This study reviews the data available on the distribution of three-banded armadillos of the genus Tolypeutes, identifying potential geographic barriers and evaluating possible biogeographic processes that may account for the present-day distribution of the species and its conservation status. The database was derived from published records, interviews, and voucher specimens, over a timescale ranging from the fossil record to specimens collected in 2013. A total of 236 localities were recorded, with 68 attributed to Tolypeutes matacus and 168 to Tolypeutes tricinctus. The vegetation within the range of the genus is predominantly a xerophytic mosaic of grassland, savannas, open woodland, and xeric thorn forest. The marine transgressions of the Miocene and the uplifting of the Brazilian Shield may have contributed to the vicariant separation of the ancestral populations of T. matacus, to the west and south, and T. tricinctus, to the north and east. The three-banded armadillo is possibly one of the most threatened of Brazilian mammals, considering the low number of recent records and the fact that it is hunted intensively throughout its range.
Tolypeutes Illiger, 1811 is one of the few genera of Cingulata that are distributed primarily in the dry forests of South America. Two species are currently recognized (Wetzel, 1985), the Brazilian three-banded armadillo, Tolypeutes tricinctus (Linnaeus, 1758) and the southern three-banded armadillo, Tolypeutes matacus (Desmarest, 1804). T. tricinctus is the only cingulate endemic to Brazil, occurring mainly in the semi-arid scrub forests and savannas of the northeastern and central regions of the country, while T. matacus is found in western Brazil, Bolivia, Paraguay, and northern and central Argentina (Wetzel et al., 2007).
Marcgrave (1648) first described aspects of the morphology and behavior of T. tricinctus. Despite this early record, this species is one of the least studied armadillos (Santos et al., 1994; Oliveira, 1995; Superina et al., 2014), and the few available publications either refer to occasional encounters (Marini-Filho and Guimarães, 2010), or are distributional notes (Silva and Oren, 1993; Santos et al., 1994; Oliveira, 1995). By contrast, there is a relative wealth of data on the ecology (Bolkovic et al., 1995; Barrientos and Cuellar, 2004), anatomy (Milne et al., 2009) and geographic distribution (Ojeda and Mares, 1989; Mares et al., 1997; Morando and Polop, 1997; Abba and Vizcaino, 2008; Pautasso, 2008) of T. matacus.
Based on the available data, Anacleto et al. (2006) modeled the ecological niche of T. tricinctus, which they used to define the potential range of this species. More recently, Zimbres et al. (2012) modeled its future distribution in relation to existing protected areas. Until now, however, there has been no systematic mapping of the known localities for T. tricinctus, nor any reliable analysis of possible zoogeographic barriers.
Inadequate geographic data can have a negative influence on the development of species conservation and management strategies (Brito, 2004), and more records would permit more accurate ecological niche modeling. Tolypeutes populations have declined considerably in recent years, mainly due to hunting pressure (Santos et al., 1994; Barrientos and Cuellar, 2004; Abba and Vizcaino, 2011; Feijó and Langguth, 2013). Despite being classified as Vulnerable (T. tricinctus) and Near Threatened (T. matacus) by the IUCN (Superina and Abba, 2010; Abba and Superina, 2010), there are few recent data on the distribution of either species or the existence of remnant populations on which to base the definition of priority areas for conservation.
The present study was based on a compilation of the data on the distribution of the two Tolypeutes species, with the objective of (i) defining their geographic ranges, (ii) identifying possible geographic or ecological barriers to the dispersal of the species, and (iii) evaluating biogeographic evidence that may account for the present-day distribution of the species.
MATERIALS AND METHODS
A geographical database was compiled for Tolypeutes based on published records, museum specimens collected prior to 2013, and interviews with local researchers. In the case of published records, only primary sources were used, which meant that studies such as those of Oliveira et al. (2003), Anacleto et al. (2006), and Zimbres et al. (2012), which were based on the compilation of published data, were not included, although the original sources that these reports cite were.
In the case of T. matacus, the distribution of which is relatively well known in comparison with T. tricinctus, a number of studies (e.g., Ojeda and Mares, 1989; Morando and Polop, 1997; Pautasso, 2008; Abba and Vizcaino, 2008) are based on the compilation location records. As the primary objective of the present study was to define range limits and possible Zoogeographic barriers, marginal records of the occurrence of this species were included in the present analysis.
The records were plotted using the ArcGis 10.2 software (ESRI, 2013) to produce detailed distribution maps based on the geographic coordinates obtained from specimen labels or published records. When exact coordinates of the locality were unavailable, those of the administrative center of the municipality in which the record was obtained were used instead. For the evaluation of the conservation status of T. tricinctus, the records for this species were classified in three categories: (i) fossil, (ii) historic (prior to 2000), and (iii) recent.
Tolypeutes specimens were examined in six scientific collections in Brazil and one in Paraguay. In Brazil, the collections are housed at the Zoology Museum of São Paulo University (MZUSP), the National Museum in Rio de Janeiro (MNRJ), the Federal University of Minas Gerais (UFMG) in Belo Horizonte, the Museum of Natural Sciences at the Pontifical Catholic University of Minas Gerais (PUCMG), also in Belo Horizonte, Brasilia University (UNB), and the Federal University of Paraiba in João Pessoa (UFPB). In Paraguay, the National Museum of Natural History (MNHNP) in Asunción was consulted.
These data are complemented by interviews with 26 Brazilian researchers who are currently working in the field within the known distribution of the three-banded armadillos, or who have worked in this area in the recent past. Each researcher was asked if he or she had encountered Tolypeutes during fieldwork, and when positive answers were obtained, the identification of the species was confirmed based on photographs or verbal descriptions provided by the Interviewees. The geographic coordinates for the locality of the encounter were recorded as for the other records, and included in the database for mapping and analysis.
RESULTS AND DISCUSSION
A total of 236 Tolypeutes localities were considered for the present study (Fig. 1). The localities are distributed mainly within the dry zone of South America, between latitudes 4° S and 42° S, which encompasses grasslands, savannas, open woodlands, and xeric thorn forest. The two species are distributed allopatrically, with their ranges being separated by a gap of some 1000 kilometers in central Brazil, which coincides with the headwaters of the Paraná River basin to the south, and the Tocantins-Araguaia, to the north.
The geographic distribution of Tolypeutes matacus was defined based on 66 marginal localities in western Brazil, southern Bolivia, northern Paraguay, and Argentina (Table 1), primarily in the Chaco dry forests of Bolivia and Paraguay, and the Pampas grasslands of Argentina. The distribution of the species is delimited to the north by the Amazon Forest and to the west by the Andes, while the Paraguay and Araguaia river systems appear to be important barriers to the east (Fig. 2). The original distribution of this species extended as far south as Carmen de Patagones (42° S) at Bahía Blanca, in the Argentine province of Buenos Aires.
The geographic distribution of T. matacus presented here includes all historic records. Abba and Vizcaino (2011) concluded that this species is locally extinct from some areas, such as Buenos Aires province, where the most recent record was obtained in 1926. In this case, the current southern limit of the species' range would appear to be between 33° and 34° S, coinciding with the provinces of Mendoza and San Luis, in western Argentina. This local extinction was probably the result of intensive hunting pressure and habitat loss within the region surrounding the Argentinian capital.
A number of well-defined geographic barriers delimit the distribution of T. matacus further north. The Paraguay River forms the eastern limit of the species, and is characterized by distinct topography and vegetation types on its two margins (Frutos and Van Den Bussche, 2002). The Chaco biome lies to the west of the Paraguay River, and is characterized by xerophytic vegetation, formed by a mosaic of grassland, savannas, open woodlands, and xeric thorn forest (Willig et al., 2000), while to the east, the vegetation is primarily subtropical humid forest, mixed with plains, grassland, and marshes (Myers, 1982; Willig et al., 2000; Frutos and Van Den Bussche, 2002). While this eastern zone is prone to flooding during the rainy season, the Chaco is semiarid throughout the year. Krumbiegel (1940) confirmed that T. matacus occurs only in dry habitats, and is never found in marshy environments, and also that it occurs pre-dominantly in lowland areas. In the Andean region, it is confined to savanna-like habitats in the depressions between mountain ranges. This is consistent with the distribution of the species recorded in the present study, between the Andes and areas of humid broadleaf forest.
Marginal localities in which the occurrence of Tolypeutes matacus has been confirmed.
The northeastern limits of the distribution of T. matacus are less well known, (Anacleto et al., 2006). In Brazil, three historic and one recent record for the species exist (Table 1). Wetzel (1985) added a locality known as “Chavantina das Mortes, Mato Grosso” based on a specimen deposited at the Brazilian National Museum (MN S-10), although we examined this specimen, and confirmed that it belongs to the genus Dasypus.
Tolypeutes tricinctus is known from 168 localities in the Brazilian Northeast, including four fossil sites (Table 2, Fig. 3). A majority (n = 85) of these sites were concentrated in the state of Bahia, followed by Pernambuco (38), Piaui (25), Ceara (8), Sergipe (6), Maranh˜o (2), and Alagoas, Paraiba, Rio Grande do Norte, and Tocantins, all with just one record each. The points were mostly associated with the São Francisco River basin. These sites are located predominantly within the domain of the Caatinga scrub, and some adjacent areas of Cerrado savanna, which suggests a preference for open and/or semi-arid habitats.
The records available to date, however, are limited primarily to the upland areas of northeastern Brazil (Fig. 3); that is, the Borborema plateau, to the northeast (Paraíba and Pernambuco), the mountains of central and western Ceará and the Parnaíba Valley to the west. The Serra Geral de Goiás plateau lies to the southwest and the Espinhaço range to the east. The southern limits of the distribution are still ill-defined, but may coincide approximately with northern Minas Gerais.
Only eight (11.1%) of the 72 records that can be dated accurately were classified as recent. Four (6.25%) of the 64 historic records were collected prior to 1950, and a further six between 1950 and 1989. The other 54 records were collected in the 1990s. It is important to note that all the recent records were collected in a relatively small portion of the species' range, in the central and southwestern extreme (Fig. 3).
Localities in which the occurrence of Totypeutes tricinctus has been mentioned. *Recent records, †Fossil Record.
The geographic distribution of T. tricinctus is even less well defined than that of T. matacus. The distribution map is characterized by the concentration of a large number of localities in the center of the species' distribution, but extensive lacunas in peripheral areas, impeding a more conclusive interpretation of possible barriers to dispersal. Despite these shortcomings, it is possible to infer the existence of probable barriers to dispersal. The available records indicate that T. tricinctus is distributed on both plains and upland areas, in contrast with the pattern found in T. matacus. The areas to the southwest and northeast of the range of T. tricinctus correspond to two major upland areas, Serra Geral de Goiás and the Borborema highlands, respectively (Fig. 3). To the east, the range is limited by the Atlantic Forest biome, and to the west, by the Amazon forest, although the southern limit remains unclear, and the apparent absence of the species south of the southern limit of the Caatinga in northern Minas Gerais may be the result of sampling deficiencies.
While Marcgrave (1648) referred to this armadillo as a common animal, geographic lacunae have long been known, and even when specimens are available, their localities may not be known (Sanborn, 1930). While 168 localities were identified in the present study, the vast majority (138) were obtained from an unpublished masters dissertation (Santos, 1993), although the occurrence of T. tricinctus was confirmed directly at only eight of these localities (see Table 2), while all the other points were identified only from interviews. Excluding these indirect records and the fossils, only 27 (16%) localities have been recorded reliably over the past 104 years (Table 2).
The Brazilian Caatinga has traditionally been considered to be a biome of low mammalian diversity and endemism (Mares et al., 1981; Willig and Mares, 1989), attracting little interest from researchers (Brito et al., 2009). Over the past decade, however, the number of mammal species known to occur in the Caatinga has almost doubled (Carmignotto et al., 2012), stimulating new interest, although most of this research has focused on either the mesic cloud forest enclaves (brejos de altitude) or transition zones with neighboring biomes (Oliveira et al., 2003; Carmignotto et al., 2012; Feijó and Langguth, 2013). Areas of typical Caatinga, the thorn scrub of the arid central portions of this biome, where T. tricinctus occurs, have still been poorly surveyed. These sampling problems may be exacerbated by the ongoing extinction of local populations.
Conservation of Tolypeutes tricinctus
The armadillos of the genus Tolypeutes are unique among the cingulates in having hard, articulated armor and the ability to roll up completely into a ball, protecting the ventral portion of the body. This makes the digging of burrow — the strategy used by other armadillos to escape their predators — unnecessary (Marcgrave, 1648; Sanborn, 1930; Smith, 2007; Deem et al., 2009), but also renders the three-banded armadillo extremely vulnerable to human hunters throughout its range (Silva and Oren, 1993; Santos et al., 1994; Marinho-Filho et al., 1997; Noss et al., 2003; Feijó and Langguth, 2013).
These animals have been hunted intensively as a source of food since the early colonization of South America in the seventeenth century (Marcgrave, 1648; Azara, 1801; Santos et al., 1994; Bolkovic et al., 1995; Noss et al., 2003; Smith, 2007; Papavero et al., 2009). The fat of T. tricinctus is also used by many populations as a remedy for asthma, diarrhea, headaches, inflammations, and earache (Alves and Rosa, 2007), which also contributes to the exploitation of the species. In addition to this hunting pressure, the habitats occupied by these armadillos, especially T. tricinctus, have suffered extensive anthropogenic impacts, which may be especially intense in the semi-arid Caatinga, due to the intrinsic characteristics of this biome (Leal et al., 2005).
The lack of recent records from many areas—including most of the northern and eastern extremes of the range—suggests that the species may now be locally extinct from a large proportion of its original geographic distribution. From a conservation perspective, there is clearly an urgent need for more detailed data on the current distribution of the species, and in particular, the occurrence of remnant populations appropriate for conservation management.
The ecology of T. tricinctus is also poorly known, which further compounds the problems for the conservation of the species. While the ecology of T. matacus may be better understood, there appear to be important differences between species which limit the potential for extrapolation. As well as being more widespread than T. tricinctus, for example, T. matacus appears to be relatively abundant in some areas (Bolkovic et al., 1995; Noss et al., 2003; Barrientos and Cuellar, 2004). However, the species has also suffered local extinction, such as that observed in the province of Buenos Aires (Abba and Vizcaino, 2011).
Bolkovic et al. (1995) recorded an insectivorous-generalist diet for T. matacus, which appears to feed primarily on beetle larvae, but consumes a larger proportion of ants and termites during the dry season, and of fruit during the rainy season. By contrast, T. tricinctus appears to be a more specialized insectivore, feeding preferentially on termites and ants (Guimarães, 1997). These ecological differences are reflected in the interspecific variation in cranial and dental morphology. The cranium of T. tricinctus is comparatively lightly built, and the teeth are much less robust than those of T. matacus. These features may represent adaptations for a more specialist diet in T. tricinctus, which could further reinforce its vulnerability to anthropogenic impacts, in particular in comparison with T. matacus.
The review of the evidence presented in this study indicates that the current situation of the species may be very critical. In fact, it seems likely that T. tricinctus is one of the most endangered mammal species in Brazil, given that its occurrence was only confirmed at eight sites, and it continues to suffer intense hunting pressure and habitat loss. The species is known to occur in six protected areas: the Serra da Capivara and Serra das Confusões national parks in Piaui, the Chapada Diamantina National Park and the Raso da Catarina Ecological Station in Bahia, the Serra Negra Biological Station in Pernambuco, and Mirador State Park in Maranhao, which together cover a total area of almost 17,000 km2, although this represents only 2% of the Caatinga biome. Based on niche modeling, Zimbres et al. (2012) concluded that the species was not adequately protected, given that less than 10% of its original range encompasses some kind of protected area, and highlighted the urgent need for the creation of new conservation units in the Caatinga, especially within the species' range. The administration and management of existing protected areas must also be reinforced (but see Torres et al., 2009).
The paleoenvironments of the Tertiary played an important role in the evolutionary history of the South American xenarthrans (Delsuc et al., 2004, 2012). During this period, the most important events in South America were the marine transgressions into much of the Chaco and the Paraná River basin (Hérnandes et al., 2005), and the epirogenic uplifting of the central Brazilian shield, which resulted in a landscape of extensive plateaus separated by major depressions (Coli, 2005) and the deepening of some river basins (Werneck, 2011).
No reliable estimates of the divergence time of the two Tolypeutes species are available, although Delsuc et al. (2012) recently published a molecular phylogeny which indicated that the subfamily Tolypeutinae [(Cabassous + Tolypeutes) Priodontes] arose sometime after the Eocene-Oligocene transition, and that Tolypeutes diverged from Cabassous at the beginning of the Miocene, which means that the two Tolypeutes would have separated at some time from the Miocene onwards.
The allopatric distribution of the two Tolypeutes species suggests an evolutionary history similar to that proposed by Delsuc et al. (2012) for Calyptophractus-Chlamyphorus, in which the marine transgressions of the Miocene are suggested to have acted as a vicariant mechanism separating the ancestral population of T. matacus, to the west, from that of T. tricinctus, to the east, which would have subsequently evolved in isolation. These transgressions may have played a profound role in the diversification of many widely-distributed taxa within the dry zone of South America (Pascoal and Jaureguizar, 1990; Coli, 2005) and is related to the emergence of new life styles, such as fossoriality (Galewsky et al., 2005; Delsuc et al., 2012). The uplifting of the Brazilian Shield at the end of the Tertiary may also have contributed to this process. Coli (2005) concluded that this was the decisive factor in the diversification of the herpetofauna of the dry zone of South America, by creating distinct environments within a limited geographic space.
The combination of these processes had a profound effect on the hydrographic basins of this region, in particular those of the Paraná and Tocantins-Araguaia rivers, contributing to the differentiation of environments in central Brazil, including those considered to be Cerrado refuges (Coli, 2005; Werneck, 2011). These same river basins could have been crucial to the isolation of the ancestral populations of the present-day Tolypeutes species. However, a more conclusive analysis of this process would require a systematic phylogeographic study supported by molecular markers for the estimation of divergence times.
The geographic distribution of T. matacus is better known than of T. tricinctus. While T. matacus has suffered local extinction at some sites, a number of populations are known to persist, primarily in the Chaco of Paraguay and Bolivia. However, the northeastern portion of the distribution of the species is poorly sampled, and requires verification, especially in the lowland areas of open habitat, such as those found in the Brazilian state of Mato Grosso. The records of T. tricinctus are concentrated in the central portion of its known range, with few data from marginal areas, which impede the reliable definition of barriers to dispersal, reinforcing the need for new surveys in keys states, such as Piauí, Maranhão, Tocantins, and southern Bahia. It will also be important to revisit the sites of historic records for the collection of confirmatory data. Given the available evidence, the establishment of new protected areas for the Brazilian three-banded armadillo may be essential for the survival of this species, which may be one of the most endangered mammals found in Brazil.
We are grateful to the curators of the scientific collections visited during the present study — Mario de Vivo (MZUSP), João Alves de Oliveira (MN), Fernando A. Perini (UFMG), Claudia Guimaräes Costa (PUCMG), Jader Marinho-Filho (UNB), Pedro Cordeiro Estrela (UFPB), and Isabel Gamarra de Fox (MNHCP), as well as the interviewees for providing valuable complementary information, Marília Marques Guimarães Marini for help with references, and two anonymous reviewers for their helpful comments on previous drafts of this paper. We would also like to thank CAPES (AF, BATPC, GSTG, PAR) for graduate stipends and CNPq for research fellowships to SFF and AL.