- Top of page
- Materials and Methods
- Supporting Information
The Neotropical region is characterized by varied weather conditions and high levels of environmental diversity, which allows for the great diversity of forest cover that ranges from rain forests to semi-arid vegetation types (Hueck, 1972). There are two main tropical rain forest areas in South America: the Amazon and Brazilian Atlantic forests. Between these rain forests, there is a large belt of mostly seasonally dry and open vegetation areas, orientated in a north-east–south-west direction (Hueck, 1972; Ab'Saber, 1977). This belt comprises the following regions: (1) the Chaco province in northern Argentina, western Paraguay and south-eastern Bolivia, (2) the savanna-like vegetation of the Cerrado province in central Brazil, and (3) the Caatinga province in north-eastern Brazil and the Misiones nucleus in the Paraná–Paraguay river basin, which are collectively considered seasonally dry tropical forests (SDTFs; Pennington et al., 2000) (Fig. 1a).
Figure 1. (a) Geographical distribution of the seasonally dry tropical forests (SDTFs) in South America. Under the Caatinga nucleus (1) and Misiones nucleus (2) of the SDTFs, the approximate geographical distributions of the Drosophila serido and D. antonietae haplogroups are shown. (b) Neighbour-joining dendrogram showing the relationships between the cytochrome c oxidase subunit I haplotypes in the D. buzzatii cluster. Only bootstrap values greater than 50% are shown (1000 replicates). The scale bar represents genetic distances calculated according to the GTR+I+G model. The coloured branches are exclusive to D. serido (red), D. gouveai (blue), D. seriema (pink), D. borborema (green) and D. antonietae (brown).
Download figure to PowerPoint
Other important open and seasonally dry vegetation areas commonly found within the Cerrado range are the ‘campos rupestres’ (montane savanna sensu Silva et al., 2006), which are disjunctly distributed in eastern Brazil. The campos rupestres comprise a mosaic of vegetation types (Taylor & Zappi, 2004) which have floristic components characteristic of both the Caatinga and Cerrado areas, in addition to other endemic components (Taylor & Zappi, 2004; Silva et al., 2006).
It has been proposed that both the humid and dry provinces in South America were modified during the Pleistocene due to global climate changes (Haffer, 1969; Vanzolini & Williams, 1970; Ab'Saber, 1977; Prado & Gibbs, 1993; Pennington et al., 2000). The Pleistocene refuge model suggests that during the cold and dry conditions of the glacial periods, expansion of the xerophytic vegetation was favoured over that of semi-deciduous evergreen vegetation, which was restricted to more humid sites (Haffer, 1969; Vanzolini & Williams, 1970). At least some of the SDTF nuclei, for example, were supposedly interconnected during the glacial periods to form the Pleistocene arc (Prado & Gibbs, 1993; Pennington et al., 2000, 2009). During the interglacial periods, which experienced warm and humid conditions, the situation was reversed. Several studies support the hypothesized impact of cyclic climate changes on the vegetation of the continent (Mayle, 2006), including some phylogeographical data (e.g. Carnaval et al., 2009; Moraes et al., 2009).
Yeast–cactus–Drosophila model systems (Starmer et al., 1990; Manfrin & Sene, 2006) represent an ecological association where necrotic cactus tissues act as a substrate for yeast flora, providing an important food source for cactophilic species of Drosophila, such as those from the Drosophila repleta group (Vilela, 1983). The necrotic cactus tissues are used for breeding and, at least during the larval stages, feeding sites by cactophilic species of Drosophila (Starmer et al., 1990).
Within the D. repleta group, the Drosophila buzzatii cluster is a monophyletic group composed of seven sibling species that are cactophilic and inhabit open vegetation areas in South America: Drosophila buzzatii, Drosophila koepferae, Drosophila antonietae, Drosophila serido, Drosophila gouveai, Drosophila seriema and Drosophila borborema (see Appendix S1 in Supporting Information) (Manfrin & Sene, 2006). Analysis of the mitochondrial cytochrome c oxidase subunit I gene (COI) suggests that there are three main evolutionary lineages within the D. buzzatii cluster: one comprising D. buzzatii and D. koepferae; a second lineage containing only D. antonietae; and an unresolved clade that contains D. gouveai, D. borborema, D. seriema and D. serido, which have partly overlapping distributions (Manfrin et al., 2001; Fig. 1, Appendix S1). The latter was referred to as the AB clade by de Brito et al. (2002). In the current study we refer to this clade as the D. serido haplogroup in reference to the first described species within this clade.
The COI data indicate that D. antonietae is the sister taxon to the D. serido haplogroup. While D. antonietae is primarily associated with areas of the Misiones SDTF nucleus, the species within the D. serido haplogroup are found throughout the Caatinga SDTF nucleus and are also discontinuously distributed outside this nucleus in the campos rupestres and the so-called ‘matas de restinga’, the dune vegetation along the Atlantic coastal line in littoral areas of Brazil (Fig. 1a, Appendix S1). Phylogeographical analysis using the COI of D. antonietae indicates that this species experienced a population expansion in the past and colonized Atlantic coastal areas from areas of the Misiones SDTF nucleus (de Brito et al., 2002). The phylogeographical hypothesis for the D. serido haplogroup species proposed by de Brito et al. (2002) suggests that the species dispersed from north-eastern Brazil. In another phylogeographical study, exclusively focused on D. gouveai and based on the cytochrome c oxidase subunit II gene (COII), the migratory routes for this species were found to be similar to those found in the study of de Brito et al. (2002) (Moraes et al., 2009). However, the sample sizes in these works were limited and primarily restricted to D. gouveai and D. serido.
The host cactus species of the D. buzzatii cluster are associated with relatively dry environments (Taylor & Zappi, 2004), and it is reasonable to assume that the palaeoclimatic changes that affected the distribution of the xerophytic vegetation in South America would also have influenced historical and demographic events in cactus species and their associated fauna. In fact, cactus vegetation enclaves found in the campos rupestres located within the Cerrado province are considered to be an indication that the Caatinga and campo rupestre vegetation areas were interconnected in the past as a result of the glacial periods (Ab'Saber, 1977) because the campos rupestres are surrounded by the fire-swept Cerrado, which most cactus genera avoid (Taylor & Zappi, 2004). Furthermore, hypotheses of migratory routes of the Caatinga vegetation expansion (Prado & Gibbs, 1993; Prado, 2000) are supported by phytogeographical data from several genera of the Cactaceae family (Taylor & Zappi, 2004). Given its obligate ecological association with various cactus species, the D. buzzatii species cluster is a useful biological model for studying evolutionary topics such as the effects of Quaternary climatic changes on genetic differentiation in the South American biota (Sene et al., 1988).
In this paper, data from the COI region were analysed using nested clade phylogeographical analysis (NCPA) (Templeton, 2004) and neutrality tests (Fu, 1997; Ramos-Onsins & Rozas, 2002) to test the phylogeographical hypothesis previously established for D. serido and D. gouveai (de Brito et al., 2002; Moraes et al., 2009) and to establish phylogeographical hypotheses for D. borborema and D. seriema. Given that these D. serido haplogroup species are associated with xeromorphic vegetation, we also investigated the effects of Quaternary palaeoclimatic changes on the demographic fluctuations of cactophilic Drosophila using Bayesian skyline plots (Drummond et al., 2005). The impact of Quaternary palaeoclimatic changes on D. antonietae, the mitochondrial sister lineage of the D. serido haplogroup (Manfrin et al., 2001), was also investigated.
- Top of page
- Materials and Methods
- Supporting Information
The distribution of genetic variation found in COI provides useful genealogical information that can be used to infer the demographic history of cactophilic species of the Drosophila buzzatii species cluster. Furthermore, these demographic histories can be interpreted in relation to Quaternary palaeoclimatic changes.
The association between the geographical distribution of the sister COI genetic lineages of the clade composed of the D. antonietae and D. serido haplogroups suggests a vicariant diversification pattern in the origin of these lineages (Manfrin et al., 2001): the former became associated with the Paraná–Paraguay river basin area, whereas the latter remained associated with central and north-eastern Brazil. Based on our data, we add a possible cause of this vicariant event: the association of these lineages with SDTFs, with allopatric divergence occurring in some interglacial period of the Pleistocene, when the SDTF nuclei were disconnected (Pennington et al., 2000), promoting the range fragmentation of their associated biota. This hypothesis is suggested by the substantial overlap in the current geographical distribution of the D. antonietae and D. serido clades within the Misiones and Caatinga SDTF nuclei, respectively (Fig. 1a), and by the estimated coalescence time for the COI haplotypes of these two lineages, which was calculated to be approximately 1.7 Ma (1.05–2.48 Ma). This agrees that the group may have diversified in response to palaeoclimatic oscillations, because the glaciation cycles began at approximately 2.6 Ma (Rull, 2008).
The effects of Quaternary global climatic changes on speciation rates are controversial (Bush & Oliveira, 2006; Rull, 2008). In recent years, several studies have shown, based on molecular-clock-based analyses, that cladogenetic events for most Neotropical taxa pre-date the Pleistocene (Moritz et al., 2000), including genera of plants with discontinuous distributions in the SDTFs of South America (however, see the genus Inga in Pennington et al., 2009). Based on our findings, we hypothesize that for those organisms that have obligate associations with host plants and with a reduced dispersal capacity, such as the cactophilic Drosophila, vicariant diversification mediated by the Quaternary climatic changes could explain some patterns of recent divergence.
Whilst the species analysed here are cactophilic and with some overlaps in their geographical distribution (Appendix S1), their estimated demographic histories were not completely congruent. This situation could be related to the complexity of the effects of the Pleistocene climates throughout South America, which may have affected the phytogeographical provinces and their biota in different and asynchronous ways that depend on the associated topography and latitude (Behling et al., 2000; Ledru et al., 2005). Population expansions were detected using NCPA in all of the D. serido haplogroup species in the regions around the Diamantina ridge within the Caatinga range in north-eastern Brazil, from where these species moved to the north-east (the Borborema Plateau), to central Brazil (the Pirineus Ridge) and to south-eastern Brazil (the Canastra and Cipó ridges) (Fig. 2).
The population expansions to the Borborema Plateau were restricted to D. serido (clade 1-4) and D. borborema (clade 4-3). Although the onset of these expansions appears to be chronologically independent for each species, considering the beginning of the expansions in the BSP graphs (Fig. 4), we can presume a single dispersal route for these species (Fig. 2), indicating that a historical connection may have existed between the southern and northernmost Caatinga. To the best of our knowledge, this is the first time that this migratory route has been proposed. All of the other migratory routes proposed in this work involve the crossing of boundaries between adjacent phytogeographical provinces (Fig. 2), and most of these routes are congruent with routes that have been previously proposed, particularly those that represent the interconnection between nuclear SDTF areas (Prado & Gibbs, 1993; Prado, 2000).
The largest contiguous areas of the SDTF are represented by the Caatinga in north-eastern Brazil and the Misiones nucleus in the Paraná–Paraguay river basin areas (Fig. 1a). According to previous studies, two migratory routes may explain the Caatinga expansion in the Pleistocene period, connecting it to the Misiones nucleus during the formation of the ‘Pleistocene arc’, as described in Pennington et al. (2000). In brief, the northern route, supposedly connecting the Caatinga to central Brazil, crossed the current savanna-like vegetation areas of the Cerrado, whereas the southern route extended around the valleys of the São Francisco, Jequitinhonha and Rio das Velhas rivers, reaching the Atlantic coast and the Cipó ridge mountains in south-eastern Brazil (Prado & Gibbs, 1993; Prado, 2000).
The signal of expansion from north-eastern to south-eastern Brazil, including central Brazil detected in D. gouveai and D. serido (clade 1-4 in Fig. 2) is geographically congruent with the northern route proposed in Prado & Gibbs (1993) to explain the expansion of SDTF vegetation during the glacial periods. Besides the geographical concordance, palynological studies indicate that the SDTF reached its maximum extension between approximately 40 and 30 ka (Mayle, 2006), which is chronologically congruent with the population expansions assigned to D. serido and D. gouveai (Fig. 4a,b). In this sense, it is reasonable to assume that the population expansions detected for these species followed the same expansions routes of the SDTF during the Pleistocene.
The current study is the first time that the migratory route connecting north-eastern and central Brazil has been described for D. serido; however, this information has been previously reported for D. gouveai using different samples and molecular markers (de Brito et al., 2002; Moraes et al., 2009). According to Moraes et al. (2009), based on COII sequences, the population expansion in D. gouveai involved two expansion events that followed mountain ranges that extend from north-eastern Brazil to central and south-eastern Brazil, calculated at approximately 450 and 170 ka, respectively. These expansions routes are geographically concordant with those detected with COI (clade 1-4 in Fig. 2). Notably, due to the confidence interval overlap, the molecular calculations presented here for D. gouveai (Fig. 4b) are not statistically discordant with those presented in Moraes et al. (2009), and the primary picture that emerges is that Pleistocene climatic changes could have influenced the demographic dynamics of these cactophilic Drosophila species.
Expansions to south-eastern Brazil encompassing the colonization of the Canastra Ridge and Cipó Ridge, along the Espinhaço mountain range, by D. gouveai (clade 1-4) and D. seriema (clade 4-3), respectively, are also in agreement with the southern route thought to explain the expansion of the SDTF vegetation range during glacial periods in Brazil (Prado & Gibbs, 1993). In summary, both the northern and southern routes may explain the expansion of the Caatinga according to the Pleistocene arc hypothesis, which was corroborated for the cactophilic Drosophila species in the current study. Interestingly, these migratory routes were also previously supported by the phytogeographical data provided by several genera of the Cactaceae family (Taylor & Zappi, 2004).
It was also inferred that one population expansion occurred from the Caatinga to the dune vegetations (‘matas de restinga’) along the Brazilian Atlantic coast (Fig. 2), particularly for D. serido. This species is genetically structured (Table 3), and the presence of haplotypes around 17° S latitude at the Atlantic coast (samples N35 and N31; Fig. 2) that are allocated to two distinct clades, (1-4 and 2-8; Fig. 3), indicates that this region could be a secondary contact zone for the two evolutionary lineages of D. serido that diverged in allopatry. This contact zone may have been formed by population expansions such as those detected in the current study. It is particularly interesting to note that this same Atlantic coast region is a secondary contact zone for other organisms with different dispersal capabilities, such as amphibians (Carnaval et al., 2009) and lizards (Pellegrino et al., 2011); this suggests a congruent phylogeographical pattern on a larger scale.
A coalescent-based analysis suggests a temporal congruence between the population size fluctuations estimated for these cactophilic Drosophila species and the Quaternary palaeoclimatic fluctuations (Fig. 4). Considering the chronology of these events, these data suggest that the D. serido, D. gouveai, D. borborema and D. antonietae populations had similar demographic histories but that D. seriema had a distinctly different history (Fig. 4).
The BSP analysis suggests that D. serido, D. gouveai, D. borborema and D. antonietae exhibited an exponential increase in population size between 45 and 75 ka (Fig. 4), coinciding with the Wisconsin glacial period. These species showed some synchronization in their demographic dynamics; nevertheless, they followed distinct migratory routes (Figs 2 & 4). The demographic history of D. seriema was quite distinct from the remaining species, exhibiting an increase in population during the Illinoian stage to the middle Wisconsin glacial periods in the 550–90 ka interval, which was followed by a period of population equilibrium and a bottleneck coinciding approximately with the Last Glacial Maximum (LGM; Fig. 4d). The bottleneck events inferred, at least for D. seriema, are in agreement with the hypothesis of a contraction in xerophytic vegetation at the beginning of the Holocene (Ab'Saber, 1977; Pennington et al., 2000).
A relevant question is: if all of the Drosophila species analysed here are cactophilic and ecologically restricted to relatively dry vegetation areas, why does only D. seriema follow the contraction of the cactus population during the Holocene? We argue that the answer to this question is related to ecological factors. Drosophila serido, D. gouveai and D. borborema each have broad geographical ranges within the Caatinga (Appendix S1). The Caatinga contains vegetation adapted to dry conditions, and the high number of endemic plant taxa in this province is indicative of its semi-arid history (Hueck, 1972). Indeed, palynological data indicate that the Caatinga vegetation has been predominantly semi-arid since at least 42 ka (Behling et al., 2000). Furthermore, geological studies also suggest that in most areas of the Caatinga, the climate has remained predominantly semi-arid during the Holocene (de Oliveira et al., 1999; Auler et al., 2004). These studies suggest that the climate in the past enabled Cactaceae species and their associated fauna to persist in this region since at least the Pleistocene to the present. Therefore, D. serido, D. gouveai and D. borborema experienced ecological conditions that maintained their large population sizes. However, the distribution of the campo rupestre vegetation, where the cactus host of D. seriema grows, is likely to have increased during glacial periods and decreased during warm, humid phases (Ab'Saber, 1977). In fact, palaeoecological (Behling, 2002; Safford, 2007) and biogeographical data (Antonelli et al., 2010) indicate that during the glacial periods, the current isolated ‘islands’ of campo rupestre vegetation were more extensive and probably interconnected. Likewise, the fauna associated with campo rupestre vegetation should be affected by the climatic fluctuations at the Pleistocene–Holocene transition, as shown here by D. seriema (Fig. 4).
To understand the effects of Quaternary climatic changes on the Neotropical phytogeographical provinces, a multidisciplinary effort is necessary, and contributions have been added from several fields, including genetic-based phylogeographical studies. These types of studies are still rare in South America (Beheregaray, 2008) and are primarily related to the taxa associated with humid environments, such as the Mata Atlântica province (e.g. Carnaval et al., 2009; Thomé et al., 2010). Here, we present phylogeographical data from the cactophilic Drosophila species of the D. buzzatii species cluster, which are associated with seasonally dry and open vegetation areas in South America. We found geographical and chronological evidence suggesting that the major vicariant events between D. antonietae and the D. serido haplogroup, as well as the demographic fluctuations in the lineages in the latter group, could be related to the causal effects of Pleistocene climatic changes on the spatial dynamics of the SDTFs in Brazil.