Systematic revision of Gyriabrus (Rodentia, Caviomorpha), a large dinomyid from the Neogene of South America

ABSTRACT Dinomyidae is a family of rodents including one extant species, Dinomys branickii, but with a high past diversity. Gyriabrus is a large dinomyid with euhypsodont cheek teeth and an occlusal pattern that changes throughout the ontogeny of the animal. Eight species of Gyriabrus have been described in the late Miocene to Pliocene of Argentina, Uruguay, Colombia and Brazil. The taxonomic status of Gyriabrus has been questioned, being considered as juvenile forms of other euhypsodont dinomyids. According to this study, Gyriabrus has a unique dental anatomy among euhypsodont dinomyids and is a valid genus. A study of the cheek teeth permitted to propose an ontogenetic series. All the species described from the late Miocene of Argentina (Ituzaingó Formation) and Uruguay (Camacho Formation) belong to Gyriabrus holmbergi (including G. glutinatus, G. rebagliattii, G. indivisus and G. teisseirei), representing different wear stages. ?Gyriabrus quadratus, G. latidens (late Pliocene? of Argentina) and ?G. royoi (late Miocene-Pliocene of Colombia), all based on isolated teeth, are considered as valid species, pending future findings that allow a deeper comparison. This study contributes to the knowledge of the systematics, taxonomy and anatomy of extinct Dinomyidae, which are essential for future studies of the evolutionary history of the group.


Introduction
Dinomyidae is a family of caviomorph rodents that is closely related to living Chinchillidae (chinchillas and vizcachas) and extinct Neoepiblemidae and Cephalomyidae, conforming the Chinchilloidea (e.g. Candela 2018, 2019;Rasia et al. 2021). Dinomyids are represented nowadays by the pacarana (Dinomys branickii Peters 1873), which occupies Andean forests of northwestern South America, including Venezuela, Colombia, Ecuador, Peru, Brazil and Bolivia (e.g. White and Alberico 1992;Patton 2015). On the contrary, the past diversity of Dinomyidae is notably higher, with more than 50 described species (e.g. Mones 1986) and a fossil record dating back to the early Miocene (e.g. Kramarz 2006;Rasia et al. 2021).
In addition to these species, Paula Couto (1983) reported from the Neogene of Brazil (Figure 1) an isolated tooth that referred to cf. Gyriabrus sp. (see also Kerber et al. 2017). In later works (e.g. Cozzuol 2006;Ribeiro et al. 2013), Gyriabrus sp. is considered part of the faunal assemblage of the Solimões Formation (late Miocene of Brazil). Cf. Gyriabrus sp. was also recorded in the late Miocene Chiquimil A (=El Jarillal Member, Chiquimil Formation) in Catamarca Province, Argentina (e.g. Herbst et al. 2000;Brandoni 2013), but other works do not mention the presence of Gyriabrus in this unit (e.g. Reguero and Candela 2011;Esteban et al. 2014Esteban et al. , 2017. Pascual (1966) mention the presence of Gyriabrus in the Arroyo Chasicó Formation (late Miocene) in Buenos Aires Province, Argentina, but no other work consider this record (e.g. Bondesio et al. 1980;Brandoni 2013).
Despite being euhypsodont, the occlusal pattern of Gyriabrus changes substantially throughout ontogeny, even in adulthood; unlike 'tetrastylines' and 'eumegamyines', with a more stable occlusal tooth pattern (e.g. Rinderknecht et al. 2018). This feature leads to some authors to consider Gyriabrus as a basal form to other euhypsodont taxa (e.g. Pascual 1966;Mones 1981), while other authors proposed that the specimens referred to Gyriabrus may represent in fact juvenile stages of other euhypsodont genera like Telicomys or Isostylomys (Rinderknecht et al. 2018).
To estimate the body mass of an extinct animal is important given that most morphological, physiological, behavioural and ecological features are closely related to and scale with size (e.g. Millen 2008; Millien and Bovy 2010). Dinomyidae are among the largest living rodents (Rinderknecht and Blanco 2015), with its only extant species, Dinomys branickii, weighting 10-15 kg (White and Alberico 1992). In the past, especially during the late Miocene-Pleistocene, the family reached giant proportions (e.g. Kraglievich 1926;Rinderknecht and Blanco 2008;Nasif et al. 2013;Rinderknecht et al. 2018). Josephoartigasia monesi Rinderknecht and Blanco 2008, the largest extinct rodent ever reported, has an estimated body mass of 350 kg (Millien 2008) to one ton (Rinderknecht and Blanco 2008). In this work, the morphology, especially the dental anatomy, of specimens assigned to Gyriabrus is analysed and compared with other euhypsodont dinomyids, in order to assess the validity of the genus and the taxonomic status of the described species. Also, the body mass of Gyriabrus is estimated and its biogeographic significance is revised.
Craniomandibular nomenclature used in diagnoses and descriptions is taken from several works (e.g. Mones 1997;Nasif and Abdala 2015).
The body mass estimation was performed following Millien and Bovy (2010) and Croft (2000), using dental measurements (see Supplemental online material S1). It was calculated only for Gyriabrus holmbergi, given it includes better preserved and more complete specimens.
Additional information is provided in the Supplemental online material (Som). Emended diagnosis: Large sized dinomyid with euhypsodont cheek teeth, with an occlusal pattern that changes throughout ontogeny, even at adult age, unlike Dinomys branickii, Tetrastylus intermedius and Isostylomys laurillardi. Pentalophodont to tetralophodont cheek teeth, with laminar lophs/lophids, lacking crenulations. Enamel band wider in the leading edge than in the trailing edge. Lophs/lophids at least twice wider than the flexa/flexids, which are filled with cement, like other dinomyids.
Comments: The tooth recovered from the Acre region (Brazil; see Figure 1), illustrated and referred to cf. Gyriabrus by Paula Couto (1983), is a left p4 with four lophids connected lingually. It probably represent a specimen with advanced wear stage given that specimens with little wear are pentalophodont and with lingually free lophids (see below). As noted by Kerber et al. (2017) the tooth shows some differences with Gyriabrus holmbergi (see below) and also with G. latidens (see below), so it could represent another species of Gyriabrus. But being an isolated tooth, it is difficult to reach a specific status with confidence and I agree with the identification of Paula Couto (1983) and Kerber et al. (2017), maintaining it as cf. Gyriabrus until further findings that confirm its genus identity.
Emended diagnosis: Species of Gyriabrus differing from ? Gyriabrus quadratus by being euhypsodont and with the fourth loph (metaloph+posteroloph) of the tetralophodont upper molars of the same width than the anterior ones; larger than ? G. royoi and with more curved crown of the P4; and from G. latidens because the metalophulid II of the p4 become reduced at advanced wear stages.

Description
Skull: The palatal sulci are well developed (Figure 3a). The most anterior contact between the maxillae and the palatines reaches the limit between M1-M2 ( Figure 3a). The anterior palatine foramina are located at the anterior level of the M1. The posterior maxillary foramina (only the right one is preserved in MACN-A 5879) are located behind the M3 (Figure 3a). The choanae are located posterior to the M3, like in other dinomyids (e.g. Fernández de Álvarez 1958; Rinderknecht and Blanco 2008;Rinderknecht et al. 2018), except for Dinomys branickii in which they can reach anteriorly the midpoint of M3 (see Nasif and Abdala 2015). The ventral zygomatic root is at the level of the P4 above the cheek teeth series (Figure 3b) as in other dinomyids (e.g. Rinderknecht and Blanco 2008;Nasif and Abdala 2015;Rinderknecht et al. 2018).
The lower diastema is shorter than the lower cheek teeth series. The mental foramen is small but well developed, located at the midpoint of the height of the mandible (Figure 3c-d). The notch for the insertion of the tendon of the masseter medialis pars infraorbitalis muscle (masseteric notch in Figure 3) is large and located at the level of the m1, in the upper half of the height of the mandible (Figures 3d, f, h). There is a small foramen in the retromolar fossa, behind the m3, that probably corresponds to the retromolar foramen ( Figure 3c). The retromolar foramen is the exit of the retromolar canal, a bifurcation of the mandibular canal (e.g. Alves and Deana 2015), and is here observed in other caviomorph rodents like Dinomys branickii, Tetrastylus intermedius and Hydrochoerus hydrochaeris (Som S2).
Teeth anatomy and ontogeny: The cheek teeth are euhypsodont (see Som S3), with laminar lophps/lophids, with the enamel band wider at the leading edge than at the trailing edge. Lophs are straight or curved, and lophids are curved or sinuous. The flexa/ flexids have half or less the width of the lophs/lophids, and are filled with cement (Figures 3a, c, e, g), a common feature to other dinomyids like Ferigolomys pacarana (Kerber et al. 2018b), species of Tetrastylus (Nasif et al. 2013;Sostillo et al. 2022), Arazamys castiglionii (Rinderknecht et al. 2011), and Isostylomys laurillardi (Rinderknecht et al. 2018).
The M1 is tetralophodont, and the M2-M3 are pentalophodont in early wear stages (Figure 4a). Later, the M2 become tetralophodont, and the M3 remains pentalophodont (Figure  4b), at least in known specimens. In the M1 with little wear, the first two lophs are fused labially, and the last two lophs are fused lingually (Figure 4a). With advanced wear, the second and third lophs of the M1 become joined labially (Figure 4b).
It is not possible to know from known specimens if the M1 is pentalophodont in earlier wear stages. In the M3 with advanced wear the first two lophs become fused labially, as in the M1-M2, and the third loph become fused lingually to the fourth loph. The last two lophs of the M3 become fused labially ( Figure 4b) and is possible that merge completely with more wear.
The p4 is pentalophodont in early stages of wear ( Figures  4c-d). With advanced wear stages, the metalophulid II tend to disappear (Figure 4e), and the last two lophids join lingually (Figures 4d-e).
The m1-m3 are pentalophodont (Figure 4c), with a neolophid as the second lophid, between the metalophulid I and metalophulid II (see Nasif 2009). The metalophulid I, neolophid and metalophulid II are joined labially (Figures 4c-e). In the m1, the neolophid become fused with the metalophulid I, leading to a tetralophodont pattern (Figures 4d-e). With advanced wear the last three lophids become joined lingually in the m1-m2 (Figures 4d-e).
It is noteworthy that following the ontogenetic series here described, once two structures are fused in one ontogenetic stage, they remain fused in the following ones (see Figure 4), as is seen in other dinomyids like Dinomys branickii (Nasif and Abdala 2015).
The lower incisor (the only known) has a subtriangular cross section, with the anterior enamelled face almost flat and smooth. The enamel reach the mesial and distal portion of the tooth. The base of the lower incisor reach the area behind the m3 (Som S3).
Comments: The holotype of Gyriabrus holmbergi (='Megamys' holmbergi) was an isolated lower molar (e.g. Ameghino 1885; Kraglievich 1932), described as pentalophodont, with the first three lophids joined labially, and the last two completely free, with the first lophid very reduced. This morphology is identical to that of lower molars of MACN-Pv 3956 and MLP 15-252 ( Figures  4c-d). Nevertheless, the material labelled as the holotype of this species is an isolated right lower incisor (MACN-A 5826) with identical morphology to the preserved incisor of other material (MLP 15-252) referred to G. holmbergi. Moreover, the specimen MACN-Pv 3956 was identified by Kraglievich (1932) and is labelled as a topotype.
The p4 of MACN-Pv 3956 was considered by Kraglievich (1932) as a possible abnormal tooth. Later, Rinderknecht et al. (2018) interpreted the interrupted lophids of this tooth as similar to the little-worn teeth observed in the living Dinomys branickii (see Nasif and Abdala 2015), what reinforced the idea that Gyriabrus could be grouping young individuals of other taxa. The fact that this p4 is not at the same plane than the rest of the cheek teeth (Figure 3f) suggests that there was some degree of malocclusion, supporting the statement of Kraglievich (1932). The same pathological morphology is observed in extant caviomorph rodents (e.g. Lagostomus The holotype, and only known material, of 'Gyriabrus glutinatus' is a left m1 with four lophids (Figure 5a) and the same occlusal pattern of the m1 in MACN-Pv 3956 (Figure 4d), so the synonymy between these two species is here proposed. With more wear the third lophid (hypolophid) become fused lingually with the fourth lophid (posterolophid), as is seen near the base of the tooth ( Figure  5a), being similar to the m1 in MACN-Pv 4728 (Figure 4e). Kraglievich (1930) described the holotype of 'Gyriabrus teisseirei' as more robust than G. holmbergi. This is not striking given that according to the ontogenetic series here described, the holotype of 'G. teisseirei' is a specimen with more advanced wear than the one previously referred to G. holmbergi, being both adults with approximately the same size (Figures 4a-b).
Similarly, the holotype of 'G. rebagliattii' has, according to Kraglievich (1932), wider molars than G. holmbergi. This can be interpreted also as specimens of the same species with different wear stages, given that a variation of proportions is observed in specimens of Isostylomys laurillardi with different wear stages (see Rinderknecht et al. 2018).
The holotype of 'G. indivisus' is a poorly preserved upper molar, a left M1 or M2, with the same occlusal morphology than the M1 (Figure 4a) or M2 (Figure 4b) of G. holmbergi, having the first two lophs (anteroloph and protoloph) joined labially, and the third (mesoloph/mesolophule) and fourth (metaloph+posteroloph) lophs joined lingually. With more wear the tooth reach a similar pattern to the M1 of G. holmbergi with advanced wear (see Figure  4b), with the second (protoloph) and third (mesoloph/mesolophule) lophs joined labially. Hence, 'G. indivisus' is here considered as a synonym of G. holmbergi. Geographic location: Villa Ballester, Buenos Aires Province, Argentina.
Holotype: PVL 1041 (Col. Rusconi 1302), isolated left M1 or M2. Comments: The tooth described by Rusconi (1945) is an upper molar, considered an M3 by Rusconi, but it is more likely to be an M1 or M2. It is very similar to the M1 of Gyriabrus holmbergi with advanced wear (Figure 4b). It is tetralophodont, with the first three lophs joined labially, and the last two lophs joined lingually ( Figure  5b). With more wear the third and fourth loph become fused labially.
According to the description and figured material of Rusconi (1945), the molar is protohypsodont, given that it has a closed root, so it probably do not belong to the genus Gyriabrus, but it is difficult to assess this with one isolated tooth. It is also possible that the tooth represent a senile individual with very advanced wear.
So, while the occlusal morphology of ?Gyriabrus quadratus is similar to that of G. holmbergi, the fact that the holotype and only known material is protohypsodont, and not euphypsodont, prevent a synonymy between both species.

Comments:
The holotype and only known material is an isolated cheek tooth, recognised by Rusconi (1945) as a M3, but the morphology and curvature of the tooth suggest that it is a p4. The occlusal morphology resembles that of Gyriabrus holmbergi with moderate to advanced wear (see Figure 4d-e). It is euhypsodont and pentalophodont, the first three lophids are joined labially, and the first two lophids, and last three ones are joined lingually (Figure 5c). This occlusal pattern does not change with wear in the specimen.
Being an isolated tooth and not being identical to other p4 of Gyriabrus holmbergi (the only other species with known p4), the validity of G. latidens is maintained, but future findings would clarify its taxonomic status.

Comments:
The P4 referred to this species is slightly smaller than the other species of the genus. It is identical in occlusal morphology to little worn P4 of Gyriabrus holmbergi (Figure 4a). It is euhypsodont and pentalophodont, with the last three lophs joined lingually (Figure 5d). In labial view the crown is almost straight and getting narrow at the base, differing from other species of Gyriabrus. The upper incisor of ?G. royoi is very similar in cross section to inferior ones of G. holmbergi, but there are no known upper incisors of the latter species to compare with.
Given the differences of the crown morphology of the holotype with Gyriabrus holmbergi, and the fact that its geographic provenance is distant from the other recognised species (Fig. 1), the validity of ?Gyriabrus royoi is here maintained, pending future findings.
The other species of Gyriabrus, known by isolated teeth, would have had a body mass close to that of G. holmbergi given that tooth measurements are similar (see Som S1).

Common features of the upper and lower dentition
One of the main problems on systematics and taxonomy of euhypsodont Dinomyidae is that most of the genera and species are known only by upper or lower teeth, or the teeth referred to the same genera or species do not belong to the same individual. There are only a few notable cases of specimens with associated upper and lower dentition, like Isostylomys laurillardi (Rinderknecht et al. 2018), Tetrastylus intermedius (Rovereto 1914), and Tetrastylus laevigatus (Sostillo et al. 2022). In addition, there are only a few works that analyse ontogenetic variations of extinct dinomyids (e.g. Nasif 2009;Rinderknecht et al. 2018;Sostillo et al. 2022).
Most of the material referred to Gyriabrus are isolated teeth, but there are some specimens that consist of palate or mandibles with full cheek dentition ( Figure 3). Despite there are no associated upper and lower dentitions of Gyriabrus, there are some features of the cheek teeth that strongly suggest they belong to the same taxa (see Figure 4): • Laminar lophs/lophids with curved or sinuous shape, and much wider than the flexa/flexids. • The molars are pentalophodont, with a tendency to tetralophodonty in the M1/m1 and M2, by fusion of the last two lophs (metaloph and posteroloph) in the upper molars and first two lophids (metalophulid I and neolophid) in the lower ones. • The last three lophs in the upper molars (mesoloph/mesolophule, metaloph and posteroloph) and first three lophids in the lower molars (metalophulid I, neolophid and metalophulid II) are joined lingually and labially, respectively. • With advanced wear, the first two or three lophs in the upper molars (anteroloph, protoloph and mesoloph/mesolophule) and the last three lophids in the lower ones (metalophulid II, hypolophid and posterolophid) become fused labially and lingually respectively.

On the validity of Gyriabrus
The morphology of the upper and lower cheek teeth described above is unique among dinomyids, and it is not observed in any other euhypsodont taxa (e.g. Fernández de Álvarez 1947, 1958Rinderknecht and Blanco 2008;Rinderknecht et al. 2011Rinderknecht et al. , 2018Nasif et al. 2013 morphology with Gyriabrus holmbergi is Doellomys parcus (see Fernández de Álvarez 1947), known by its M1-M2. Tetrastylus is discarded as the adult form of Gyriabrus, given that both taxa are equivalent in size, but it has the same occlusal pattern of Telicomys. Thus, Gyriabrus represents a valid genus, with a well-known species (G. holmbergi) and three species (?Gyriabrus quadratus, G. latidens, ?G. royoi) known by isolated teeth, that cannot be studied or compared with more detail.

Caviomorph diversity of the 'Mesopotamiense' and biogeographic relevance
The caviomorph assemblage of the 'Mesopotamiense' (i.e. the 'Conglomerado Osífero' or lower levels of the Ituzaingó Formation; see Brandoni 2013) is one of the most diverse of South America, with 44 genera and more than 68 species (see Candela 2005). Recent studies (e.g. Candela and Noriega 2004;Vucetich et al. 2005;Nasif et al. 2013;Rasia and Candela 2018;Kerber et al. 2018a) drastically reduced the number of caviomorph taxa for this unit. This study synonymise the four species of Gyriabrus described for the 'Mesopotamiense', again reducing the number of species for the Ituzaingó Formation, but also recognise the validity of the genus Gyriabrus, which has been recently questioned (Rinderknecht et al. 2018).
In addition, the presence of Gyriabrus in northern South America (Colombia), Acre (Brazil) and in southern South America (Argentina and Uruguay) supports a biogeographic connection of these areas during the late Miocene (see also Cozzuol 2006;Kerber et al. 2017;Rasia and Candela 2018).

Conclusions
The unique dental morphology of Gyriabrus indicates that it is a valid genus. Specimens with complete dentition allowe to reconstruct the dental ontogeny of Gyriabrus holmbergi. Despite there are no associated upper and lower dentitions, several features strongly suggest they belong to the same taxa.
?Gyriabrus quadratus, G. latidens, and ?G. royoi, based on isolated teeth, are considered valid species, pending future findings that allow a deeper comparison.
The presence of Gyriabrus in northern South America (Colombia), the Acre region (Brazil) and southern South America (Argentina and Uruguay) suggest a paleobiogeographic connection between these areas. Also, the presence of Gyriabrus holmbergi in the late Miocene Ituzaingó and Camacho formations supports a close affinity between this two units.
Gyriabrus holmbergi had a body mass of around 24.3 kg (±8), and was larger than the living Dinomys branickii.
This study contributes to the knowledge of the systematics, anatomy and biogeography of extinct Dinomyidae, which are essential for future studies of the phylogeny and evolutionary history of the group.