La región Patagónico-Fueguina comporta un rico ensamble de roedores sigmodontinos. En este trabajo presentamos una síntesis del conocimiento sobre los procesos de diversificación del grupo durante el Neógeno tardío. La diversidad de sigmodontinos comprende 16 géneros y cerca de 24 especies; sin embargo, la mayor parte de las mismas pertenecen a las tribus Abrotrichini y Phyllotini. Varios géneros de abrotriquinos son endémicos de la región, mientras que los filotinos están en general representados por especies de amplia distribución fuera de Patagonia. Se pueden reconocer dos grandes ensambles eco-geográficos de sigmodontinos: un grupo nor-oriental de tierras bajas, con especies mayoritariamente asociadas a las formaciones vegetales arbustivas del Monte, y otro sud-occidental de tierras de mediana a alta elevación, que agrupa típicos elementos patagónicos, incluyendo abrotriquinos y una diversidad de formas andinas. El patrón de disminución latitudinal en el número de especies de sigmodontinos es más complejo que aquel tradicionalmente supuesto de una pauperización norte a sur. La porción continental más austral de Patagonia es tan pobre en número de especies como la porción norte de Tierra del Fuego (seis especies) sugiriendo que la insularidad es insuficiente para explicar el ensamble isleño. Los ciclos glaciales podrían haber jugado un papel principal en el control de la riqueza específica de sigmodontinos y mamíferos en general. El registro fósil de sigmodontinos en Fuego-Patagonia está restringido al Pleistoceno tardío y Holoceno. Eventos destacables en esta historia incluyen extinciones regionales de especies sud-orientales de amplia distribución durante el Holoceno tardío y una restructuración de las comunidades posiblemente debida a cambios ambientales de origen antrópico reciente. Dos patrones filogeográficos principales pueden ser gruesamente asociados con los grupos nor-oriental y sud-occidental previamente reseñados. Mientras que el ensamble nor-oriental comprende básicamente especies sin estructura filogeográfica, el sud-occidental involucra varios linajes que muestran profundos quiebres. La fauna de sigmodontinos de Fuego-Patagonia está conformada mayoritariamente por especies que colonizaron desde bajas latitudes y por otras diferenciadas in situ.
Relative to its surface area and latitudinal placement, the Patagonian–Fuegian region supports a high diversity of land mammals. From the Río Negro Province in the north to the Isla Grande de Tierra del Fuego in the southern tip (Tierra del Fuego, hereafter), about 80 species have been recorded (Table S1, see Supporting Information). More than 50% of them are rodents, and a single subfamily, the Sigmodontinae (Cricetidae), is represented by 16 genera and about 24 species. Interestingly, sigmodontine diversity is largely restricted to the tribes Abrotrichini and Phyllotini (two of at least nine tribes of Sigmodontinae), and the former has a substantial part of its diversity associated with the region.
The variety and abundance of Fuego–Patagonian field mice, as well as the poor representation of other extra-Patagonian speciose groups, such as marsupials and bats, have long been recognized in the literature (e.g. Darwin, 1839; Allen, 1905; Osgood, 1943; Pearson, 1983). The study of Fuego–Patagonian sigmodontine diversity and evolutionary history has increased in intensity in recent years, as documented below. However, there have been few efforts to provide a synthesis of our knowledge of the group in the region.
Sigmodontines have a long fossil record in South America, with the oldest remains aged at about 5 Ma (Prevosti & Pardiñas, 2009). In contrast, the more ancient Patagonian fossils are from late Pleistocene deposits (e.g. Pearson, 1987; Pardiñas & Teta, 2008), thus providing a very short time window to explore evolutionary processes. The knowledge of the Patagonian living assemblages is uneven, positively biased towards the northwestern forest–steppe ecotone (the area around the city of San Carlos de Bariloche); in contrast, the central and austral tablelands are particularly under-represented in the literature (e.g. Pearson & Pearson, 1982; Monjeau et al., 1998; Pardiñas et al., 2003). Most of the taxonomic work on Patagonian field mice was produced between the end of the 19th and the first decade of the 20th centuries (cf. Osgood, 1943). Indeed, many genera that characterize Patagonia –Chelemys, Euneomys, Loxodontomys, Reithrodon, among others – are in need of systematic revision. Finally, until very recently, there were no more than a few phylogeographical studies of sigmodontine genera of the region (e.g. Hillyard et al., 1997; Kim et al., 1998).
The Abrotrichini, a recently recognized tribe extirpated from the classical Akodontini, is a clade that encompasses four mostly Patagonian genera (Notiomys, Geoxus, Chelemys and Pearsonomys) of long-clawed fossorial and semi-fossorial forms, and related mice of the genus Abrothrix (e.g. Pearson, 1984; Patterson, 1992; D'Elía et al., 2007). These five genera comprise nine species distributed along the southern Andes and neighbouring arid lands, and are adapted to environmental conditions ranging from the southern Andean forests to the Patagonian steppe. Although the genus Pearsonomys is endemic to the Pacific Valdivian forest (Patterson, 1992; D'Elía et al., 2006b), Abrothrix olivaceus is one of the most widespread abrotrichines, reaching 56°S, the southernmost recorded sigmodontine. Abrotrichine genetic geographical structure also reinforces the concept of a long association history of this tribe with Patagonia (Lessa et al., 2010, in press; see below).
Phyllotines show an important diversity in Patagonia, with six species belonging to five genera. During the last decade, successive molecular-based phylogenetic analyses (Smith & Patton, 1999; D'Elía, 2003; Steppan, Adkins & Anderson, 2004) have progressively reduced the number of species and genera recognized in this tribe. In contrast with the abrotrichines, phyllotine genera have substantial fractions of their distributions outside Patagonia (Table 1). Indeed, Loxodontomys is the only phyllotine genus that has most of its range in Patagonia. Remarkably, this tribe is today absent from Tierra del Fuego even though two of its genera, Eligmodontia and Phyllotis, reach the Magellan Strait. In general, phyllotines seem to be more strictly related to arid and semi-arid environments (Hershkovitz, 1962; Mares, 1980), whereas abrothrichines are also associated with southern Andean forests.
Euneomys and Reithrodon are two very peculiar sigmodontines that were once considered to be members of the Phyllotini (Steppan, 1995), but have been shown to be distantly related to this tribe, as well as to each other (D'Elía, 2003; Steppan et al., 2004). Both are medium to large in size, short tailed, densely furred, with hypsodont laminated molars, herbivorous in diet and nocturnal habits (Pearson, 1983, 1987, 1988). Data on their distribution and abundance along Fuego–Patagonia suggest some ecological differences between these rodents, with Euneomys dominating the harsh central and southwestern areas and Reithrodon occupying mainly oriental lowlands and northwestern steppes. The environmental preferences displayed by Euneomys might reflect an ancient connection to typical Patagonian habitats; fossil and genetic data strongly support this statement (see below). The long-tailed rat Irenomys, a unique sigmodontine genus restricted to the Nothofagus forest, is sister to Euneomys, according to molecular phylogenetic analysis (D'Elía et al., 2006a). The morphological distinctiveness between them, however, seems to be indicative of a deep divergence and a long history of association (probably arising in the Pliocene) of Irenomys with southern forest environments.
Two additional sigmodontine tribes are represented in southern South America, Akodontini and Oryzomyini, the most diverse sigmodontine groups in the subcontinent (Musser & Carleton, 2005; Weksler, 2006). Interestingly, Fuego–Patagonian representatives of Akodontini and Oryzomyini are scarce (Fig. 1), in accordance with their preference for tropical–subtropical to temperate environments. The ubiquitous genus Akodon is the only akodontine genus reaching mainly the northeastern portion of Patagonia, where it is represented by the species A. iniscatus and A. neocenus (Pardiñas, 2009). Typical genera that live in temperate grasslands and have southern expressions, such as Necromys or Oxymycterus, are not found beyond 39°S (Pardiñas et al., 2004). Only one oryzomyine, the versatile hantavirus reservoir Oligoryzomys longicaudatus, is widely represented in Fuego–Patagonia (Palma et al., 2005; Carbajo & Pardiñas, 2007). The low diversity of oryzomyines observed in Fuego–Patagonia also characterizes other nonforested environments, such as the Pampas and the Argentine Monte Desert.
Sigmodontine rodents and other small mammals that inhabit Patagonian dry lands (including the Monte Desert and Patagonian steppe) can be divided into two main, partially overlapping, assemblages. The approximate geographical boundary between them runs along the West Central Patagonian hills (‘Patagónides’) to meet the Río Deseado valley and Deseado Massif (Fig. 2). One of these main assemblages is found in the northeastern Patagonian lowlands and comprises species typically linked to the Monte Desert, such as A. neocenus, Calomys musculinus and Graomys griseoflavus (Figs 2, 3). This group is characterized by the predominance of akodontines and phyllotines and the total absence of abrotrichines. Contrary to a general perception (Ojeda, Blendinger & Brandl, 2000), only some (e.g. Akodon azarae, Holochilus brasiliensis), but not all, members of this assemblage are ‘marginal’ in Patagonia. This misperception largely results from limited sampling, and the unappreciated fact that many typical Monte Desert taxa are found beyond the limits of this biome. For example, C. musculinus and G. griseoflavus reach 48–50°S (Pardiñas et al., 2003; Udrizar Sauthier et al., in press), whereas the Monte Desert does not occur beyond 43°S (León et al., 1998).
In contrast, a predominantly medium- to high-land assemblage, mostly comprising abrotrichines, some phyllotines (Loxodontomys and Phyllotis xanthopygus) and distinct sigmodontine lineages (such as Euneomys), dominates the rest of the Patagonian steppe and grasslands (Figs 2, 3). This group is more diverse than the northwestern group, includes several Patagonian endemics (e.g. Notiomys; Fig. 3) and shares species with the Valdivian and Magellanic forests (e.g. Loxodontomys, Geoxus) and the northern portion of Tierra del Fuego (Patagonian grassland biome).
Not all taxa strictly fit into this main division. For example, both Oligoryzomys and Reithrodon are widespread in Fuego–Patagonia, a distributional pattern that probably reflects recent dispersion events (see below). Moreover, important portions of central and coastal Patagonia display a complex altitudinal mosaic in which the two main rodent assemblages are present side by side. In these landscapes, largely allopatric species – such as the pairs A. iniscatus and A. olivaceus, or Euneomys chinchilloides and G. griseoflavus– are found in sympatry (but not syntopy). In addition, we are beginning to understand the fundamental role of river valleys as corridors. These landscape elements seem to have favoured dispersal, especially from east to west. The Río Chubut, the only Patagonian river studied in some detail in this context (Udrizar Sauthier, 2009), clearly shows a transitional westward dispersal of typical eastern faunistic elements.
These two main assemblages appear to have responded differently to geobiotic Neogene events. The northeastern group comprises species without phylogeographical structure in Fuego–Patagonia, whereas the southwestern group exhibits several phylogeographical breaks within the region (Lessa et al., 2010, in press).
The Valdivian and Magellanic subpolar forests, associated with the Andean foothills, encompass a reduced assemblage of sigmodontine rodents (Fig. 3). Several are widespread forms, such as A. longipilis and Loxodontomys micropus (Pearson, 1983). Forest endemics include the genera Irenomys, an arboreal specialized rat, Pearsonomys, and the species A. sanborni (Osgood, 1943). Forest assemblages are still poorly known, particularly on the Argentine side. Apparently, some typical sigmodontines, such as Irenomys, are limited to forest environments north of the La Plata and Fontana lakes (45°S; Pardiñas et al., 2004). South of this latitude, an endemic form of Abrothrix, A. lanosus, seems to be associated with a narrow strip of humid forest shrublands, reaching the southern coast of Tierra del Fuego (Feijoo et al., 2010). The Magellanic tundra assemblage is even poorer than the Magellanic forests. Southernmost islands host Abrothrix species related to A. olivaceus from which putative endemic forms, such as A. hershkovitzi or A. llanoi, are hard to distinguish; in addition, the widespread Oligoryzomys and Euneomys (cf. Osgood, 1943) are also present.
Southern South America narrows as the latitude increases. As might be expected, a progressive pauperization of mammalian species is observed towards higher latitudes, a general pattern already recognized by several authors (e.g. Osgood, 1943; Texera, 1973). However, until very recently, distributional data were insufficient to examine this pattern in detail. The sigmodontine assemblages of Tierra del Fuego exemplify this point very well. Osgood (1943) recognized six sigmodontine species in the Archipelago, and highlighted differences in richness with respect to southern mainland populations. He also emphasized insularity as an explanation for the reduced species' pool on the island, and pointed out the role of the Magellan Strait in enforcing isolation.
With substantially more data at hand, we provide the following outline of the pattern of sigmodontine diversity (Fig. 4): (1) the geographical pattern of species' number decay at higher latitudes is more complex than was previously envisioned; (2) several continental species disappear or are virtually absent before reaching the Magellan Strait; (3) the pattern differs between closed (Valdivian and Magellanic forests) and open (Monte, Steppe and Patagonian grasslands) biomes; and (4) very recent extinctions have occurred within the region.
The southernmost islands collectively support depauperized assemblages with no more than four species (Texera, 1973; Patterson, Gallardo & Freas, 1984). However, a very low species' number is also observed in the southern mainland, especially close to the Atlantic Ocean. For example, southern Santa Cruz steppe–grassland assemblages are as poor as those in northeastern Tierra del Fuego (six species in both cases). Moreover, several widespread Patagonian mainland species, such as P. xanthopygus and Eligmodontia morgani, have scattered populations associated with particular habitats in the vicinity of the Magellan Strait (Pardiñas et al., 2009). Nevertheless, Eligmodontia might be present in suitable habitats of the northern portion of the island. This possibility is partially supported by its presence in the late Pleistocene archaeological assemblage of the Fuegian site Tres Arroyos 1 (U. F. J. Pardiñas, unpubl. data).
The most significant pauperization of Tierra del Fuego sigmodontine assemblages is mostly linked to forested environments. This can be illustrated by comparing records from Punta Arenas (a Nothofagus forest environment in the southern mainland) with those from the southern portion of Isla Grande de Tierra del Fuego. According to available data (cf. Osgood, 1943), the forest-dwelling genera Chelemys, Geoxus and Loxodontomys occur in the vicinity of Punta Arenas; in contrast, the dense forests of Isla Grande de Tierra del Fuego are exclusively occupied by A. olivaceus.
A significant decline in sigmodontine species' number seems to be associated with the Deseado Massif and the Río Deseado valley (Fig. 2), where several widespread species, such as A. iniscatus and G. griseoflavus, have their southernmost populations. A second break roughly coincides with the valley of the Río Santa Cruz, a southern distributional limit to sigmodontines such as Notiomys, as well as for other Patagonian mammals (e.g. the armadillo Zaedyus; cf. Allen, 1905).
The glacial history of the southern tip of South America may have contributed substantially to shape the sigmodontine diversity pattern at high latitudes. The forest areas of Tierra del Fuego were almost totally glaciated during the last glaciation and several earlier maxima (Rabassa, 2008). Thus, regional extinctions during glacial advances, coupled with variation in persistence and recolonization ability, might account for the observed distributional patterns. Genetic data (see below) suggest that A. olivaceus populations persisted during the last glaciation in a southern refugium, allowing the subsequent recolonization of Tierra del Fuego. Although the exact location of this proposed refugium is not clear, potential areas include northeastern Tierra del Fuego, which remained unglaciated through the most recent glaciations; other possible islands include Cape Horn and Isla de los Estados, or areas currently below sea level, especially on the eastern continental shelf.
Glacial advances – especially those of the early and middle Pleistocene – also deeply affected the southern mainland (Clapperton, 1993; Rabassa, 2008). It is likely that, during deglaciation episodes, major rivers running towards the Atlantic Ocean probably played an important role as barriers for recolonization processes (Turner et al., 2005). However, specific studies are needed to test this hypothesis. Faunal pauperization south of the Deseado Massif seems to be a product of glacial impact at these high-latitude extreme habitats.
Late Pleistocene diversity and the Pleistocene–Holocene transition
The oldest South American sigmodontine rodent fossils are found in sedimentary rocks of the Monte Hermoso Formation (c. 5 Ma, early Pliocene) in southeastern Buenos Aires Province, Argentina (Pardiñas, D'Elía & Ortiz, 2002; Prevosti & Pardiñas, 2009). These fossils are limited to a few fragmentary specimens, but show that several sigmodontine tribes were present at that time in the continent, including Akodontini, Phyllotini and Reithrodontini (Reig, 1978; Pardiñas & Tonni, 1998). Fossil sigmodontine rodents are relatively frequent in Pliocene–Pleistocene deposits of the eastern Pampean region, including the first records in early–middle Pleistocene times of several of the extant species (Reig, 1978, 1986; Pardiñas, 1999a). In turn, no mammals that may be clearly linked to these latest Miocene–Pliocene or early Pleistocene records of the Pampean region have been found in Patagonia (Tonni & Carlini, 2008).
Late Pleistocene mammal remains are relatively abundant in Patagonia, especially in archaeological contexts representing the interval between 13 and 10 radiocarbon kiloyears ago (14C ka) (Tonni & Carlini, 2008, and references cited therein). Although there are some small mammals associated with these sites, they are generally biased towards larger mammals.
Sigmodontine samples for the time interval between 13 and 8 14C ka are represented by less than a dozen archaeological and palaeontological sites scattered throughout Fuego–Patagonia and collectively represent nearly all living species of the region (Pardiñas, 1999b; Figs 5, 6). At 10 14C ka, the El Trébol fossil samples in northwestern Patagonia are dominated by the presence of sigmodontines from open grassy areas and shrublands, such as Reithrodon auritus and L. micropus, respectively (Pardiñas & Teta, 2008). Between 10 and 8 14C ka, the assemblages of Cueva Traful I and Cueva Epullán Grande are characterized by a lower species' number and the absence of primary Nothofagus forest sigmodontines, such as Irenomys tarsalis and Geoxus valdivianus (Pearson & Pearson, 1993; Pardiñas, 1999b; Pardiñas & Teta, 2008; Fig. 6). The only available sample from central Patagonia comes from a small cave of 12 14C ka that yielded an assemblage mostly dominated by typical sigmodontines from open shrubby and grassy areas, such as Eligmodontia, A. olivaceus and the rock-dweller P. xanthopygus. These data indicate a local landscape dominated by sparse shrubby vegetation mixed with bunchgrass patches and large rocky exposures (Teta et al., 2009). In southern Patagonia, the assemblages recovered at Piedra Museo 1 (10.4 14C ka; Fig. 6) and Los Toldos (c. 9–8 14C ka) are consistent with a relative expansion of grassy steppes under cold and humid climatic conditions. Latest Pleistocene conditions in the southern tip of South America are represented by samples from Cueva del Milodón (13 14C ka) and the Fuegian site Tres Arroyos 1 (12 14C ka), both in Chile. Small mammal fossils from these assemblages are indicative of open areas under very cold and windy conditions with minor Nothofagus evidence in the mainland.
In summary, inhospitable and cooler conditions, with scarce vegetation cover and extensive open bare areas, may have been widespread across Patagonia during the late Pleistocene and most of the early Holocene, at the time of the first human arrival (Pardiñas & Teta, 2008; Teta et al., 2009). In addition, it seems that, during the Pleistocene, areas of central Patagonia did not host species related to the Monte Desert (e.g. such as A. iniscatus, C. musculinus and G. griseoflavus).
Holocene stability and progressive colonization from the northeast
Sequences covering the entire Holocene are scarce in Patagonia. In two classical archaeological sites of northwestern Patagonia, Cueva Traful I and Cueva Epullán Grande (Fig. 6), Holocene samples suggest relative stability during the last 10 14C ka (Pearson & Pearson, 1993; Pardiñas, 1999b). Minor variations have occurred since the middle Holocene and indicate a progressive expansion of some northeastern forms towards the southwest, as well as brief expansions of mesic microenvironment-adapted species during cold and humid pulses (Pearson & Pearson, 1993; Pardiñas, 1999b). For example, in the Cueva Traful I sequence, forest dwellers, such as Geoxus valdivianus and I. tarsalis, are well represented around 9.4–8 14C ka and 2.7–2.2 14C ka, respectively (Fig. 6), in agreement with the more humid conditions and expanded tree coverage suggested by palynological data for these periods (Heusser, 1993). Steppe vegetation in Cueva Traful I is dominant, at least during the last 6.2 ky, possibly linked to a summer rainfall reduction between 8.5 and 5 ka (Markgraf, 1983); the earliest record of the arid land genus Eligmodontia is found in association with these changes (Fig. 6). Similar situations are also documented for the middle to late Holocene sequences of central Chubut (Udrizar Sauthier, 2009) and northwestern Santa Cruz (Pardiñas, 1998, 1999b). The middle Holocene was characterized by more humid conditions in northern Patagonia, contrasting with the northern Santa Cruz assemblages, which indicate a severe water deficit during the period between 7.5 and 4.7 14C ka (Pardiñas, 1999b).
Eastern forms, typically adapted to xeric shrub steppes of the Monte Desert, progressively expanded towards the west mostly during the middle to late Holocene. In northwestern Santa Cruz, the first appearance of A. iniscatus and G. griseoflavus is recorded around 7.6 14C ka, together with a frequency increase of Eligmodontia spp. (Fig. 6). In northwestern Patagonia, small mammal communities were enriched during the middle Holocene with the addition of A. neocenus in Cueva Epullán Grande (Pardiñas, 1999b). Other expansive events occurred along the Río Chubut, involving G. griseoflavus and Calomys spp. (Udrizar Sauthier, 2009).
The late Holocene (after 3.5 ka bp) was a highly variable period in terms of climate, with a rich fossil record that includes several extralimital occurrences for some sigmodontine rodents (Pardiñas, 1999b; Teta, Andrade & Pardiñas, 2005). Except for some sequences in northern or southernmost Patagonia, no unequivocal signals of climatic events of global occurrence, such as the Medieval Climatic Anomaly or the Little Ice Age, have been detected (but also, see Rebane, 2002). Humid and cold pulses, in some cases related to glacier advances, presumably allowed the eastern expansion of some species, such as L. micropus and Chelemys macronyx, that are adapted to mesic microenvironments in the shrubby steppes of northern Patagonia (Teta et al., 2005). Similarly, a cold and humid interval around 1.2–1 ka facilitated the ingression of A. lanosus, Chelemys macronyx and L. micropus into the eastern xeric steppes of the southern tip of the continent. In turn, warm and humid pulses, possibly associated with the Medieval Climatic Anomaly, might have allowed the range expansion of the amphibious rat H. brasiliensis along and across northern Patagonian streams (Fernández et al., in press; Pardiñas & Teta, in press). The variety of situations listed above suggests that, during the Holocene, northwestern Patagonian micromammal communities were the result of species-specific responses to environmental changes; massive replacements of species' pools across the west–east environmental gradient are not recorded.
Recent extinctions and the impact of historical human activities
One main conclusion that emerges from the study of available fossils of sigmodontine rodents is that micromammal communities have remained relatively stable through most of the last 10 ky. In contrast, dramatic changes in assemblage composition have been recorded during the last century, involving deep restructuring, local or regional extinctions and explosive increments of some opportunistic species. As there have been no significant climatic fluctuations during the last century, at least in comparison with those recorded during the Holocene, these changes must be mainly connected with human impact (Pardiñas, 1999b; Pardiñas et al., 2000; Andrade & Teta, 2003; Teta et al., 2005). Since the end of the 18th century, the massive introduction of sheep and, to a lesser extent, cattle occurred over nearly all of Patagonia, reaching a maximum in the 1940s and 1950s (Aagesen, 2000). Human impact also included shrub extraction and deforestation. Finally, in the lower valley of the Río Chubut, anthropogenic activities deeply modified the original environments, turning extensive natural areas into cultivated fields during the last 150 years (Pardiñas et al., 2000; Udrizar Sauthier, 2009). In the entire region, these changes were coupled with fire regime alterations, desertification and the introduction of exotic species (e.g. Veblen et al., 1999; Kitzberger & Veblen, 2003).
In central Patagonia, local extinctions included the disappearance of A. longipilis, L. micropus, Notiomys edwardsii and O. longicaudatus from the central portion of the Río Chubut valley (Udrizar Sauthier, 2009). Loxodontomys micropus also disappeared from the southern edge of the Somuncurá plateau (Andrade, 2009). In turn, opportunistic species, such as Calomys spp. and O. longicaudatus, took advantage of these changes, and their populations increased in size and geographical distribution. In some areas of the Río Chubut, Calomys accounts for up to 95% of the total small mammals recorded in owl pellets (Pardiñas et al., 2000). In western grassy steppe areas, shrub expansion and grassland reduction produced by cattle overgrazing and the introduction of some exotic shrubs favoured increases of A. olivaceus and Eligmodontia spp. (see Pearson, 1983; Monjeau, 1989; Teta et al., 2005), and probably allowed the expansion of A. iniscatus into northeastern Patagonia. However, not all local extinctions can be linked to human impact. For example, H. brasiliensis disappeared from northern Patagonia during the last few hundred years (< 400 years), perhaps in association with the Little Ice Age (Pardiñas & Teta, in press).
A less studied, but no less intriguing, phenomenon is the drastic decrease in abundance of Euneomys spp. in some areas of northwestern Patagonia after 10 000 or more years of predominance (Pearson, 1987; Pearson & Pearson, 1993; Pardiñas, 1999b; Fig. 6). Several hypotheses have been proposed to explain the causes of this event, including changes in fire regimens and the introduction of exotic pathogens (Pearson, 1987; Rebane, 2002). Introduced livestock may, however, be the most likely explanation for the decline of Euneomys, through the replacement of bunchgrass with spiny shrubs (Rebane, 2002) and a drastic reduction in open areas (Veblen & Markgraf, 1988). Of course, this phenomenon did not occur uniformly across Patagonia, and Euneomys is still dominant in some open, hostile, rocky areas of central and southern Patagonia (Pardiñas, 1999b; Pardiñas et al., 2003; Andrade, 2009).
Remarkably, although rodents have played a central role in studies on the effects of Quaternary glaciations on the biota, as in tropical South America (e.g. Lessa, Cook & Patton, 2003), Eurasia (e.g. Michaux, Libois & Filippucci, 2005), Africa (e.g. Nicolas et al., 2006) and North America (e.g. Runck & Cook, 2005), few studies (Smith et al., 2001; Cañón et al., 2010; Lessa et al., 2010) have been directed at addressing this issue in southern South America. More generally, few genetic-based studies have advanced hypotheses on the effect of Neogene glaciations on the sigmodontine fauna of southern South America. From the available information, it is clear that we are far from having a full understanding of the response of this group to historical climate change, although some generalizations may be advanced from the observed geographical patterns of genetic diversity.
Patagonian–Fuegian sigmodontines display three main phylogeographical patterns (Cañón et al., 2010; Lessa et al., 2010; Fig. 7). A set of species shows low levels of genetic variation and lacks phylogeographical structure. These single-clade species are those distributed mainly in central and northeastern Patagonia (A. iniscatus, C. musculinus, Eligmodontia typus and G. griseoflavus) and also include species extending to the southern end of the continent (Eligmodontia morgani) or to Tierra del Fuego (R. auritus and O. longicaudatus; see Belmar et al., 2009). The second broad observed pattern is represented by six species that exhibit phylogeographical structure within the Patagonian–Fuegian region. Thus, A. longipilis, A. olivaceus, Chelemys macronyx, E. chinchilloides, L. micropus and P. xanthopygus show genetic variation that is geographically structured within the study area. All these species belong to the southwestern ecogeographical assemblage that characterizes the Fuego–Patagonian steppe and, in some cases, adjacent forested areas. The third pattern is exhibited solely by Geoxus valdivianus, a species that shows two distinct clades, one in northern and the other in southern continental Patagonia; these clades are not sister to each other and differ by more than 10%, raising the possibility that they may represent different species (Lessa et al., 2010). The genetic data available thus far on A. lanosus (Feijoo et al., 2010) and N. edwardsii (Pardiñas et al., 2008) are insufficient for inferences of their genetic structure.
Among the species of the second group, the observed divergence between clades varies from 2% (Chelemys macronyx; Alarcón et al., in press) to around 5% (A. longipilis; Lessa et al., 2010). Similarly, the number and distribution of phylogeographical units are also variable. Abrothrix longipilis is the most diverse species with three parapatric clades within the study area: (1) Tierra del Fuego and southern Patagonia; (2) central Patagonia; and (3) northern Neuquén in northern Patagonia (Lessa et al., 2010; see also Palma et al., 2010). The other species present two clades. Patagonian haplotypes of A. olivaceus belong to a single shallow clade (Smith et al., 2001; Rodríguez-Serrano, Cancino & Palma, 2006), and those from Fuegian specimens form a distinct clade (Lessa et al., 2010). The phylogeographical breaks of E. chinchilloides and L. micropus are broadly congruent with that shown by A. longipilis at middle latitudes of Patagonia (Cañón et al., 2010; Lessa et al., 2010; see Figure 7 for examples of phylogeographical patterns).
The existence of phylogeographical structure within the Patagonian–Fuegian region strongly suggests a history of differentiation that may have occurred, at least in part, within the region (Lessa et al., 2010). Similarly, other studies in the Patagonian region have uncovered geographical structure suggesting local differentiation as well (Avila, Morando & Sites, 2008; Zemlak et al., 2008; Jakob, Martinez-Meyer & Blattner, 2009; Cosacov et al., 2010). Further, for several sigmodontine species, genetic data provided no evidence of range shifts towards the north during the Last Glacial Maximum. The available data suggest the survival of large populations within their current distribution ranges, or at least within the region. In contrast, southward colonization, initially suggested by Smith et al. (2001), probably took place in species such as G. griseoflavus, as noted above. Although the exact number and location of glacial refugia remain unknown (Cañón et al., 2010), it is clear that some of the refugia for at least some species must have been located at higher latitudes. In the case of A. olivaceus, this includes Tierra del Fuego.
Importantly, several phylogeographical units show signals of demographic expansion (Cañón et al., 2010; Lessa et al., 2010), which is often taken as indicative of a history of presumably postglacial colonization (Hewitt, 2000; Lessa et al., 2003). However, mitochondrial DNA-based estimates of expansion times for these clades, using species-specific Bayesian estimates of mutation rates, fall within the last 500 000 years (late Quaternary) and generally are older than the Last Glacial Maximum (Lessa et al., 2010).
In summary, the emerging pattern for the recent biogeographical history of sigmodontines in Patagonia and Tierra del Fuego includes both recent (although not necessarily post-Last Glacial Maximum) colonization from lower latitudes, as well as differentiation within the region. Multiple refugia, including some at higher latitudes, need to be invoked to explain the distribution of current genetic diversity harboured by sigmodontine populations (for a synthesis of this issue across diverse Patagonian taxa, see Sérsic et al., 2011).
THE EVOLUTION OF FUEGO–PATAGONIAN SIGMODONTINE ASSEMBLAGES: SUMMARY AND PROSPECTS
Our attempt to provide a comprehensive review of the history of sigmodontine rodents in Fuego–Patagonia has been made possible by significant advances in the study of current and historical distributions of these small mammals and their environments, coupled with taxonomic, phylogenetic and phylogeographical efforts. The data at hand are sufficient to indicate that: (1) the diversification of Fuego–Patagonian sigmodontines has involved both local differentiation and colonization from northern sources over a time scale of at least one million years; (2) most demographic changes reflected by patterns of genetic variation trace back to the last 500 000 years, but few of these are likely to be post-glacial; (3) local extinctions, colonization and changes in abundance have occurred through the late Pleistocene and Holocene in association with climate change; and (4) similar classes of changes have occurred most recently, often as a result of human-related activities that have impacted local habitats.
There are limitations to these inferences related to insufficient and uneven coverage of the vast Fuego–Patagonian region in terms of both current diversity (including genetic data) and fossil and subfossil data, as well as a lack of comprehensive taxonomic studies of the taxa involved. These limitations, as well as others inherent to the data analysed, preclude a greater integration of fossil and molecular data.
Some apparent unconformities between fossil and genetic inferences of species' history serve to illustrate the limitations of the available data. For instance, the fossil record indicates that C. musculinus entered Patagonia in the last few thousand years, whereas genetic estimations suggest older times of expansion. The incompleteness of the fossil record may explain this incongruence. However, genetic inferences are based on simplified models and, at this point, a single locus, and should be interpreted with much caution.
It is clear that much more detailed work will be needed in order to refine the general outline provided here and to establish rigorous ties to geological events suffered by Fuego–Patagonia during the Neogene (Clapperton, 1993; Rabassa, 2008). The role of rivers and their changes during glacial cycles is not fully understood, but was possibly important both for dispersal and as potential barriers. Extensive flooding episodes also characterized the late Neogene history of the region and were large enough to produce extensive gravel beds – the famous ‘Rodados Patagónicos’– from tablelands to the coast (Clapperton, 1993; Martínez & Kutschker, 2011). Geocryogenic processes, especially those that surely affected central plateaus during glacial advances, need to be studied in detail. However, the recorded wedge ice casts in northeastern Patagonia suggest that much of the nonglaciated territory was also under extremely harsh environmental conditions (Trombotto, 2008, and references cited therein). Recently published evidence on hyperarid conditions during several Pleistocene periods (Bouza et al., 2007) adds a new piece to this complex puzzle. Finally, we still have very few data to adequately understand the potential role played by the Atlantic Ocean continental shelf, which was largely exposed during glacial advances (Clapperton, 1993; Rabassa, 2008; Ponce et al., 2011).
Genetic data on Fuego–Patagonian sigmodontine rodents have accumulated rapidly in recent years. Some patterns are beginning to emerge, but it is clear that multilocus data and substantially expanded geographical sampling are necessary to identify potential refugia and to distinguish them from recolonized areas, examine the relative importance of shared versus idiosyncratic species' responses to long-term climate change and to reduce the uncertainty associated with a single mitochondrial gene. The interplay between presumably neutral divergence associated with phases of geographical isolation and adaptive divergence in response to environmental variation is only beginning to be examined in Fuego–Patagonia (e.g. Ruzzante et al., 2011).
Finally, the integration of sigmodontine results with those obtained for other components of the Fuego–Patagonian biota is of much need to gather a general picture that, for example, may identify the location of Pleistocene refugia. In this sense, a recent and intense surge of interest in the biogeography of southern South America (e.g. Ruzzante et al., 2011; Sérsic et al., 2011), which has developed among different research groups, beckons an era of profound learning of this unique and marvellous part of the world.
The authors are indebted to the organizers of the 2009 Workshop in La Plata, Argentina, J. Rabassa, D. Ruzzante, E. Tonni and A. Carlini, for the opportunity to present part of this contribution. This review was based on data that were obtained, processed or discussed with many students and assistants, including A. Formoso, D. Podestá, G. Cheli, D. Udrizar Sauthier, J. Sánchez, G. Mendos, G. Massaferro, J. Guzmán, C. Cañón, O. Alarcón, J. Martínez, A. Parada and C. Sierra. Access to archaeological samples was made possible by many archaeologists, especially L. Borrero, F. Martin, E. Crivelli-Montero, M. Fernández, L. Miotti, M. Massone, M. Salemme, E. Moreno, C. Belelli, R. Goñi, C. Aschero, J. Belardi, R. Barbarena, A. Sanguinetti de Bórmida, T. Civalero, J. Gómez Otero, M. Silveira and M. Boschín. Daniel Ruzzante, Mariana Morando and one anonymous reviewer made valuable suggestions in an earlier version of this work. Economic funds for field and laboratory activities were provided by Consejo Nacional de Investigaciones Científicas y Técnicas (PIP 6179) and Agencia (PICT 32405 and PICT 2008–0547) (to UFJP), Comisión Sectorial de Investigación Científica-Universidad de la República (PEDECIBA) and The National Geographic Society (CRE 7813–05) (to EPL) and Fondo Nacional de Desarrollo Científico y Tecnológico (11070157 and 1110737) (to GD). Our deepest gratitude is expressed to these persons and institutions.