(Fig. 1 and Fig. 2)

1897 Prosotherium garzoni Ameghino: 426 (original description), Fig. 12.

1897 Prosotherium triangulidens Ameghino: 427 (original description); regarded as synonymy herein.

1897 Prosotherium robustum Ameghino: 427 (original description); regarded as synonymy herein.

1901 Prosotherium quartum Ameghino: 371 (original description); regarded as synonymy herein.

1945 Propachyrucos ameghinorum Simpson: 551 (original description), fig. 1 and plate 1; regarded as synonymy herein.

1952 Propachyrucos simpsoni Chaffee: 524 (original description), plate 17, figs. 1–2; regarded as synonymy herein.

2015 Propachyrucos cf. Pr. smithwoodwardi Cerdeño and Reguero: fig. 8A–H; regarded as synonymy herein.

Lectotype. MACN A 52-455, fragment of skull with right and left I1, right P1–2, and left P1–3, M1–3 (Fig. 1A–B). This specimen was informally proposed as lectotype by Patterson (1952) and Reguero (1999).

Paralectotype. MACN A 52-456, left mandibular fragment with p2–4 and m2–3 (Fig. 1C–F).

Diagnosis. As for the genus.

Occurrence. Sarmiento Formation, Chubut Province; Agua de la Piedra Formation, Mendoza Province. Late Oligocene, Deseadan SALMA, Argentina.

Referred material in this study. From Agua de la Piedra Formation, Mendoza Province: MCNAM-PV 4189, left maxillary fragment with P4–M3; MCNAM-PV 4654, left maxillary fragment with M1–2; MCNAM-PV 3960, right mandibular fragment with p3–m2; MCNAM-PV 3962, left mandibular fragment with m1–3; MCNAM-PV 3964, right mandibular fragment with p2–m3 (m1 broken); MCNAM-PV 4191, left m1?; MCNAM-PV 4223, left mandibular fragment with p2; MCNAM-PV 4642, right m3; MCNAM-PV 4644, left p3; MCNAM-PV 4793, right mandibular fragment with p2–4; MCNAM-PV 4805, left p4 or m1; MCNAM-PV 4190, right mandibular fragment with p3–m2; MCNAM-PV 4711, right mandibular fragment with p2–4; MLP 79-XII-18-38, left maxillary fragment with P1–2; MLP 93-XI-21-54 g, right M1 or M2. From Sarmiento Formation, Chubut Province: AMNH 29574 (holotype of Pr. ameghinorum), nearly complete skeleton; AMNH 29604 (holotype of Pr. simpsoni), mandibular fragment with complete left and right i1–m3, including alveoli of di3–dc–dp1; MACN A 52-462 (lectotype of P. quartum, see comments below about it), left mandibular fragment with alveoli of i1–2, di3–dc–dp1, and p2–3 (p2 partially preserved inside the alveolus); MACN A 52-464 (holotype of P. triangulidens), fragment of palate with left P1–M3; MACN A 52-465 (holotype of P. robustum), fragment of palate with right P1–M3; MPEF 5753, mandibular fragment with symphysis and alveoli of both i1–2 and dp1, right p2–m3 and left p2; MPEF 5754, left mandibular fragment with p2–4; MPEF 5755, right mandibular fragment with p2–p4; MPEF 5761, set including probably right P4, two right M1 or M2, three right p3 or p4, and three right m1 or m2.

Remarks. Ameghino (1897) described Prosotherium garzoni based on a fragment of skull and a left mandibular ramus (MACN A 52-455–456; Fig. 1A–F). The author illustrated the palate as having both complete dental series (Ameghino, 1897, fig. 12), but currently this specimen has only the right P1–2, isolated in a separate fragment, and the left P4 is missing. Ameghino (1897) differentiated this species for the presence of a mesiolabial (parastylar) groove in the upper premolars and of a deep lingual groove in the upper molars, along with large upper and lower diastemata. In the same work, Ameghino (1897) described two other species, P. robustum (Fig. 1G–H) and P. triangulidens (Fig. 1I–J) based only on upper teeth. The author differentiated P. robustum for the larger size, the rectangular morphology of the P1, with a groove in the lingual face, and by the absence of the mesiolabial groove in the premolars; in turn, he differentiated P. triangulidens by the triangular morphology of the premolars with an almost transversally implantation, and the absence of the mesiolabial groove in the premolars. Later, Ameghino (1901) based P. quartum on a mandibular fragment without preserved teeth (Fig. 1K–N), characterised by bearing p1 (alveolus), and upper teeth with labial groove. As the latter are now lost, we here designate the mandibular fragment as a lectotype. This mandibular fragment is catalogued as MACN A 52-462, but it was erroneously labelled as MACN A 52-467, which corresponds to M1 or M2 referred to P. triangulidens. Loomis (1914) considered the difference in size between P. triangulidens and P. robustum as insufficient grounds for being recognized as different species, and so he synonymized the two. With regard to P. quartum, Loomis stated that the presence of the p1 was not a distinctive characteristic, because it is present in P. garzoni as well; he synonymized P. quartum with P. triangulidens because of their similar size (Table 1).

When Simpson (1945) described Propachyrucos ameghinorum AMNH 29574, a nearly complete skeleton from Scarritt Pocket (Chubut Province), he based his conclusions only on the presence of a complete dental series, but indicated that the skull was almost equal to that of Prosotherium (Simpson, 1945, p. 553). Strikingly, Simpson did not compare this specimen with the type of P. garzoni or with the data published by Loomis (1914). From the same locality, Chaffee (1952, Plate 17, fig. 3) later assigned the specimen AMNH 29608 (maxillary and mandibular fragments; see Suppl. App. 1.1 on interpretation of lower teeth) to P. garzoni, after having compared it with the specimen AMNH 14154 of this species (Loomis, 1914). At the same time, Chaffee (1952) described Propachyrucos simpsoni based on an almost complete lower jaw with both left and right i1–m3.

In his influential but unpublished study of Oligocene typotheres, Reguero (1999) recognized the validity of both Prosotherium and Propachyrucos. He synonymized Prosotherium triangulidens, P. robustum, and P. quartum with P. garzoni, arguing that their differences in size were insufficient for establishing different species. Reguero (1999) also synonymized Propachyrucos simpsoni and Pr. ameghinorum with Prosotherium garzoni. He stated that, aside from a small difference in size, the dental morphology of Pr. ameghinorum (AMNH 29574; see Section 3.2) is very similar to that of P. garzoni. In addition, he considered Pr. simpsoni (AMNH 29604; see Section 3.2) as synonymous with P. garzoni based on the presence of poorly developed canine and p1 (herein interpreted as dc and dp1), the similar size, the lingual face of the m2–3 straighter than in Propachyrucos smithwoodwardi, and the trigonid of p3–4 larger than the talonid.

Over the last twenty years, several studies have accepted Reguero's (1999) interpretation, so that Pr. ameghinorum, Pr. simpsoni and P. triangulidens have disappeared from the literature (e.g., Cerdeño and Reguero, 2015, Kramarz and Bond, 2017, Reguero and Cerdeño, 2005 and Seoane et al., 2017). Recently, Cerdeño and Reguero (2015) assigned several specimens from Quebrada Fiera (Mendoza Province) to Propachyrucos cf. Pr. smithwoodwardi. These authors only compared these new materials to Pr. smithwoodwardi and P. garzoni, implicitly accepting the synonymies proposed by Reguero (1999). However, as Reguero's (1999) nomenclatural acts remain unpublished and unofficial (see Section 3.1), only the synonymies of P. robustum and P. quartum with P. triangulidens proposed by Loomis (1914) are at present formally valid.

Discussion. After revising the type material, we have found some minor differences in the premolars of these specimens. In Prosotherium garzoni, the P2–3 (P4 is missing) have a well-developed parastylar groove and the lingual face of P1 is straight to slightly concave (Fig. 1A–B). The parastylar groove is also present in the type of P. robustum (Fig. 1G–H), but shallower, and the lingual face of the P1 is concave. In turn, the type of P. triangulidens (Fig. 1I–J) lacks the parastylar groove, and the lingual face of the P1 is concave. This variability is also seen among other specimens referred to Prosotherium: ACM 3731 (Fig. 2A–B) has the P1 with straight to slightly convex lingual face, and P2–4 lack the parastylar groove; MACN A 52-458 shows the P1 with concave lingual face and P2–3 without parastylar groove; MACN A 52-468 has lost the P1, but the parastylar groove is barely visible on the P2–3 (little worn, representing a young individual). The type specimens of P. garzoni and P. robustum, MACN A 52-455 and MACN A 52-465, respectively, are also relatively young individuals, as they preserve remnants of the central fossettes. This feature is not observable in the type of P. triangulidens, MACN A 52-464 (Fig. 1I–J), because teeth are broken at the alveolar level, where premolars appear to be shorter and wider; this condition is explained by the change of the crown shape that occurs with ontogeny, as it has been observed in other hegetotheriids (e.g., Sostillo et al., 2018 and Vera and Ercoli, 2018), as well as in mesotheriids (Billet et al., 2008, Cerdeño and Schmidt, 2013 and Gomes Rodrigues et al., 2017).

Ameghino (1901) differentiated P. quartum from P. garzoni by the fact that the former is larger and possesses p1. Both Loomis (1914) and Reguero (1999) recognized the presence of p1 in specimens of P. garzoni, which we here re-interpret as dp1. We suggest that the presence of this tooth (along with di3 and dc) varies with ontogeny. For example, the specimen MACN A 52-461, which represents an early ontogenetic stage, bears the right p2 without wear inside the alveolus and p3 is little worn and retains a small fossettid and a posterolingual groove. In addition, this specimen has three small alveoli between i2 and p2, which we herein interpret as for di3, dc, and dp1, respectively (Fig. 2C–D; Suppl. App. 1.2; see also Reguero, 1999). The condition in MACN A 52-461 (p3 more worn than p2) allows us to infer that the erupting sequence occurred posteriorly–anteriorly, as established for Paedotherium and contrary to Tremacyllus (Cerdeño et al., 2017 and Sostillo et al., 2018), although Vera and Ercoli (2018) established the opposite for one specimen of Tremacyllus. The lectotype of P. quartum MACN A 52-462 represents a later ontogenetic stage, in which the alveoli of di3–dc–dp1 are closing (Fig. 1K–N), as these teeth would not have replacement; the p2 is broken in the alveolus and we cannot infer to what extent it was erupted. Other specimens such as ACM 3731 (Fig. 2E–F) and MPEF 5753 (Fig. 2G–H) represent the following ontogenetic stage, where there is no trace of the alveolus of the canine (although that region is not well preserved), and the p2 is in use. The next stage is represented by MACN A 52-459 and MPEF 5755 (Fig. 2I–J), in which the alveolus of dp1 is almost resorbed, as Gomes Rodrigues et al. (2017) showed for the mesotheriine Typotheriopsis (but see Cerdeño and Schmidt, 2013 about the interpretation of deciduous dentition — DP2/dp3 — in mesotheriines). The paralectotype MACN A 52-456, in which there is no trace of the dp1 alveolus, appears to represent the latest ontogenetic stage of the materials we examined (Fig. 1E–F). In terms of size, we do not find differences to justify the specific differentiation of P. quartum; the dimensions of the alveolus of p2 are similar to the measurements of this tooth in the pool sample of P. garzoni (Suppl. App. 1.3).

Based on dental measurements and on the observed ontogenetic stages, we agree with Loomis (1914) and Reguero (1999), and consider P. triangulidens, P. robustum, and P. quartum as junior synonyms of P. garzoni (Table 1).

In reference to Propachyrucos ameghinorum, Simpson (1945, p. 553) recognized the similarity of its skull to that of Prosotherium. In addition, he superficially compared the postcrania with the specimens of Prosotherium collected by Loomis, but he did not consider the published measurements (Loomis, 1914, p. 70). After Simpson (1945), the holotype of Pr. ameghinorum, AMNH 29574, bears I3 and C represented by tiny alveoli, and he proposed that the teeth were probably lost in the lifetime; the alveolus of I2 is a little larger and the author stated that this tooth would be small but functional. In turn, the preserved i3–c–p1 are small, cylindrical and procumbent teeth. Comparing with P. garzoni, Pr. ameghinorum AMNH 29574 shares the presence of parastylar groove in upper premolars, upper molars with deep lingual groove, some vestigial anterior teeth with diastemata between them and similar size (Simpson, 1945, fig. 1; Fig. 3A–B, Suppl. App. 1.2–3). These similarities lead us to reinterpret these anterior teeth in AMNH 29574 as deciduous: DI2–DI3–DC and di3–dc–dp1, contrary to Simpson (1945). The P1 has the lingual face concave like the holotypes of P. robustum and P. triangulidens (Fig. 1G–J). The P2–4 have a shallow groove between parastyle and paracone like the holotype of P. robustum (Fig. 1G–H). The di3–dc–dp1 are more procumbent than the corresponding teeth in Propachyrucos (see Section 3.2). In addition, the lengths of i1–m3 and p2–m3 in Pr. ameghinorum is 53.1 mm and 30.9 mm, respectively (Simpson, 1945, table 1), closer to those in P. garzoni ACM 3137 (53 mm and 33.2 mm, respectively) and MACN A 52-456 (p2–m3, 30.1 mm; Fig. 1C–D) than to those in Pr. smithwoodwardi MACN A 52-451 (41 mm and 23.5 mm, respectively) (Ameghino, 1897, p. 425, and see Section 3.2).

In summary, when we compare Pr. ameghinorum, Pr. smithwoodwardi and P. garzoni, we find that the features present in Pr. ameghinorum are the same as in the juvenile specimens of P. garzoni, most notably including the di3–dc–dp1 separated by diastemata, the presence of the parastylar groove in upper premolars and lingual groove in upper molars, the concave lingual face of P1, and their similarity in size. While all these characteristics support synonymizing Pr. ameghinorum with P. garzoni, it should be noted that we did not directly examine the holotype of Pr. ameghinorum (AMNH 29574). Accordingly, we could not corroborate the presence of the mental foramina in AMNH 29574 (and Simpson [1945] did not describe the mandible). This characteristic foramen is present in several mandibular fragments assigned to Prosotherium, as MACN A 52-456, 52-459, 52-461, ACM3731, MPEF 5753, 5754, 5755, although, in MACN A 52-459 and MPEF 5755, the foramen is not as well preserved as in the paralectotype MACN A 52-456.

As said above, Reguero (1999) also synonymized Propachyrucos simpsoni with P. garzoni. Chaffee (1952) described the species from a mandibular fragment with left and right i1–m3, but alveoli of i3–c–p1 (AMNH 29604), and as a confer referred material he assigned a mandibular fragment with left and right i1–2, p1, dp2–4 and m1 (AMNH 29620). The type of Pr. simpsoni AMNH 29604 (Chaffee, 1952, Plate 17, figs. 1–2; Fig. 3C–D) shares with P. garzoni the i1–2 clearly procumbent, reduced alveoli of i3–c–p1, the robustness of the dentary, and the morphology of the p3–4, in which the talonid is smaller than the trigonid (Fig. 3C–D; Chaffee, 1952), while in Propachyrucos the lobes are similar in size (Table 2). In addition, Pr. simpsoni and P. garzoni are similar in size (L i1–m3, 53 mm and L p2–m3, 32.1 mm in Pr. simpsoni; L i1–m3, 53 mm and 33.2 mm in P. garzoni ACM 3731; see also Suppl. App. 1.4). The reduced alveoli in AMNH 29604 are herein reinterpreted as for deciduous teeth, di3–dc–dp1, as in the case of Pr. ameghinorum, which adds to the other mentioned similarities between Pr. simpsoni and P. garzoni; the interpretation of alveoli for deciduous teeth is congruent with the presence of a remnant of central enamel in the occlusal face of the teeth (Chaffee, 1952, Plate 17, figs. 1–2), which indicates a still young individual.

Concerning AMNH 29620, we have doubts about Chaffee's (1952) interpretation of dp2–4, because they are high (described as deep in the mandible almost reaching the bottom) and without roots or sign of permanent teeth below them; besides, the supposed dp2 (Chaffee, 1952: fig. 4) does not show a morphology of p2 (the trigonid would be longer, the ectoflexid shallower and lingual face straighter); this illustration is more compatible with a p4–m3 series, in which m2 is barely worn and m3 is erupting; however, as we have not revised this specimen, we cannot check this possibility and the relative position of these teeth in the preserved mandible and with respect to the supposed p1 (Chaffee described the latter as bilobed, so it could really correspond to p2 or p3, but not to p1, which is elliptic in Hegetotheriidae). So, we base our conclusions on Pr. simpsoni only on the type AMNH 29604, keeping in mind that AMNH 29620 could have been wrongly described.

To test whether differences in size between the synonymized species Pr. ameghinorum, Pr. simpsoni, P. garzoni, and Propachyrucos smithwoodwardi are significant, we performed one ANOVA for each tooth position and Tukey's multiple comparison test (see Section 2 and Suppl. App. 1.2 for details). The analyses revealed that Pr. smithwoodwardi is significantly different from Pr. ameghinorum, Pr. simpsoni, and P. garzoni in all teeth (p2, p3, p4, m1 and m2). In addition, when we compare the length of the series i1–m3 and p2–m3, we find that Pr. smithwoodwardi is significantly smaller (near 20%) than the other three species (which, in turn, are very similar to each other; see absolute values above). On the other hand, the analyses show that Pr. ameghinorum is significantly different from P. simpsoni in the p2 and significantly different from P. garzoni in the m1 and m2 (Suppl. App. 1.2). As a whole and considering that differences between Pr. simpsoni with P. garzoni are not significant, the statistical analyses support the synonymy between Pr. ameghinorum and the two latter species, being separated from Pr. smithwoodwardi.

Based on the previous comparisons and discussion, we formally propose the synonymy of Pr. ameghinorum, Pr. simpsoni and P. garzoni, supporting Reguero's (1999) interpretation, with priority of name for P. garzoni (Table 1).

A few additional comments should be made regarding the specimens from the Agua de la Piedra Formation (Quebrada Fiera, Mendoza Province, Argentina) identified by Cerdeño and Reguero (2015) as Propachyrucos cf. Pr. smithwoodwardi. The authors justified this assignment mainly because of the lesser development of the paracone fold of the upper molars with respect to P. garzoni. Also, they referred several mandibular fragments with premolars with long and more or less labially rounded trigonids, together with short and triangular talonids, and m1–2 showing rounded trigonid and triangular to trapezoidal talonid. Despite the fact that the authors did not compare directly with the type material of P. garzoni, they found that the sample of Quebrada Fiera showed some features more similar to P. garzoni and others more similar to Pr. smithwoodwardi, and even the size showed some variation with respect to Pr. smithwoodwardi (e.g., MCNAM PV 3960 being larger). However, the authors stated that these differences were not sufficient to recognize two pachyrukhine taxa with certainty. The present revision of this material confirms that it corresponds to a single taxon, but also reveals that the upper dentition matches that of P. garzoni, whereas the upper dentition is unknown for Pr. smithwoodwardi (see below); therefore, we consider that all specimens from Quebrada Fiera referred to as Propachyrucos cf. Pr. smithwoodwardi (Cerdeño and Reguero, 2015) instead belong to Prosotherium garzoni (Table 1).

(Fig. 4A–K, N–O)

1897 Propachyrucos smithwoodwardi Ameghino: 425 (original description), fig. 11.

Holotype. MACN A 52-451, right mandibular fragment with alveoli of c–p1 and p2–m3 (Fig. 4A–E); the anterior part figured by Ameghino (1897) is presently missing and is reproduced here (Fig. 4E) after Ameghino's illustration.

Diagnosis. As for the genus.

Occurrence. Sarmiento Formation, late Oligocene, Deseadan SALMA. Chubut Province, Argentina.

Referred material. MACN A 52-452, left mandibular fragment with p3–m2 (Fig. 4F–G); MACN A 52-453, left mandibular fragment with alveoli of c–p1 and p2, and an isolate right m1 or m2 (Fig. 4H–I, J–K, respectively).

Remarks. Ameghino (1897) characterized this species as having the complete dental series, relatively large i3 and canine (not reduced), and lower mandibular rami than in Pachyrukhos. The specimen MACN A 52-451 (Fig. 4A–E), though incomplete (as noted by Reguero, 1999, p. 189), matches the original illustration and description of Pr. smithwoodwardi, and it thus corresponds to the holotype of the species.

Following the comparisons described above, our study distinguishes Propachyrucos smithwoodwardi from Prosotherium garzoni and reassigns many species originally attributed to Propachyrucos to the genus Prosotherium. The following discussion focuses then on Pr. smithwoodwardi, Pr. crassus, whose transfer to the genus Prohegetotherium (Reguero and Cerdeño, 2005) has been recently rejected (Kramarz and Bond, 2017), Pr. aequilatus, and the species described by Roth (1899) from Colloncuran levels.

Discussion. Regarding Propachyrucos crassus (Fig. 4L–M), we studied all material assigned to this species (see Suppl. App. 1.1), which consists of three mandibular fragments constituting the syntypes (MACN A 52-448–52-450), which were differently interpreted. Reguero and Cerdeño (2005), based on Reguero's (1999) work, considered MACN A 52-448–52-449 as bearing c–p2 and MACN A 52–450 as p1–p2, whereas Kramarz and Bond (2017) considered MACN A 52–448 with p1–p3 and the three fragments possessing the p1 (or its alveolus) lingual to the tooth row. When we examined the lower jaws of Deseadan Hegetotheriidae, we found that all specimens present bilobed p2 and elliptical p1 without any groove (i.e. Sallatherium altiplanense, Reguero and Cerdeño [2005, fig. 2e]; Pr. smithwoodwardi [Fig. 4A–E]; P. garzoni [Fig. 1, Fig. 2 and Fig. 3]; specimen assigned to Prohegetotherium sculptum by Chaffee [1952, plate 16, figs. 2–3]; Pr. ameghinorum [Fig. 3B]). Based on this evidence, we agree with Kramarz and Bond's (2017) interpretation and consider the first bilobed tooth as p2, so MACN A 52-458–52-459 have p1–3 and MACN A 52-450 has p1–2 (Fig. 4L–M). This established, we find that the elliptical p1 lingually placed and the bilobed p2 with long and narrow trigonid lingually concave is a combination of features unique to this taxon. However, the elliptical cross section of p1 is shared by almost all Hegetotheriidae. Besides, Pr. crassus can be excluded from Propachyrucos and Prosotherium based on the following features: larger size, lingual position of the p1, p2 with long and narrow trigonid, lingually concave, and p2–3 with trigonid smaller than talonid (Table 2). On the other hand, following Kramarz and Bond's (2017) revision of Prohegetotherium sculptum, there is no known lower dentition of this species to compare with Pr. crassus, which supports their conclusion about maintaining this species of uncertain generic position, and clearly apart from Propachyrucos. The features of Pr. crassus (lingual position of p1, morphology of the p2 and general size) are shared with the specimen AMNH 29605 described by Chaffee (1952, Plate 16, fig. 2–3) as Prohegetotherium sculptum, and could represent the same taxon. It is evident that additional materials and the revision of all remains attributed to Prohegetotherium sculptum are necessary to clarify this taxonomy.

Concerning Pr. aequilatus, it is only known by its holotype (Ameghino, 1901), a left mandibular fragment with p4–m3 (MACN A 52-454; Fig. 4N–O; Suppl. App. 1.4). Mones (1986: p. 161) included this specimen as Pr. smithwoodwardi, whereas Reguero (1999) recognized it as type of Pr. aequilatus. The p4 has a rounded trigonid, barely larger than the triangular talonid, and a straight lingual face; m1–2 have the trigonid smaller than the talonid, and the lingual face is also straight; m3 has a deep ectoflexid and a shallow labial talonid groove; its lingual face is slightly convex, with a posterior furrow that lingually delimits the third lobe. The m1–2 differ from Pr. smithwoodwardi, in which the trigonid is smaller than the talonid (Table 2) and the lingual face is slightly convex; in addition, Pr. aequilatus is 15–20% larger and the mandibular ramus is higher than in Pr. smithwoodwardi. In turn, the talonid smaller than the trigonid in p4 coincides with P. garzoni, but the statistical analyses show Pr. aequilatus significantly different from both P. garzoni (for m1 and m2) and Pr. smithwoodwardi (for p4, m1 and m2) (see Suppl. App 1.2). Therefore, Pr. aequilatus cannot be clearly assigned to any of these species/genera and we propose to consider the name as a nomen dubium.

With respect to Propachyrucos depressus, the only specimen in the MLP collection (MLP 12-2915a, Fig. 5A–F) comprises two symphyseal fragments, one right mandibular fragment with p4–m3, one left mandibular fragment with probably p3–p4, and an isolated right m3. Due to the presence of two right m3 and two symphyses, at least two individuals are present and thus the specimens constitute syntypes, instead of the holotype of the species as mentioned by Vera (2019: fig. 4I). The original description (Roth, 1899) does not detail how many specimens the author had in hand. The symphysis fragment that preserves the anterior dentition bears the right and left i1, right i2, alveolus of left i2 and both subtriangular alveoli of a posterior tooth (likely p2), separated by a short diastema (Fig. 5A–B). The absence of i3–c–p1 and the highest degree of procumbence discard these specimens from belonging to Propachyrucos; their assignment to Prosotherium is also discarded because of their smaller size (Suppl. App. 1.3–4). The isolated m3 has the third lobe rounded (Fig. 5C–D), a characteristic of Pachyrukhos and different from the triangular outline in Paedotherium (Cerdeño and Bond, 1998 and Ercoli et al., 2018). In addition, the m1–2 preserved in one mandibular fragment (Fig. 5E–F) present the posterolingual projection observed in Pachyrukhos (Seoane and Cerdeño, 2019), which is not present in Paedotherium or Tremacyllus (Cerdeño and Bond, 1998 and Ercoli et al., 2018). Therefore, the syntypes of Propachyrucos depressus are herein considered as belonging to Pachyrukhos (Table 1 and see Suppl. App. 1.5, figure S1, for comparison with Pachyrukhos). Because of the missing p2 in the syntypes, we cannot ascertain to which of the two species of Pachyrukhos recognized by Seoane and Cerdeño (2019) it would belong, even though the Colloncuran age points to Pachyrukhos moyani Ameghino, 1885.

Out of the three teeth (p3, p4 and m3) described by Roth (1899) as Propachyrucos medianus, only the two premolars were located (MLP 12-3160; Fig. 5G–I), although Vera (2019) only illustrated one tooth as holotype and considered it m1 or m2, without performing any anatomical comparison. These teeth are similar in size to Pachyrukhos (Seoane and Cerdeño, 2019) and Paedotherium (Ercoli et al., 2018) and are slightly larger than in the holotype of Pr. smithwoodwardi, but smaller than Prosotherium (Suppl. App. 1.3–4). Their trigonids are subtriangular and smaller than the talonids (also triangular; Table 2), and the distal face is straight with a small posterolingual projection, so their morphology is closer to that of Pachyrukhos (Seoane and Cerdeño, 2019) than to that of Paedotherium (Ercoli et al., 2018), Tremacyllus, where the talonids are smaller than trigonids (Cerdeño and Bond, 1998 and Ercoli et al., 2018), or Propachyrucos. Since Pascual et al.’s (1978) work, several authors accepted the transference of this species to Pachyrukhos (Kramarz et al., 2011, López et al., 2018 and Vera, 2019). The premolars of Pachyrukhos medianus (MLP 12-3160) are larger than those of Pachyrukhos moyani and coincide with the material from the Collon Curá Formation identified as Pachyrukhos sp. 1 by Seoane and Cerdeño (2019).

The holotype of Pr. robustus, MLP 12-3161 (Fig. 5K–L), is a right mandibular fragment with p4–m2, only the m1 complete, and all occlusally broken. Roth (1899) described them as such a partial series (p4–m2), but Vera (2019) interpreted the series m1–3 and the complete tooth as m2 (although it is included as m1 in table 2 and fig. 3B). It is not easy to determine who is right, as the last molar is posteriorly incomplete and cannot be confidently identified as the m3. The complete molar (m1 or m2) of Pr. robustus is 60–70% larger than those (m1 or m2) of Propachyrucos and Pachyrukhos (Suppl. App. 1.4), and therefore we agree in discarding MLP 12-3161 from these two genera. Our comparisons show that Pr. robustus is similar to Hegetotherium mirabile Ameghino, 1887 (see Seoane and Cerdeño, 2019), with which MLP 12-3161 shares a subtriangular to semicircular trigonid, smaller than the triangular talonid; deep and wide ectoflexid; and slightly convex lingual face, but lacks the posterolingual projection present on lower molars of H. mirabile (Seoane and Cerdeño, 2019). In turn, Vera (2019) established the affinities of MLP 12-3161 with Hemihegetotherium, closing it to Hm. tantillum Vera, 2019. However, Vera (2019) characterized the m1–2 of Hm. tantillum by having the convex distal face, trigonids similar in length to talonids, and medial and distal projections, but none of these characteristics is present in Pr. robustus MLP 12-3161. The complete molar of this specimen has the distal face straight, trigonid shorter than talonid (Table 2), and lacks the mesial and distal projections (Fig. 5K–L). Hence, we consider Propachyrucos robustus as a nomen dubium, highlighting the similarity between this specimen and H. mirabile.

In addition to the type material of these species, we found several unpublished specimens in the MLP catalogued as Propachyrucos sp.: MLP 12-2889a, 59-II-26-91, 61-IV-11-139–152, 61-IV-11-163–166, 61-IV-11-186–192, 61-IV-11-207 and 61-IV-11-316–317 (Suppl. App. 1.1). Most of them share the diagnostic characters of Pachyrukhos (Seoane and Cerdeño, 2019): lower teeth with both lobes subtriangular and trigonid smaller than talonid, except in the p3, where the trigonid is rounded and as large as the talonid (Fig. 6A–B); p2 with reduced trigonid (when it is present) (Fig. 6C–D); m1–2 with posterolingual projection (Fig. 6E–F); m3 with wide and shallow posterolingual groove and rounded third lobe (Fig. 6E–F); upper premolars with a smooth furrow in the posterior part of the lingual face and molars without any groove (Fig. 6G–H). Based on these features, we identify MLP 12-2889a as Pachyrukhos moyani, and the remaining material as Pachyrukhos sp. The specimen MLP 61-IV-11-146 is unique in having a quadrangular trigonid that is separated from the talonid by opposite labial and lingual grooves, suggesting that this specimen belongs to Interatheriidae (cf. Protypotherium) rather than to Hegetotheriidae.

Most of these specimens come from El Pajarito (early Miocene, Colhuehuapian SALMA, Chubut Province; Goin et al., 2007): MLP 61-IV-11-139–152, MLP 61-IV-11-163–166, MLP 61-IV-11-186–192, MLP 61-IV-11-207 and MLP 61-IV-11-316–317. Instead, MLP 12-2889a, which comes from the Collón Curá Formation (early middle Miocene, Neuquén Province) and MLP 59-II-26-91 comes from Laguna de La Bombilla (Chubut Province). This specimen (Fig. 6I–N) deserves a special comment. It is constituted by three isolated teeth, a right p4, a right m3, and a right incomplete unworn premolar. The p4 (Fig. 6I–J) shows the condition observed in Pachyrukhos, both lobes triangular and the posterior one more developed; the m3 (Fig. 6K–L) is trilobed, with the first lobe triangular and with a small mesial projection, and the third lobe rounded; the lingual face is convex, but with a shallow concavity opposite to the labial talonid groove. According to the size (Suppl. App. 1.4) and the presence of a trilobed m3 with rounded third lobe, MLP 59-II-26-91 could belong to Propachyrucos or to Pachyrukhos, but the morphology of p4 allows its assignment to Pachyrukhos sp. The recognition of Pachyrukhos at Laguna de La Bombilla sheds light on the age of this site. It was firstly interpreted as an early–middle Oligocene locality by Marshall et al. (1983), but recently Miquel and Bellosi (2010) postulated that the Sarmiento Formation outcrops at this locality, and this formation covers from middle Eocene to early Miocene (41.6–18.7 Ma) according to Ré et al. (2010). The presence of Pachyrukhos points to an early Miocene age for the bearing-fossil deposits at Laguna La Bombilla.