Parachleuastochoerus Golpe-Posse, 1972

Alba, David M., Siarabi, Souzanna, Arranz, Sara G., McKenzie, Sharrah & Casanovas-Vilar, Isaac, 2025, Dental remains of ‘ ParaChleUaStoChoerUS’ valentini (Suidae: Tetraconodontinae) from the early Late Miocene of Sant Quirze (Vallès-Penedès Basin, NE Iberian Peninsula): taxonomic and phylogenetic implications, Swiss Journal of Palaeontology (9) 144, pp. 1-58 : 13-46

Genus Parachleuastochoerus Golpe-Posse, 1972 ‘Parachleuastochoerus‘ valentini (Filhol, 1882)

( Figs. 3 , 4 , 5 , 6 , 7 , 8 , 9 , 10 , 11 )

Referred material See Table 1.

Description

Upper incisors Te I1 is represented by two specimens with a moderately worn crown and a partial root

( Fig. 3A–B ; one already figured by Pickford, 2014: fig. 13A) plus a complete ( Fig. 3E ) and two incomplete

( Fig. 3C–D ) germs. Te crown is labiolingually compressed, slightly tilted mesially, and somewhat higher than long and pointed when unworn. Te lingual side is marked concave, with a thick and obliquely oriented endocrista that runs from the junction of the precrista and lingual cingulum up to the crown apex. Te lingual cingulum is moderately developed and obliquely oriented, ending in a cuspule-like swelling where it meets the thick postcrista, which displays minor crenulations when unworn. Te precrista is thinner than the postcrista and progressively curves until merging with the lingual cingulum. Te cervix displays a moderately developed mesiolingual preanticline and a distolabial endosyncline. Te root is only slightly waisted at the cervix and curves distally toward its apex.

Tree additional upper incisors are here interpreted as I2s. Te best preserved one ( Fig. 3F ), which includes the almost unworn crown and the entire root, was figured as an i3 of L. splendens by Bataller (1924: pl. 5, fig. 9). Te other specimens ( Fig. 3G–H ) only preserve the crown but are similarly worn, showing an entirely comparable morphology. Te crown is smaller and proportionally lower-crowned and more labiolingually compressed than that of the I1. It displays a triangular profile in labial/lingual views, with the pointed apex being located toward the mesial half of the crown. Te labial crown wall is convex, whereas the lingual is concave except at its swollen central portion. Te lingual cingulum is subtle and discontinuous at the level of the lingual swelling, which is V-shaped (apically tapering) but does not constitute a well-defined endocrista. Te crown markedly protrudes from the cervix distally and, especially, mesially. Te precrista and postcrista are straight to mildly convex; the former is slightly shorter and steeper than the latter, which appears slightly crenulated (albeit less so than in the I1). Te cervix shows no marked anticlines and the root is more labiolingually compressed, lingually curved, and distally tilted than in the I1.

A very worn incisor with partial root ( Fig. 3I ), only preserving the cervix on the labial side, is tentatively identified as an I3 because the root is more mesially waisted at the cervix and more distally tilted than in the I2. No occlusal details can be ascertained due to the advanced degree of wear.

Upper canines Te C 1m is represented by two relatively worn specimens ( Fig. 3K–L ), one of them figured by Pickford (2014: fig. 15A). Tey are similar to one another when differences in wear are considered, although one of them ( Fig. 3K ), which shows a more obliquely oriented wear facet ( Fig. 3L ), appears somewhat stouter. Te root is very short compared with the crown, which is largely covered by cementum. Te cervix displays marked endoanticline and ectoanticline. Te crown is somewhat labiolingually compressed and slightly curved distolabially. It displays marked basal protrusions both mesially and distally. Te mesial protrusion, located at the level of the presyncline, appears to be apically continued by a distinct precrista that is largely eroded by wear. Te distal protrusion is more marked and continued by a distinct keel up to the preserved crown apex. In one of the specimens ( Fig. 3L ), marked labial and lingual apicobasal grooves are present on the cementum surface. Te mesial wear facet displays an elliptical to oval shape that more strongly tapers toward the distal side.

A very slightly worn tooth preserving the basal-most portion of the root is here identified as a C1f ( Fig. 3J ). Te crown is labiolingually compressed and reminiscent of that of the mesial upper premolars (see below), except for being labially more bulging and higher crowned. It displays a single main cusp located toward the mesial half of the crown, as well as very distinct mesial and distal styles that considerably protrude from the cervix. Te labial crown wall is very convex except for moderately developed clefts that flank the mesial and distal styles. Te lingual wall is flat and displays multiple vertical grooves directed toward the crown apex. Te lingual cingulum is centrally discontinued and otherwise very subtle except at the level of the prestyle. Tere are two (mesial and distal) roots fused at least at their base.

Upper premolars A single P1 is available from the sample ( Fig. 4A ) because, of the four upper premolars from SQ-TF previously identified by Pickford (2014) as such, three are here considered P2s (Pickford, 2014: fig. 14B, G, I; see our Fig. 4B–C, G ) and the remaining one (Pickford, 2014: fig. 14A) is considered a DP2 (see later). One of these P2s ( Fig. 4G ), which is only partially preserved, was formerly identified by Golpe-Posse (1972) as an I2.

Te crown of the single available P1 is slightly worn but is missing its apex and it only preserves the basal portion of the mesial root. Te crown has an elongate (buccolingually compressed) occlusal contour that becomes markedly convex at the mesial and distal ends, with the latter being lingually twisted. Te crown is moderately high-crowned and displays a rather trenchant appearance, with its single main cusp (paracone) being clearly located on the mesial half of the crown. A slightly sinuous and steep paraprecrista descends from the paracone apex toward the very faint prestyle located on the mesial end of the crown. Tis crest displays minor serrations and shapes a marked concave profile in buccal/lingual views, so that the paracone appears mesially tilted. Te parapostcrista emerges in distal direction from the distal aspect of the paracone and is less steep, but thicker, than the paraprecrista. It displays a distinctly protruding portion visible in buccal/lingual views that does not form a distinct (cuspule-like) metacone. Instead, it is continued by a thinner and distolingually oriented portion of the crest that ends close to the distolingual corner of the crown. Te crown walls are only moderately convex around the paracone, otherwise being slightly concave. Tere are no distinct cingula and the crown only slightly protrudes mesially from the cervix, which displays a marked endoanticline (the ectoanticline cannot be adequately assessed). Te cervix extends much farther rootward distally than mesially, at least on the lingual side. Two roots, fused at least along their basal-most portion, were apparently present originally, with the basal portion of the mesial root being vertically oriented.

Besides the P2s previously described by Pickford (2014) as P1s ( Fig. 4B–C, G ), two new P2s ( Fig. 4D–E ) from SQ-TF and another one from CF2 ( Fig. 4F ) are described here. Four specimens preserve the complete crown, which is only slightly worn ( Fig. 4B–D ) except in one of the specimens ( Fig. 4F ). Te distal root is preserved in two P2s ( Fig. 4D, G ), while the mesial is preserved in a partial specimen ( Fig. 4E ). Te P2s from SQ-TF are larger than but display a similar morphology to the P1 described above, coupled with some differences. Tus, the P2s displays a comparatively lower occlusal relief, a more lingually tilted mesial portion with a more distinct prestyle, and an expanded and/or lingually tilted distal-most portion. Tere is also some variation in the course of the parapostcrista, although it invariably terminates close to the distobuccal end of the crown. Only in a single specimen ( Fig. 4G ), the parapostcrista appears to bifurcate at its distal end. Te two roots are distinct except at their basal-most portion, the mesial one being verticalized and the distal being apparently slightly stouter and either vertical or slightly tilted distally.

Te distal premolars include two P3s ( Fig. 4H–I ) and five P4s ( Fig. 4J–N ) from SQ-TF. Te former display a moderate degree of wear and, in one of the specimens ( Fig. 4H ), the roots are completely preserved. Te P4s include a lingual crown fragment ( Fig. 4J ) and four specimens that completely preserve the crown, which is almost unworn in one specimen ( Fig. 4K ) and moderately worn in the remaining ones. Among the latter, one of the specimens ( Fig. 4L ) preserves the roots socketed in the maxillary bone; the other two ( Fig. 4M–N ), which only preserve the basal-most portion of the roots, were reported by Pickford (2014: fig. 12C) and are probably antimeres of the same individual; the left one is also unambiguously associated with an M1–M3 series (see later). Te two P3s from SQ-TF display a suboval

( Fig. 4H ) to subtriangular ( Fig. 4I ) occlusal contour that is longer than broad and broadest at its distal-most portion, with a markedly convex mesial contour and an almost straight distal one. Te paracone is inflated and centrally located. Te paraprecrista is steep and progressively becomes concave in buccal/lingual views until merging with a well-developed and slightly mesiolingually tilted prestyle at the mesial end of the crown. Te postcrista is short and slightly distobuccally directed, ending in a well-developed poststyle that is higher than the prestyle and located close to the distal end of the crown. Nevertheless, due to wear it is unclear whether a distinct paracone was originally present distally from the paracone. Te prestyle is flanked by moderately developed buccal and lingual clefts, whereas a similarly developed cleft separates the buccal base of the paracone from the poststyle. Tere is no buccal cingulum, whereas a continuous lingual cingulum (albeit narrower at the paracone level) extends from the prestyle until the distolingual corner of the crown, where it becomes ledge-like and bears a poorly developed protocone (basically, a thickening of the lingual cingulum). Tis premolar is three-rotted, with a mesially tilted mesial root and more vertically oriented distal roots that are fused at their base.

Te P4s from SQ-TF can be best ascertained based on an almost unworn crown with no dentine exposure

( Fig. 4K ). Tree additional specimens, which preserve the complete and only moderately worn crown ( Fig. 4L–N ), display a similar occlusal morphology. Te crown displays a suboval to subtriangular and lingually tapering occlusal contour that is much broader than long and longer buccally than lingually. Te lingual half of the crown appears slightly tilted mesially, owing to a more or less marked angulation of the distolingual crown corner. Tis angulation appears related to the position of the protocone, which is not located between the paracone and metacone, but more transversely aligned with the former. Te protocone, which is the most voluminous cusp, is less peripheral than the buccal cusps. Te metacone is subequal in size (lower and smaller) compared with the paracone. Both buccal cusps are closely packed, so that their dentine exposures become confluent at early wear stages. Nevertheless, they are well distinct, with their bases being separated by a transverse groove that runs from the protofossa onto the apical portion of the buccal crown wall. Tis groove is well distinct but narrow in the least worn specimen ( Fig. 4K ), more indistinct in another one ( Fig. 4L ), and broader and cleft-like in the two likely antimeres ( Fig. 4M–N ). Te protofossa is deep and broader distally than mesially, being centrally partly interrupted by the crests radiating from the paracone and protocone apices. Te mesial and distal cingula are well developed, display secondary crenulations of the enamel when unworn, and terminate in well-developed styles at the mesiobuccal and distobuccal corners of the crown. None of the specimens displays a lingual cingulum, but the two likely antimeres display a moderately developed but discontinuous buccal cingulum. Tis premolar is three-rotted, displaying a robust lingual root and two distinct and somewhat diverging lingual roots, which are slenderer and only fused at their basal-most portion.

Upper molars Among the many upper molars included in the sample, only a few are associated: they include an M1–M3 tooth series ( Fig. 5L, T, Z ) associated with the P4 ( Fig. 4N ) of the same individual (of which only the M2 and M3 were figured by Pickford, 2014: fig. 14F, H) and a maxillary fragment with M1–M2 and associated M3 ( Figs. 5K, S, Y , 6A ) that was also figured by Pickford (2014: fig. 15B–C). Despite having different catalog numbers, these specimens likely belong to a single individual based on similarities in size, shape, and wear stage. Te remaining are isolated, either complete or partial, consisting of multiple M1s ( Fig. 5A–J ), M2s ( Fig. 5M–R, U–W ), and M3s

( Fig. 5X , A’–F’), of which only two M1s were previously figured by Pickford (2014: figs. 14E, 16B). Tis sample of upper molars mostly comes from SQ-TF, except for an M1 ( Fig. 5B ) and an M3 ( Fig. 5 F’) from CF2, as well as another M3 ( Fig. 5X ) from SQ-PN. All the molar positions include some lightly worn crowns or unworn germs enabling an adequate description of their occlusal morphology. One of the M1s ( Fig. 5A ) was formerly identified by Golpe-Posse (1971) as an M2.

Te M1s and M2s ( Fig. 5A–W ) show similar proportions and occlusal shape, being characterized by a subrectangular occlusal contour (slightly longer than broad), with two biconvex and distinct lobes, a straighter mesial contour, and a slightly protruding distal one. Te distal lobe is generally slightly narrower and more asymmetric than the mesial (due to a greater distolingual than distobuccal expansion of the crown base). Te four main cusps are pyramidal in shape and display marked crests and Fürchen, with a moderate development of enamel crenulations when unworn. Te lingual cusps are slightly more distally located than the mesial. Te mesial cingulum is well developed and displays a distinct protopreconule at about its median portion, but is continued neither buccally nor lingually. Te transverse valley separating the two lobes is partly interrupted by a centrally located hypopreconule that is more developed than the protopreconule. Te buccal metaectoconule is variably developed (generally small, but sometimes double, completely enclosing the transverse valley), whereas the lingual hypoectoconule is small and slightly more distally located at the end of the hypoectocrista. Te metaectoconule is sometimes continued distally by a variably developed distolingual cingulum that may merge with the distal one. Te latter is at least as well developed as the mesial but more protruding and buccolingually more restricted, displaying some development of secondary enamel cuspules.

Te M3s ( Fig. 5X –F’) are generally lightly to moderately worn, with very restricted dentine exposure (if at all) and conspicuous evidence of Fürchen and some crenulations of the enamel (most evident in the unworn crown germ; Fig. 5 E’). Besides the aforementioned likely antimeres

( Fig. 5Y–Z ), based on occlusal similarities four additional specimens might constitute two additional antimere pairs ( Fig. 5 A’–B’ and Fig. 5 C’–D’) and be associated with some of the preceding molars, although this cannot be conclusively determined. Te M3 is similarly broad to the M2 on the mesial lobe, but much longer, displaying a more elongate and distally tapering, subtriangular, and asymmetric occlusal contour (longer lingually than buccally). Te central lobe is narrower than the mesial and separated from it by moderately to marked buccal and lingual constrictions of the occlusal contour. Te pentacone-bearing distal lobe is much smaller but generally well distinct from the central, being mostly located on the lingual half of the crown and either distally directed or somewhat buccally tilted. Te two mesial cusps are slightly larger and more inflated than the hypocone and the metacone, while the lingual cusps are more distally located than their corresponding buccal counterparts (more clearly so than in the M1 and M2). Te mesial cingulum and the centrally located protopreconule are even better developed than in the preceding molars. Only in a couple of M3s, the mesial cingulum continues along the buccal crown wall forming a buccal cingulum (which is somewhat or totally interrupted at the level of the metacone), whereas there is no distinct lingual cingulum. All the M3s display a distinct hypoectoconule, generally more developed than that of the M2s and that the M3 metaectonocule, which consists of one or more cuspule-like enamel thickenings (integrated in the buccal cingulum when present). Te pentacone is located toward the distal end of the crown and generally well developed (albeit smaller than the other main cusps), whereas the pentapreconule is more indistinct and smaller than the other preconules. Only in some specimens the pentacone is smaller and more indistinct ( Fig. 5 E’), even being preceded by two secondary cusps that might be interpreted as the pentaectoconule and an incipient hexaconule

( Fig. 5 C’–D’).

Lower incisors Te lower incisors are represented in the sample by multiple i1s ( Fig. 7A–F ), i2s ( Fig. 7G–J ), and i3s ( Fig. 7K–O ) from SQ-TF, of which only an i1 ( Fig. 7A ) had previously been figured by Pickford (2014: fig. 13B). Te latter specimen was interpreted by Golpe-Posse (1971) as an I1 (which is probably a typo) and by Pickford (2014) as an i2. Te latter interpretation was probably influenced by the mesially tilted distal margin; however, as noted by Pickford (2014), the distal margin is worn, while a missing crown portion on the mesial side, close to the apex, further contributes to give this tooth a weird appearance. Based on the vertical endocristid and the apicobasal alignment between the crown and the root, we concur with McKenzie et al. (2023a, 2024) that this incisor is better interpreted as an i1.

Some of the remaining i1s included in the sample more adequately enable the description of the unworn crown shape of this incisor ( Fig. 7C–E ), which is characterized by a symmetrical incisal edge with subtle serrations and two (mesial and distal) distinct mamelons (preconulid and postconulid, respectively, with no protoconid in between). Te crown is tall relative to its basal dimensions, which are broader than long but progressively become labiolingually compressed toward the apex. Te labial wall is uniformly convex, whereas the lingual side is concave and displays a well-developed endocristid. Te prestylid and poststylid are thick and similarly well developed, whereas the endocristid is very vertically aligned (only minimally mesially tilted), very thick, and lingually protruding, markedly tapering along its apical-most portion until fading away at or close to the unworn incisal edge. Te prefossid and endofossid are thus well separated by the endocristid, with the former becoming indistinct slightly farther away from the cervix than the endofossid. Te endocristid originates from a moderately developed swelling of the lingual crown base, located above the marked endosynclinid. Te preanticlinid and postanticlinid are deep and similarly developed to one another, variously pointed depending on the specimen. However, in all cases they are markedly asymmetric because the ectosynclinid extends much farther rootward than the endosynclinid. Te crown base is not markedly waisted at the cervix and the root is vertically aligned with the crown and mesiodistally compressed, although slenderer in the newly reported specimen that partially preserves the crown ( Fig. 7B ) than in the previously published specimen ( Fig. 7A ).

Te i2s available in the sample are less completely preserved than the i1s (only the basal-most portion of the root is preserved in some specimens; Fig. 7H, J ) but appear quite similar in size and proportions, merely being somewhat larger and slightly more mesiodistally compressed at the base, but similarly high-crowned. Despite overall similarities, the i2 can be distinguished from the i1 by the more mesially tilted crown relative to the root. Furthermore, the unworn ( Fig. 7G ) and slightly worn ( Fig. 7J ) incisal edge is constituted by two distinct mamelons that, unlike those of the i1, are markedly asymmetrical, with the postconulid being lower than the preconulid. Other differences of the i2 relative to the i1 include a slightly more mesially tilted endocristid and a more curved poststylid—resulting in an overall more asymmetrical crown and a more inclined incisal edge (even when unworn)—and slightly shallower and less asymmetric anticlinids (as the ectosynclinid is less extended onto the root).

Te i3 is represented by five specimens ( Fig. 7K–O ). Tree of these incisors ( Fig. 7K, M, O ) were originally interpreted as I2s by Golpe-Posse (1971), some of them being subsequently identified as an i3 ( Fig. 7K ) and as an I3 ( Fig. 7O ) by Pickford (2014). Most recently, McKenzie et al. (2024) concurred with Pickford (2014) that the former specimen is an i3 and mentioned another (previously unpublished) specimen ( Fig. 7L ) that similarly preserves a lightly worn crown and most of the root. Te remaining specimen is severely worn ( Fig. 7M ) but may be identified as an i3 based on the morphology of the root and basal crown portion, whereas the two crown germs reported here ( Fig. 7N–O ) are consistent in crown shape and proportions with the more complete specimens, except for being slightly more labiolingually compressed. An alternative identification of these two crown germs as di3s is not impossible, given that this tooth locus is unknown for ‘ Pa. ’ valentini , but is unlikely given that they are no more brachyodont or markedly smaller than the remaining i3s. Te i3 markedly differs in shape and proportions from the i1 and i2, being smaller overall, generally more mesiodistally compressed, and much lower-crowned, further displaying a more asymmetric and mesially tilted crown. Te short prestylid and the longer poststylid are curved and confluent at the mesially located apex of the crown, which is markedly convex and lacks distinct mamelons. A sharp and thick endocristid originates from a marked and distolingually located basal swelling. Te endocristid is only slightly tilted mesially and generally terminates at or close to the apical portion of the postcristid, separating the narrow endofossid from the more spacious prefossid. A less distinct secondary cristid is present in some specimens in the prefossid, running from the prestylid toward the crown apex. Te crown is somewhat waisted at the cervix, which shapes a distinct and moderately deep preanticlinid, no discernible postanticlinid, and similarly developed and mesially tilted ectosynclinid and endosynclinid. Te root is mesiodistally compressed and progressively tapers toward its apical portion, which is mesially curved.

Lower canines Te three c1ms available from the whole sample were all reported by Bataller (1924: pl. 5, figs. 1–3), who attributed them to L. splendens . Tey are registered at the MGSB with a single catalog number despite belonging to more than a single individual, being thus distinguished here by the addition of an old collection number within brackets. Two of these specimens ( Fig. 8A–B ) are only partially preserved, lacking their apical-most portion, although one of them is only missing the tip and preserves a long and oblique wear facet on the distal face of the tooth along half of its preserved length ( Fig. 8A ). Te third specimen ( Fig. 8C ) is completely preserved, albeit somewhat damaged at its basal-most portion, further displaying a small wear facet restricted to the apical-most portion of the distal side of the crown. Te c 1m is very hypsodont (preserved chord length in the best-preserved specimen is 9.7 cm) and ever-growing (thus lacking a distinct root). From base to apex, the crown is very curved distalward and slightly curved labially, only minimally tapering toward the apex, which appears pointed due to wear on the distal side. Te cross section is scrofic 1 and very asymmetric, with the labial side being the narrowest, the lingual side being somewhat broader than the distal (10–24%) and much broader than the labial (63–79%), and the distal side being also clearly broader than the labial (33–63%). Te distal side is devoid of enamel and rather flat to mildly convex, so that no bump attributable to any cementum ridge can be discerned. Te labial side is moderately

1 Two main morphologies of of c 1m are distinguished in suids: in scrofic canines, the lingual side is the broadest and the labial is the narrowest, whereas in verrucosic canines the labial and lingual sides are similarly broad and broader than the distal (Cherin et al., 2020). Terefore , in scrofic canines the narrowest side is the labial, whereas in verrucosic canines it is the distal ( Van der Made et al., 1999) .

concave, whereas the lingual is flatter except toward its mesial-most portion, where a shallow but distinct sulcus runs along the whole height of the crown, contributing to delineate a broad and markedly convex mesial precristid. Fine longitudinal striations of the enamel can be discerned along the labial and lingual sides, being particularly evident on the lingual side of one of the specimens ( Fig. 8A ). Another specimen ( Fig. 8B ) shows more marked transverse striations at certain portions of the crown that correspond to enamel hypoplasias. Te third specimen ( Fig. 8C ), in contrast, displays narrow, enamelfree bands on the basal-most portion of the lingual side, of which one is apically continued throughout the whole height of the crown; these narrow bands do not appear to result from taphonomical damage, but it is unclear whether they may be qualified as pathological (albeit they are lacking in the other specimens).

Te single c1f ( Fig. 8D ) similarly comes from SQ-TF but remained unpublished until now. It completely preserves the very slightly worn crown as well as most of the root (except for its apical-most portion). Both the crown and the root are markedly compressed labiolingually. Te crown is pointed and distally recurved, with its apex being aligned with the distal margin of the crown and the root. Te labial and mesial crown walls are markedly convex, whereas the labial crown wall is flatter, being separated from the distal side by a very thick, blunt, and distolingually oriented postcristid that runs from the crown apex toward its base. Together with the more indistinct and distolabially oriented ectocristid, the postcristid delimits a shallow distolabial basin on the lower half of the crown. Te root is not markedly waisted at the cervix, which is somewhat irregular labially (with no distinct ectosynclinid) and displays a moderately developed postsynclinid distally, a shallow endoanticlinid lingually, and a moderately developed mesiolabial preanticlinid. Te root is labiolingually compressed, distally curved (especially on its mesial side), and moderately apically tapering, and if complete it would have been somewhat higher than the crown.

Lower premolars Te available sample of lower premolars is not very abundant but records the four premolar loci, including the p1 ( Fig. 9A–B ), the p2 ( Fig. 9C–F ), the p3 ( Fig. 9G–I ), and the p4 ( Fig. 9J–K ). Only a single p1 ( Fig. 9A ) from SQ-TF was previously figured by Pickford (2014: fig. 14D). Te p1s preserve the complete and unworn crowns but only the basal-most portion of the two roots ( Fig. 9A–B ), whereas the p2 is represented by a complete and only slightly worn specimen ( Fig. 9C ), as well as a partial one ( Fig. 9D ) and two (mesial and distal) fragments ( Fig. 9E–F ). Te latter specimens currently have two different catalog numbers but originally constituted a single, complete tooth that Bataller (1924: pl. 11, fig. 18) attributed to an indeterminate carnivoran.

Te two mesial premolar positions are elongate (very buccolingually compressed) and moderately trenchant but low-crowned, with the pointed protoconid being located on the mesial half of the crown. Te p1

( Fig. 9A–B ) displays a subelliptical contour with a distinct constriction on the lingual side at the level of the protoconid, a markedly convex mesial end, and a slightly distolingually tilted distal portion. Te crown walls are moderately convex around the protoconid, which is very buccolingually compressed, and rather concave distally from it. A blunt, slightly lingually tilted, and very steep protoprecristid descends from the protoconid apex toward a low but moderately well-developed mesial stylid. Apically, this protoprecristid displays a moderately convex profile in lingual/buccal views, basally becoming convex before reaching the prestylid. Te protopostcrista is less inclined than the protoprecrista and displays some serrations and cuspule-like developments until bifurcating into two distinct cristids. Te buccal one progressively curves until reaching the distal end of the crown and, along its midway, displays a buccolingually compressed hypoconid that is slightly higher than the prestylid. In turn, the lingual cristid is directed toward the distolingual crown margin and, together with the former cristid, delimits a distinct but narrow and distally open fossid. A moderately developed buccal cleft is present at the level of the protopostcristid. Te cervix displays moderately developed anticlinids and the crown extends farther rootward distally than mesially on both the buccal and lingual sides. Although the roots are not preserved, it can be ascertained that two distinct roots were originally present.

Te p2s ( Fig. 9C–F ) are similar in shape to the p1s (subelliptical and distolingually tilted occlusal contour, buccolingually compressed and mesially located protoconid, steeply inclined protoprecristid ending in a low and moderately developed prestylid, and cervix extending farther rootward distally than mesially on both sides). However, they are larger and further display some occlusal differences: the protoprecristid displays a straighter profile and is more buccally tilted than in the p1; the protopostcristid does not bifurcate and is obliquely oriented toward the distobuccal end of the crown, where it ends on a rather indistinct hypoconid. Like the p1, the p2 displays two distinct roots, the mesial being stouter and more mesially tilted than the distal one.

Te distal premolars are represented by a relatively complete p3 that preserves the damaged crown and most of the roots ( Fig. 9G ), two more partial p3s ( Fig. 9H–I ), and two slightly worn p4s ( Fig. 9J–K ) of which one preserves most of the roots. None of these specimens had previously been figured, although Pickford (2014) reported one of the p3s ( Fig. 9H ). Te most complete p3

( Fig. 9G ) displays a moderately elongate subelliptical contour, with a mild lingual constriction at about the level of the protoconid as well as markedly convex and slightly lingually tilted mesial and distal ends. One of the partial p3s ( Fig. 9I ) displays an even narrower occlusal contour on the mesial half of the crown, whereas the other

( Fig. 9H ) appears broader and suggests an original oval (i.e., distally expanded) rather than subelliptical contour. All the specimens display a thick, straight, and steeply inclined protoprecristid ending on a low and moderately developed mesial prestylid. In the more complete p3, the protopostcristid, albeit considerably worn, is straight and mesiodistally aligned, ending in a rather indistinct hypoconid close to the distal end of the crown. Te lingual crown walls are only convex around the protoconid and otherwise concave, with mildly developed lingual and buccal clefts at the protopostcristid level. Tis premolar displays three roots, the distal ones being slenderer but as mesially tilted as the mesial.

Te p4s ( Fig. 9J–K ) are about as long as the more complete p3 but much stouter (relatively broader), displaying a suboval contour that is only minimally broader distally than mesially and very rounded overall, with only a very subtle lingual constriction at about crown midlength. Te protoconid is centrally located, tall, and pointed, with inflated buccal and lingual walls. Te protoprecristid is very thick, slightly lingually tilted, and as steeply inclined as in the p3. It descends from the protoconid apex to the mesial stylid, which is somewhat higher and better developed than in the p3. Te prestylid is flanked by distinct cingular developments that are restricted to the mesial crown margin, being separated from the base of the protoconid by narrow but marked clefts. In one of the specimens ( Fig. 9K ), the somewhat worn protopostcristid is distolingually oriented and almost as thick as the protoprecristid, albeit shorter. In the other, almost unworn p4

( Fig. 9J ), a smaller but distinct cuspulid (metaconid) can be still distinguished mesiolingually from the protoconid. In both specimens, the short protopostcristid descends distolingually until reaching a trefoil-like structure located at the distal portion of the crown. Tis structure, which consists of a distal cuspulid-like portion and two transverse (buccal and lingual) cristid-like extensions, may be globally interpreted as the hypoconid. It is higher than the prestylid and separated from the protoconid and metaconid by deep and wide clefts at both sides of the protopostcristid (the buccal one being slightly better developed than the lingual). A marked transverse groove runs along the lowest portion of both clefts, even partially interrupting the connection between the end of the protopostcristid and the hypoconid. Tese clefts display no cingulids, which are entirely restricted to the distal crown margin at both sides of the hypoconulid. Te p4 also displays three roots, which are similar but larger than those of the p3.

Lower molars Although lower molars are well represented in the sample, only a few are associated. Tese include those socketed in two mandibular fragments from SQ-TF, one with m1–m3 ( Figs. 6C , 10C, M, W ) and the other with m2–m3 ( Figs. 6D , 10L, V ), plus the two m1 antimeres of a juvenile mandible from SQ-PN ( Figs. 6B , 10F–G ); an m2 mesial fragment ( Fig. 10R ) and partial m3

( Fig. 10Y ) that, according to collection data, come from SQ-TFC, might also belong to the same individual. Te remaining lower molars, either complete or partial, come from SQ-TF and consist of isolated specimens, including m1s ( Fig. 10A, B, D, E, H, I ), m2s ( Fig. 10J–K, N–Q, S, T ), and m3s ( Fig. 10U, X, Z –B’)—note that these are our preferred anatomical identifications, but due to size variation among individuals, the smaller isolated m2s might be alternatively identified as m1s of large individuals, or vice versa. Several of the lower molars housed in the MGSB

( Figs. 10B, E, I, J, P, U, Z ) were figured by Bataller (1924: pl. 6, figs. 1–9, 11), who misidentified some of them as upper molars and assigned them all to S. palaeochoerus . In turn, Pickford (2014: fig. 16E–F) figured two of the lower molars ( Fig. 10A, K ). For all lower molar positions, there are complete unworn and lightly worn specimens enabling the adequate description of their occlusal morphology. Te least worn specimens ( Fig. 10A, J, K, N, U, Z ) display very marked Fürchen and considerable development of enamel wrinkling. Te latter becomes easily eroded by wear even at minimal dentine exposure, and enamel appears thick in more worn specimens.

One of the specimens identified here as an m1

( Fig. 10A ) was previously identified as an m2 by Golpe-Posse (1972) and Pickford (2014), although this is debatable because these two molar positions display a similar occlusal morphology and considerably overlap in size. As a result, only those molars belonging to mandibular fragments ( Fig. 6B–D ) can be unambiguously identified as either m1s ( Fig. 10C, F, G ) or m2s ( Fig. 10L–M ), indicating that the latter is somewhat longer and broader than the former. Tis distinction does not apply when interindividual variation is considered, as one of the unequivocal m1s ( Fig. 10C ) is about the same size as one of the m2s ( Fig. 10L ). Otherwise, the specimens identified here as m1s ( Fig. 10A–I ) tend to taper mesially, whereas m2s

( Fig. 10J–P ) tend to taper distally, although with some exceptions. Both molar positions display a subrectangular occlusal contour with two (mesial and distal lobes), separated by buccolingual constrictions that are more marked buccally than lingually. Te four main cuspids are pyramidal in shape and transversely aligned in pairs. Te protoprecristid is normally thick and curves along the mesial portion of the crown toward the end of the shorter and lesser developed metaectocristid and metaprecristid, sometimes merging with the former, more rarely with the latter, and in other instances being separated from them both by a groove. A beaded and normally continuous mesial cingulid is present in all the specimens, being more extensive and marked along the mesiobuccal corner of the crown. Te protoendocristid bifurcates into two distinct and short cristids, the mesial one being directed toward the similarly developed metaprecristid, partly obliterating the protofossid but being separated from one another by a longitudinal groove. Te distal portion of the protoendocristid and the metaendocristid are obliquely oriented and distally enclose the protofossid. Te metaendocristid ends in a rather distinct metaendoconulid, which is located mesiolingually from the larger hypopreconulid. Te obliquely oriented protopostcristid and the more distally oriented metapostcristid end at the transverse valley that separates the two lobes and that is largely interrupted by a large, centrally located, and often buccolingually expanded hypopreconulid. Te transverse valley is lingually open with a single exception ( Fig. 10N ), in which a small entoendoconulid is present lingually from the hypopreconulid. In contrast, the transverse valley is buccally enclosed by a cingulid that is not continuous with the mesial cingulid and displays a markedly beaded appearance (with multiple enamel thickenings that generally do not constitute a distinct hypoectoconulid). Te two distal cuspids are mesiodistally aligned with their mesial counterparts and display a symmetrical cristid pattern, being separated by a very narrow (groovelike) hypofossid. Te latter is distally enclosed by a well-developed and distally projecting pentaconid, which partly overlaps the distal cingulid and is slightly tilted lingually relative to the hypopreconulid.

Te m3s ( Fig. 10U –B’) resemble the other lower molars in occlusal morphology but differ in size, occlusal contour, and the greater development of the distal-most (pentaconid-bearing) portion of the crown, which constitutes a distinct third lobe. In the two individuals in which the m3s are associated with other molars ( Fig. 6C–D ), the m3 is similarly broad but much longer than the m2. As in the other lower molars, the mesial lobe of the m3 is separated from the central lobe by a more marked buccal than lingual constriction of the occlusal contour, whereas this is not the case regarding the distal lobe. Tere is some variation in the degree of distal tapering: in one of the m3s ( Fig. 10U ), the central lobe is only moderately narrower than the mesial and the distal tapers more abruptly; in contrast, other specimens ( Fig. 10W– X ) taper distally more progressively, resulting in a more triangular contour. Otherwise, the occlusal shape of the mesial and central lobes is entirely comparable to the two lobes of the m1 and m2, and will not be reiterated here. Te distal lobe, in contrast, displays a distinct pentapreconulid that is mesiodistally aligned with the larger hypopreconulid, as well as a large pentaconid that is located at the distal end of the crown and smaller than the remaining four main cuspids. Te pentaconid is linked to the pentapreconulid by a short and generally thick pentaprecristid, whereas the pentaectocristid displays several enamel thickenings but no distinct pentaectoconulid. One or two secondary cuspulids may also be present lingually from the pentaconid and pentaprecristid. In all the specimens, the distal lobe is distally directed or only buccally tilted very slightly.

Upper deciduous teeth All the upper deciduous tooth positions are represented in the sample. None of the upper deciduous incisors had previously been figured, although two of the ones here identified DI1s ( Fig. 11A, E ) were formerly reported by Golpe-Posse (1971) as I3s. In total, the described sample includes five DI1s

( Fig. 11A–E ), a single DI2 ( Fig. 11F ), and a single DI3

( Fig. 11G ); the identification of the two latter incisors is tentative, as they display a similar morphology, although the one identified as a DI2 is larger and higher-crowned than the DI3.

Te DI1s ( Fig. 11A–E ) are smaller but similar in shape to the I1s, except that the former display a more labiolingually compressed crown and a somewhat slenderer root. Te mesiolabial crown wall is convex, whereas the lingual portion is slightly concave except at its base. Te precrista and postcrista are moderately marked and progressively converge to configure a rounded incisal edge. Te lingual cingulum is subtle and crown base below it does not appear swollen. An obliquely oriented endocrista (double in one of the specimens; Fig. 11C ) originates from the cingulum and fades away at about crown midheight, thus only partly separating the prefossa from the endofossa. Te cervix shapes a moderately deep preanticline mesiolingually and a more indistinct endosyncline distally. Te root is much higher than the crown and moderately compressed labiolingually at its base, progressively curving distally and tapering apically, with the apex further being slightly tilted lingually. Te crown is only very slightly waisted at the cervix, particularly on the distal end.

Te upper incisors interpreted as DI2 ( Fig. 11F ) and DI3 ( Fig. 11G ) display a similar morphology, with the crown being much more labiolingually compressed and more mesially and distally protruding from the cervix than in the DI1. Te DI2 resembles the I 2 in displaying a triangular profile in labial/lingual views, as well as a convex but centrally inflated lingual crown wall that displays an apically tapering but rather indistinct endocrista. However, the DI2 differs from the I 2 in being smaller, lower-crowned, and more asymmetrical (with the apex more clearly positioned on the mesial half of the crown, and a much longer and less steeply inclined postcrista that displays a concave rather than convex profile in labial/lingual views), and further displaying a better developed prestyle at the base of the precrista but an even more indistinct lingual cingulum). Te DI3 shows essentially the same morphology as the DI2 but is lower-crowned and more mesially protruding from the cervix, and it further displays somewhat more distinct lingual cingulum and endocrista (the latter delimiting a more spacious and concave endofossa). Te DI3 preserves most of the root, which is labiolingually compressed, markedly waisted at the cervix, and very distally tilted.

Te upper deciduous premolars include two DP2s

( Fig. 11H–I ), of which one was previously identified as a P2 by Golpe-Posse (1971) and Pickford (2014: fig. 14A); two DP3s, including a complete one ( Fig. 11J ) previously reported but not figured by Pickford (2014) as well as a mesial crown fragment ( Fig. 11K ); and four DP4s

( Fig. 11L–O ), two of them previously reported (and one figured) by Pickford (2014: fig. 16A). Te two DP2s are very lightly worn and display an occlusal morphology entirely comparable to one another—somewhat intermediate between those of the P1 and the P2, but differing from these permanent loci in the less mesially located paracone, the more asymmetrical occlusal contour, and the lower occlusal relief. Te occlusal contour of the DP2 is elongate and vaguely comma-shaped, owing to the mesiolingually directed prestyle and the lingually tilted distal third of the crown. Te prestyle is well developed and surrounded by cingular developments that do not extend either lingually or buccally. Te main cusp (paracone) is quite centrally located (albeit slightly toward the mesial half of the crown), pointed, and labiolingually compressed. Two similarly steep crests originate from the paracone apex: the paraprecrista is mesiodistally aligned except on its mesial-most third, which curves toward the prestyle; the parapostcrista, in turn, displays minor cuspule-like thickenings and a greater degree of curvature, being initially distobuccally directed and subsequently distolingually oriented until merging with the moderately well-developed and crest-like metacone. Te latter is well distinct in lingual/buccal views and approximately as high as the prestyle. Te buccal crown wall is moderately convex, slightly bulging at the paracone level and basally protruding at the level of the metacone. Te lingual crown wall, in contrast, is concave except at the level of the paracone, forming a well-developed basinlike structure at the protruding distolingual corner of the crown, which is surrounded by a distinct cingulum. Te cervix extends farther rootward on the distal than on the mesial half of the crown, but less markedly so than in the P1 and P2. Tis premolar displays two distinct roots that are only merged at the base: the mesial is vertically directed, whereas the distal is stouter and distally tilted.

Te DP3 ( Fig. 11J–K ) displays a subtriangular (longer than wide) occlusal contour with marked buccal and lingual constrictions at about crown midlength, which define two distinct lobes. Te mesial lobe is somewhat elongate and markedly convex mesially, whereas the distal lobe is much wider than the mesial and also more asymmetrical, as a result of a more marked distolingual than distobuccal crown expansion. Tere is a moderately developed prestyle on the mesial end of the crown, continued along the buccal and lingual walls by subtle cingula that fade away before reaching the distal lobe. Te paracone is centrally located on the mesial lobe and pyramidal in size. Tree main cristids originate from its apex: a marked and mesiolingually directed paraprecrista, which ends at the lingual cingulum, close to the prestyle; and the more tenuous but similarly obliquely oriented paraendocrista and parapostcrista (which are distolingually and distobuccally directed, respectively). A secondary and obliquely oriented cristid is also present mesiobuccally from the paracone, although it is shorter than the paraprecrista and, unlike the latter, does not reach the paracone apex. Te paraendocrista and parapostcrista reach the transverse valley and are continued into the distal lobe by the ectoprotocrista and ectometacrista, respectively. Te distal cusps are similarly developed and close to one another (i.e., not very peripheral). Te distal cristids that originate from them are discernible but shorter and less well-developed than the mesial ones, although the protoendocrista seemingly reaches the moderately developed distal cingulum. Te latter is not continued either buccally or lingually. Te whole occlusal surface displays moderately abundant enamel wrinkling.

Te DP4s ( Fig. 11L–O ) resemble in general occlusal pattern the M1s and M2s (see above), including a subrectangular occlusal contour that is longer than broad, with two distinct lobes, four main cusps (with the buccal ones being more mesially located than their lingual counterparts), and well-developed mesial and distal cingula. However, besides being smaller and lower-crowned, the DP4s further display some differences characteristic of this tooth locus that enable their distinction from the permanent molars, namely: the straight and markedly oblique mesial contour; the slightly less peripheral distal cusps, which appear closer to one another than the mesial ones; and the lesser developed (almost inconspicuous) protopreconule and metaectoconule (whereas, in contrast, the hypopreconule is as developed as in the permanent upper molars).

Lower deciduous teeth Te lower deciduous teeth from the sample record all tooth positions except for the di3 (unless MGSB48863 and IPS31060, here identified as i3s, are indeed deciduous, which we consider unlikely). Te di1 is represented by three specimens ( Fig. 11P–R ), two of which ( Fig. 11Q–R ) were previously identified as permanent i1s by Golpe-Posse (1971). In turn, the di2 is represented by two specimens ( Fig. 11S–T ), one of them

( Fig. 11S ) originally considered a permanent i2 by Golpe-Posse (1971). Te shape of the unworn incisal edge of the di1 cannot be ascertained, but the general shape of this deciduous incisor ( Fig. 11P–R ) resembles that of the i1 at a smaller size (see above), being both characterized by a relatively high crown that is vertically aligned with the root, as well as a distinct endocristid that is only slightly tilted mesially and originates from a moderately developed basal bulge. Te most remarkable differences refer to the less marked difference in rootward extension of the ectosynclinid as compared with the endosynclinid, as well as the more marked waisting of the root base below the cervix in the di1. Te di2 ( Fig. 11S–T ) differs from the di 1 in the markedly mesial tilting of the crown and endocristid, with the distal margin of the crown displaying a rather continuous curvature with the root across the cervix. In these regards, the differences between di2 and di1 parallel those between their permanent counterparts at a smaller size. Nevertheless, the di2 more markedly differs from the i2 by displaying a slenderer crown as well as a more marked distal curvature of the crown and root. Te two di2s from SQ-TF display a similar shape, characterized by a very broad endocristid that originates from a moderately developed basal bulge and progressively tapers apically, narrow prefossid and endofossid, a preanticlinid somewhat deeper than the postanticlinid, and a moderately mesiodistally compressed and lingually curved root. However, these two specimens further denote some variation in this tooth locus, as one of them

( Fig. 11S ) displays a mildly convex postcristid, whereas the other ( Fig. 11T ) displays a conspicuously concave poststylid that runs in parallel to the profile of the prestylid in lingual/labial views.

With regard to the deciduous lower premolars, the dp2 and the dp3 are only represented by a partial germ

( Fig. 11U ) and a distal crown fragment ( Fig. 11V ), respectively, whereas the dp4 is represented by three complete crowns ( Fig. 11W–X, Z ) and two mesial fragments

( Fig. 11 A’–B’) from SQ-TF, plus the socketed specimen of the juvenile mandible ( Figs. 6B , 11Y ) from SQ-PN. Te distal end of the dp2 is broken away ( Fig. 11U ) but it would have originally displayed a subelliptical or suboval occlusal contour. Te protoconid is centrally located and, despite the apparent lack of wear, displays a rounded morphology, owing to the convex profile of both the protoprecristid and protopostcristid in buccal/lingual views. Te protoprecristid is long, steeply inclined, blunt, and sinuous, as it originates from the mesiolingual aspect of the protoconid apex and then progressively curves in mesiobuccal direction until reaching a well-developed prestylid at the mesial end of the crown. Te protopostcristid is as developed and inclined as the protoprecristid but shorter, ending at a well-developed hypoconid that is much higher than the prestylid, mesiodistally aligned with the protoconid, and distally continued by a straight cristid aligned with but less steeply inclined than the protopostcristid. Tere is no metaconid distally from the protoconid. Te buccal crown wall is mildly convex and the lingual a bit flatter, except at the level of the prestylid and, especially, the protopostcristid, where a marked buccal cleft and a shallower lingual one flank the mesial aspect of the hypoconulid. No buccal or lingual cingulids can be discerned on the preserved portion of the crown.

Te dp3 ( Fig. 11V ) is more incompletely preserved than the dp2. It resembles the morphology of the p3 at a smaller size, being either interpretable as a L mesial fragment or a R distal fragment, although the lack of a distinct stylid on the preserved end of the crown favors the latter option. Te dp3 appears larger and higher crowned than the dp2, with a steeply inclined and only slightly distobuccally curved protopostcristid that reaches the distal end of the crown without displaying a distinct metaconid or hypoconid. Te crown base appears distolingually protruding, while the crown walls appear convex at the protoconid level and concave along the protopostcristid, particularly on the buccal side, although no conspicuous cleft can be discerned.

Te dp4 ( Fig. 11W –B’) has the trilobate morphology characteristic of this tooth locus, with an elongate and slightly mesially tapering occlusal contour, a moderately rounded mesial end, and a slightly projecting distal margin. Te buccal crown base configures more conspicuous convexities (and, hence, deeper constrictions between consecutive lobes) than the lingual. Te six main cuspids are pyramidal in shape, with sharp cristids and deep Fürchen, and are transversely aligned in pairs. On the mesial end of the crown, a thick and obliquely oriented paraprecristid is directed toward the primoprecristid, terminating in some specimens in a small and indistinct parapreconulid. Mesially from it, a buccolingually restricted and non-projecting mesial cingulid may be discerned. Te paraendoconulid, protoendoconulid, and hypopreconulid are poorly developed and smaller than the pentaconid. Te latter is distinct but lower and less extensive than the six main cuspids, being centrally located (even if slightly buccally tilted) on the projecting and rounded distal cingulid. Buccal and lingual cingulids are only mildly developed along the distal lobe. Minor crenulations of the enamel can still be discerned in some specimens, particularly on the cristids and cingulids.

Comparisons

Upper incisors Te size variation displayed by the I1s from SQ-TF largely encompasses that displayed by the I1s from CB attributed to the same taxon (Additional file 1: Table S2). In contrast, they are larger on average than the I1s of V. steinheimensis (albeit with considerable overlap) and more clearly larger than those of C. simorrensis and R. matritensis . In terms of shape (Additional file 1: Table S3), comparisons of the lingual morphology to other I1s of ‘ Pa. ’ valentini are restricted to a single specimen from CB reported by McKenzie et al. (2024), which displays a less distinct endocrista—however, an unambiguous assignment of this incisor (IPS1749) to ‘ Pa. ’ valentini is not possible, as V. steinheimensis is also recorded there. Te figured I1 of V. steinheimensis from Gratkorn is also too worn to ascertain the lingual morphology, although overall it appears lower-crowned than those of ‘ Pa. ’ valentini at comparable wear stages. Te I1s from Göriach and elsewhere attributed to R. matritensis display a more lingually tilted crown than those of ‘ Pa. ’ valentini , but otherwise are similar in terms of crown height and endocrista development.

Te I2s from SQ-TF superficially resemble those of L. splendens , which display overlapping dimensions (even though the latter are larger on average; Van der Made, 1996b: pl. 36, figs. 9–10, 13). However, the described specimens display multiple morphological details that discount this alternative attribution: the higher, more labiolingually compressed, and less lingually recurved crown; the more asymmetrical profile in labial/lingual views (the tip is more mesially located); and the more diffuse endocrista and much fainter lingual cingulum. Accordingly, they are assigned to ‘ Pa. ’ valentini . Tese I2s are larger, more elongate, and lower-crowned than two incisors from CB (IPS92710 and IPS92857) identified by McKenzie et al. (2023a: fig. 3i–j) as I2s but here reinterpreted as DI1s of ‘ Pa. ’ valentini (see below). We therefore conclude that the specimens described here are the first I2s reported for ‘ Pa. ’ valentini . Tey closely resemble in size and proportions (Additional file 1: Table S2) the I2s of V. steinheimensis , whereas those of R. matritensis from Göriach are somewhat smaller (albeit with some overlap). In occlusal shape (Additional file 1: Table S3), the described specimens of ‘ Pa. ’ valentini differ from those of V. steinheimensis in the less marked lingual cingulum and more labiolingually compressed root, as well as from those of R. matritensis in the more indistinct endocrista and lingual cingulum. Te I2s from SQ-TF are also similar in size and proportions to the incisor from the same site tentatively identified here as an I3, being larger on average than the I3s of V. steinheimensis and R. matritensis (Additional file 1: Table S2). It is noteworthy that the moderately worn I3 of V. steinheimensis from Gratkorn displays a moderately distinct endocrista, more similar to that displayed by the I2s of ‘ Pa. ’ valentini (not ascertainable in the I3 due to wear) and the I3 from Göriach than the I2s from the latter site (Additional file 1: Table S3).

Upper canines One of the C1ms from SQ-TF described here ( Fig. 3L ) was previously reported by Pickford (2014: p. 187), who qualified it as “extraordinary”. Te new specimen described here ( Fig. 3K ) shows some minor differences (it is relatively broader labiolingually, no clear labial or lingual grooves can be discerned on the cementum surface, and the mesial wear facet against the c 1m is more steeply inclined). However, it generally agrees in morphology, confirming the main features highlighted by Pickford (2014): oval cross section, marked mesial and distal crests overhanging the small root, and deep lingual and labial anticlines. Te C1ms from SQ-TF overlap in mesiodistal length with two specimens from Kleineisenbach attributed to ‘ Pa. ’ valentini by Pickford (2016) but are labiolingually less compressed, unlike those of V. steinheimensis and R. matritensis , which display similar proportions but are comparatively smaller (more markedly so those of the latter species; Additional file 1: Table S2). In terms of shape (Additional file 1: Table S3), the two C1ms from SQ-TF closely resemble those from Kleineisenbach attributed to ‘ Pa. ’ valentini . Tey differ from those of V. steinheimensis and R. matritensis in displaying a shorter crown with a more distinct (albeit short) root and marked mesial and distal protrusions of the crown base, as well as a more marked distal keel.

Te specimen from SQ-TF identified as a C1f of ‘ Pa. ’ valentini markedly differs from the C1fs from CCN20 (IPS106329) and CB (IPS1715 and IPS93150) that were attributed to this species by McKenzie et al. (2023a: fig. 3d, 2024: fig. 4e–f). Te latter are much higher-crowned and display a single and stouter root, thus more closely resembling the C1fs of Pr. palaeochoerus from CB ( McKenzie et al., 2024: fig. 4a–d) and elsewhere ( Van der Made et al., 1999: pl. 2, fig. 4; McKenzie et al., 2023a: fig. 3b–c; Alba et al., 2024: fig. 6b). In contrast, the C1f from SQ-TF displays tetraconodontine affinities (see below). For this reason, the aforementioned specimens from CCN20 and CB are here reassigned to Pr. palaeochoerus , implying that the SQ-TF specimen is the first described C1f of ‘ Pa. ’ valentini . In size and proportions, this specimen resembles the C1fs of C. simorrensis from Przeworno (which are only slightly longer) as well as those of R. matritensis from Göriach—the C1f of V. steinheimensis being, to our knowledge, unknown (unless the Przeworno sample is assigned to this species). In occlusal shape, the C1f from SQ-TF resembles those of C. simorrensis and R. matritensis in being double-rooted and displaying a premolar-like crown, i.e., elongate (relative to breadth) and lower-crowned (relative to length). Te specimen described here more closely resembles the C1f of C. simorrensis from Przeworno in displaying strong mesial and distal styles, although further similarities are difficult to ascertain because these specimens were only figured in labial view. However, the C1f from SQ-TF appears to display a more mesially tilted apex, with the precrista and postcrista being straighter in labial view (instead of moderately convex and concave, respectively), as well as a deeper ectoanticline. Te crown base also appears more protruding mesially and distally, and the degree of fusion of the two roots might have been greater (as suggested by the more vertical orientation of the fused basal portions of the roots), although the latter cannot be conclusively ascertained due to incomplete preservation. In the C1f of R. matritensis , the styles are much weaker and the occlusal contour is different from that of the SQ-TF specimen (more uniformly convex labially and less flat lingually, being swollen at the level of the main cusp).

Upper premolars Te single P1 from SQ-TF is similar in size and proportions to those of ‘ Pa. ’ valentini from CB ( Fig. 12A ; Additional file 1: Table S2), whereas those of V. steinheimensis and, especially, C. simorrensis and R. matritensis are larger and relatively broader on average. Te shape of the SQ-TF specimen also most closely resembles the two P1s of ‘ Pa. ’ valentini previously reported by McKenzie et al. (2024) from CB, except for a few minor occlusal details (fainter prestyle and less distinct metacone in the former). In contrast, the P1s of ‘ Pa. ’ valentini more extensively differ from those of V. steinheimensis , C. simorrensis , and R. matritensis in being more trenchant and higher-crowned, and displaying a more mesially located paracone, milder constrictions of the crown wall, a more indistinct metacone, and a longer and more curved parapostcrista, further lacking a distinct lingual cingulum. Te P1s of V. steinheimensis more markedly differ by displaying a more developed prestyle, as well as a more inflated paracone with deep buccal and lingual clefts both mesially and distally from it. However, the specimens attributed to C. simorrensis and R. matritensis from Göriach are also distinguishable from those of ‘ Pa. ’ valentini based on the features listed above—although the P1s from Simorre attributed to R. matritensis by Pickford and Laurent (2014) display a somewhat more mesially located metacone and a less developed lingual cingulum than the Göriach specimens.

Te P2s from SQ-TF and CF2 display very similar proportions to the P1 of the same taxon at a larger size (Additional file 1: Table S2), but are much slenderer than the P2 from CCN20 (IPS124328; BLI = 54%) previously attributed to ‘ Pa. ’ valentini by McKenzie et al. (2023a: fig. 4e). Te latter specimen displays a somewhat trenchant appearance, but its overall morphology is not particularly tetraconodontine-like, and its wide proportions and marked distolingual extension of the crown are much more consistent with being a large P2 of Pr. palaeochoerus (cf. McKenzie et al., 2024: fig. 14e), to which it is reassigned here. Te specimens described herein thus represent the first P2s reported for ‘ Pa. ’ valentini . Tey are similar in size to but slenderer on average than those of V. steinheimensis , whereas the P2s of C. steinheimensis and R. matritensis are generally both larger and relatively broader than those of ‘ Pa. ’ valentini ( Fig. 12B ; Additional file 1: Table S2). Te described P2s further differ from those of the other tetraconodontines included in the comparative sample in the same features as the P1s. Tus, compared with V. steinheimensis , ‘ Pa. ’ valentini displays a higher-crowned P2 with a more mesially located paracone and less marked constrictions of the crown walls, a more obliquely oriented parapostcrista, a more indistinct metacone, and lack of lingual cingulum. Te same differences apply as compared with R. matritensis from Göriach and Villefranche d’Astarac and, as far as it can be ascertained, C. simorrensis , even though such differences appear more marked in the case of V. steinheimensis (as it is also the case for the P1). Some P2s from Simorre and Elgg attributed to R. matritensis by Pickford and Laurent (2014) and Pickford (2016), respectively, display a somewhat more mesially placed paracone and lesser developed lingual cingulum, but also differ from those of ‘ Pa. ’ valentini in the lower occlusal relief and better developed metacone.

Te two P3s from SQ-TF largely overlap in size and proportions ( Fig. 12C ; Additional file 1: Table S2) with those of ‘ Pa. ’ valentini previously reported from multiple sites, being in contrast longer and relatively slenderer than those of V. steinheimensis , and generally narrower in absolute and especially relative terms than in C. simorrensis and R. matritensis . In terms of occlusal shape (Additional file 1: Table S3), the P3s from SQ-TF appear more mesially tapering than those of ‘ Pa. ’ valentini from CB ( McKenzie et al., 2024), but cannot be adequately compared with them because they are too worn and somewhat damaged. Te SQ-TF specimens are more similar to the single P3 of ‘ Pa. ’ valentini from CCN20 ( McKenzie et al., 2023b), which displays a well-developed prestyle, an inflated paracone, and an almost continuous lingual cingulum that bears a low and crest-like protocone, but which differs by possessing a stouter occlusal contour (with a more convex buccal side and a more distolingual expansion). Similar variation in occlusal contour and proportions is displayed by the P3s of ‘ Pa. ’ valentini from elsewhere (Pickford, 2014)—including the P3 from Keineisenbach ( Pickford, 2016)—overall fitting well with the variation of this species. Te P3s of V. steinheimensis display a similar morphology but differ from those of ‘ Pa. ’ valentini in the possession of more indistinct metacone and protocone, so that the distolingual extension of the cingulum is less accentuated. In turn, the P3s of C. simorrensis as well as R. matritensis from Göriach and elsewhere display a variably developed distobuccal extension of the crown, with specimens of C. simorrensis generally displaying a less developed lingual cingulum and a more indistinct protocone than in ‘ Pa. ’ valentini . However, C. simorrensis and R. matritensis differ from both ‘ Pa. ’ valentini and V. steinheimensis in displaying a stouter and higher-crowned P3 at comparable wear stages, which is conspicuously hypertrophied relative to the P4 and the upper molars. In most specimens, the development of the metacone cannot be adequately ascertained because this cusp is rapidly obliterated by wear (constituting a confluent oblique wear facet with the distal aspect of the paracone). However, the P3s from SQ-TF suggest that the metacone would have been less developed in ‘ Pa. ’ valentini than in the other taxa, at least based on two lightly worn specimens from Simorre. Tese two specimens were attributed to different species ( R. matritensis and C. simorrensis ) by Pickford and Laurent (2014) despite showing comparable occlusal proportions and shape, including a stout contour, high relief, a moderately developed distolingual extension, and a tall metacone that is located closer to the paracone and appears better developed than in ‘ Pa. ’ valentini .

Te P4s from SQ-TF match relatively well the P4 size and proportions of ‘ Pa. ’ valentini from the Vallès-Penedès Basin and elsewhere ( Fig. 12D ; Additional file 1: Table S2), only extending slightly its upper breadth range. Tose of V. steinheimensis , despite differences in occlusal contour (see below), display similar proportions but are overall smaller (albeit with some overlap). Te P4s of C. simorrensis and R. matritensis considerably overlap in proportions with ‘ Pa. ’ valentini but tend to be stouter on average and are also generally smaller. In occlusal shape (Additional file 1: Table S3), the P4s from SQ-TF resemble the more worn specimens from CB attributed to ‘ Pa. ’ valentini by McKenzie et al. (2024) —characterized by a very lingually tapering contour, a protocone less peripheral than the buccal cusps, a metacone subequal in size to the paracone and separated from it by a vertical groove on the buccal crown wall, moderately well-developed prestyle and poststyle, and an absent to poorly developed and discontinuous buccal cingulum—except that the newly reported specimens display a mesially tilted lingual crown half. Other specimens attributed to ‘ Pa. ’ valentini further display additional variation in occlusal contour (more or less lingually tapering or variably mesiobuccally protruding) and/or other minor details (slightly developed buccal and lingual cingula or less mesially located protocone), but otherwise display a similar morphology. Overall, the P4s of ‘ Pa. ’ valentini from SQ-TF and elsewhere differ from those of V. steinheimensis in the less mesially located protocone (not completely aligned transversely with the paracone), the larger metacone less closely packed with the paracone and separated from it by a buccal groove, the more marked distal cingulum, and the generally more lingually tapering occlusal contour. In both ‘ Pa. ’ valentini and V. steinheimensis , the P4 is almost as high and much broader than the P3, whereas in R. matritensis the P4 is higher than in these species but comparatively less hypertrophied (i.e., much lower-crowned and only slightly wider) than the P3. Te P4s of C. simorrensis and R. matritensis further differ from those of both ‘ Pa. ’ valentini and V. steinheimensis in several occlusal details, including the more or less constricted buccal contour (only insinuated in some specimens of ‘ Pa. ’ valentini ) and the generally somewhat better developed lingual cingulum. Tey further differ from V. steinheimensis (thus more closely resembling ‘ Pa. ’ valentini ) in the less mesial position of the protocone (generally not completely aligned with the paracone) and the presence of a vertical groove separating the apices of paracone and metacone on the buccal crown wall. As in ‘ Pa. ’ valentini , the P4 samples available for C. simorrensis and R. matritensis show some variation in occlusal contour (degree of lingual tapering), position of the protocone (more or less peripheral and/or aligned with the paracone), size and position of the metacone (more or less vestigial and/or closely packed to the paracone), and development of the lingual cingulum. However, no consistent differences can be discerned between the P4s attributed to C. simorrensis and R. matritensis by Pickford and Laurent (2014). Only the P4s of R. matritensis from the type locality (Puente Vallecas) and Paşalar stand out by displaying a protocone completely aligned with the paracone and virtually lacking a metacone. Even if it is assumed that the apices of both cusps would have been distinct when unworn (Golpe-Posse, 1972; Pickford & Laurent, 2014), the case is that they are distinct at comparable wear stages (including the remains of a buccal groove) in other P4s variously attributed to C. simorrensis or R. matritensis , including the large sample from Göriach.

Upper molars Te M1s from SQ-TF and CF2 fit well in size and proportions with those previously reported from the Vallès-Penedès Basin and elsewhere ( Fig. 12E ; Additional file 1: Table S2). In turn, the M2s are larger and sometimes slenderer than the M1s, and further agree in dimensions and proportions with previously reported specimens of ‘ Pa.’ valentini ( Fig. 12F ; Additional file 1: Table S2). In contrast, the M1s and M2s of the studied sample are larger than those of V. steinheimensis (more markedly so in the case of the M2), with no relevant differences in proportions. Te same applies when compared with the samples of R. matritensis and C. simorrensis , only with some overlap, particularly in the case of the Mira specimen (Golpe-Posse, 1971, 1972; Pickford, 2014), which is the largest M2 attributed to C. simorrensis . Te M2 from the Sansan tetraconodontine, here provisionally identified as ‘ S. ’ choerotherium (nomen dubium), is smaller than those of C. simorrensis and, especially, ‘ Pa. ’ valentini but overlaps in length with those of R. matritensis and V. steinheimensis (despite being slightly narrower). In occlusal shape (Additional file 1: Table S3), the described M1s and M2s from SQ-TF and CF2 resemble those of ‘ Pa. ’ valentini previously figured in the literature, but also those of V. steinheimensis . In contrast, the M1s and M2s attributed to R. matritensis and C. simorrensis , leaving aside size differences, generally display better developed preconules and a more distinct conule on the distal cingulum, as well as a more distally tapering distal lobe (particularly the M2), although there is considerable variation in these regards.

Te M3s from SQ-TF and CF2 largely overlap in size and proportions with those of ‘ Pa. ’ valentini from the same basin and elsewhere ( Fig. 12G ; Additional file 1: Table S2), being similar in proportions to but generally larger than the M3s of V. steinheimensis (albeit with some overlap). Te described M3s are also larger and slenderer than those attributed to R. matritensis (albeit with some overlap in proportions) and C. simorrensis (albeit with some overlap in size, especially as a result of the inclusion of the Mira specimen in C. simorrensis ; see McKenzie et al., 2024). Te lectotype M3 of C. doati from Bonnefond, a nominal species here considered a nomen dubium following McKenzie et al. (2024), falls close to the variation of ‘ Pa. ’ valentini but is slightly broader and is also closer in dimensions to the M3 from Mira ( Fig. 12G ; Additional file 1: Table S2). Te M3 from Kleineisenbach attributed by Pickford (2016) to ‘ Pa. ’ valentini is the smallest currently recorded for this species and overlaps in size and proportions with those of V. steinheimensis , further falling close to the metrical variation of the other tetraconodontines included in the comparative sample but not far from other small M3s of ‘ Pa. ’ valentini . Te M3 from Sansan tetraconodontine—i.e., ‘ S. ’ choerotherium (nomen dubium)—is smaller than all the tetraconodontines included in the studied and comparative samples except R. matritensis . In occlusal shape (Additional file 1: Table S3), the described M3s are consistent with the attribution to a single species and more closely resemble those previously attributed to ‘ Pa. ’ valentini —despite some variation in occlusal proportions and contour, as previously noted by McKenzie et al. (2023a, 2024). In contrast, the M3s of ‘ Pa. ’ valentini differ from those of V. steinheimensis in the possession of a less tapering central lobe and a generally better developed and more distinct distal lobe, which bears a lesser developed pentapreconule but a better developed and distally directed or buccally tilted pentacone. Te M3s of ‘ Pa. ’ valentini also differ from those attributed to R. matritensis , C. simorrensis (including the Mira specimen), and ‘ S. ’ choerotherium (nomen dubium) from Sansan in the lesser developed pentaconules (including the pentapreconule), the less marked constriction between the mesial and central lobes (contrasting with the more distinct distal lobe), and the lack of lingual tilting of the central and distal lobes relative to the mesial. Compared with these species, ‘ Pa. ’ valentini also displays a narrower but more distinct and distally protruding distal lobe, which generally bears a lesser developed distal cingulum and displays a much more developed, distinct, and not lingually tilted pentacone—albeit there is also considerable variation in the development of the M3 distal lobe in these taxa, from very poorly developed to moderately protruding distolingually. Te M3s of ‘ Pa. ’ valentini markedly differ from the lectotype of C. doati (Additional file 1: Table S3), as previously remarked by McKenzie et al. (2024), who considered that the latter is a nomen dubium —potentially synonymous with either C. simorrensis or V. steinheimensis —although the broad proportions and metrical similarities with the Mira specimen are more consistent with an attribution to C. simorrensis .

Lower incisors Te i1s described here cannot be compared with any previously known specimen of ‘ Pa. ’ valentini , given that the incisor identified as such by Pickford (2014) from the SQ-TF sample is here considered an i2. Nevertheless, our metrical comparisons (Additional file 1: Table S2) indicate that the described i1s are larger and, on average, more mesiodistally compressed than those of V. steinheimensis . Te former are also generally larger than the few specimens attributed to R. matritensis and C. simorrensis , which nevertheless appear more similar in proportions to those of ‘ Pa. ’ valentini than to those of V. steinheimensis . In terms of occlusal shape (Additional file 1: Table S3), no differences can be noticed compared with the previously figured i1s of V. steinheimensis and C. simorrensis from Mira, all of which are nevertheless too worn to ascertain the apical morphology of the crown. Nevertheless, the root of the single completely preserved i1 from TF-SQ appears much more tapering on its apical half than the CB specimen figured by McKenzie et al. (2024).

Te i2s from SQ-TF appear larger and relatively much broader than the i2 from Wartenberg bei Erding (MD = 8.5 mm, BL> 9.6 mm) attributed to ‘ Pa. ’ valentini by Pickford (2016); however, this is largely because this specimen is broken and does not preserve the base of the crown on the lingual side. Yet the i2s described herein are also somewhat larger than those from CCN20 and CB previously attributed to ‘ Pa. ’ valentini by McKenzie et al. (2023a, 2024), which also overlap to a large extent in both size and proportions with the few i2s available for R. matritensis and C. simorrensis , and the more extensive sample of V. steinheimensis —excluding an incisor from Steinheim (GPIT MA 1178–42; MD = 5.0 mm, BL = 7.4 mm) that was identified as an i2 by Pickford (2016), but which appears too small and is here considered a di2 following Hünermann (1968). Comparisons of occlusal shape (Additional file 1: Table S3) with the i2s of R. matritensis and C. simorrensis are hindered by their advanced wear, as it is also the case of various figured i2s of ‘ Pa. ’ valentini and V. steinheimensis . Nevertheless, compared with the least worn i2s of V. steinheimensis figured by Chen (1984) and Pickford (2016), those from SQ-TF display a more asymmetrical crown, due to the more curved poststylid, which is also more distinct (as marked as the prestylid). In these regards, the described specimens most closely resemble the lightly worn i2s of ‘ Pa. ’ valentini from CB ( McKenzie et al., 2024), which are slightly smaller and display an even more asymmetrical crown (with a more curved poststylid and a more mesially tilted endocristid).

In dimensions and proportions (Additional file 1: Table S2), the described i3s considerably overlap with those previously attributed to ‘ Pa. ’ valentini as well as V. steinheimensis , even though the latter are somewhat narrower labiolingually on average. In terms of shape (Additional file 1: Table S3), the i3s from SQ-TF also more closely resemble those of ‘ Pa. ’ valentini , which display a higher (relative to basal dimensions), slenderer (less labiolingually inflated), and more mesially tilted crown than the figured i3s of V. steinheimensis . Te single available i3 of R. matritensis from Göriach is even more distinctive by displaying a more labiolingually elongate (mesiodistally compressed) crown and root.

Lower canines. Te c1ms from SQ-TF allow us, for the first time, to describe the morphology of this tooth locus for ‘ Pa. ’ valentini (see Discussion for further considerations in this regard). Although these specimens were originally attributed to L. splendens by Bataller (1924), their marked scrofic morphology does not fit with the c 1m proportions of the latter species, which generally displays verrucosic or mildly scrofic canines ( Van der Made, 1996a). In particular, 56% (39/70) of the c1ms of L. splendens measured by Van der Made (1996b: fig. 48) display verrucosic proportions (with the distal side narrower than the labial), whereas the remaining 44% have moderate scrofic proportions (with a La/Di ratio above 80% in 30 specimens and a single specimen with a La/Di ratio around 75%). In contrast, the three described c1ms display lower La/Di ratios (61%, 70%, and 75%), and indeed they display a narrower labial side than all the c1ms of L. splendens measured by Van der Made (1996b: fig. 48). Tese differences are even more marked regarding the seven c1ms of L. splendens from Sant Quirze measured by Van der Made (1996b: fig. 48), which display La/Di ratios between 110 and 140% and a labial side more than twice wider than the specimens here attributed to ‘ Pa. ’ valentini . Te latter, like those of other tetraconodontines included in the comparative sample, display a clearly scrofic morphology, with a very asymmetric cross section characterized by a narrowest labial side and a broadest lingual side, which is 64–79% broader than the labial (Additional file 1: Table S2). In cross-sectional dimensions and proportions (Additional file 1: Table S2), they extensively overlap with the c1ms of V. steinheimensis , which display a lingual side 46–111% broader than the labial, albeit being on their upper size range. Compared with the SQ-TF specimens, a c 1m of R. matritensis from Göriach is slightly smaller and displays a greater asymmetry between the lingual and the labial sides (the former about twice the latter), but the latter is not the case for an apical fragment from the same site, whereas no comparable measurements are available for other specimens attributed to either C. simorrensis or R. matritensis . Despite overall similarities in c 1m shape among the aforementioned taxa (Additional file 1: Table S3), the c1ms of V. steinheimensis and C. simorrensis display a more marked concavity along the labial side than the SQ-TF specimens, and at least the most complete specimen of V. steinheimensis from CB ( McKenzie et al., 2024) further differs by a more markedly labial curvature toward the tip. No suitable c 1m pictures of R. matritensis from Göriach are available—as the specimen depicted by Hofmann (1893) only preserves the tip—but it may be ascertained based on a c 1m from Simorre attributed to this species by Pickford and Laurent (2014). Te c1ms of C. simorrensis differ from all other taxa considered here, including R. matritensis , by the presence of a longitudinal ridge of cementum along the distal side, which mostly relies on the comparison between the C. simorrensis specimens from Carpetana ( Pickford, 2013b) and the aforementioned specimen from Simorre. However, as noted by Pickford and Laurent (2014) based on Stehlin (1900), this feature can also be ascertained in a c 1m from the type locality of C. doati (see Discussion for additional remarks in this regard).

Te single c1f from SQ-TF displays similar size and proportions (Additional file 1: Table S2) than those previously attributed to ‘ Pa. ’ valentini except for being slightly narrower in absolute terms. Taken overall, despite overlapping in proportions, the c1fs of ‘ Pa. ’ valentini are larger than the few c1fs reported for V. steinheimensis and C. simorrensis . On morphological grounds, the described specimen is the first c1f of ‘ Pa. ’ valentini enabling an adequate description of the occlusal shape of this tooth locus, because those previously described by McKenzie et al. (2023a, 2024) from CCN20 and CB are heavily worn. Tese specimens display a somewhat stouter root than the SQ-TF c1f but, as far as it can be ascertained, are consistent in shape. Although root development cannot be properly evaluated in the few available specimens of V. steinheimensis , the latter display a very different crown shape, being somewhat lower-crowned and possessing along the distal facet a much deeper fossid flanked by sharper and more distinct ectocristid and postcristid. Tis is difficult to evaluate based on the available pictures of the c1fs of C. simorrensis from Przeworno 2 and R. matritensis from Göriach, which appear higher-crowned than those of V. steinheimensis but differ from those of ‘ Pa. ’ valentini by being less labiolingually compressed and, in the case of the Przeworno 2 specimens, by displaying a much slenderer root.

Lower premolars Te newly reported p1 from SQ-TF closely resembles in size and shape that previously described by Pickford (2014) from the same site, except for the slightly developed prestylid and better developed cuspule-like developments along the protopostcristid. When considered together, these two p1s are slightly longer than but display similar proportions to those from CB attributed to the same taxon ( McKenzie et al., 2024; Fig. 13A ; Additional file 1: Table S2). McKenzie et al. (2023a: fig. 9e) further identified a lower premolar from CCN20 (IPS95813) as a p1 of ‘ Pa. ’ valentini , but this specimen is most likely a dp2 of Pr. palaeochoerus and hence excluded from the comparative sample. Te p1s of V. steinheimensis are shorter and generally also relatively stouter (albeit with some overlap) than those of ‘ Pa. ’ valentini . In turn, the single p1 of C. simorrensis resembles in size and proportions the more abundant p1 sample of R. matritensis from Göriach, which overlap in length with both V. steinheimensis and ‘ Pa. ’ valentini but are relatively stouter (and, with a single exception, also broader in absolute terms). In terms of occlusal shape (Additional file 1: Table S3), the two p1s from SQ-TF most closely resemble those of ‘ Pa. ’ valentini from CB ( McKenzie et al., 2024), despite some variation in the inclination of the protopostcristid in both samples. Furthermore, the bifurcation of the protopostcristid at the level of hypoconid can only be ascertained, even if less clearly than in the SQ-TF specimens, in some but not all of the premolars from CB identified as p1s of ‘ Pa. ’ valentini . However, an alternative identification of the latter as p2s of ‘ Pa. ’ valentini is not supported based on size, suggesting that the distal shape of the p1 is intraspecifically variable in ‘ Pa. ’ valentini . Indeed, similar variation can be ascertained in V. steinheimensis , with some p1s from Steinheim and Gratkorn displaying a bifurcated protopostcristid with a distolingually placed hypoconid and others from Steinheim and CB displaying an undivided or less clearly divided protopostcristid with a generally more centrally located hypoconid. Taken overall, this tooth locus is not very diagnostic between ‘ Pa. ’ valentini and V. steinheimensis , although the former tends to display a more elongate p1 crown with a less steeply inclined protopostcristid. Te p1s of R. matritensis from Göriach and C. simorrensis from Carpetana display a morphology reminiscent of that of ‘ Pa. ’ valentini , with a distally bifurcated protopostcristid, except for the partial fusion of the roots, the apparent lack of a distinct hypoconid, and the above-mentioned less elongate occlusal contour. Te differences are more marked in the case of the C. simorrensis p1s, which further differ because the protopostcristid bifurcates more mesially (closer to the protoconid) and the crown displays a more oval contour (with the distal half being much wider than the mesial).

Te p2s from SQ-TF can be compared with that of the Fonte do Pinheiro mandible, here attributed to ‘ Pa. ’ valentini , which overlaps in length (MD = 19.0 mm; Van der Made, 1989 provided no BL measurement). Te p2s from SQ-TF further overlap considerably in dimensions, but minimally in proportions ( Fig. 13B ; Additional file 1: Table S2), with those of V. steinheimensis —excluding IPS92720 from CB ( McKenzie et al., 2024), which was originally interpreted as a p2 (on the grounds of its small size) but is probably best interpreted as a p3 based on occlusal shape and proportions (see below). Despite the overlap in dimensions, the p2s of ‘ Pa. ’ valentini tend to be relatively more elongate than those of V. steinheimensis . Te p2s of C. simorrensis and R. matritensis from Göriach are relatively broader (and also absolutely broader in the case of R. matritensis ) than those of ‘ Pa. ’ valentini and V. steinheimensis . Te p2 from Sansan, probably attributable to the same species represented by the type maxilla of ‘ S. ’ choerotherium (nomen dubium), is smaller than those of C. simorrensis and, especially, R. matritensis from Göriach, but stouter than all the other specimens and broader in absolute terms than those of V. steinheimensis and ‘ Pa. ’ valentini . With regard to occlusal morphology (Additional file 1: Table S3), the p2s from SQ-TF appear comparable in occlusal contour and general morphology with the single p2 previously available for ‘ Pa. ’ valentini from Fonte do Pinheiro. In contrast, the p2s of V. steinheimensis differ from those of ‘ Pa. ’ valentini in the presence of more or less marked buccal and lingual constrictions of the occlusal contour at the protoconid level, as well as the generally bifurcated protopostcristid with a sometimes better-developed hypoconid—although there is variation in these features. Te p2s of R. matritensis from Göriach resemble those of ‘ Pa. ’ valentini in the morphology of the protopostcristid and distinctiveness of the hypoconid, but differ in their less elongate and more distally expanded occlusal contour. Tese differences further apply to the two p2s from Villefranche d’Astarac figured by Pickford and Laurent (2014), including the lectotype mandible of C. simorrensis and another mandible assigned to R. matritensis , which further insinuate an incipient bifurcation of the protoprecristid. In contrast, the p2s from the mandibles of C. simorrensis from Carpetana resemble those of ‘ Pa. ’ valentini in protopostcristid and hypoconid development. However, they more clearly differ from the latter than the above-mentioned specimens from Göriach and Villefranche d’Astarac in the more buccally tilted and less steeply inclined protoprecristid, the more marked buccolingual constriction of the occlusal contour at the protoconid level, and the greater expansion of the distal portion of the crown. A p2 from Sansan—originally identified as a P1 of C. simorrensis by Ginsburg (1977), subsequently identified as a p2 of the same species by Pickford (2012), and most recently attributed to Retroporcus sindiensis ( Lydekker, 1878) by Van der Made (2020) —is too worn to allow for meaningful comparisons with ‘ Pa. ’ valentini other than illustrating the much stouter occlusal contour of Sansan specimen (see the Discussion for further comments about its taxonomic attribution).

Te single p3 from SQ-TF that is complete enough to take all measurements (MGSB48817) and the p3 mesial fragment (MGSB48859) newly reported here appear narrower than the other incomplete specimen (IPS96062 [IPS1857]) from the same site ( Table 2), previously reported (but not figured) by Pickford (2014), as well as other p3s of ‘ Pa. ’ valentini previously reported from elsewhere ( Fig. 13C ; Additional file 1: Table S2). Te SQ-TF complete specimen falls instead within the p3 variation range of V. steinheimensis , which is characterized by a narrower crown in both absolute and relative terms. Te above-mentioned lower premolar IPS92720 from CB, previously identified as a p2, is smaller than both species but fits well with the p3 proportions of V. steinheimensis . Tese comparisons suggest the size and proportions of the p3 are quite variable in both species, so that other occlusal details are probably more reliable to distinguish between them (see below). In turn, the p3s of C. simorrensis and R. matritensis considerably overlap in dimensions with those of ‘ Pa. ’ valentini (except for the SQ-TF specimen reported here) but are broader on average, although differences in both size and proportions are more marked as compared with V. steinheimensis . It is noteworthy that the p3 sample of R. matritensis from Göriach displays considerable variation (particularly in length and, hence, proportions), in agreement with the condition displayed by V. steinheimensis as well as ‘ Pa. ’ valentini (especially when the SQ-TF specimen is considered). Regarding occlusal shape (Additional file 1: Table S3), the previously figured p3s of ‘ Pa. ’ valentini vary considerably in occlusal contour (elliptical to oval, and variably constricted lingually at the protoconid level)—in further agreement with their aforementioned metrical variation. However, they otherwise display a consistent occlusal shape, characterized by: a pointed, centrally located, and distally tilted protoconid; a thick, blunt, straight, and moderately steep protoprecristid that ends in a low and mildly developed prestylid; and a similarly developed protopostcristid, which is flanked by shallow buccal and lingual clefts and usually bifurcates, giving rise to a very restricted, shallow, and distally open fossid with a poorly-developed and distobuccally located hypoconid. Te p3s from SQ-TF fit well with this morphological pattern, except that the protopostcristid does not bifurcate (only ascertainable in the more complete specimen) and that the protoprecristid appears more steeply inclined; the occlusal contour of the complete p3 is most similar to that of the Fonte do Pinheiro mandible figured by Roman (1907). Te p3s of V. steinheimensis also show some variation in occlusal contour (more or less elongate, subelliptical to suboval, with or without lingual constriction) and distal development of the protopostcristid and hypoconulid (non-bifurcated and centrally located vs. bifurcated and distobuccally placed). However, they are generally slenderer than those of ‘ Pa. ’ valentini , from which they further differ in possessing a better developed prestylid, more concave crown walls along the protoprecristid, much deeper buccal and lingual clefts at the level of the protopostcristid, generally a more distinct (albeit poorly developed) metaconid just distally or distolingually from the protocristid (only ascertainable on unworn to lightly worn specimens), and a more rootwardly extended cervix on the distal crown portion. Te p3s of both R. matritensis and C. simorrensis are somewhat variable in occlusal contour but more closely resemble those of ‘ Pa. ’ valentini in most of the aforementioned features. Te only exception is the greater distal than mesial rootward extension of the cervix in the specimens of R. matritensis and C. simorrensis . In this regard, the p3 from Kleineisenbach attributed to ‘ Pa. ’ valentini by Pickford (2016) is most compatible with this species, differing from C. simorrensis and R. matritensis in the less pronounced rootward extension of the cervix distally. Furthermore, R. matritensis and C. simorrensis differ from both ‘ Pa. ’ valentini and V. steinheimensis by displaying a hypertrophied (longer, broader, and much higher) p3 as compared with the remaining premolars (including the p4), thus paralleling the hypertrophy of the P 3 in the upper dentition.

Te two p4s from SQ-TF considerably but not completely overlap in size and proportions with the small sample of p4s previously available for ‘ Pa. ’ valentini

( Fig. 13D ; Additional file 1: Table S2). In particular, one of the described specimens fits well in both size and proportions, whereas the other is shorter and relatively broader. Compared with the larger sample available for V. steinheimensis , one of the SQ-TF specimens is larger (both longer and broader) but displays similar proportions, whereas the smaller p4 from SQ-TF is broader in both absolute and relative terms—overall suggesting that the SQ-TF specimens fit worse with the variation of V. steinheimensis than that of ‘ Pa. ’ valentini . Te p4s of R. matritensis and C. simorrensis show very similar size and proportions, being broader on average (with substantial overlap) than those of V. steinheimensis , but generally showing broader proportions at comparable p4 length. In contrast, ‘ Pa.’ valentini shows a broader range of variation, with some p4s (including those from SQ-TF) more closely resembling the proportions of R. matritensis and C. simorrensis , and others (from Saint-Gaudens and CB) being slenderer (even more so than those of V. steinheimensis ). Tis illustrates that the p4 proportions of ‘ Pa. ’ valentini (and the associated occlusal contour) are not particularly diagnostic, unlike other details of its occlusal morphology. In the latter regard (Additional file 1: Table S3), the two p4s from SQ-TF resemble those of ‘ Pa. ’ valentini from elsewhere in multiple features, including the presence of a distinct metaconid apex just distolingually from the protoconid (in some of the least worn specimens, such as mandible from El Buste, attributed by Pickford & Laurent, 2014 to C. doati ), the distolingually oriented protopostcristid, and the distally and centrally located hypoconid. Te p4s of V. steinheimensis similarly display a metaconid distolingually placed from the protoconid (in unworn specimens) and a well-developed and centrally located hypoconid, but they differ from those of ‘ Pa. ’ valentini in the generally more oval (distally broader) occlusal contour (although it is variable in ‘ Pa. ’ valentini , as illustrated by the Fonte do Pinheiro specimen), the stronger and usually higher prestylid (associated to better developed cingular developments and/or deeper clefts along the protoprecristid), and the more concave crown walls along the protopostcristid (usually with a broader and/or deeper buccal cleft). Te p4s of R. matritensis from Göriach and elsewhere, as well as those attributed to C. simorrensis , are somewhat variable in occlusal contour (generally subrectangular to suboval) but nevertheless stouter and less elongate than in the two aforementioned taxa, differing from the generally elliptical p4 occlusal contour of ‘ Pa. ’ valentini (but see the Fonte do Pinheiro specimen). Furthermore, in R. matritensis and C. simorrensis the prestylid is less well developed, the crown walls along the protoprecristid are more inflated (less concave), and the buccal cleft along the protopostcristid is shallower than in V. steinheimensis , most closely resembling ‘ Pa. ’ valentini in these regards. Nevertheless, although in R. matritensis and C. simorrensis the p4 appears much smaller than the p3, the former tooth is much broader than the m1, differing from the condition of both ‘ Pa. ’ valentini and V. steinheimensis , where the p4 is similar in breadth to the m1.

Lower molars Te described m1s and m2s display similar proportions and overlap to some extent with one another in breadth and, particularly, length, although the m2s are larger on average. Te studied sample almost completely overlaps in m1 ( Fig. 13E ) and m2 ( Fig. 13F ) size and proportions (Additional file 1: Table S2) with ‘ Pa. ’ valentini from elsewhere, only minimally expanding the range of variation previously recorded for this species (including those attributed to C. doati by Pickford, 2016). Te described sample further overlaps to a large extent with the m1 and m2 size and proportions recorded for V. steinheimensis , but surpasses the upper size range of the latter, thereby indicating a better fit with the metrical variation of ‘ Pa. ’ valentini . Size differences are more clearcut as compared with R. matritensis and C. simorrensis , which generally display smaller m1s and m2s than ‘ Pa. ’ valentini , although with some overlap. In contrast, there seem to be no noticeable differences in m1 and m2 proportions among the species included in the comparative sample. Te described m1s and m2s display a generalized tetraconodontine occlusal morphology—with marked Fürchen and profuse but moderately developed enamel wrinkling when unworn or lightly worn, and a generally more conspicuous (but not abrupt) buccal than lingual constriction between the mesial and the distal lobes. In these regards (Additional file 1: Table S3), they are fully comparable to the variation displayed by ‘ Pa. ’ valentini but also V. steinheimensis or even (leaving proportions aside) R. matritensis and C. simorrensis .

Te four m3s from the studied sample almost overlap completely with the previously documented variation of ‘ Pa. ’ valentini in size and proportions ( Fig. 13G ; Additional file 1: Table S2), except for extending slightly the upper range of occlusal proportions. In contrast, despite a considerable overlap, the m3s of ‘ Pa. ’ valentini are larger on average than those of V. steinheimensis (despite considerable overlap), with the latter species almost completely overlapping the joint range of variation in m3 dimensions and proportions of C. simorrensis and R. matritensis (with the latter tending to fall on the lower size range of C. simorrensis and V. steinheimensis ). In occlusal shape (Additional file 1: Table S3), the described m3s show some variation in occlusal contour—ranging from those with a central lobe subequal to the mesial one and an abruptly tapering distal lobe ( Fig. 10U ) to a more triangular contour that progressively tapers distalward

( Fig. 10W–X )—although in all instances they display an only slightly buccally tilted distal lobe. Tis variation can be accommodated within that previously recorded for the m3s of ‘ Pa. ’ valentini , which even include some specimens in which the distal lobe is more distinct from the central one or more markedly tilted buccally. Te m3s of V. steinheimensis also show variation in occlusal contour and degree of distal tapering, but are generally more clearly subtriangular and uniformly tapering, so that the distal lobe tends to be less developed, less distinct from the central lobe, and more markedly oriented distobuccally. Te m3s of R. matritensis from Göriach and C. simorrensis generally taper distally less markedly than in V. steinheimensis and some specimens of ‘ Pa. ’ valentini , with the central lobe being subequal in size to the mesial one, as in other specimens of ‘ Pa. ’ valentini . Furthermore, although in R. matritensis from Göriach and C. simorrensis the buccal tilting and breadth of the distal lobe is variable, the latter is generally broader than in V. steinheimensis and less distinct than in ‘ Pa. ’ valentini . A similar morphology is displayed by other m3s attributed to R. matritensis , where the buccal tilting of the distal lobe is very variable and also ranges from almost non-existent to very marked (even when considering specimens from the type locality).

Upper deciduous teeth Te described DI1s ( Fig. 11A–E ) are slightly longer and relatively narrower (Additional file 1: Table S2) than those from CB previously attributed to ‘ Pa. ’ valentini by McKenzie et al. (2024). However, as explained above, we consider that two upper incisors from this site—previously identified as I2s of ‘ Pa. ’ valentini by McKenzie et al. (2024) but which strikingly differ from the I2s from SQ-TF—are indeed larger DI1s of the same species. Te possibility remains that the smallest tetraconodontine DI1s from CB belong instead to V. steinheimensis , also recorded there, but the largest specimens fit sufficiently well in size and shape (Additional file 1: Table S3) with those described herein so as to be attributed to the same species (i.e., ‘ Pa. ’ valentini ). Te lack of additional deciduous incisors reported from the literature for the species included in the comparative sample precludes more detailed comparisons. If correctly identified, the DI2 and DI3 from SQ-TF reported herein represent the first description of these tooth loci for ‘ Pa. ’ valentini , suggesting that they display similar shape and proportions, only differing in the slightly larger and higher crown of the DI2.

Te two DP2s from SQ-TF are slightly larger than but resemble in proportions those from CB previously attributed to ‘ Pa. ’ valentini , being more clearly larger and slenderer than those from the same site assigned to V. steinheimensis (Additional file 1: Table S2). Te SQ-TF specimens also fit better in occlusal morphology (Additional file 1: Table S3) with previously figured DP2s of ‘ Pa. ’ valentini , whereas those of V. steinheimensis display an even more asymmetrical contour and a more buccolingually expanded distal crown portion, a better developed metacone (much higher than the prestyle), a concave buccal crown wall along the paraprecrista, and more marked buccal and lingual clefts at the level of the parapostcrista. In turn, the single complete DP3 from SQ-TF closely resembles in size and proportions the specimens previously attributed to ‘ Pa.’ valentini and is, in contrast, slightly broader than those of V. steinheimensis (Additional file 1: Table S2). Besides these slight metrical differences, the DP3 described herein more closely resembles those of ‘ Pa. ’ valentini than those of V. steinheimensis in the more marked constriction between the mesial and distal lobes (resulting in a rather pearshaped, less subtriangular occlusal contour). Finally, the described DP4s overlap to a large extent in size and proportions with those of ‘ Pa. ’ valentini but are in contrast larger than those of V. steinheimensis ( Fig. 13H ; Additional file 1: Table S2), albeit with some overlap. A DP4 of C. simorrensis from Villefranche d’Astarac overlaps with both taxa. In occlusal shape (Additional file 1: Table S3), the described DP4s also fit better with those previously attributed to ‘ Pa. ’ valentini than to those of V. steinheimensis . As previously noted by McKenzie et al. (2024), the DP4s of these species differ in the generally straighter and more uniformly inclined mesial contour as well as the less transversely aligned distal cusps of ‘ Pa. ’ valentini . A DP4 from Villefranche d’Astarac attributed to R. matritensis by Pickford and Laurent (2014) displays an intermediate morphology, more closely resembling ‘ Pa. ’ valentini in the mesial contour but V. steinheimensis in the alignment of the distal cusps.

Lower deciduous teeth Te di1s described herein

( Fig. 11P–R ) resemble in size and proportions the few specimens previously attributed to ‘ Pa. ’ valentini and V. steinheimensis in the literature (Additional file 1: Table S2). In terms of occlusal shape (Additional file 1: Table S3), the di1s from SQ-TF fit well with those previously attributed to ‘ Pa. ’ valentini , whereas those assigned to V. steinheimensis are similar but seemingly display a more labiolingually compressed apical portion of the crown. A lower deciduous incisor of R. matritensis from Göriach, originally interpreted as a possible di2 or di3 (Hofmann, 1893), is morphologically similar (as far as it can be appreciated from the figure) to both species, being thus interpreted as a di1. In turn, the di2s from SQ-TF are also similar in dimensions to those previously assigned to ‘ Pa. ’ valentini and V. steinheimensis (Additional file 1: Table S2), although the small sample sizes preclude more meaningful comparisons. In terms of shape (Additional file 1: Table S3), the described di2s agree well with a previously reported specimen of ‘ Pa. ’ valentini from CB, overall differing from those of V. steinheimensis from the same site in the more marked mesial tilting of the crown. Nevertheless, the new specimens reported here differ from the CB ones in displaying a thicker endocristid (more similar to specimens previously assigned to V. steinheimensis ). As noted by McKenzie et al. (2024), the identifications of the CB di2s must be considered tentative as currently available samples are too scarce to adequately assess intra- and interspecific variation. Despite of this, the SQ-TF sample shows that differences in di2 endocristid thickness are probably intraspecifically variable within ‘ Pa. ’ valentini , as it is also the case of other features such as the lingual contour of the poststylid.

Te partial dp2 from SQ-TF (BL> 6.9 mm) is much broader than that attributed to V. steinheimensis (Additional file 1: Table S2), further differing from the latter in occlusal shape (Additional file 1: Table S3) by displaying a stouter contour, a better developed prestylid, a less pointed protoconid, and a larger hypoconid flanked by more marked buccal and lingual clefts. McKenzie et al. (2023a: fig. 11p) tentatively identified a lower premolar from CCN20 (IPS114102) as a dp2 of ‘ Pa. ’ valentini , but its morphology fits better with a large dp2 of Pr. palaeochoerus . Given that the SQ-TF partial specimen vaguely resembles the p2s and p3s of ‘ Pa. ’ valentini , but differs from them in several occlusal details, the smaller size, and the lower crown, the most reasonable option is to identify it as the actual dp2 of this species and reassign the CCN 20 specimen to Pr. palaeochoerus . Something similar occurs with the dp3, as McKenzie et al. (2024: fig. 13s–v) identified several partial lower premolars from CB (IPS92881, IPS92779, IPS93153, and IPS92388) as tetraconodontine but, based on the higher occlusal relief of the dp3s of V. steinheimensis , C. simorrensis , and R. matritensis (Additional file 1: Table S3), the CB specimens most likely correspond to dp2s and dp3s of Pr. palaeochoerus . In contrast, the distal crown fragment from SQ-TF (BL> 8.0 mm) more closely resembles the distal morphology of the p3 of ‘ Pa. ’ valentini at a smaller size, being thus interpreted as a dp3 of this taxon, even though it is slightly broader than previously reported specimens of both ‘ Pa.’ valentini and V. steinheimensis (Additional file 1: Table S2). Finally, the dp4s from SQ-TF and SQ-PN closely resemble in size and proportions the dp4s from elsewhere attributed to ‘ Pa. ’ valentini and are larger than those attributed to V. steinheimensis , even though the two species overlap to a large extent ( Fig. 13H ; Additional file 1: Table S2). A dp4 of R. matritensis from Göriach is smaller than those of ‘ Pa. ’ valentini but overlaps in proportions, more closely resembling in size those of V. steinheimensis . In terms of occlusal morphology (Additional file 1: Table S3), the described specimens also closely resemble previously reported specimens of ‘ Pa. ’ valentini from elsewhere while differing from those of V. steinheimensis in a few details, such as the less protruding mesial cingulid, the more distally protruding distal cingulid (less rounded distal contour), and the more transversely aligned pairs of crests. Although a dp4 of R. matritensis is also available from Göriach, the only available published picture does not allow us to compare it with the described specimens.

Cladistic analysis

Our parsimony analysis recovered five most parsimonious cladograms with a tree length of 52 steps (CI = 0.642, HI = 0.359, RI = 0.672, RCI = 0.431). Tey all retrieved Hyotherium as the basal-most ingroup taxon and recovered the monophyly of both the analyzed suines and tetraconodontines. Among tetraconodontines, the bootstrap 50% majority-rule consensus yielded a polytomy between Parachleuastochoerus s.s. (i.e., based on the type species of the genus), a clade including Tetraconodon + Nyanzachoerus , and another clade including the three taxa newly coded in this paper ( Fig. 14 ). Te strict consensus yielded the same results for the latter taxa but collapsed the clade of Tetraconodon + Nyanzachoerus (Additional file 1: Fig. S1 ). Among the newly coded tetraconodontines, ‘ Pa. ’ valentini clustered with the Göriach sample (variously attributed to R. matritensis or C. simorrensis s.l.) exclusive of V. steinheimensis , indicating that the former are more closely related to one another, but also more so to V. steinheimensis than to Parachleuastochoerus s.s. Tis topology is not very robust, as it only took an additional step to entirely collapse the tetraconodontine clade with suines—whose monophyly appears in contrast much more robust, like that of tetraconodontines + suines to the exclusion of Hyotherium . Tetraconodontines+ suines are defined by 10 synapomorphies (of which seven unambiguous), suines by six (four unambiguous), tetraconodontines by eight (four unambiguous), and the clade including the three tetraconodontines newly coded here by seven (although only three unambiguous; Additional file 1: Table S6). In contrast, ‘ Pa. ’ valentini and the Göriach sample are united by a single unambiguous synapomorphy.

Figures

Fig. 3 Permanent upper incisors and canines of ‘Parachleuastochoerus’ valentini described in this paper: A IPS96058 [VP1013],R I1; B IPS30986, R I1; C MGSB48853.1, L I1 partial germ; D MGSB48853.2, R I1 partial germ; E MGSB48842, R I1 germ; F MGSB30487, L I2; G MGSB48862, damaged L I2; H MGSB48559, R I2 crown; I IPS96051 [VP1032], R I3; J MGSB48918, L C1f; K MGSB49727,R C1m; L IPS31069 [VP1009], L C1m. All the specimens come from SQ-TF and are depicted, from left to right, in lingual, mesial, labial, and distal views, except for the C1f, which is depicted, from left to right, in occlusal, lingual, and labial views

Fig. 4 Permanent upper premolars of ‘Parachleuastochoerus’ valentini described in this paper: A IPS95962 [4290], R P1; B IPS96073 [VP998], R P2; C IPS96074 [VP999], L P2; D MGSB48854, R P2; E MGSB48860,L P2 mesial fragment; F IPS31028 [IPS1980], R P2 from CF2; G IPS96072 [IPS1858], R P2 distal fragment; H MGSB48586, R P3; I MGSB48587,R P3; J IPS138283, R P4 lingual fragment; K MGSB48937, L P4; L MGSB48836, L P4; M IPS96056 [IPS1825], R P4; N IPS31065a [VP1014],L P4. All specimens come from SQ-TF unless explicitly indicated; they are depicted, from left to right, in occlusal,lingual, and labial views, except for a partial P4, which is only depicted in occlusal view

Fig. 5 Permanent upper molars of ‘Parachleuastochoerus’ valentini described in this paper: A IPS96055 [VP1015],L M1 crown; B IPS31026 [IPS1881], L M1 from CF2; C MGSB48939,R M1; D MGSB48831,partial R M1; E IPS96081 [IPS1794],R M1; F MGSB48915, damaged L M1; G MGSB48824, partial L M1; H MGSB49725, R M1 mesial fragment; I IPS63283,R M1 mesial fragment; J IPS30988,R M1 distal fragment; K IPS31061 [VP1007], R M1; L IPS31065b [VP1014], L M1; M MGSB48914, R M2; N MGSB48832, L M2; O MGSB48814,R M2 partial germ; P MGSB48816, L M2; Q MGSB48835, damaged L M2; R MGSB49726.2,R M2 buccal fragment; S IPS31061 [VP1007], R M2; T IPS31065c [VP1014], L M2; U IPS31078 [55], R M2 distal fragment;V MGSB48833, fragmentary L M2?; W MGSB48841,L M2? mesial fragment; X IPS38614, R M3 distal fragment from SQ-PN; Y IPS31061 [VP1007],R M3; Z IPS31065d [VP1014], L M3; A’ MGSB48829,R M3; B’ MGSB48830,L M3; C’ MGSB48815, R M3; D’ MGSB48813, L M3; E’ IPS95947 [514], R M3 germ; F’ IPS31027 [IPS1882],L M3 from CF2. All specimens come from SQ-TF unless explicitly indicated and are depicted in occlusal view

Fig. 6 Dentognathic remains of ‘Parachleuastochoerus’ valentini described in this paper: A IPS31061 [VP1007], R maxillary fragment with M1–M2 with associated M3, in occlusal (left), lingual (top right), and buccal (bottom right) views; B IPS27505 [IPS1063], partial juvenile mandible with L dp4–m1 and R m1 from SQ-PN,in occlusal (left), right buccal (top right), and left buccal (bottom right) views;C MGSB30480, R mandibular fragment with m1–m3, in lingual (top), buccal (middle), and occlusal (bottom) views; D MGSB30490,L mandibular fragment with m2–m3, in occlusal (left), lingual (middle), and buccal (right) views. All specimens come from SQ-TF unless explicitly indicated

Fig. 7 Permanent lower incisors of ‘Parachleuastochoerus’valentini described in this paper:A IPS96057 [VP1008, 56], L i1; B MGSB48848,R i1; C MGSB48851,R i1 crown; D MGSB48853.3, L i1 germ; E MGSB48853.4,R i1 germ; F MGSB48913,L i1; G MGSB48850, L i2 germ; H MGSB48843, L i2; I MGSB48936.1,L i 2? partial crown;J MGSB48849, L i2; K IPS96087 [VP1004, 60], R i3; L IPS95966 [5456],L i3; M IPS96086 [IPS1859], L i3; N MGSB48863, damaged L i3 germ; O IPS31060 [VP1003], L i3. All specimens come from SQ-TF and are depicted, from left to right, in lingual, mesial, labial, and distal views

Fig. 8 Permanent lower canines of‘Parachleuastochoerus’valentini described in this paper: A MGSB30471 [24], R c1m; B MGSB30471 [23], L c1m; C MGSB30471 [22], L c1m; D MGSB48852,L c1f. All specimens come from SQ-TF and are depicted,from left to right, in lingual, mesial, labial, and distal views

Fig. 9 Permanent lower premolars of ‘Parachleuastochoerus’valentini described in this paper:A IPS96063 [VP1002], L p1; B MGSB48856, L p1; C MGSB48855,L p2; D MGSB48857,partial R p2 crown; E MGSB48858, R p2 mesial fragment; F MGSB48861, R p2 distal fragment; G MGSB48817, damaged L p3; H IPS96062 [IPS1857], partial R p3; I MGSB48859, L p3 mesial fragment; J MGSB48588, L p4; K MGSB48951, L p4. All specimens come from SQ-TF and are depicted, from left to right, in occlusal, lingual, and labial views

Fig. 10 Permanent lower molars of ‘Parachleuastochoerus’ valentini described in this paper: A IPS96060 [VP996], L m1 germ; B MGSB30489, L m1; C MGSB30480,R m1; D MGSB48821,R m1; E MGSB30475, R m1; F IPS27505 [IPS1063],L m1 from SQ-PN; G IPS27505 [IPS1063], R m1 from SQ-PN; H MGSB48919,L m1 (damaged); I MGSB48822, L m1 partial germ; J MGSB48873,R m2 germ; K IPS96061 [VP997], L m2 germ; L MGSB30490, L m2; M MGSB30480,R m2; N MGSB48818, R m2; O MGSB48820,L m2; P MGSB48819, R m2; Q MGSB48823, R m2 mesial fragment; R IPS119090 from SQ-TFC, L m2 mesial fragment; S MGSB48840, R m2 distal fragment; T MGSB48827, R m2 distal fragment;U MGSB30488,L m3 germ; V MGSB30490, L m3; W MGSB30480,R m3; X MGSB48949,R m3; Y IPS119091, partial R m3 from SQ-TFC; Z MGSB48828,R m3 distal fragment; A’ MGSB48825,L m3 distal fragment;B’ MGSB48826, R m3 distal fragment. All specimens come from SQ-TF unless explicitly indicated and are depicted in occlusal view

Fig. 11 Deciduous teeth of‘Parachleuastochoerus’valentini described in this paper: A IPS94956 [VP1005], L DI1; B MGSB48865,L DI1; C MGSB48868, R DI1; D MGSB48869, L DI1; E IPS94954 [IPS1860], R DI1; F IPS95997 [VP988], L DI2; G MGSB48866,L DI3; H IPS95945 [5462], L DP2; I IPS96083 [VP1006],R DP2; J IPS96070 [VP1011], R DP3; K IPS96077, R DP3 mesial fragment; L MGSB48839,L DP4; M IPS31080 [VP1010], L DP4; N IPS95961 [4292], R DP4; O IPS96069 [VP1001], R DP4 crown; P MGSB48864.1, L di1; Q IPS31024 [IPS1908], R di1 from CF2;R IPS96085 [VP1000],L di1; S IPS96084 [VP1028], R di2; T MGSB48864.2, L di2; U MGSB48936.2, L dp2 partial germ; V MGSB48936.3, R dp3 distal fragment; W IPS96075, R dp4; X MGSB48838, R dp4; Y IPS27505 [IPS1063],L dp4 from SQ-PN; Z IPS31073 [VP1012], L dp4; A’ IPS95965, L dp4 mesial fragment; B’ IPS96076,R dp4 mesial fragment. All specimens come from SQ-TF unless explicitly indicated. Incisors are depicted, from left to right, in lingual, mesial,labial, and distal views, whereas premolars are depicted,from left to right, in occlusal, lingual, and distal views, except for DP3s,DP4s,and dp4s, which are depicted only in occlusal view

Fig. 12 Bivariate plots of BL vs. MD for the upper cheek teeth of ‘Parachleuastochoerus’ valentini from Sant Quirze localities as compared with those of this species and other tetraconodontines from elsewhere: A P1; B P2; C P3; D P4; E M1; F M2; G M3; H DP4. Metrical data for the material from Sant Quirze localities are reported in Table 1, while those for the comparative sample have been taken from the literature (see Materials and methods for details on the published sources).Convex hulls outlining the variation of several species (without considering the remains from Sant Quirze localities) are depicted

Fig. 13 Bivariate plots of BL vs. MD for the lower cheek teeth of ‘Parachleuastochoerus’ valentini from Sant Quirze localities as compared with those of this species and other tetraconodontines from elsewhere: A p1; B p2; C p3; D p4; E m1; F m2; G m3; H dp4. Metrical data for the material from Sant Quirze localities are reported in Table 1, while those for the comparative sample have been taken from the literature (see Materials and methods for details on the published sources).Convex hulls outlining the variation of various species (without considering the remains from Sant Quirze localities) are depicted

Fig. 14 Bootstrap 50% majority-rule consensus tree.Those ingroup clades with bootstrap support≥50% are depicted, with bootstrap support indicated below nodes and Bremer’s indices above.The analysis found five most parsimonious trees, whose strict consensus (see Additional file 1: Fig. S1) failed to recover one of the clades weakly supported by bootstrap (Bremer’s index =0)

Fig. 1 A–B Orthophotomap of the Sant Quirze area from 1945 (A) and 2022 (B), indicating the approximate position of the various sites discussed in this paper. C–D Two views of the SQ-TF site in the 1920s. Abbreviations: CCA,Can Canals; CF2, Can Feliu2; CP, Can Poncic; CPL, Can Pallars i Llobateres; MDU, Mas Duran; SQ-PN, Poble Nou de Sant Quirze; SQ-TF,Trinxera del Ferrocarril-Sant Quirze.The UTM coordinates (31N / ETRS89) employed for each site are the following: CCA (424360, 4597110) and MDU (423670, 4598280) correspond to the homonymous farmhouses; CP (421900, 4598475) and CPL (421450, 4598750) approximate location after Alba et al. (2018); CF2 (422480, 4598150) is located based on the current toponym of Can Feliu; SQ-PN has been placed between the Can Barra farmhouse (423430, 4597810) and Los Rosales (423800, 4597620) neighbourhood;finally, SQ-TF has been situated according to Galindo i Torres et al.’s (2001) map (424100, 4597060), approximately coinciding with Sant Quirze C based on Quintana Cardona (1995), although according to the old literature (see main text) it would have extended>200 m along the railway (i.e., approximately 424190, 4596840). Base orthophotomaps for panels A–B © Institut Cartogràfic i Geològic de Catalunya, downloaded from VISSIR v.3.35 and reproduced with permission by means of a Creative Commons license CC BY 4.0 (Institut Cartogràfic i Geològic de Catalunya 2024). Panels C–D reproduced from Bataller (1924)