Introduction

Cuniculus Brisson, 1762 (= Agouti Lacépède, 1799; Coelogenus Cuvier, 1807; and Coelogenys Illiger, 1811 [ICZN 1998: opinion 1894]; see Patton 2015) includes caviomorph rodents of the family Cuniculidae (= Agoutidae, e.g., McKenna and Bell 1997), which is nowadays represented by the extant species Cuniculus paca (Linnaeus, 1766) and Cuniculus taczanowskii (Stolzmann, 1885) (see Stolzmann 1885; Moojen 1952; Pérez 1992; Nowak 1999; Eisenberg and Redford 1999; Woods and Kilpatrick 2005; Castro et al. 2010; Patton 2015). The recently proposed extant species “Cuniculus hernandezi” Castro, López, and Becerra, 2010 was deemed a nomen nudum (Ramírez-Chaves and Solari 2014), requiring further analysis. The current distribution of Cuniculus encompasses the tropical portion of the South and Central Americas (Pérez 1992; Queirolo et al. 2008 and references therein), but reports of Cuniculus fossils are restricted to the territory of Brazil.

The first description of an extinct species of this genus was made by the Danish naturalist Peter Wilhelm Lund based on fossils collected during his explorations of Lapa da Cerca Grande, Lagoa Santa, Minas Gerais, Brazil (Lund 1837: 317). Lund erected the species Cuniculus (= Coelogenys) rugiceps (Lund, 1837) (see Lund 1837: 331–332, pl. 3) and described it as larger than Cuniculus paca and presenting some morphological particularities, such as rugosities on the atlas (not observed in extant species) and stronger rugosities on the surface of the skull bones (Lund 1837: 317). During Lund's life, Cuniculus taczanowskii was not yet described, thus all Lund's discussions regarding extinct species were in relation to C. paca. However, the osteological differences between C. paca and C. taczanowskii are relatively subtle (see Pérez 1992; Ríos-Uzeda et al. 2004) and Lund's distinctions between the fossils and C. paca are applicable to C. taczanowski too. In subsequent works, Lund (1839a, b, 1841a) mentioned more fossils of cuniculids recovered from his excavations in different caves from the Lagoa Santa region. He assigned a portion of the new fossils to Cuniculus (= Coelogenys) rugiceps and emphasized again its larger skull proportions compared to C. paca (Lund 1839a: 42, 1841a: 102). Lund assigned another portion of the fossils to two new species: (i) Cuniculus (= Coelogenys) laticeps (Lund 1839a: 42, 1841a: 102), for which the posterior portion of the zygomatic arch was more laterally projected than in C. paca, giving a flattened appearance to the skull; and (ii) Cuniculus (= Coelogenys) major (Lund 1839a: 42, 1841a: 102), which was characterized solely by its larger dimensions relative to Cuniculus laticeps and Cuniculus rugiceps, and consequently to the extant species too.

Later, Lund (1839a: 79–80, 1841a: 139–140) argued that the differences between C. rugiceps and C. laticeps might be due to sexual dimorphism, and synonymized both names under C. laticeps. In another work, Lund reinforced the validity of the species C. laticeps and the larger one, C. major (Lund 1839b: 228–229).

In 1840, Lund presented additional distinctive characters for C. laticeps in the explanation of the pl. 20 (Lund 1840: 54–55, 1841c: 270–271): (i) well-developed infraorbital foramen; (ii) narrow dorsal root of the zygomatic process of the maxilla; (iii) zygomatic arch backward projected (considered as the posterior elongation of the jugal); and (iv) a “smooth skull”. In addition, the skull illustrated in pl. 20 shows that the dimensions of C. laticeps are very similar to C. paca (Lund 1840: pl. 20: 1, 2, 1841c).

After Lund, Winge (1887) in a review of the Lagoa Santa fossil rodents considered the two extinct species maintained by Lund (1939a) (Coelogenys laticeps and Coelogenys major) as subspecies of Cuniculus (= Coelogenys) paca. Later, Rusconi (1933) considered the validity of Cuniculus (= Agouti) major, assigning new material from Lapa do Caetano (= “Gaetano”, Rusconi 1933), Lagoa Santa, to this taxon. Subsequently, Paula Couto (1953, 1975) and Cartelle (1999, 2002, 2012) listed C. major and C. laticeps among the extinct species identified from Lagoa Santa caves, and Vasconcelos et al. (2015) reported new material assigned to C. major from Prudente de Morais, Lagoa Santa. Besides Lund's (1837, 1839a, b, 1840, 1841a, c) and Winge's (1887) works, extinct cuniculids were mentioned in some publications, but few works have been devoted to thoroughly discussing extinct cuniculid taxonomy (e.g., Rusconi 1933).

Although extinct cuniculids were among the first fossil mammals studied during Lund's pioneering work in the 19^th century, the little attention that has been paid to them resulted in a confusing and unresolved taxonomic arrangement. Moreover, in Lund's (1837, 1839a, b, 1840, 1841a, c) works there are no type specimens (Oliveira et al. 2010), which makes the comparison of new findings with the extinct species described by him difficult (although this issue has being partially resolved; see Hansen 2012). In general, fossils assigned to Cuniculus have not been described in detail, leading to scarce information about their anatomy and morphology, which hampers the critical review of taxonomic classification. Consequently, we are in need of a careful taxonomic revision of the group that will allow us to investigate basic aspects of their paleobiology and geographic distribution. Moreover, with the exception of the material reported by Vasconcelos et al. (2015), previous records did not come from excavations with stratigraphic control, and, consequently, our understanding regarding the chronological distribution of extinct cuniculids is also poor.

Within this context, we aim to clarify different aspects related to the extinct cuniculids: (i) to describe new cuniculid findings from Cuvieri Cave (Lagoa Santa, Minas Gerais, southeast Brazil; Fig. 1) and Toca de Cima dos Pilão Cave (Serra da Capivara, Piauí, northeast Brazil; Fig. 1); (ii) to provide taxonomic remarks on the extinct Cuniculus species; (iii) to provide a paleobiogeographic discussion of the extinct Cuniculus species; and (iv) to contextualize the taphonomy and chronology of Cuniculus fossils recovered from Cuvieri Cave. Our study suggests that there is only one extinct species, which should be C. rugiceps, and the distribution of this taxon, previously restricted to Lagoa Santa, southeast Brazil, should be extended up to northeastern Brazil. The taphonomic analyses suggest that the skeletons of C. rugiceps from Cuvieri Cave were accumulated by pitfall entrapment and represent at least seven individuals. Furthermore, the stratigraphically controlled excavations associated with U-Th chronology allow the establishment of a Late Pleistocene minimum age for C. rugiceps specimens from Cuvieri Cave.

Institutional abbreviations.—AMNH, American Museum of Natural History, New York, USA; FMNH, Field Museum of Natural History, Chicago, USA; FUMDHAM, Fundação Museu do Homem Americano, São Raimundo Nonato, Piauí, Brazil; FZBRS, Fundação Zoobotânica do Rio Grande do Sul, Porto Alegre, Rio Grande do Sul, Brazil; LEEH, Laboratório de Estudos Evolutivos Humanos da Universidade de São Paulo, São Paulo, Brazil; MCN.M, Coleção de Mastozoologia do Museu de Ciências Naturais da Fundação Zoobotânica do Rio Grande do Sul, Porto Alegre, Rio Grande do Sul, Brazil; MCN.P, Museu de Ciências Naturais da Universidade Federal do Paraná, Curitiba, Paraná, Brazil; MNRJ, Museu Nacional do Rio de Janeiro, Rio de Janeiro, Brazil; MZUSP, Museu de Zoologia da Universidade de São Paulo, São Paulo, Brazil; UF, Florida Museum of Natural History, Gainesville, USA; USNM, Smithsonian National Museum of Natural History, Washington, USA; ZMK, Zoological Museum, Natural History Museum of Copenhagen, Copenhagen University, Denmark.

Other abbreviations.—MNI, minimum number of individuals; NISP, number of identifiable skeletal parts.

Material and methods

Studied sites.—Lagoa Santa karst: The fossils recovered from Cuvieri Cave, Minas Gerais, Brazil (coordinates: UTM 7.846.105N and 603.756E; fuse 23K; Córrego Alegre datum; Fig. 1) are housed in the LEEH. This small cave has a main horizontal passage with three pits. The remains discussed here were excavated between 2002 and 2008 from Locus 3, which is characterized by a vertical shaft with an abrupt pit about 8 m deep (Fig. 1B). The excavation techniques adopted systematically recorded the spatial distribution of the remains and the stratigraphic sequence of the deposit (see Hubbe et al. 2011 for details).

Serra da Capivara karst: The fossils recovered from Toca de Cima dos Pilão (coordinates: UTM 9.019.281N and 768.738E; fuse 23L; South American datum; Fig. 1) are housed in FUMDHAM. This cave consists of two open chambers partially separated by detached blocks (Guérin et al. 1996; Fig. 1C). The fossils reported here were excavated from Salão Terezinha between 1986 and 1990 (Guérin et al. 1996).

Descriptive and comparative procedures for new fossil material.—To describe and compare the cuniculid material excavated from Cuvieri Cave and Toca de Cima dos Pilão we performed both qualitative and quantitative analyses.

In the qualitative analysis, we described the morphology of the skull portions, dentaries (anatomic terms following Woods and Howland 1979 and Pérez 2010) and lower cheek teeth from Cuvieri Cave, and isolated molars from Toca de Cima dos Pilão. Furthermore, we compared the fossils from both caves with the extinct cuniculid fossils excavated by Lund, C. paca, C. taczanowskii (for details see SOM, Supplementary Online Material available at  http://app.pan.pl/SOM/app61-Mayer_etal_SOM.pdf) and “Cuniculus hernandezi”. Note that although “Cuniculus hernandezi” (Castro, López, and Becerra, 2010) is a nomen nudum (Ramírez-Chaves and Solari 2014) we decided to incorporate “C. hernandezi” in our analyses for morphological comparisons and biogeographic purposes.

Fig. 1.

Map showing the location of the Brazilian studied areas (A), and maps of Cuvieri Cave (Lagoa Santa, Minas Gerais) (B), and Toca de Cima dos Pilão Cave (Serra da Capivara, Piauí) (C). In Cuvieri Cave, fossils were excavated from Locus 3, and in Toca de Cima dos Pilão, fossils were excavated from Salão Terezinha. Map in B courtesy of Laboratório de Estudos Evolutivos Humanos and Grupo Bambuí de Pesquisas Espeleológicas, map in C adapted from Acervo da Fundação Museu do Homem Americano.

All analyses were based on comparisons with material deposited in museum collections, with the exception of the comparisons with C. taczanowskii that were based on Ríos-Uzeda et al. (2004: fig. 2) and “C. hernandezi” that were based on Castro et al. (2010: fig. 6). Among the material collected by Lund (from 1837 to 1845), we selected the most complete specimens that maintain the original assignation to “C. paca forma laticeps” and “C. paca forma major” by Winge (1887; for details see Appendix 1 and SOM 2). The morphological descriptions prioritize comparisons with C. paca because it has the widest geographic distribution within Cuniculidae and is represented in many collections, which makes C. paca a good extant reference for comparisons with future cuniculid fossil discoveries. Comments regarding C. taczanowskii and “C. hernandezi” are presented only when C. taczanowskii and/or “C. hernandezi” present conspicuous morphological differences to C. paca.

In the quantitative analyses, we used digital callipers (accurate to 0.01 mm) to measure six linear distances on dentaries (following Driesch 1976; SOM 1: fig. 1A), the mesio-distal length and linguo-labial width of teeth, and six linear distances on skull bones (SOM 1: fig. 1B). Regarding the dentary analysis, measurements were taken from specimens from Cuvieri Cave, Lund's collection and the extant C. paca and C. taczanowskii (for details see SOM 2: table 1). A principal component analysis was performed to compare these sets of materials using the pooled-within species covariance matrix based on the six dentary linear distances. For the extant species (C. paca and C. taczanowskii) and the extinct C. paca laticeps, specimens with missing data were excluded from the analysis. We were unable to include C. paca major in our analysis, because three linear distances were missing from all specimens. For the material from Cuvieri Cave we averaged the information available for all specimens and used these averages in our analysis since no single specimen allowed the measurement of all distances. Regarding the tooth analysis, we measured the lower cheek teeth from Cuvieri Cave and isolated molars from Toca de Cima dos Pilão and then compared the measurements between both sites. Concerning the skull measurements, we measured skulls from fossils found in Cuvieri Cave with the goal of preliminarily characterizing this material, and to serve as a metric reference for future attempts to recognize extinct cuniculids.

Lastly, we need to make a methodological clarification. The fossil material recovered from Cuvieri Cave excavations, either plotted on site or retrieved later from sieving, received an individual number (for details regarding the excavation procedure see Hubbe et al. 2011). For the sake of brevity, wherever such fossil fragments have been refitted together, the new repaired specimens are simply referred to as Fossil 1, Fossil 2, hellip; etc., throughout the text, and the original specimen numbers are presented within parenthesis when they first appear in the text. The same terminology is adopted when a single fossil is represented by a nearly complete bone element.

Taxonomic and paleobiogeographic procedures.—Our taxonomic and paleobiogeographic considerations were based on: (i) Lund's (1837, 1839a, b, 1840, 1841a, c) works on extinct cuniculids; and (ii) anatomical data from our descriptions and comparisons of findings from Cuvieri Cave, and Toca de Cima dos Pilão with Lund's fossil collection. The taxonomic remarks take into consideration the principles of the International Commission of Zoological Nomenclature (ICZN). The paleobiogeographic notes are based on records from the literature and the identification of material from northern (Acre) and southern (Paraná) regions of Brazil (SOM 1: fig. 2).

Taphonomic and chronologic contextualization of material from Cuvieri Cave.—To identify the bone accumulation processes and characterize the biostratinomic history of the sample, we analysed the mode of fossil occurrence (articulated/scattered), the horizontal and vertical spatial distribution of specimens, the main bone surface modifications (i.e., fragmentation, tooth marks, and weathering stages) and stratigraphic data. Fragmentation was estimated for each specimen as percentage intervals relative to the respective whole skeletal element that it represents. Specimens comprising at least 90% of the respective skeletal element were noted as almost complete; specimens encompassing less than 50% were noted as fragments; and specimens of intermediary condition were noted as incomplete. Observations of tooth marks produced by carnivores and rodents followed Lyman (1994) and weathering stages followed Behrensmeyer (1978).

To characterize the materials from Cuvieri Cave quantitatively regarding individual representation, we adopted the Number of Identifiable Skeletal Parts (NISP) and the Minimum Number of Individuals (MNI). NISP values were obtained from the total number of specimens identified taxonomically and MNI values were calculated from the most abundant skeletal element identified taxonomically, taking into consideration the symmetry of the mammalian body (Lyman 1994). To avoid ambiguity in the quantitative units used in the taphonomic analysis, the terms specimen and element refer to “a bone or tooth, or fragment thereof” and “a single complete bone or tooth in the skeleton of an animal”, respectively (Shotwell 1955, 1958; see Lyman 1994 for discussion). Since extinct cuniculids were described mainly based on cranial characters, we are in need of a full description of the postcranial skeleton of extinct cuniculids. In spite of this, we preliminarily assigned taxonomically the postcranial elements from Cuvieri Cave to extinct cuniculids based on their morphological similarity to C. paca and their larger size compared to C. paca before calculating MNI and NISP. The description of the postcranial skeleton of C. rugiceps will be presented in a forthcoming publication.

To provide a chronological contextualization for the Cuvieri Cave remains we considered the fossils' stratigraphic positions relative to the chronological information available.

Taxonomic remarks

In this section we will review the description of all erected extinct cuniculid species (Cuniculus rugiceps Lund, 1837, Cuniculus laticeps Lund, 1839, and Cuniculus major Lund, 1839), and make the point that C. laticeps and C. major were poorly characterized, while C. rugiceps (Lund, 1837) should be the only extinct species of Cuniculus. Moreover, we consider the synonymy between C. rugiceps and C. laticeps proposed by Lund (1839a) inadequate, because, as shown below, we consider that C. laticeps is not a well-established species for several reasons. First, two out of four morphological characters described by Lund for C. laticeps (i.e., well-developed infraorbital foramen and narrow dorsal root of the zygomatic process of the maxilla; Lund 1840: 54–55, pl. 20: 1, 1841c: 270–271) can be observed in recent specimens of C. paca (e.g., MCN.M 1052 and specimens figured by Mendes-Oliveira et al. 2012: fig. 4A, B) and C. taczanowskii (figured by Ríos-Uzeda et al. 2004: fig. 3). There are also examples of variation in these structures during the ontogeny of C. paca (Mendes-Oliveira et al. 2012: fig. 4A–D). The other two characters, related to the zygomatic arch (zygomatic arch laterally projected and jugal posteriorly elongated), are not clear on Lund's figured material of C. laticeps (Lund 1840: 54–55, pl. 20: 1, 1841c: 270–271) and in the material assigned to C. paca laticeps housed in ZMK (Appendix 1; SOM 2). Second, there is not a well-marked distinction between C. laticeps and the other extinct and extant species regarding size. The plate that illustrated both C. laticeps and C. paca skulls (Lund 1840: 54–55, pl. 20: 1, 2, 1841c: 270–271) shows they have very similar dimensions. Note, however, that our quantitative analysis of the dentaries shows that C. laticeps may be slightly larger than extant species, but without a more representative sample size, few solid conclusions can be drawn. Third, it seems that specimens assigned to C. laticeps have different characteristics. For example, there are specimens housed in ZMK assigned to C. paca laticeps that present morphological and size similar to C. paca, while others are more robust, resembling more the first taxon described by Lund (1837)—C. rugiceps

On the other hand, we consider there is solid evidence to support the validity of C. rugiceps. Regarding the morphology of the material originally assigned to C. rugiceps prior to its synonymy with C. laticeps, based on the original drawings in Lund's (1837) works we can specify that C. rugiceps had a more posteriorly elongated jugal than C. paca (relative to the position of the squamosal/jugal contact in lateral view; Fig. 2). Note that although this character was described by Lund (1840) only in C. laticeps after to the synonymy, a more posteriorly elongated jugal is not observable in the respective figured specimen of C. laticeps in Lund's (1840: 54–55, 1841c: 270–271) work, as mentioned above. This contradiction reinforces our position about the mistaken character of the synonymy of C. rugiceps and C. laticeps.

In relation to size, the jugal fragment of the zygomatic arch assigned to C. rugiceps before its synonymy (Fig. 2) is clearly larger than the corresponding portion of the C. laticeps' skull figured in Lund's (Lund 1837: 331–332, pl. 3: 1, 1840: 54, pl 20: 1, 1841c: 270) work. Although the C. rugiceps and C. laticeps specimens figured were represented at different scales, both skulls of C. laticeps and the jugal of C. rugiceps were illustrated along with comparative specimens of C. paca (Lund 1837: pl. 3, 1840: pl. 20: 1–4, 1841c), allowing us to indirectly compare the sizes of C. rugiceps and C. laticeps. The differences in size between C. rugiceps and C. laticeps are most probably not associated with ontogenetic variation. This is because Lund specified that the jugal of C. paca compared to C. rugiceps (Lund 1837: 331–332, pl. 3) and the skull of C. paca compared to C. laticeps (Lund 1840: 55, pl 20: 2, 1841c: 271) represent senile individuals, and the differences observed in the size of the C. rugiceps and C. laticeps specimens exceed the measured sexual variations observed in extant C. paca (see Richard-Hansen et al. 1999).

Based on this evidence, we consider that the first extinct cuniculid species erected by Lund, C. rugiceps, originally described as presenting a larger size and stronger rugosities than C. paca (Lund 1837: 331–332), should be revalidated and maintained as distinct from C. laticeps. In turn, we suggest that C. laticeps or C. paca laticeps (Coelogenys laticeps, Lund 1839b: 228; Coelogenys laticeps, Lund 1839a: 42 [Lund 1841a: 102]; “Cuniculus paca forma laticeps”, Winge 1887: 64–66; Cuniculus paca laticeps, Paula Couto 1950: 43, footnote 4; Coelogenys paca laticeps, Paula Couto 1950: 169, footnote 146; Agouti laticeps, Cartelle 1999: 32) is a species inquirenda. We consider the possibility that this taxonomic unit could be based on a mix of materials of C. rugiceps and C. paca. A wide-ranging review of the material assigned to C. laticeps is required to clarify this issue.

Finally, the last extinct cuniculid species erected by Lund, C. major, was designated solely based on a statement mentioning the larger size of this species when compared to C. rugiceps and C. laticeps (Lund 1839b: 228–229). No illustration of this taxon was provided. However, the jugal that comprises the posterior portion of the zygomatic arch assigned to C. major (i.e., ZMK 1/1845: 2200) presents a size and morphology equivalent to the jugal of C. rugiceps figured prior to its synonymy with C. laticeps (Fig. 2). In this sense, in our opinion, there are no characters to validly maintain more than one extinct cuniculid species, and following the Principle of Priority (ICZN 1999: article 23), C. major is considered here as a synonym of C. rugiceps (Fig. 2).

Furthermore, we would like to clarify that based on the discussion above and our analyses, we assigned the material excavated from Cuvieri Cave and Toca de Cima dos Pilão to the large extinct species C. rugiceps. In the following sections we provide an in-depth description of the new material as well as biogeographic, taphonomic and chronologic remarks regarding the remains found in Cuvieri Cave.

Systematic palaeontology

Fig. 2.

Fossils from the Lagoa Santa, Minas Gerais, Brazil, attributed to extinct pacas (ages unknown). A, B. Specimens assigned to “Cuniculus paca forma major” Lund, 1839. A. ZMK 1/1845:2029, mandible in occlusal (A₁) and lateral (A₂) views. B. ZMK 1/1845:13450, maxilla with P4–M3 series in occlusal view (B₁); fragment of jugal in lateral view (B₂). C. Atlas (ZMK 1/1845:2989) assigned to “Cuniculus paca forma laticeps” by Winge (1887), which Lund (1837) described as Cuniculus rugiceps. Arrow is pointing to the rugosities on the dorsal surface. D. Syntypes of Cuniculus rugiceps from Lagoa Santa with the original label of Herluf Winge (D₁) and Lund's (1837) illustration (D₂) comparing the right jugal of C. rugiceps (“Fig. 3”) with C. paca (“Fig. 4”) (Lund 1837: pl. 3, 1840 [1841c]: pl. 20).

Fig. 3.

Fossils of the extinct paca Cuniculus rugiceps (Lund, 1837) from Cuvieri Cave (minimum age of ∼30 ky). A. Fossil 7 (CVL3 P13160), dorsal portion of the skull in dorsal (A₁), ventral (A₂), right lateral (A₃), left lateral (A₄), and posterior (A₅) views. B. Fossil 8 (CVL3 P13168), right zygomatic arch in dorsal (B1), lateral (B₂), and anterior (B₃) views. C. Fossil 4 (CVL3 P13342), right dentary in lateral (C₁) and medial (C₂) views. D. Fossil 2 (CVL3 P11221), anteroventral face of the mentonian region of a right dentary showing rugosities. White arrows, discrete notch on the dorsal edges of the orbits; black arrows, large and deep notch on the bottom of the infraorbital foramen.

Fig. 4.

Fossils of the extinct paca Cuniculus rugiceps (Lund, 1837) from Cuvieri Cave (minimum age of ∼30 ky) (A, B) and modern specimen of Cuniculus paca Linnaeus, 1766 from Rio Grande do Sul (C). A. Fossil 7 (CVL3 P13160), cranial roof in dorsal (A₁) and lateral (A₃) views, detail (A₂). B. Fossil 8 (CVL3 P13168), right zygomatic arch in dorsal (B₁) and lateral (B₂) views, detail (B₃). C. MCN.D 207, skull in dorsal (C₁) and lateral (C₃) views, details (C₂, C₄).

Table 1.

Dentary measurements of the extant species of Cuniculus and the extinct Cuniculus rugiceps from Cuvieri Cave. Abbreviations: N, number of specimens analysed; Min/Max, minimum/maximum value measured; 1, maximum length; 2, tooth row length; 3, length between the distal margin of alveoli and base of incisive; 4, diastema length; 5, distance between mandibular condyle and ventral margin of angular process; 5a, vertical branch height (for details regarding the distances measured see SOM 1: fig. 1).

Fig. 5.

Lower cheek teeth of the extinct paca Cuniculus rugiceps (Lund, 1837) from Cuvieri Cave (minimum age of ∼30 ky). A₁, B–F, ontogenetic sequence of cheek teeth. A. Fossil 6 (CVL3 P13344) (image inverted). A₁, dp4–m2, the m3 is lacking due to fragmentation. A₂, dentary in medial view showing p4 replacing the dp4 (arrow). B. Fossil 4 (CVL3 P13342) (image inverted), showing dp4–m3. C. Fossil 1 (CVL3 4052), showing the p4 erupting and the m3 completely functional. D. Fossil 5 (CVL3 P13145), showing p4–m3. E. Fossil 3 (CVL3 P13149), showing p4–m3. F. Fossil 2 (CVL3 P11221), showing p4–m3. G, H. Isolated molars from Toca de Cima dos Pilão deposits; in occlusal (G₁, H₁) and lingual (G₂, H₂) views. G. FUMDHAM 188-19036, possible right m3. H. FUMDHAM 188-19521, possible right m2.

Discussion

Quantitative analysis.—To quantitatively characterize the extinct and extant species we performed a principal component analysis comparing dentaries from extinct Cuniculus rugiceps and “Cuniculus laticeps”, and extant Cuniculus paca and Cuniculus taczanowskii. Our analysis shows that the averaged measurements of C. rugiceps (Lund, 1837) dentaries are clearly larger than the other species, thus supporting our interpretation that C. rugiceps is a valid species. On the other hand, the two “C. laticeps” (= C. paca laticeps) dentaries match in size with the larger C. paca or are slightly larger than the extant species. Based on the limited number of dentaries of “C. laticeps” it is hard to sustain any conspicuous difference between this taxonomic unit and extant species regarding the second axis of major variation. One of the “C. laticeps” specimens is outside the distribution of the extant species, so there might be a tendency towards shape differences beyond size (Fig. 6). As mentioned earlier, further studies are required to characterize the size of “C. laticeps”.

Fig. 6.

Principal component analysis performed using the pooled-within species covariance matrix based on six dentary linear distances. Graydotted lines and associated numbers show the variables that have high correlations (r > 0.5) with each of the PCs. The gray numbers represent the measurements described in SOM 1.

To quantitatively compare teeth from Cuvieri Cave and Toca de Cima do Pilão we measured their mesio-distal length and linguo-labial width. The comparison shows similar values for teeth from both localities (SOM 2: table 2). Based on the similar values (associated with the compatible morphology; see description and comparison), we assigned the material excavated at Toca de Cima dos Pilão to the extinct species C. rugiceps.

To provide a metric reference for further analysis and future attempts to recognize C. rugiceps we measured six distances on the skull. The measurements of the skull remains from Cuvieri Cave are characterized preliminarily by two widths on the frontal bone and four linear distances between structures on the zygomatic arch (SOM 2: table 3).

Biogeographic remarks.—The current distribution of Cuniculus encompasses the tropical and subtropical portions of South America and part of Central America. Cuniculus paca has the broader range, covering an area from Mexico to southern Brazil and northeast Uruguay (Pérez 1992; Queirolo et al. 2008 and references therein; Fig. 7). Cuniculus taczanowskii is restricted to the Andean region, covering Peru, Ecuador, Colombia, Bolivia, and northwest Venezuela (Ríos-Uzeda et al. 2004; Tirira et al. 2008; Fig. 7). “Cuniculus hernandezi” is to date restricted to its type localities in Colombia (Castro et al. 2010; Fig. 7). Fossils of extant Cuniculus were found only in presumed Quaternary deposits (Auler et al. 2006), and, as far as we know, are only of C. paca. All C. paca fossils were reported within the tropical region of Brazil, including the states of Minas Gerais (Lund 1837, 1839a; Souza-Cunha and Guimarães 1986; Hubbe et al. 2011); Bahia (Lessa et al. 2008); Mato Grosso do Sul (Salles et al. 2006); Goiás (Salles et al. 1999); São Paulo (Chahud 2005); Paraná (Kerber et al. 2013; see SOM 1: fig. 2); Acre (no published data, see SOM 1: fig. 2) and Pará (Toledo et al. 1999; Fig. 7). Guérin et al. (1996) cited the presence of Agouti paca (= Cuniculus paca) in Serra da Capivara, state of Piauí, northeastern Brazil, but did not refer to any material. However, at least some of these reports could be referable to C. rugiceps, and a detailed revision is needed. For example, according to Lessa et al. (2008), material of Cuniculus collected in Bahia has larger dimensions than the Recent exemplars of C. paca.

To date, reports of extinct cuniculids, which according to our interpretation belong solely to C. rugiceps (Lund, 1837), were restricted to the Lagoa Santa caves (including Cuvieri Cave) (Lund 1837, 1839; Winge 1887; Rusconi 1933; Vasconcellos et al. 2015; this work). Our morphological and quantitative analyses suggest the at least two specimens from Toca de Cima dos Pilão cave belong to C. rugiceps (Fig. 5G₁–H₂; SOM 2: table 2). Thus, the newly diagnosed C. rugiceps remains from Serra da Capivara represent the northernmost limit of C. rugiceps' range, expanding the geographic distribution of this taxon almost 1500 km away from Lagoa Santa. The record from Toca de Cima dos Pilão cave also contributes to knowledge of the ancient fauna from the Serra da Capivara region, since this rodent was not recognized in the sample analysed in a recent review of the late Quaternary caviomorph fauna (Kerber et al. 2016). The available records of C. rugiceps are restricted to the tropical region of Brazil.

Fig. 7.

Distribution of the extinct Cuniculus rugiceps and extant species of Cuniculus and its fossil record. References in the text.

Taphonomic contextualization

Bone accumulation and main modifications.—Based on the analysis of the bone assembly found in Locus 2, Hubbe et al. (2011) argued that the main process of bone accumulation inside Cuvieri Cave was most likely pitfall entrapment. Our findings for Cuniculus rugiceps (Lund, 1837) agree with this interpretation. First, we found some C. rugiceps articulated skeleton segments during excavations. In Locus 3A, cranial fragments are associated with a dentary, long bones and vertebrae, while in Locus 3B there was a clustered set of a sacrum, lumbar vertebrae and a femur (Fig. 8A, B). Second, most of the C. rugiceps remains were excavated from the subdivision A of Locus 3 (92% of the total amount of specimens, ∼323 bones/m³; Fig. 8) and the remaining specimens (8%, ∼42 bones/m³) came from the southern region of Locus 3B, with very few fossils in the northern region. This data reinforces the interpretation of pitfall entrapment because C. rugiceps fossils occur more densely in the area located immediately below the vertical shaft entrance in direct communication with the main passage (∼8 m above; Figs. 1, 8A: black arrow). One possible explanation, although not the only one, is that this area is the most likely one where a carcass would be deposited if an animal fell and died after falling into Locus 3. One possibility that we can rule out is bone transport leading to specimen concentration in Locus 3A, because the deposit stratigraphy and sediment characteristics do not indicate the occurrence of water flows capable of transporting bones.

The integrity of the fossils was similar for the three fragmentation categories. Almost complete, incomplete and fragments represented 228 (31%), 282 (38%), and 230 (31%) specimens, respectively. The majority of the fragmentation and movement of the remains is probably due to falling and trampling by other individual C. rugiceps and/or other taxa represented in the deposit (Hubbe 2008; Mayer 2011). The falling of detached limestone blocks, which were recorded embedded in sedimentary facies during excavation, possibly acted as another relevant factor (Fig. 8A).

Only rodent gnawing marks were detected on C. rugiceps remains (Fig. 8C). These marks were detected on less than 1% of the remains and probably occurred relatively soon after death inside Locus 3, as indicated by observations of modern local fauna that reveals fresh bones are recurrently gnawed by small rodents (Lund 1836: 243–244). The low percentage of remains gnawed indicates that the processes of bone accumulation by predators and scavengers (i.e., rodents) did not play a major role in the deposition of C. rugiceps skeletons in Locus 3.

Fig. 8.

Taphonomic aspects of the extinct paca Cuniculus rugiceps (Lund, 1837) from Cuvieri Cave (minimum age of ∼30 ky), Lagoa Santa. A. Strati graphic distribution of plotted specimens (specimens recovered by sieving are not represented) relative to the flowstone dated. Clusters of dots in the lower layers of Locus 3A and 3B represent bones preserved in anatomical association. Black arrow indicates the position of the 8 m deep pitfall. B. Excavation record of specimens in anatomical association (B₁) with schematic indication of bone elements (B₂). C. Left femur in posterior view (CVL3P13911), showing bone modifications caused by rodent gnawing (arrows; see the remarkable circular opening of diaphysis). D. Right innominate in medial view (CVL3B2165a), showing initial cracking caused by weathering (arrows).

Bone modifications associated with weathering were identified on 15% of C. rugiceps specimens, which represent modifications compatible only with stage 1 weathering (Behrensmeyer 1978; Fig. 8D). Despite the low intensity of these weathering marks, it indicates that at least part of the bones remained on the surface for some time before burial. Moreover, since caves are underground environments, the weathering probably occurred at a slower rate than would occur on surface (as was the experiment conducted by Behrensmeyer 1978), and potentially even bones that do not present weathering modifications could have been exposed on the surface of Locus 3 for a relatively long time. This suggests that the falling and entrapment of other animals and the falling of limestone blocks allied with rodent scavenging probably played the major roles in modifying the deposits. Currently, a broader taphonomic study based on all taxa recovered from Locus 3 is underway and will add new information to this issue.

Quantification of fossils.—The NISP calculated for C. rugiceps is 740 and several skull and postcranial elements are represented. The left calcaneus is the most abundant element in the sample and served as the basis to calculate the MNI, resulting in at least seven individuals. There is some discussion about the adequacy of these indexes in relation to different bone accumulation processes (Badgley 1986; Castro and Langer 2011). The NISP method is appropriate to count individuals from assemblages whose remains suffered wide-ranging transportation and are deposited as scattered pieces, displaying a low probability of association. In contrast, the MNI method assumes a high probability of association and is suitable to infer the number of individuals contained in assemblages formed in natural traps (Badgley 1986). As such, we argue that MNI is a better estimate for the number of C. rugiceps individuals in Locus 3, since falling and entrapment is the most likely cause of the bone accumulation.

Some evidence suggests that the seven individuals of C. rugiceps did not fall into Locus 3 in a single event. First, the specimens exposed during excavations were vertically distributed in at least two distinct sedimentary facies encompassing ∼1.10 m of the ∼1.80 m total excavated depth in Locus 3 (Fig. 8A). Despite some burial perturbations of a lesser range being identified through the mechanical refitting of medium-sized mammal bones, the breadth of the movements are mainly near the horizontal plane or subparallel to inferred facies boundaries (Mayer 2011), reinforcing trampling as an important taphonomic process. Since there was no mixing of the sedimentary facies, the idea of independent entrapment of the individuals appears plausible. Second, considering the behavioural aspects of C. paca, the solitary habits of this cursorial rodent support the idea of individual independent entrapment in Locus 3. Moreover, the presence of C. paca inside caves is commonly recorded (Lund 1838, 1841b; Nowak 1999; Zeppelini Filho et al. 2003) and this behaviour could be shared by the extinct C. rugiceps, which would facilitate eventual falling of individuals over relatively long periods of time.

Chronology.—The chronological information available so far for C. rugiceps is insufficient to allow any solid interpretation about the temporal distribution of the species. The flowstone formed above the deposits containing the fossils of C. rugiceps from Cuvieri Cave (Fig. 8A) suggests a Late Pleistocene minimum age of ∼30 ky for our findings. However, since bone remains from Lagoa Santa caves range from older than 300 ky to as young as the present, more dating will be required to allow a better understanding of the chronological distribution of C. rugiceps in the fossil record (Auler et al. 2006; Hubbe et al. 2011).

Conclusions

We propose that the only valid name for the extinct Cuniculidae is Cuniculus rugiceps. The new fossil remains of C. rugiceps presented by us from Cuvieri Cave, Minas Gerais state, Brazil suggest that at least seven individuals of C. rugiceps accumulated asynchronously by pitfall entrapment inside Locus 3, and were buried after some undetermined time of exposure on the surface. The stratigraphically controlled excavations and dating of the Cuvieri Cave deposits allowed the establishment of a Late Pleistocene minimum age (∼30 ky) for these specimens. Furthermore, remains of C. rugiceps from Toca de Cima dos Pilão cave, Piauí state allowed us to tentatively assign a tropical geographic distribution for this species.