, USA) method (Walsh et al., 1991), amplified by the polymerase-chain reaction (PCR) (Saiki et al., 1988), and sequenced using an automated sequencer (ABI-Prism 377, Perkin Elmer, Foster City, CA, USA) in the Laboratory of Evolutionary Genetics of the Museum of Vertebrate Zoology. The data set varied from 327 to 801 bp of the cytochrome b (cyt b) gene. One hundred and sixty new sequences were added to an existing data base of 194 sequences of Amazonian and Atlantic Forest small mammals gathered by J. L. Patton and collaborators (Patton et al., 1996, 2000; Mustrangi & Patton, 1997; Smith & Patton, 1999; Patton & Costa, in press; Costa et al., in press). A list of specimens and locality data used in the present report can be found in Costa (2001). Variation in the cyt b gene was evaluated in order to estimate the levels of sequence divergence between those taxa occurring throughout the Amazon, the Atlantic Forest, and forests in the Cerrado and Caatinga regions. Sequence divergence was calculated using the Kimura two-parameter (K2p) algorithm (Kimura, 1980) as implemented in PAUP* version 4.0b8 (Swofford, 1999). Hierarchical relationships among haplotypes, populations and formal taxa were

74 L. P. Costa inferred using the cladistic approach of maximum parsimony using PAUP*. Trees were constructed using the heuristic search option via stepwise addition, with ten replicates and random addition sequence of taxa. The support for internal branches was evaluated by decay indices (Bremer, 1988) and bootstrap analyses (Felsenstein, 1985), with 100 bootstrap replicates each consisting of a full heuristic search as described above. I compared areas and the clades identified by superimposing cladograms on maps of the geographical location of samples. The concordance of area cladograms and the depth of the nodes in these phylogenies permitted me to make inferences on how, when and where the taxa differentiated and thereby to address the historical importance of the central Brazilian forests in connecting Amazon and Atlantic Figure 2 Area relationships of eight taxa of small mammals. (a) All Amazonian areas form a clade relative to the Atlantic Forest (Philander). (b) The Atlantic Forest is deeply nested in a more inclusive clade in four cases (M. nudicaudatus, M. murina, O. megacephalus group and O. trinitatis group). (c) In three cases (Rhipidomys, Caluromys and Marmosops) the Atlantic Forest has a paraphyletic status. The samples from the central area (CA) are also deeply nested in all phylograms in which they appear. Tree topologies are based on bootstrap consensus trees; therefore clades shown are supported by at least 50% bootstrap values. The black dots indicate clades supported by more than 75% bootstrap values.

Phylogeography of Neotropical forest small mammals 75 Forest populations of rodents and marsupials. The question of phylogenetic distinctiveness of the rain forest domains was examined under a cladistic evolutionary approach. In other words, I asked if sister-group relationships were concentrated within or between the Amazon and the Atlantic Forest (Fig. 1). To explore the discrepancy in time of splitting events in the many taxa utilized in the analyses, I estimated branch lengths of the trees by maximum likelihood using third position changes only, constraining and not constraining a molecular clock. Then, I performed a likelihood ratio test (LRT) of the molecular clock, and used only the taxa for which the LRT was not significantly different (Huelsenbeck & Rannala, 1997). Calibration of the clock was carried out for marsupials using the estimates of Mustrangi & Patton (1997) for Marmosops, and for the rodents using the parameters of Smith & Patton (1999). RESULTS The phylogeographical relationships between the Amazon and the Atlantic Forest From the eleven groups analysed, three (Micoureus, Gracilinanus and Oryzomys macconnelli group) showed polytomic relationships, and for that reason could not be used in the inference of sister-group relationships between geographical clades in the two rain forest domains. Among the eight taxa remaining (Fig. 2), in only one group, Philander (Fig. 2a), was the Atlantic Forest (AF) clade the sister group to all other areas. In four of eight cases, the Atlantic Forest clade was deeply nested in a more inclusive clade (Fig. 2b), while in three cases it showed up as a composite area (Fig. 2c). A more detailed example of the last case can be seen in Fig. 3, which depicts the phylogeographical relationships of five species of the climbing rat Rhipidomys. There are two species in the Atlantic Forest, Rhipidomys mastacalis in the north and an undescribed species in the south. These two species are not sister groups. Rather, the southern species has a sister relationship with R. leucodactylus Tschudi, a species from the west-central Amazon, while R. mastacalis is closely related to R. nitela Thomas, a species occurring in the central Brazilian area, the two forming the sister group to the species of the south-western Amazon, R. gardneri Patton, da Silva and Malcolm. Another way to examine the evolutionary distinctiveness of the domains is to compare the genetic distances among haplotypes of widely distributed species. For these small mammals, K2p distances are commonly greater among localities within domains than across domains. An example of this is the woolly mouse opossum, Micoureus demerarae (Thomas). This species occurs throughout the Amazon and the central Brazilian forests, reaching the Atlantic Forest in coastal Brazil (Fig. 4a). Haplotypes from the most eastern Amazon locality are closer to those from the northern Atlantic Forest than they are to haplotypes from other Amazonian localities (Fig. 5). Similarly, haplotypes of the brown foureyed opossum, Metachirus nudicaudatus (Figs 4b & 5), and Figure 3 Phylogeographical relationships among cytochrome b haplotypes of five species of the climbing rat, genus Rhipidomys. The map shows the range of the genus Rhipidomys and the geographical distribution of the samples. Note that the two species that occur in the Atlantic Forest are not sister groups. Bold numbers at nodes are bootstrap values and decay indices, respectively. Percentages are average Kimura two-parameter distances.

76 L. P. Costa Figure 4 Maps of the range and geographical distribution of the samples of species of Micoureus (a) and Metachirus nudicaudatus (b). The phylogenetic relationships among haplotypes (shown to the right) indicate the existence of northern and southern components in the Atlantic Forest. For Micoureus, solid lines circumscribing samples correspond to species boundaries and dashed lines indicate clades within species; symbols correspond to each of the five species. For Metachirus, solid lines circumscribe intraspecific phylogeographical units represented by different symbols and dashed lines indicate clades within such units. Bold numbers at internal nodes are bootstrap values and decay indices, respectively; percentages are average Kimura two-parameter distances. the murine opossum, Marmosa murina (Linnaeus) (Figs 5 & 6a), from different localities in the Atlantic Forest are genetically more similar to those compared from the Amazon than the Amazonian haplotypes are to each other. This pattern can also be observed for species within the same complex. As an example, Oryzomys megacephalus Fisher from the northern Amazon and central Brazil is more closely related to O. laticeps (Lund) from the Atlantic Forest than to O. perenensis (Allen) from the western Amazon (Fig. 7). In terms of the Atlantic Forest itself, there are northern and southern components, as exemplified by Rhipidomys (Fig. 3), Micoureus (Fig. 4a) and Metachirus Burmeister (Fig. 4b). In both Rhipidomys and Micoureus, clades corresponding to two different species replace one another from north to south, while in Metachirus nudicaudatus, there are two geographically structured clades of the species in the Atlantic Forest. The break in the distribution of species and clades of these taxa is nearly coincidental. Nevertheless, the K2p distances within each of these clades are different, suggesting that vicariant events affected taxa at different times. Cyclic climatic fluctuations are one plausible explanation for this coincidental pattern in space but not in time. The phylogeographical affinities of the Cerrado and Caatinga domains The samples from the intervening region between the Atlantic Forest and Amazonia, including the Cerrado and Caatinga domains and for simplification called here the central area (CA), were deeply nested in all phylogenies

Phylogeography of Neotropical forest small mammals 77 Figure 5 Maps of the geographical location and average Kimura 2-parameter distances of selected haplotypes of three taxa of didelphid marsupials (Micoureus demerarae, Metachirus nudicaudatus and Marmosa murina). Note the small genetic distances between samples from the Atlantic Forest and the Amazon, especially when compared with the genetic distances separating samples from localities within the Amazon itself. The Amazon and the Atlantic Forest are shown in dense stippling, the Cerrado hatched and the Caatinga in sparse stippling. obtained (Figs 2 7). This fact attests to the importance of this broad area in the diversification of rain forest small mammals. Animals inhabiting this region could have their closest relatives in the Atlantic Forest, in the Amazon, or be basal relative to the two rain forest domains (Fig. 8). The results, derived from analyses of all trees in Figs 2 7, are summarized in Table 1. Five of eleven groups could not be used in this analysis, because two of them are absent in one of the areas and three showed polytomic relationships (for details on these taxa, see Table 1). As one might expect when dealing with forest animals living in a somewhat more foreign environment, the samples from the CA were never basal to their relatives in the forest domains. More often the CA clade was the sister to an Atlantic Forest clade than to an Amazonian clade, as is the case for Marmosa murina (Figs 2b & 6a), Caluromys (Figs 2c & 6b), Rhipidomys (Figs 2c & 3) and the Oecomys trinitatis group (Fig. 2b). I also looked at individual localities in the CA to see if the affiliation of this area and the two rain forest domains was only a matter of geographical distance from the source in either the Atlantic Forest or the Amazon (Fig. 9). Under the assumption of isolation by distance, samples from localities in the CA situated near the Amazon should share a more recent common ancestor with Amazonian samples, whereas those from localities closer to the Atlantic Forest should be related to those in Atlantic Forest. To explore this hypothesis, I took into account all localities within the Cerrado and Caatinga, and those in the transition areas between these two domains and the lowland rain forests, but excluded those from the central Amazon and Atlantic Forest themselves. At each CA locality, only taxa that had relationships with either the Amazon or the Atlantic Forest were considered. These comparisons (Fig. 9) do not present the clear-cut trend hypothesized above. Rather, in some cases specimens in localities near the Atlantic Forest share a more recent common ancestor with Amazonian groups, whereas in the transition zone between the Cerrado and the Amazon, specimens were often related to Atlantic Forest individuals. It is also interesting to note that localities to the southwest of the CA tend to be more affiliated to the Amazon, while localities in the northeast are more often related to the Atlantic Forest (Fig. 9). This observation suggests possible routes and directions of interchange between the Amazon and the Atlantic Forest. Patterns and timing of diversification One of the aims of this paper was to search for congruent patterns of diversification linking subareas in each domain. Nevertheless, while some genera or group of species show a strong hierarchical structure over the whole sampled geographical area, others, although geographically structured, show little resolution with respect to the relationships

78 L. P. Costa Figure 6 Maps showing the range and geographical distribution of the samples of species of Marmosa murina (a) and Caluromys (b). The phylogenetic relationships among haplotypes are shown to the right. For Caluromys, solid lines circumscribing samples correspond to species boundaries and dashed lines indicate clades within species; symbols correspond to each of the two species. For M. murina, solid lines circumscribe intraspecific phylogeographical units represented by different symbols and dashed lines indicate clades within such units. Bold numbers at internal nodes are bootstrap values and decay indices, respectively; percentages are average Kimura two-parameter distances. among species or geographical clades within species. Two closely related species groups of the rice rat and representatives of two sister genera of didelphids illustrate this situation in Fig. 10. The Oryzomys megacephalus group (Fig. 10a) is comprised of three species that replace one another from east to west, with two (O. laticeps and O. megacephalus) more closely related species in the Atlantic Forest and northern Amazon south central Brazil, and a more distant one (O. perenensis) in the western Amazon. In contrast, rice rats of the Oryzomys macconelli group (Fig. 10b), although strongly divergent and geographically bounded, do not show the same level of resolution among the five species included in the complex. Similarly, the geographical clades of Marmosa murina (Fig. 10c) are structured into three sharply defined clades hierarchically organized, while the geographical clades of Micoureus demerarae (Fig. 10d) are linked in the form of a basal polytomy. Other taxa examined present similar contrasts. Among eight taxa that are widespread in the Amazon, Atlantic Forest and CA, five show resolved area relationships while four do not (Table 2). Among the taxa that have restricted distributions or are absent in one of the domains, two have resolved area relationships while one does not (Table 2). As groups of closed related taxa (e.g. sister groups Marmosa Micoureus and Gracilinanus Marmosops) can present resolved or unresolved area relationships, independently of the geographical range of these taxa, I conclude that the lack of area

Phylogeography of Neotropical forest small mammals 79 Figure 7 Phylogeographical relationships and geographical distribution of three species of the rice rat Oryzomys of the group megacephalus. Oryzomys megacephalus from northern Amazon and central Brazil is more closely related to O. laticeps from the Atlantic Forest than to O. perenensis from western Amazon. Bold numbers at internal nodes are bootstrap values and decay indices, respectively; percentages are average Kimura two-parameter distances. Solid lines circumscribing samples on the map correspond to species boundaries and dashed lines indicate the northern and southern clades within O. megacephalus. Figure 8 Area cladograms depicting possible phylogeographical affinities of the central area (CA) under three hypotheses: the samples from the CA could be more closely related to those from the Atlantic Forest (left), to Amazon samples (middle), or be basal relative to the two forest biomes (right). cladogram resolution is neither a matter of phylogenetic constraint nor of distributional pattern. In Fig. 11 I present clock-constrained trees for all taxa for which the clock was not rejected by the LRT. Clearly, the gene trees of different taxa coalesce at different times. Although cladogenesis and gene divergence are not necessarily coincident (Edwards & Beerli, 2000), the magnitude of the differences in coalescence times observed in Fig. 11 suggests that speciation in many taxa could have been affected by different historical events. Although molecular clocks are not always reliable indicators of absolute time (Ayala, 1997), taxon-specific local clocks are probably widespread (Edwards & Beerli, 2000). Therefore, one can infer from Fig. 11 that several splits occurred prior to the Pleistocene (shaded area). Alternatively, there is some concordance which indicates concurrent vicariant events: the split of the Atlantic Forest species Philander frenatus (Olfers) from all others in the genus is nearly coincidental with the split of the three species of Gracilinanus. Similarly, there is a simultaneous split between the north-western Amazonian species of Rhipidomys (R. macconelli de Winton and R. wetzeli Gardner) and western Amazonian species of Marmosops [M. neblina Gardner and M. noctivagus (Tschudi)] from other species in their complex. DISCUSSION Cracraft & Prum (1988) hypothesized historical relationships among putative centres of endemism in the Neotropics through successive vicariant events for four clades of birds. Interestingly, the only discordant pattern between

80 L. P. Costa Taxa Central area Sister to the Atlantic Forest Sister to the Amazon Basal to the Atlantic Forest and to the Amazon Table 1 Sister-group relationships of six groups of small mammals distributed in the Amazon, Atlantic Forest (AF) and central area* Oryzomys megacephalus group Oecomys trinitatis group Marmosa murina Philander Caluromys Rhipidomys *Marmosops and Metachirus are absent in the central area, and the relationship is polytomic for Micoureus, Gracilinanus and taxa in the Oryzomys macconnelli group. Figure 9 Pie-charts showing sister-group relationships of samples at individual localities in the central area of Brazil. At each locality, only taxa that had relationships with either the Amazon or the Atlantic Forest were considered. Samples from each locality were scored as either sister to Amazon or Atlantic Forest samples. The size of the pie is proportional to the number of samples examined at each locality. phylogeny and geography involved the Atlantic Forest, which appeared in their analysis as a composite, or biogeographical ÔhybridÕ area, having some component taxa nested within the Amazonian clade while others were outside this clade. Patton et al. (1997) found the Atlantic Forest as basal or polytomic in relation to other areas in the Amazon basin and Central America, for six generic or infrageneric groups of didelphid marsupials and two species complexes of the murid rodent Oryzomys Baird. In my expanded analyses, including the same taxa examined by Patton et al. (1997) as well as additional taxa, the Atlantic Forest was often nested in a more inclusive clade (Fig. 2b), but similar to Cracaft and Prum s results, it also showed up as a composite area (Fig. 2c). The main difference between my work and the previous analyses by Patton and collaborators is the inclusion of samples from the region between the Amazon and Atlantic Forest. As a result, if earlier we viewed the Atlantic Forest as a disconnected area, long separated from the Amazon, this view has drastically changed with the inclusion of samples from the central Brazilian forests. Given the geographical distribution of clades and the levels of divergence found, the central Brazilian area does not behave as a separate region, but is complementary to either Amazon or Atlantic forests. As a consequence, the central Brazilian forests are an integral part of the evolutionary history of lowland small mammals. Their role as present and past habitats for forest species must be recognized, and representatives from this area should always be included in the analyses of the evolutionary history of lowland rain forest faunas. In terms of the affinities of the central Brazilian area, there is a slight trend towards a closer association of the small mammals living in forest habitats in the central region with those from the Atlantic Forest, as revealed by the analyses of sister group relationships (Table 1). Nevertheless, the analyses of individual localities summarized in Fig. 9 did not indicate the same result. Hence, at this point the data suggest that the CA region has affinities with both the Atlantic Forest and the Amazon. This result is not unexpected, because many taxa contemplated in this study belong to old lineages (Fig. 11) that experienced numerous contractions and expansions of the Cerrado vegetation over the Pleisto-

Phylogeography of Neotropical forest small mammals 81 Figure 10 Differences in hierarchical structure among the cladograms of taxa of the rice rat Oryzomys and the didelphid species Marmosa murina and Micoureus demerarae. Relationships among species of the O. megacephalus group and the geographically structured clades of M. murina are fully resolved, while the same level of resolution is not observed for the species in the O. macconnelli group and the intraspecific clades of M. demerarae. Symbols on the maps and cladograms correspond to the different species of Oryzomys, and to intraspecific phylogeographical units of M. murina and M. demerarae. Numbers at nodes are bootstrap values (above) and decay indices (below). Table 2 Pattern of distribution and area relationships for eleven groups of small mammals Pattern of distribution Area relationships Resolved Not resolved Widespread in the Amazon, Atlantic Forest and central area Oryzomys megacephalus group Marmosa murina Caluromys Oecomys trinitatis group Rhipidomys Oryzomys macconnelli group Micoureus demerarae Philander Restricted or absent in Gracilinanus Marmosops one of the areas Metachirus cene and preceding geological eras. Nevertheless, a decisive assessment of the affinities of the CA with the rain forest domains will have to wait until more phylogenies are available. The partitioning of the Amazon into several highly divergent phylogeographical areas has already been documented (Cracraft & Prum, 1988; Patton et al., 1997, 2000). In contrast, the division of the Atlantic Forest as indicated by the phylogenetic analyses in this paper has been unrecognized for small mammals until very recently. Vanzolini (1988), Bates et al. (1998) and Costa et al. (2000) have noted a latitudinal break in the distribution of Atlantic Forest vertebrates, which separates the north from the south in reptiles, lowland birds and lowland mammals, the latter two using parsimony analysis of endemicity. Although these analyses found a strong sister-group relationship between the northern and the southern Atlantic Forest clades, in my analysis I found a different pattern: often, taxa occurring in northern and southern Atlantic Forest were not sister taxa. Species of Rhipidomys and Micoureus illustrate this situation in Figs 3 and 4a. The species relationships of the genus Marmosops are likely

82 L. P. Costa Figure 11 Clock-constrained phylogenetic trees for six genera of didelphid marsupials (above) and three genera of sigmodontine rodents (below). Only taxa for which the clock was not rejected by the likelihood ratio test were included. The results indicate that most of the splits probably occurred prior to the Pleistocene (shaded area). another example (see Mustrangi & Patton, 1997). These results further strengthen the likelihood that the Atlantic Forest is a composite area. One of the main observations from the data presented here is the lack of congruence of area cladograms across taxa, a result concordant with recent data for other groups of taxa. To date, the question of area relationships between and within the Atlantic Forest and Amazonia has addressed primarily avian taxa. Bates et al. (1998), studying Neotropical lowland passerine birds and using raw geographical distributions of species and subspecies, found many different hypotheses of area relationships and concluded that a single network of Neotropical area relationships is not likely. Zink et al. (2000) examined six lineages of birds occurring in the arid lands of North America and obtained different area cladograms depending on how widespread and missing taxa were coded. Nonetheless, no cladogram was obtained in which all lineages were congruent. Are these results just an artefact of the materials and methods utilized or do they reflect reality? I argued above that the lack of congruence in cladograms obtained for the many taxa examined here was not because of phylogenetic constraints or the patterns of distribution of taxa. Other reasons for the lack of congruence in phylogeographical patterns were thoroughly discussed by Zink et al. (2001) and include differences in levels of gene flow among populations, different responses of species to the same environmental or geological feature that could have acted or not as a barrier to gene flow, or different Ôresponse timesõ of species effectively isolated by the same barrier, because of differences in their effective population size. All the above explanations assume vicariance as the historical process shaping the distribution of taxa and ultimately leading to phylogeographical congruence. Nevertheless, the effects of dispersal confounding the signal in a phylogeographical analysis cannot be ignored (Avise, 2000). As pointed out by Ronquist (1997), dispersal barriers appear and disappear throughout earth history and may have different effects on the evolutionary history of different species. Zink et al. (2000) evaluated the relative roles of vicariance and dispersal for a series of avian lineages from the arilands of North America, and found that although vicariance was the dominant mode of evolution, at least 25% of speciation events could have been derived from dispersal across a preexisting barrier. In contrast to the scarcity of investigations about the significance of potential forest corridors linking populations of mammals and Neotropical faunas in general, botanists have addressed the question of corridors for plants since the early

Phylogeography of Neotropical forest small mammals 83 1960s, and two possible migration routes between the Amazon and the Atlantic forests have been suggested. Rizzini (1963) and Andrade-Lima (1964) refer to a migration route through a mesophytic forest corridor that would have crossed the Caatinga at certain periods since the late Tertiary. The montane forests (brejos) that presently exist within the semi-arid region would be in fact relicts of an ancient and wider forest cover (Andrade-Lima, 1982). Bigarella et al. (1975) added a second route, which they called the Southeast-Northwest bridge, based on floristic similarity between the eastern Amazon and south-eastern Atlantic Forest. As suggested by Oliveira-Filho & Ratter (1995), this route could have run through central Brazil, either as a continuous forest corridor or as a series of forest patches between which island-hopping would have occurred. Finally, Por (1992) has noted three historical ÔpathwaysÕ of connection between Amazonia and the Atlantic Forest: a major southern route through the Paraná River basin, a secondary route to the northeast around the horn of Brazil, and a minor route via the gallery forests along rivers of the central Brazilian Cerrado (Fig. 12). All three routes could have been important for small mammals, because connections between these two rain forest domains are seen in the phylogeography of individual species. For both Micoureus demerarae (Figs 4a & 5) and Caluromys philander (Linnaeus) (Fig. 6b), populations from coastal Brazil are linked with those in eastern Amazonia (Fig. 12), while for Metachirus nudicaudatus (Figs 4b & 5) and Caluromys lanatus (Olfers) (Fig. 6b), the linkage is between coastal Brazil and south-western Amazonia (Fig. 12). For Marmosa murina, the path of linkage is equivocal as each of the three routes between the coastal and Amazon forests are supported (Fig. 6a & 12). Levels of sequence divergence in each of these cases are relatively low (1 5%), suggesting that the connections are more recent, and quite likely result from Pleistocene vegetation shifts in the region. In addition to the factors responsible for incongruence in area cladograms discussed above, many other events could have affected the history of the lineages discussed here. In general, the old idea of biological and geological stability of the tropics (Sanders, 1969) has been replaced in a view of historical dynamism. There are ecologically orientated hypotheses, like the Ecological Gradients Hypothesis (Endler, 1982); there are hypotheses dealing with river dynamics (Wallace, 1952; Salo et al., 1986); there are those that incorporate tectonic history (Patton et al., 2000); still others link the glacial cycles with sea-level changes creating marine incursions or continental freshwater lakes (Klammer, 1984; Marroig & Cerqueira, 1997) and finally there is the idea of climatic changes favouring the formation of forest refuges (Haffer, 1969; Vanzolini & Williams, 1970). These various hypotheses are not mutually exclusive and, indeed, several could have influenced the evolutionary history of any particular lineage. It should not be surprising that separate taxon lineage could have been affected by different events. There is also a temporal component to be accounted for, specifically the age of each lineage and when the splitting Figure 12 The distribution of lowland rain forests in South America. The Amazon is divided into the four regions initially recognized in 1852 by Alfred Russel Wallace. The ÔpathwaysÕ connecting the Amazon and the Atlantic Forest of coastal Brazil hypothesized by Por (1992) are indicated. For these, the thickness of the arrows indicates the degree of past biogeographical connection. Below are two hypothetical cladograms depicting area relationships between the Amazon (AM) and the Atlantic Forest (AF). Top: south-eastern AM exhibits connection to Atlantic Forest; bottom: south-western AM exhibits connection to Atlantic Forest. Beside each cladogram are examples of didelphid species with phylogeographical patterns that fit each hypothesis. events occurred. It is apparent from Fig. 11 that there are major differences in times of lineage diversification and that most of these small-mammal lineages are very old, some going back at least 8 Myr. These age estimates are so old that one of the more influential hypotheses formulated to explain Neotropical diversity, the Pleistocene Refugia Hypothesis (Haffer, 1969; Vanzolini & Williams, 1970), cannot be responsible for most of the speciation events in these lineages, because times of divergence are mostly pre- Pleistocene. A molecular systematic study of passerine birds

84 L. P. Costa (Zink & Slowinski, 1995) also showed that diversification rates were lower in the Pleistocene than for the preceding period. All the reasons discussed above indicate that many regional events, varying in scale and age, were responsible for the patterns we observe today. Therefore, I suggest that the incongruence of area relationships reflects reality. These findings also reinforce the idea that speciation in the Neotropics will not be explained by any single model of vicariance or climatic changes. As a result, and given the general scarcity of phylogenies for Neotropical lineages, at this time we should focus on recording specific biogeographical patterns rather than seeking to identify processes of diversification of the whole Neotropical fauna. ACKNOWLEDGMENTS I thank each of the following colleagues for providing specimens used in the research and or for aiding in the field collections: Y. L. R. Leite, J. L. Patton, L. H. Emmons, L. Geise, R. T. de Moura, M. D. Engstrom, A. L. Gardner, L. Lessa, J. R. Malcolm, A. Paglia, M. J. de J. Silva, N. da Silva, M. N. F. da Silva, R. M. Timm, R. S. Voss, T. A. Yates, N. Cáceres, M. T. da Fonseca, A. M. Fernandes, D. C. Bianchini, L. G. Vieira, F. P.C. Santos, R. L. Dias and M. A. L. Sábato. J. L. Patton, D. B. Wake, R. Gillespie, Y. L. R. Leite and P. Stott read the manuscript and provided helpful comments for its improvement. M. F. Smith, S. Morshed and Y. Chaver provided essential help in the laboratory and Y. L. R. Leite aided in the preparation of figures. Financial support for either laboratory or field work was obtained from the National Geographic Society, World Wildlife Foundation, Berkeley Chapter of Sigma-Xi, Museum of Vertebrate Zoology of the University of California, Berkeley, John D. and Catherine T. MacArthur Foundation, and grants from the National Science Foundation to J. L. Patton. Permits for collection of specimens were kindly provided by IBAMA (Instituto Brasileiro do Meio Ambiente) and by the Instituto Estadual de Florestas de Minas Gerais. During four years of my doctoral studies, I was supported by a fellowship from CAPES (Fundação Coordenação de Aperfeiçoamento de Pessoal de Nível Superior). REFERENCES Ab Saber, A.N. (1977) Os domínios morfoclimáticos da América do Sul. Primeira aproximação. Geomorfologia (São Paulo), 53, 1 23. Andrade-Lima, D. (1964) Contribuição à dinâmica da flora do Brasil. Arquivos do Instituto Ciências da Terra, 2, 15 20. Andrade-Lima, D. (1982) Present-day forest refuges in Northeastern Brazil. Biology diversification in the tropics (ed. G.T. Prance), pp. 245 251. Plenum Press, New York. Avise, J.C. (2000) Phylogeography: the history and formation of species, p. 447. Harvard University Press, Cambridge, MA, USA. Avise, J.C., Arnold, J., Ball, R.M., Bermingham, E., Lamb, T., Neigel, J.E., Reeb, C.A. & Saunders, N.C. (1987) Intraspecific phylogeography the mitochondrial DNA bridge between population genetics and systematics. Annual Review of Ecology and Systematics, 18, 489 522. Ayala, F.J. (1997) Vagaries of the molecular clock. Proceedings of the National Academy of Science, USA, 94, 7776 7783. Bates, J.M., Hackett, S.J. & Cracraft, J. (1998) Area-relationships in the neotropical lowlands: an hypothesis based on raw distributions of passerine birds. Journal of Biogeography, 25, 783 793. Bigarella, J.J., Andrade-Lima, D. & Riehs, P.J. (1975) Considerações a respeito das mudanças paleoambientais na distribuição de algumas espécies vegetais e animais no Brasil. Anais da Academia Brasileira de Ciências, 47, 411 464. Bremer, K. (1988) The limits of amino-acid sequence data in angiosperm phylogenetic reconstruction. Evolution, 42, 795 803. Brieger, F.G. (1969b) Contribuição à fitogeografia do Brasil com referência especial as orquídeas. Anais do XX Congresso Nacional de Botânica, 41 44. Coimbra-Filho, A.F. & Câmara, I.G. (1996) Os limites originais do bioma Mata Atlântica na Região Nordeste do Brasil, p. 86. FBCN Fundação Brasileira para a Conservação da Natureza, Rio de Janeiro. Costa, L.P. (2001) Phylogeographic studies on Neotropical small mammals, PhD Thesis, University of California, Berkeley. Costa, L.P., Leite, Y.L.R., Fonseca, G.A.B.d. & Fonseca, M.T.d. (2000) Biogeography of South American forest mammals: endemism and diversity in the Atlantic Forest. Biotropica, 32, 872 881. Costa, L.P., Leite, Y.L.R. & Patton, J.L. (in press) Phylogeography and systematic notes on two species of Gracilinanus (Marsupialia, Didelphidae). Proceedings of the Biology Society of Washington. Cracraft, J. (1985) Historical Biogeography and patterns of differentiation within the South American avifauna: areas of endemism. Ornithological Monographs, 36, 49 84. Cracraft, J. (1988) Deep-history biogeography retrieving the historical pattern of evolving continental biotas. Systematic Zoology, 37, 221 236. Cracraft, J. & Prum, R.O. (1988) Patterns and processes of diversification: speciation and historical congruence in some Neotropical birds. Evolution, 42, 603 620. De Oliveira, P.E., Franca Barreto, A.M. & Suguio, K. (1999) Late Pleistocene Holocene climatic and vegetational history of the Brazilian caatinga: the fossil dunes of the middle Sao Francisco River. Palaeogeography Palaeoclimatology Palaeoecology, 152, 319 337. Edwards, S.V. & Beerli, P. (2000) Perspective: gene divergence, population divergence, and the variance in coalescence time in phylogeographic studies. Evolution, 54, 1839 1854. Endler, J.A. (1982) Pleistocene forest refuges: fact or fancy? Biological diversification in the tropics (ed. G.T. Prance), pp. 641 657. Columbia University Press, New York. Felsenstein, J. (1985) Confidence limits on phylogenies: an approach using the bootstrap. Evolution, 39, 783 791.

Phylogeography of Neotropical forest small mammals 85 Haffer, J. (1969) Speciation in Amazonian forest birds. Science, 165, 131 137. Huelsenbeck, J.P. & Rannala, B. (1997) Phylogenetic methods come of age: testing hypotheses in an evolutionary context. Science, 276, 227 232. Kimura, M. (1980) A simple method for estimating evolutionary rate of base substitutions through comparative studies of nucleotide sequences. Journal of Molecular Evolution, 16, 111 120. Klammer, G. (1984) The relief of the extra-andean Amazon basin. The Amazon: limnology and landscape ecology of a mighty tropical river and its basin (ed. H. Sioli), pp. 47 83. Dr W. Junk Publishers, Dordrecht. Kuhlmann, E. (1977) A Vegetação. Geografia do Brasil Região Nordeste, Vol. 2, pp. 85 110. IBGE, Rio de Janeiro. Ledru, M.-P. (1993) Late quaternary environmental and climatic changes in central Brazil. Quaternary Research, 39, 90 98. Marroig, G. & Cerqueira, R. (1997) Plio-Pleistocene South American history and the Amazon Lagoon Hypothesis: a piece in the puzzle of Amazonian diversification. Journal of Comparative Biology, 2, 103 119. Moojen, J. (1948) Speciation in the Brazilian spiny rats (Genus Proechimys, Family Echimyidae). University of. Kansas Publications, Museum of Natural History, 1, 301 406. Mori, S.A. (1989) Eastern, Extra-Amazonian Brazil. Floristic inventory of tropical countries: the status of plant systematics, collections, and vegetation, plus recommendations for the future (ed. D.G. Campbell and H.D. Hammond), pp. 545. New York Botanical Garden, Bronx, NY, USA. Mori, S.A., Boom, B.M. & Prance, G.T. (1981) Distribution patterns and conservation of eastern Brazilian coastal forest tree species. Brittonia., 33, 233 245. Mustrangi, M.A. & Patton, J. (1997) Phylogeography and systematics of the slender mouse opossum Marmosops (Marsupialia, Didelphidae). University of California Publications, Zoology, 130, 1 86. Oliveira-Filho, A.T. & Ratter, J.A. (1995) A study of the origin of central Brazilian forests by the analysis of plant species distribution patterns. Edinburgh Journal of Botany, 52, 141 194. Patton, J.L. & Costa, L.P. (in press) Molecular phylogeography and species limits in rain forest didelphid marsupials of South America. Predators with pouches: the biology of carnivorous marsupials (eds M. Jones, C. Dickiman and M. Archer). Australia. Patton, J.L., da Silva, M.N.F. & Malcolm, J.R. (2000) Mammals of the Rio Juruá and the evolutionary and ecological diversification of Amazonia. Bulletin of the American Museum of Natural History, 244, 1 306. Patton, J.L., Reis, S.F. & Silva, M.N.F. (1996) Relationships among didelphid marsupials based on sequence variation in the mitochondrial cytochrome b gene. Journal of Mammalian Evolution, 3, 3 29. Patton, J.L., Silva, M.N.F., Lara, M.C. & Mustrangi, M.A. (1997) Diversity, differentiation, and the historical biogeography of nonvolant small mammals of the neotropical forests. Tropical forest remnants: ecology, management, and conservation of fragmented communities (ed. W.F. Laurance and R.O. Bierregaard Jr), pp. 455 465. University of Chicago Press, Chicago. Patton, J.L., Silva, M.N.F. & Malcolm, J.R. (1994) Gene genealogy and differentiation among arboreal spiny rats (Rodentia: Echimyidae) of the Amazon Basin: a test of the riverine barrier hypothesis. Evolution, 48, 1314 1323. Por, F.D. (1992) Sooretama: the Atlantic rain forest of Brazil, p. 130. SPB Academic Publishing, The Hague. Prado, D.E. & Gibbs, P.E. (1993) Patterns of species distributions in the dry seasonal forests of South America. Annals of the Missouri Botanical Garden, 80, 902 927. Raven, P.H. & Axelrod, D.I. (1974) Angiosperm biogeography and past continental movements. Annals of the Missouri Botanical Garden, 61, 539 673. Redford, K.H. & Fonseca, G.A.B. (1986) The role of gallery forests in the zoogeography of the Cerrado s non-volant mammalian fauna. Biotropica, 18, 126 135. Rizzini, C.T. (1963) Nota prévia sobre a divisão fitogeográfica do Brasil. Revista Brasileira de Geografia, 25, 1 64. Rizzini, C.T. (1979) Tratado de fitogeografia do Brasil. Editora Universidade de São Paulo, São Paulo. Ronquist, F. (1997) Dispersal-vicariance analysis: a new approach to the quantification of historical biogeography. Systematic Biology, 46, 195 203. Saiki, R.K., Gelfand, D.H., Stoffel, S., Scharf, S.J., Higuchi, R., Horn, G.T., Mullis, K.B. & Erlich, H.A. (1988) Primerdirected enzymatic amplification of DNA with a thermostable DNA polymerase. Science, 239, 487 491. Salo, J., Kalliola, R., Hakkinen, I., Makinen, Y., Niemela, P., Puhakka, M. & Coley, P.D. (1986) River dynamics and the diversity of Amazon lowland forest. Nature, 322, 254 258. Sanders, H.L. (1969) Marine benthic diversity and the stabilitytime hypothesis. Diversity and stability in ecological systems (ed. G.M. Woodwell and H.H. Smith), Brookhaven Symposia in Biology, Brookhaven National Laboratory, Upton, NY, USA. Vol. 22, pp. 71 81. Silva, J.M.C. (1995) Birds of the Cerrado Region, South America. Steenstrupia, 21, 69 92. Smith, M.F. & Patton, J.L. (1993) The diversification of South American murid rodents: evidence from mitochondrial DNA sequence data for the akodontine tribe. Biology Journal of the Linnean Society, 50, 149 177. Smith, M.F. & Patton, J.L. (1999) Phylogenetic relationships and the radiation of sigmodontine rodents in South America: evidence from cytochrome b. Journal of Mammalian Evolution, 6, 89 128. Swofford, D.L. (1999) PAUP*. Phylogenetic analysis using parsimony (*and other methods). Sinauer Associates, Sunderland. Vanzolini, P.E. (1988) Distributional patterns of South American lizards. Proceedings of a workshop on Neotropical distribution patterns (ed. P.E. Vanzolini and W.R. Heyer), pp. 317 342. Academia Brasileira de Ciências, Rio de Janeiro. Vanzolini, P.E. & Williams, E.E. (1970) South American anoles, the goegraphic differentiation and evolution of the Anolis chrysolepis species group (Sauria, Iguanidae). Arquivos de Zoologia, São Paulo, 19, 291 298. Vivo, M. (1997) Mammalian evidence of historical ecological change in the Caatinga semiarid vegetation of northeastern Brazil. Journal of Comparative Biology, 2, 65 73.

86 L. P. Costa Wallace, A.R. (1952) On the monkeys of the Amazon. Proceedings of the Zoological Society of London, 20, 107 110. Walsh, P.S., Metzger, D.A. & Higuchi, R. (1991) Chelex 100 as a medium for simple extraction of DNA for PCR-based typing from forensic material. Biotechniques, 10, 506 513. Zink, R.M., Blackwell-Rago, R.C. & Ronquist, F. (2000) The shifting roles of dispersal and vicariance in biogeography. Proceedings of the Royal Society Biological Sciences Series B, 267, 497 503. Zink, R.M., Kessen, A.E., Line, T.V. & Blackwell-Rago, R.C. (2001) Comparative phylogeography of some aridland bird species. Condor, 103, 1 10. Zink, R.M. & Slowinski, J.B. (1995) Evidence from molecular systematics for decreased avian diversification in the Pleistocene Epoch. Proceedings of the National Academy of Sciences, 92, 5832 5835. BIOSKETCH Leonora Pires Costa has a PhD in integrative biology from the University of California at Berkeley. Her current areas of research interest include evolution, phylogeography and systematics of Neotropical small mammals. She has worked on ecology and systematics of small mammals for the last 14 years, with extensive field work carried out in the Atlantic Forest, the Amazon, Cerrado, and Caatinga regions in Brazil. She is now a curatorial and research associate at the Zoology Department of the Institute of Biological Sciences, Federal University of Minas Gerais, Brazil.