Large mammal burrows in late Miocene calcic paleosols from central Argentina: paleoenvironment, taphonomy and producers

Size and plan view pattern

Observed horizontal diameter (Dh) ranges from 0.15 to 1.50 m ( n = 112) and the frequency distribution suggests a roughly normal distribution where three subpopulations can be distinguished (Fig. 6). The small subpopulation has a Dh from 0.15 to 0.34 m (8 %), the intermediate subpopulation has a Dh from 0.39 to 0.98 m (84 %), and the large subpopulation exhibits a Dh from 1.05 to 1.50 m (8 %).

In plan view exposure, which is found only at SG and LLP localities (n = 78), a number of morphologies can be distinguished (Fig. 7). (1) The more common are straight to slightly curved burrows (89 % of cases), which exhibits a Dh= 0.15–1.15 m, showing a uniform inclination of internal laminae (ranging from ≈ 0° to 27°), the maximum height difference between the proximal and distal portion of a burrow is 0.6 m, and the maximum preserved length is 5.18 m (Figs. 7A, 7B). Some burrow fills in this category display a decrease in inclination of internal laminae toward more distal positions (i.e., from 27° to 8°). (2) A sinuous burrow that displays two opposite curves in plan view was recorded in 5% of the cases (Figs. 7C, 7D). The horizontal diameter of sinuous burrow ranges from 0.42 to 0.80 m, dip of internal laminae is subhorizontal to slightly inclined (up to 8°), and the maximum observed length is 8 m. (3) The third plan view pattern is a C-shaped curve observed in 6% of the burrows, with an horizontal diameter ranging from 0.44 to 0.72 m (Figs. 7E 7F), which commonly appears as a ramp with a height difference of up to 0.55 m, the inclination of internal laminae can be uniform (from 3°to 12°) or show a shallowing toward the distal position (from 14° to subhorizontal).

In a few cases, the distal portion of burrow showed a lateral expansion of up to 23% of the Dh, commonly having a subhorizontal lamination (Figs. 7G, 7H). Other burrow fills exhibit a rounded end with no enlargement that can be accompanied by an upward bending of mudstone laminae against the walls of the burrow.

Cross-sectional shape and body mass

The analysis of the well defined cross-sectional shape of burrows (n = 24 from all localities) suggest a distinction between elliptical (with the major axis subhorizontal) and subcircular cross-sections. Elliptical cross-sections are more common (n = 18) and the corresponding Dh ranges from 0.39 to 1.50 m (belonging to the intermediate and large subpopulations, Fig. 6), with an average Dv/Dh ratio of 0.55. The burrows with elliptical cross-section include a few cases (n = 4) with a flat bottom and convex top. The subcircular cross-sectional shape (n = 6) is represented in the intermediate subpopulation with a Dh ranging from 0.39 to 0.56 m, and an average Dv/Dh ratio of 0.88.

Morphometric analyses suggest that 90.13 % of the variation is explained by the first two principal components (Fig. 8B), and deformation grids range from elliptical (score = −0.12) to subcircular (score = 0.17) (Fig. 8A).

Body mass estimates of the producers of the burrow on the basis of the cross-sectional area (using the method by Wu et al., 2015) suggest that there are two ranges (Table S1). Most of the estimates (n = 18) belongs to the intermediate subpopulation with a range from 37 to 439 kg, whereas the remaining estimates comes from the large subpopulation (n = 7) with a range 708 to 1,623 kg. Burrows with subcircular cross-section from the intermediate subpopulation, are linked with a producer having body mass from 92.84 to 186.0 kg.

Orientation and inclination

Readings of plunge azimuths of burrow fills from all localities are variable but most values are located in the northeast to southeast quadrants (i.e., between N20° and N140°) (Figs. 9A, 9C). The average dip angle of all measured burrows with respect to the paleohorizontal is 7.25°and ranges from nearly 0 to 27° (Fig. 9B). Most orientation data come from the intermediate subpopulation (Dh = 0.39–0.98 m) and especially from LLP locality.

Composition of burrow fills

The vertebrate burrows are easily spotted in the field because of the distinctive laminated structure of the infilling sediments that contrasts with the structureless host rock. The filling is composed of an alternation of laminated claystone and siltstone with massive fine-grained sandstone and siltstone containing floating claystone intraclasts. Laminated intervals are a few millimeters to about 50 mm thick, whereas massive intervals tend to be thicker. Most of the burrow fills display a laminated interval in the lowermost part of the fill, with the upper part massive, especially in the Salinas Grandes de Hidalgo locality (Figs. 10A–10C). A few burrows display a poorly defined lamination to massive structure throughout (Fig. 10D). Claystone and siltstone laminae at the bottom of the structure typically deflect upwards against the burrow wall, which is a good criterion to distinguish burrow fills that are mostly exhumed by erosion. Laminae tend to be horizontal but successive laminated packages resting at low angles were also identified. Individual laminae are normally graded (typically siltstone grading to claystone), and locally disrupted giving a brecciated aspect. Both synsedimentary faulting and deformation were identified (Fig. 10F). A pseudomeniscate structure was identified in two cases (one from Salinas Grandes de Hidalgo and the other from Laguna La Paraguaya). This structure is composed of massive siltstone or fine-grained sandstone arranged in adjacent crescent-shaped bodies with the convex margin pointing downslope that span the full width of the fill (Fig. 10E). Horizontal width of individual pseudomeniscate bodies taken parallel to the burrow axis is 120 mm.

Faunal remains found in burrow fillings

In general, bone remains found inside a burrow can be considered as belonging to its producer or occupant only if they are articulated, disarticulated but still closely associated, nearly complete, are commonly found in a terminal portion and fit the size (cross-sectional diameter) of the burrow (e.g., Smith, 1987; Groenewald, Welman & MacEachern, 2001; Damiani et al., 2003). The remains found inside the studied burrows do not fulfill any of these criteria. In most cases, these bone remains have been passively introduced and it is uncertain if they belong to the producers. The remains are essentially fragmentary, disarticulated, with evidence for abrasion and weathering (Fig. 12A, Fig. S1); suggesting that they spent some time at the surface and then were introduced into the burrows by currents along with other sedimentary particles. The fragmentary and disarticulated state of Doellotatus sp. and one of the specimens of Paedotherium minor and the considerably small size of the animals (body mass about 1–2 kg, Table 1) in comparison with the containing burrows; further suggest that these remains were introduced by currents. In the case of Proscelidodon sp., the bones are disarticulated but associated, which suggest that they can belong to a single specimen, and the partial horizontal diameter of the burrow match the size of this ground sloth. The only articulated remains are fragments of the dorsal carapace of Glyptodontidae that occur in burrows large enough to hug these animals (Dh = 0.78 to 1.50 m) (Table 1). In consequence, the unique remains that can belong to the producer of the burrows are Proscelidodon sp. and those of Glyptodontidae.

Fossorial mammals of the Cerro Azul Formation and size of burrows

The mammals with fossorial habits recorded in the Cerro Azul Formation include xenarthrans, notoungulates and rodents (e.g., Goin, Montalvo & Visconti, 2000; Cerdeño & Montalvo, 2001; Urrutia, Montalvo & Scilato-Yané, 2008). Among the Xenarthra, the Glyptodontidae, Dasypodidae and Mylodontidae display fossorial adaptations. The same is true for Mesotheridae and Hegetotheriidae (Notoungulata); and Caviidae, Octodontidae, and Chinchillidae (Rodentia). Below we discuss the potential producers for each size class of the burrows (Table 1) as expressed by the horizontal diameter and cross-sectional area of the burrows.

For the small subpopulation (Dh = 0.15–0.34 m, 8% of cases), with a body mass ranging from 1 to 13 kg, the likely candidates are the notoungulate Paedotherium minor, the dasypodids Doellotatus, Chorobates, Proeuphractus, and Chasicotatus; and the rodent Lagostomus. Paedotherium (Hegetotheriidae) is a medium-sized rodent-like ungulate native to South America. This taxon is very common in the Cerro Azul Formation, both in La Pampa and Buenos Aires provinces (e.g., Montalvo, Tomassini & Sostillo, 2016). Articulated remains of this genus have been found within Pliocene burrow casts (about 0.16–0.22 m wide) from the Atlantic coast of Buenos Aires province (e.g., Genise, 1989; Scognamillo, 1993; Elissamburu, Dondas & De Santis, 2011) and a morphofunctional analysis of its postcranial skeleton suggest a digging capacity (Elissamburu, 2004).

The Dasypodidae show a neotropical geographic distribution and were important components of the late Miocene-Pliocene South American fauna (Scillato-Yané, 1982; Ortiz Jaureguizar, 1998). Dasypodids exhibit fossorial habits and were abundant during the late Miocene in the Pampean region of Argentina, suggesting preference for open environments and well drained soils (Scillato-Yané et al., 2013). Most dasypodids recorded in the Cerro Azul Formation were small- to medium-sized, with body mass in the range 1–10 kg for Doellotatus, Chasicotatus, Proeuphractus and Chorobates (Table 1). In particular, the holotype of Chasicotatus ameghinoi is a nearly complete carapace about 150 mm wide (Scillato-Yané, Krmpotic & Esteban, 2010), which match the lower size range of the small subpopulation. Modern dasypodid burrows are usually simple ramps lacking significant enlargements (e.g., González, Soutullo & Altuna, 2001; Abba, Udrizar & Vizcaíno, 2005), which is similar to the architecture of the fossil burrows.

In the same localities of Paedotherium-bearing burrows from the Atlantic coast of the Buenos Aires province, there are also burrows containing articulated remains of Lagostomus that partially overlap in diameter with those containing Paedotherium remains (Genise, 1989; Elissamburu, Dondas & De Santis, 2011). The extant Lagostomus maximus (plains vizcacha) is well known for its digging adaptations and for living in communal burrow systems (e.g., Jackson, Branch & Villarreal, 1996). Plains vizcacha burrow systems show an average entrance horizontal diameter of 0.26 m and a range of 0.17–0.37 m (Llanos & Crespo, 1952), which matches the range of the small subpopulation. However, extant L. maximus burrow systems have several entrance ramps that typically converge into a central chamber or a much more complex architecture (e.g., Llanos & Crespo, 1952; Rafuse et al., in press), which contrast with the simple ramp type morphology of the fossil burrows. The 43 mm in diameter subcircular burrow identified in the fill of a larger burrow at Salinas Grandes de Hidalgo (# 638) is probably related to a caviomorph rodent (Caviidae or Octodontidae).

For the dominant intermediate subpopulation (Dh = 0.39–0.94 m, 83% of measured burrows), with an estimated body mass ranging from 37 to 438 kg, the likely candidates are the Mesotheriinae (Mesotheriidae, Notoungulata); Eosclerocalyptus, Coscinocercus, and Aspidocalyptus (Xenarthra, Glyptodontidae); Macrochorobates and Macroeuphractus (Xenarthra, Dasypodidae); and Proscelidodon (Xenarthra, Mylodontidae). The fossil remains found in this size range that are likely candidates are those of Glyptodontidae and Proscelidodon sp. (Table 1). There are two Mesotheriinae species recognized for the late Miocene of central Argentina: Pseudotypotherium subinsigne and Typotheriopsis silveyrai (Cerdeño & Montalvo, 2001). These species exhibited a small to medium size (20.88 to 60.13 kg after Croft, Flynn & Wyss, 2004) (Table 1). The Mesotheriidae shows modifications in the appendicular skeleton that suggest a scratch-digging habit and fossorial adaptations and are envisaged as having used its hypsodont teeth to cut roots and break the substrate, to aid digging with claws (Shockey, Croft & Anaya, 2007).

Kraglievich (1934) and Quintana (1992) suggested that glyptodonts were not functionally suited for digging. However, a geometric morphometric study of the limb bones of five glyptodont species of Miocene and Pleistocene age and comparison with extant armadillos led Vizcaíno et al. (2011) to conclude that were generalized diggers, as modern Dasypodini and Euphractini. Generalized diggers are species that dig short burrows for protection or in search of food and that feed on the surface or just below it by making ‘food probes’ (Abba, Udrizar & Vizcaíno, 2005). In order to asses if glyptodonts were likely producers of the fossil burrows we compared the width of the dorsal carapace and the dorsal carapace height / width ratio with comparable values of the fossil burrows. Dorsal carapace width of Miocene-Pliocene glyptodonts range between 0.40 and 0.77 m (Perea, 2005; Vizcaíno et al., 2011; Zurita et al., 2011), well in the range of horizontal diameter of the fossil burrows. Information on the ratio between carapace height and width for Miocene-Pliocene glyptodonts is incomplete, and similar data for Pleistocene South American glyptodonts (Duarte, 1997; Zurita et al., 2010) average 0.87 (range = 0.78–0.91; n = 4). In our case study, glyptodonts are considered good candidates for constructing the subcircular burrows of the intermediate subpopulation, which are 0.39–0.56 m wide and display an average Dv/Dh ratio of 0.88. Regarding the large dasypodids Macrochorobates and Macroeuphractus, the available body mass estimates suggest a range of 10 to 100 kg (Vizcaíno & Fariña, 1999) and little is known about their paleoecology.

Among the mylodontids, the Scelidotherinae, endemic to South America (McDonald, 1987; Taglioretti et al., 2014); are only represented for the Huayquerian—Chapadmalian SALMAs (late Miocene to early Pliocene) by Proscelidodon, a ground sloth related to open environments with grasslands, under temperate and warm climate (Miño Boilini et al., 2011; Pujos et al., 2012; McDonald & Perea, 2002). A digging habit was inferred for Proscelidodon after a morphofunctional study of a Montehermosian (latest Miocene-early Pliocene) forelimb (Aramayo, 1988). Body mass estimates are only available for Pleistocene scelidotherines (Table 1) and range from 584 to 1,057 kg (De Esteban-Trivigno, Mendoza & De Renzi, 2008; Bargo et al., 2000; Fariña, Vizcaíno & Bargo, 1998). These would be maximum estimates for late Miocene scelidotherines because the primitive Mylodontidae were smaller and there seems to be a trend toward progressively larger sizes in the Pleistocene (e.g., McDonald & Perea, 2002). Large Pliocene-Pleistocene fossil burrows near Mar del Plata city (Buenos Aires province) have been attributed to mylodonts on the basis of the finding of bone remains inside the fill (Frenguelli, 1955) and using the surface ornamentation of the burrows (Zárate et al., 1998; Dondas, Isla & Carballido, 2009).

For the large subpopulation, with a Dh ranging from 1.05 and 1.50 m (9% of cases) and an extrapolated body mass of 700–1,600 kg, the more likely producer is Proscelidodon sp. and, secondarily, the Glyptodontidae.

To summarize, the studied fossil burrows can be attributed to several producers, according to their horizontal diameter. The more likely producers of the studied fossil burrows are: (1) for the small subpopulation, the smaller dasypodids (Doellotatus, Chasicotatus, Proeuphractus and Chorobates) on the basis of body mass, the fossorial habit and architecture of modern dasypodid burrows and, secondarily, Paedotherium minor. (2) For the intermediate and large subpopulations, the Glyptodontidae and Mylodontidae (Proscelidodon sp.) are good candidates as these were the largest representatives of the late Miocene burrowing fauna of the Cerro Azul Formation. The Glyptodontidae were generalized diggers, like modern dasypodids, and exhibited a carapace fitting especially the subcircular burrows. Proscelidodon sp. is also a likely candidate of the elliptical and larger burrows. For the intermediate subpopulation, probably the large dasipodids (Macrochorobates and Macroeuphractus) and Mesotheriinae should be considered.

Sedimentology

Thick, monotonous, massive continental successions of siltstone showing moderate to good sorting with associated paleosols, as those described for the Cerro Azul Formation, are typical of loess deposits, whose dominantly eolian origin is well established (e.g., Johnson, 1989; Pye, 1995). The presence of pedogenic calcite is indicative of well-drained soil profiles in sub-humid, semi-arid, and arid climates with low rainfall (less than 800 mm/yr) and high evapotranspiration (see review in Sheldon & Tabor, 2009). Previous estimation of mean annual precipitation for the development of the paleosols of the Cerro Azul Formation is 449 ±147 mm (Cardonatto et al., 2016). Paleosols showing a Bt horizon and blocky or prismatic peds can be compared with mollisols (Cardonatto et al., 2016). Some paleoenvironmental constraints can also be derived from the composition of the mammal fauna, and the stable isotopic composition of enamel teeth. Vertebrate remains of the Cerro Azul Formation, mainly notoungulates and rodents, suggest that these sediments were deposited in open landscapes like steppes or herbaceous plains (Montalvo et al., 2008). Carbon isotope composition from late Miocene herbivorous enamel teeth from Salinas Grandes de Hidalgo and nearby localities indicates a dominance of C3 plants in lowland areas (MacFadden, Cerling & Prado, 1996), which are favoured in climates with a cool growing season (Ehleringer, Cerling & Helliker, 1997)

Orientation of burrows

Comparison with orientation data from modern Dasypodidae burrows can help to interpret the orientation pattern of fossil burrows. As xenarthrans are imperfect homeotherms, their body temperatures do change with the environment (e.g., McNab, 1980; McNab, 1985). It has been suggested that the burrow entrance orientation of armadillos avoid prevailing winds and both uniform and preferential orientation has been documented (e.g., McDonough & Loughry, 2008). The cases of no preferential orientation are related to the invasive armadillo Dasypus novemcinctus from southern USA (Texas, Alabama, Oklahoma) and Belize (Clark, 1951; Zimmerman, 1990; Platt, Rainwater & Brewer, 2004; Sawyer et al., 2012). All these cases are mostly related to forested areas. Studies documenting a preferred orientation of Dasypodidae burrows are from Argentina, Uruguay and Brazil, involving open environments and several species (Crespo, 1944; Carter & Encarnaçao, 1983; González, Soutullo & Altuna, 2001; Abba, Udrizar & Vizcaíno, 2005; Ceresoli & Fernandez-Duque, 2012). The pioneer study by Crespo (1944) included three localities from western Argentina, ranging from 27°37″S to 34°13″S with annual precipitation ranging from less than 200 mm to 500 mm. The vegetation ranges from low bushes, to shrubland and psammophilous grassland with sparse trees. These localities belong to the Monte and Espinal biogeographic provinces (e.g., Roig, Roig-Juñent & Corbalán, 2009) and the included armadillo species are: Chaetophractus vellerosus, C. villosus and Zaedyus pichiy. A compilation of the entrance orientation data from the three localities of Crespo (1944) suggests a dominant entrance orientation toward the west (Fig. 9D). This distribution is remarkably similar to the fossil burrows if we assume that entrance orientation was at 180°of dipping azimuth (Fig. 9C). Dominant surface wind patterns in northern Argentina are humid and sometimes hot winds from the east and north (e.g., Barros et al., 2015), whereas cold winds are from the south. In consequence, the orientation pattern described by Crespo (1944) from open environments of the semiarid region of Argentina can be interpreted as preferential orientation of entrances avoiding dominant hot and cold winds. Similar patterns of armadillo burrow entrance orientation avoiding prevailing winds were documented by Carter & Encarnaçao (1983) in Minas Gerais, Brazil; González, Soutullo & Altuna (2001) in Uruguay (Fig. 9E); Abba, Udrizar & Vizcaíno (2005) in Buenos Aires province of Argentina (Fig. 9F); and Ceresoli & Fernandez-Duque (2012) in Formosa province, northern Argentina. Alternative explanations for this preferential orientation are that, as the armadillos seek food following an odour in the wind, they tend to approach a site from downwind and dig in the lee side (Carter & Encarnaçao, 1983) and to maximize sun exposure during cold winters (Ceresoli & Fernandez-Duque, 2012). In particular, the most adequate example to evaluate the orientation of the fossil burrows is the data from dasypodid burrows by Crespo (1944), which were collected in open semiarid settings similar to those of the late Miocene of central Argentina. In consequence, it is possible to propose that the late Miocene wind pattern of central Argentina was similar to the present one with hot winds from the east and north and cold winds from the south.

Associated trace fossils

The trace fossil assemblage of the Cerro Azul Formation is of low diversity and abundance and dominated by insect trace fossils (Celliforma, Rosellichnus, Fictovichnus, Rebuffoichnus and Teisseirei), and was compared with the Celliforma ichnofacies (Cardonatto et al., 2016). The Celliforma ichnofacies is typical of well-drained calcareous paleosols developed under low vegetation coverage (Genise et al., 2010; Genise et al., 2016). The reduced size of associated rhizoliths suggests that the vegetation was dominated by scrubs with minor participation of herbaceous plants.

The local occurrence of cemented Coprinisphaera at LLP and additional occurrences of fossil dung-beetle brood balls (Quirogaichnus coniunctus Laza, 2006) from the formation in a nearby locality (Laza, 2006) is indicative of the presence of the Coprinisphaera ichnofacies, suggesting herbaceous communities and wetter climatic conditions (Genise et al., 2016) for the easternmost locations of the formation.

Burrowing habits in large late Miocene mammals

Mammal burrows are typically constructed as shelters from environmental extremes and predators, and also for food storage, foraging and reproduction (e.g., Reichman & Smith, 1990; Kinlaw, 1999). From these common uses of burrows, protection from environmental extremes and predators are more likely for the studied fossil burrows and no evidence supporting the remaining functions is available. Top predators during deposition of the Cerro Azul Formation are the Phorusrhacidae (Cenizo, Tambussi & Montalvo, 2012; Vezzosi, 2012) that occupied the role of large carnivorans, as well as the Sparassodonta (Goin, Montalvo & Visconti, 2000).

However, the main factor controlling the occurrence of large mammal burrows during the late Miocene (Fig. 1) is herein related to environmental changes. It has been suggested that different mammal groups acquired fossorial habits during the Cenozoic as a response to the expansion of open, savanna-like environments under cold, dry and seasonal climates (Nevo, 1979; Nevo, 1995; Nevo, 2011). During the late Miocene (the Huayquerian SALMA), southern South America experienced a global cooling as response to the increase in the Antarctic ice sheet (Zachos et al., 2001) and the uplift of the Andes (e.g., Amidon et al., 2017), which favored cold and seasonally dry climatic conditions. This regional framework is confirmed by the inferences on the sedimentology, faunal remains and invertebrate ichnology of the studied succession. This evidence suggests open environments, with well-drained soils and low vegetation coverage, and a semiarid and seasonal climate. Considering that the more likely candidates for the largest burrows are xenarthrans (Glyptodontidae and Mylodontidae), which are imperfect homeotherms (e.g., McNab, 1980; McNab, 1985), the necessity and convenience for excavating an underground refuge is clear. In addition to escape from predation, these animals used burrows to avoid extremely cold or warm climatic conditions in order to conserve energy and water, and to breed because of the particular physiology of xenarthrans (Vizcaíno et al., 2001).