The fossil preserves the distal two-thirds of the right jaw including the condyle, the coronoid process and the full molar series, in addition to an isolated right canine and p1 (Figs 1, 2A–C). This material represents a smaller specimen than the holotype (see molar and mandibular sizes; Tables 1, 2; Fig. 1), probably a young-adult individual, judging from the incomplete development of the coronoid process and the presence of a recently erupted (unworn) m4. Following the individual dental age stages (IDAS) proposed by Anders et al. (2011), this specimen falls into IDAS 3 category, which is defined by the presence of fully erupted dentition with wear restricted to the inner occlusal surface —absence of interior enamel and persistence of the external enamel border— on the m1. In the particular case of this specimen, only the m1 has lost by wear the paraconid and protoconid, remaining in the molar the truncated bases of these cuspids, with the enamel confined to the periphery; on the talonid of this tooth, the wear is restricted to the hypoconid apex.

Dentary. The dentary preserves part of the body —the upper half of the alveolar process bearing the p3 roots and the m1 to m4— and the almost complete ramus, except for the anterior part of the ventral border and the angular process (Figs 1, 2A–C). The lateral side of the body shows the posterior portion of the rim of a mental foramen located between the m1 and m2. In the holotype, the size and position of the mental foramina varies in each hemimandible. The foramen present in the specimen IBIGEO-P 110, is in the same position as the third mental foramen on the right body of the holotype (Fig. 1B). Bellow the boundary between the m3 and m4, the lateral aspect of the body presents a deep fossa, which is also present in the type specimen (PVL 4187), particularly emphasized on the left dentary. This pit, together with a shallow fossa extended anteriorly toward the alveolar border of the m1, could be related with the attachment of the buccinator muscle (following the pattern described in Didelphis; Hiiemae and Jenkins 1969). Behind the m4, the retromolar space is short, in agreement with the age stage inferred for this specimen. The medial side of the mandible is more fragmented (Fig. 2B) because part of the covering bone was lost, exposing the large roots of the last molar deeply implanted in the dentary. Medial to the roots of the p3, the mandibular bone expands medially, suggesting the location of the posterior extension of the symphysis which in the holotype reaches the point below the p3 and m1 limit. Below the m2 and m3 there are subtle tuberosities, also present in the holotype where they are located more anteriorly, below the m1 and m2. These structures could be related with the attachment of the geniohyoideus and genioglossus muscles, although in Didelphis it occupies a more ventral position (Hiiemae and Jenkins 1969, fig. 4B).

The ramus conserves the condylar process complete while the coronoid process is partially preserved. The coronoid process, shorter and lower than in the holotype, shows a thick and straight coronoid crest that forms an obtuse angle with the main axis of the dentary (Figs 1, 2). This crest extends anteriorly and basally up to the level of the m3–m4 limit. The dorsal border of the coronoid process appears to be rounded and ends in a posterior beak, also present in the holotype and several other sparassodonts (e.g., Arctodictis, Cladosictis). The posterior border of the coronoid process is anteriorly slanted and together with the coronoid crest determine a triangular-shaped coronoid process. In lateral view, the coronoid process is smooth, except for a subtle crest and fossa that could correspond to the attachment of the ventral extension of the M. zygomaticomandibularis. The prominent posterior shelf of the masseteric fossa, the area of attachment of the superficial and deep masseter muscles (Turnbull 1970), is narrower than in the holotype.

In medial view (Fig. 2B), the mandibular ramus shows the medial wall of the mandibular foramen, located approximately in line with the tallest point of the coronoid process. Dorsal and anterior to the mandibular foramen, the dentary shows a horizontal crest. In the holotype, this mandibular area is better preserved and shows a more subtle crest extending from the m2 to the posterior end of the m4 –not as posterior as in IBIGEO-P 110. We interpret this structure as part of the mylohyoid line for the attachment of the M. mylohyoideus, following its similar location in the dog (Evans and de Lahunta 2013; fig. 6–22). Dorsoposterior to the mandibular foramen, the ramus bears a rounded fossa, apparently absent in the type specimen where the area is partially damaged. In IBIGEO-P 110 this depression occupies a similar position to that described by Ercoli et al. (2017; fig. 2D) in the CarnivoraGalictis cuja where the lateral pterygoid muscle inserts in a position anterior and medial to the neck of the condylar process. Moreover, both the holotype and IBIGEO-P 110 exhibit a small pit just anterior to the articular head of the condylar process which is in agreement with the location of the pterygoid fovea in Didelphis (the insertion of the inferior head of the lateral pterygoid muscle; Hiiemae and Jenkins 1969; Diogo et al. 2016). For Callistoe, we interpret this pit and the fossa posterior to the mandibular foramen as possible areas related with the lateral pterygoid muscle attachment.

The condylar process is a bar located at the same level as the alveolar dorsal border. Due to postmortem deformation, it is not transverse to the main axis of the dentary, as expected anatomically, but orientated posterolaterally (Fig. 2C). The articular head narrows laterally and occupies the dorsal and posterior aspect of the process. The posterior side of the neck shows a deep fossa opened medially which would be related with the attachment of the lateral ligament, the fibrous part of the temporomandibular joint (Evans and de Lahunta 2013). The mandibular notch is open posteriorly and with a straight ventral side as in the holotype, but it is proportionately shorter, forming an almost right angle with the posterior border of the coronoid process.

Dentition. In addition to the complete teeth recovered in the dentary (m1 to m4), the dentition includes detached canine and p1, and roots of the p3 implanted in the alveoli. The teeth size in this specimen is about 20% smaller than the holotype (Table 2), which may be due to intraspecific variability or sexual dimorphism. Considering that the molar series in the new specimen measures 80% of that of the holotype, the assignment of the isolated premolar to the p1 is based on this same length proportion in relation to the p1 in the holotype. As mentioned, the teeth are only slightly worn, being the m1 the most affected tooth (see below).

The canine fragment mostly represents the extralveolar portion of the tooth (Fig. 3 A–C). It is large, pointed, and labiolingually compressed. The tooth is formed by dentine and enamel; the latter is preserved as a thin and very small sheet exposed on the lingual apex of the crown (Fig. 3C). This contradicts the initial assumption that the canine in Callistoe vincei lacked enamel (Babot et al. 2002); instead, it seems that the enamel is restricted to the most apical end of the tooth forming an enamel cap that wears down quickly. The canine exhibits a main wear surface which is present on the distolabial side and caused by the attrition with the lingual side of the upper canine. This facet is long, slightly concave, and well demarcated by distal and labial sharp borders. Several longitudinal striae, sulci, and grooves extend along the entire length of the canine, mainly on the lingual and distal sides. The deepest is the medial groove, followed by two shallower distal sulci. The remaining grooves are represented by striae close to each other, more marked towards the root of the tooth. The presence of a deep lingual sulcus is a common feature in Sparassodonta, emphasized in the larger members of the group (e.g., Borhyaena, Australohyaena; Arminiheringia, Proborhyaena; Paraborhyaena; Babot et al. 2002; Engelman et al. 2020).

The first premolar is almost complete, except for the mesial root, the most mesiobasal border of the crown, and the tip of the distal cuspid (Fig. 3D, E). The crown is asymmetric —the mesial border is convex and short and the distal one is concave and longer— and more convex on the labial half. The medial side is flatter than the lingual one. The mesial portion exhibits a small and labially located cuspulid from which arises a crest towards the apex of the protoconid. Lingual to the mesial cuspulid, but unconnected, there is a mesial cingulid broken for the most part. The distal portion of the p1 bears a partially preserved talonid probably supporting a small cuspid. The lingual side of the talonid exhibits a wear surface that partially removed a subtle cingulid.

The third premolar was lost before collection and is only represented by its roots. In the specimen IBIGEO-P 110 this tooth must have been completely erupted given that in Sparassodonta the third premolar erupts about the same time as the fourth molar (Forasiepi and Sánchez-Villagra 2014). The roots are rounded, subequal in diameter, and separated by a relative wide septum. In the holotype the p3 roots are very large —wider and longer than in the m1— and are greatly exposed over the alveolar border, aligning the apex of tooth crown almost at the same height as the apical portion of the m4.

In this specimen the molar row is complete and very well preserved, which is key because previously known specimen of C. vincei, preserve only one complete molar, the m4 (Figs 1B, 4A, B). The overall tooth size and the trigonid increase rapidly in size from the m1 to the m4 (Table 2). Some other traits of the trigonid, like the sharpness of the mesial keel and the deepness of the hypoconulid notch, also become more pronounced toward the posterior end of the molar row. The angle formed between the paracristid and the long axis of the dentary also increased posteriorly, because the protoconid is progressively set more labially. The length of the precingulid however, becomes shorter posteriorly. The talonid decreases in height and size from the m1 to m4 and the hypoflexid becomes deeper in the same direction. From the m1 to m3 the talonid conserves the same tricuspid pattern, but in the m4 the basin is greatly reduced and the cuspid arrangement is highly modified (see below). The basal extension of the molar enamel defines two lobes above the mesial and distal roots. From the m1 to the m3, the mesial lobe is more dorsally located than the distal lobe. In the m4, the enamel basal line is approximately straight, clearly visible in the type specimen. The molar roots protrude from the alveolar margin, —but in lesser extent than in the type specimen. In PVL 4187 the exposed root surface in the m3 and m4 is larger than in IBIGEO-P 110.

On the m1 the trigonid does not preserve individual cusps, the paraconid and protoconid have been obliterated by wear against the distal side of the P3 (Fig. 4C–E). The exposed surface of the cusp bases indicates that the protoconid was the largest of the two. It was located labially to the paraconid, so that the paracristid was oblique, unlike the condition in e.g., Pharsophorus and Australohyaena, where this crest is straight and the paraconid, protoconid and hypoconid are in line with the mesiodistal axis of the tooth. The metaconid, which is present on the m3, could have been also present in the m1. This inference derives from a wear facet located on the mesiolingual corner of the trigonid; but given the absence of a metaconid in the m2, it is possible but unlikely. On the trigonid, the mesial border is not preserved and the presence of the hypoconulid notch cannot be described. The precingulid arises from a tiny cuspulid located slightly distolabial to the most mesial border of the tooth; it is partially worn and its most distal end diverges into two short dorsal and basal cingulids. The talonid is complete. It bears very well-preserved hypoconid, hypoconulid, and entoconid. The talonid width is nearly equal to the paraconid base but narrower than the protoconid. The largest cusp is the hypoconid which is mesiodistally elongated and exhibits apical wear. From this cuspid a very short cristid obliqua reaches the distal wall of the protoconid in a middle position from where this notched crest runs upward, climbing the distal slope of the protoconid. The cristid obliqua is not as lingually directed as in the m2 and m3, defining a shallower hypoflexid than in the posterior molars. Distal and basal to the hypoconid, the talonid is surrounded by a labial postcingulid reaching the base of the protoconid where it limits a very narrow shelf. The hypoconulid is the lowest cuspid; it is labiolingually elongated, and occupies a distal and median position. The entoconid is conical, connected to the hypoconulid by a short postcristid which bears a tiny postentoconulid; the entoconid is disconnected from the protoconid, i.e., the entocristid is absent. The basin is shallow with a talonid notch broadly open lingually, reaching down the CEJ.

On the m2 the trigonid/talonid length ratio is higher than on the m1; i.e., the trigonid is larger than on the m1 but the talonid is only slightly larger (Fig. 4C–E). Therefore, the overall increase in length on the m2 is accomplish almost solely by the larger trigonid. Moreover, the acute angle formed by the paracristid and the mesiodistal axis of the tooth is also greater; i.e., the protoconid is more labially located. The mesial portion of the trigonid exhibits a deep hypoconulid notch which separates the mesial border of the paraconid from the precingulid. This cingulid, which surrounds the labial aspect of the paraconid base almost until the paracristid notch, is broader near the hypoconulid notch; it weakens towards the base of the paraconid along its mesiolabial face where it ends in two tiny cuspulids. The paracristid notch (carnassial notch) is visible here although not as well developed as in the posterior molars. The protoconid is a triangular cusp with a flat distal face and a complex lingual morphology. As in other mammalian sectorial teeth, the lingual aspect of the protoconid is reinforced by a blunt ridge extending from the labial apex of the cusp, descending abruptly to reach the lingual margin of the crown. This crest divides the trigonid into two areas: a mesial concave surface and a less excavated distal surface. The mesial area is functionally part of the paracristid notch determining a broad and distinctive basin lingual to it. The distal surface produces a concave area running parallel to the postprotocristid that is much less developed than the mesial one. The sharpness of the paracristid and relative bluntness of the postprotocristid are determined to a large degree by the development of these features of the lingual aspect of the protoconid. The postprotocristid exhibits an enamel bulge in the same position as the metaconid in the m3. But considering the absence of a postprotocristid notch, we conclude that this bulge is not a true metaconid, just the undifferentiated portion of the crown not under the influence of a putative metaconid enamel knot. The talonid on the m2 is similar to that of the m1 in the general size and in the position and size of the cuspids. However, some structures are better preserved. The cristid obliqua forms a distinct notch when it reaches the protoconid; from this point the crest rises upward on the flat distal face of this cusp. Another notch is visible between the hypoconid and the hypoconulid, which are close together. The labial postcingulid is similar in shape and size to that of the m1. Between the hypoconulid and the entoconid, appears to be an elongated postentoconulid and a notch between both structures. The entoconid is a pointed cusp disconnected from the protoconid. Towards the base, it exhibits a minute cuspulid. The talonid basin is shallow and opens lingually in a well-demarcated talonid notch.

The third molar is larger than the mesial teeth (Fig. 4F–H; Table 2). This difference is due only to a trigonid increase in size; the talonid is smaller than in the m2, both in absolute terms and in relation to the trigonid. The hypoconulid notch is conspicuous. Towards the base it bears a short cingulid disjointed from the precingulid. The precingulid has a main short and crested segment, with two tiny cuspulids. The distal part is a weak shelf running to a point basal to the paracristid notch. As in the precedent teeth, the distal face of the protoconid is flat. The postprotocristid is sharp; it ends in a notch formed at the junction with a small and conical metaconid located in a somewhat basal position, below the level of the paraconid apex. The talonid is appressed against the m4; the three main cuspids are well-preserved, being the hypoconid and the hypoconulid slightly larger than the entoconid. Unlike the condition of the m2, in the m3 all three talonid cusps are cone-shaped and appressed to each other. The cristid obliqua runs lingually to the protoconid, in a similar position to that of the precedent molars. This crest climbs the distal wall of the protoconid, next to the basal level of the metaconid. The talonid basin is small and it is virtually closed by the entoconid, although this cuspid does not reach completely the distal wall of the protoconid. The hypoconulid is distally located, in a central position of the distal talonid border. The labial postcingulid originates at the base of the hypoconulid; it surrounds the talonid towards the base of the hypoflexid, where this cingulid bears two tiny cuspulids.

The m4 is the largest and best preserved tooth (Figs 4I–K, 5). The roots are exposed on the medial side of the dentary, showing the depth of tooth insertion in the mandibular bone (Fig. 2B); the roots measure more than the half of the tooth crown (27 mm vs. 13 mm, respectively). The paracristid notch is a deep V-shaped efficient capturing structure where forces concentrate at the end of the stroke (Evans 2005). The hypoconulid notch is deeper than in the precedent molars; it lodges the hypoconulid and the entoconid of the m3. On the base, it bears a short cingulum which does not reach the precingulid, which is smaller and shorter than in the m3. The postprotocristid ends toward the base of the protoconid, but contrary to that observed on the m3, the protocristid notch and the metaconid are absent. The talonid is more reduced and modified in relation to the precedent molars and it has a variable morphology among the known specimens of Callistoe (Figs 4, 5). In the new specimen here described, we observe two evident cuspids on the talonid: the hypoconid and a labial accessory cuspid. The hypoconid is tiny and very close to the protoconid. The cristid obliqua is absent, given the proximity of the hypoconid with the protoconid, but its vertical extension at the distal face of the protoconid is present, as in the anterior molars. The labial accessory cuspid is linked, via a very short cristid, with the distal border of the talonid; from this point of contact, the labial postcingulid arises and runs mesially towards the hypoflexid. The hypoconulid is absent. The lingual side of the talonid is surrounded by the postcristid which ends in a tiny cuspule located at the base of the protoconid. Slightly distal, the postcristid exhibits an enamel bump that could represent a rudimentary entoconid, but this feature is so subtle as to lack any functional significance and we prefer to regard the entoconid as absent. In the holotype the talonid exhibits the hypoconid and the labial accessory cuspid. As in the specimen IBIGEO-P 110 the hypoconid is small and it is next to the distal wall of the protoconid where it reaches the crest on the distal face of the protoconid; labially, the hypoconid joins the accessory cuspid by a very short cristid. Distal to this accessory cuspid, the labial postcingulid emerges and borders the labial side of the talonid. The postcristid runs toward the base of the protoconid, closing the lingual side of the talonid. The two subtle cuspules observed in IBIGEO-P 110 at the lingual border of the talonid are absent in the type specimen.

Two names taken from dentistry are often used to refer to tooth wear: abrasion, caused by the contact between food/dust/detritus and tooth; and attrition, wear caused by the contact between teeth (Butler 1952). Although both mechanisms are often combined during mastication, if abrasion is dominant the impressions within enamel are often rounded and have no clear marginal edges (Koenigswald 2018); attritional marks are distinguished by flat areas usually with striations parallel to the movement direction (Butler 1952). This last mechanism forms facets and sharps teeth, while abrasion blunts them (Ungar 2015).

The occlusal patterns and their impressions on the tooth surface were intensely studied from more than 100 years for several purposes, e.g., evolution of tribosphenic molar and study of cusp homology, direction of chewing movements, type of ingested diet (among a prolific literature, e.g., Osborn 1888; Butler 1952, 1971; Patterson 1956; Mills 1966; Crompton and Hiiemae 1970; Crompton 1971; Greaves 1973; Crompton and Kielan-Jaworowska 1978; Fortelius 1985; van Valkenburgh 1988; Lucas 2004; Teaford 2007; Davis 2011; Smits and Evans 2012; Ungar 2015; Koenigswald 2018; Schultz et al. 2018). In the following section we describe the mesowear facets in the lower dentition of Callistoe, using traditional works of Crompton (1971), Crompton and Kielan-Jaworowska (1978), and the more recent revision of Davis (2011).

The p1 exhibits a distolingual facet exposing a dentine core surrounded by an enamel border. Toward the crown base the wear affected a distolingual cingulid which remains as a subtle ridge (Fig. 3E).

In the m1 the wear caused the loss of the paraconid and protoconid, remaining only the cusp bases (Fig. 6A, B). The exposed dentine surface is visible in occlusal view; it is surrounded by an enamel border. The labial side exhibits the remains of the cutting facet 2 of Crompton (1971), which on the m1 is brought about by the initial contact with the DP3 (a sectorial tooth in sparassodonts; Forasiepi and Sánchez-Villagra 2014) and in adult age with the distolingual side of the P3. In sparassodonts, the deciduous premolar is functional in juvenile specimens; this is particularly clear in Arminiheringia sp. (MLP 82-V-1-1) in which the m1 and m2 are heavily worn, while the P3 only exhibits facets restricted to the main cusp (Forasiepi and Sánchez-Villagra 2014). The facet 2 also includes the precingulid which exposes the dentine at the most mesial part. The wear facet 1 is restricted to the distolabial part of the distal face of the protoconid and appears to extend to the mesiolabial side of the talonid (Fig. 6B). It is formed by the contact with the paracingulum and paracone of the M1. The crown of the M1 is not preserved in other specimens of Callistoe vincei (the holotype and PVL 4207) but the paracingulum was probably present judging by its existence in the M3. On the labial side of the talonid, the postcingulid is barely worn down by an incipient contact with the mesiolingual face of the metacone of the M1 (facet 4 of Crompton 1971). The facet originated by the contact with the distolingual face of the paracone (facet 3 of Crompton 1971) is also present in this molar. There are no traces of lingual wear facets (5 and 6 of Crompton 1971) on the talonid of this specimen. In the typical tribosphenic pattern these facets are brought about by the contact with the mesial and distal sides of the protocone, cusp which is much reduced in Callistoe. In the talonid, the apical wear areas caused by abrasion and identified as small rounded zones devoid of enamel, are restricted to the hypoconid and entoconid. Notably, in this specimen of Callistoe vincei the m1 is not as worn as in the juvenile specimen of Arminiheringia sp. (MLP 82-V-1-1), where the first molar is heavily worn out while the m4 is not yet completely erupted, suggesting that in MLP 82-V-1-1 the m1 probably erupted earlier than in Callistoe.

The m2 exhibits abrasion at the apex and mesial edge of the paraconid, the apex of the protoconid and upper third of the postprotocristid, the apex of hypoconid and entoconid, and the lingual side of the paracristid notch. Worn areas caused by attrition corresponds to facets 1, 2 (covering the paracristid, the labial wall of the trigonid and the precingulid), and 4 of Crompton (1971) (Fig. 6C); these are brought about by the contact with the paracingulum and mesial side of the paracone of the M2, postmetacrista of the M1, and the mesiolingual face of the metacone of the M2, respectively. Moreover, on the mesiolabial slope of the hypoconid the facet 3 is clearly present, formed by the contact with the distal face of a reduced paracone of the M2. As in the m1, there is a considerable difference in the degree of wear between the m2 of CallistoeIBIGEO-P 110 and the juvenile Arminiheringia sp. (MLP 82-V-1-1), where the second lower molar lost almost the entire crown.

On the m3, the contact between food and tooth —abrasion— is visualized as facets on the paraconid, lingual side of the paracristid and postprotocristid, paracristid notch, and the apex of the metaconid and hypoconid. The attritional facets are feeble, i.e., did not cause complete enamel denudation. They are represented by Crompton’s facet 1 identified as a vertical and planar surface on the distal face of the protoconid, facet 2 (mainly on the labial face of preprotocristid and precingulid), and facet 3, poorly developed and restricted to the apical portion of the labial slope of the hypoconid (Fig. 6D).

The m4 is the tooth less affected by wear. The abrasion is limited to the most apical portion of the paraconid and protoconid, the lingual side of paracristid and base of the paracristid notch, and the lingual border of the postprotocristid. Attritional facets are facet 1, seen as a thin band on the distolabial portion of the protoconid and facet 2, restricted to the preprotocristid. The talonid was not affected by wear.