Cranium and associated mandible DKO-38. This laterally compressed specimen was originally described when cranium and mandible were still joined (de Bonis et al., 1990). It belongs to an aged individual lacking the left facial part and part of the cranial roof (Fig. 1A). The large and sharp upper and lower canines, the strong, thickened and projected supraorbital torus, and the well-defined sagittal crests indicate a male individual (Koufos et al., 2003 and Szalay and Delson, 1979). The right orbit is well preserved, but it is deformed by lateral compression of the skull and became egg-shaped (Fig. 1A₄). The zygomatic bone lies deep below the orbit. The upper dentition preserves the right tooth row, the left broken canine, and the P4 (Fig. 1A₃, ₅). The upper canine, although worn, is large, sharp, with large lingual wear surface, strong distal projection and triangular basal occlusal outline. The occlusal surface of the molars is extremely worn. The mandible preserves the right mandibular ramus and part of the left one, the latter lacking the ascending portion (Fig. 1B₁). The external aspect of the symphysis is slightly distorted but, as in M. pentelicus, it is rounded, without lateral constriction. The mandibular corpus is relatively deep (Table 2), with straight ventral margin. The ascending ramus is not high, the distance from the coronoid process to the ventral margin of the corpus corresponding to 51.0 mm. The condyle is small (length = 11.3 mm; breadth = 6.1 mm) and the coronoid process is low. The height measured from the coronoid process to the condyle is 10.3 mm. The lower dentition preserves both series, lacking only the incisors (Fig. 1B₁, ₂). The lower canine, bearing a large wearing facet on the distal surface and a large distal projection, is strong and sharp. The p3 has a clear honing facet extending to its mesio-lingual root. The molars are markedly worn.

Cranium and associated mandible DIT-21. The specimen, crushed and strongly compressed laterally, was originally numbered as DIT-80 (de Bonis et al., 1990: appendix 1). It belongs to a rather old individual bearing a markedly worn dentition (Fig. 2). Attempts to separate the mandible from the cranium were unsuccessful. During our intervention, a small part from the right maxilla went broken and thus it was possible to photograph and measure the teeth. The right orbit is partially preserved, but it is deformed. The right mandibular ramus, although crushed, is better preserved than its counterpart. It is relatively high and likely thick, as its measured thickness is slightly smaller than the real one due to crushing and compression (Table 2). The upper dentition preserves the right C–P4 and M3 and the left C–M3. The well-preserved left upper canine is large, worn and sharp, with a triangular basal outline in occlusal view. The right P3 and P4 are nearly entirely worn and their occlusal surface consists of a large dentine pit surrounded by thin enamel. The preserved M3 is also worn and has an almost rounded occlusal outline. The lower dentition is nearly complete, lacking the left i1 and i2, as well as the right p3 (Fig. 2F). The lower canines are strong and sharp, with developed distal projections and strongly worn distal surfaces. The well-preserved left p3 is elongated and narrow; its wear resulted into a large distal fovea and a honing facet extending to its mesial root, as well as into an entirely worn labial cuspid. The p4 and m1 are either entirely worn (what remains of the right m1 is just a root), or indiscernible/unreadable. However, the m3 is less worn than the other molars and preserves a small hypoconulid.

Left maxillary fragment with C–M1 and M3 DIT-25. It is a small maxillary fragment with extremely worn teeth (Fig. 1C). The canine is large and similar to the other specimens. A clear groove, extending to the root, can be seen on its mesial surface. The canine root is long and strong (mesial height of the root = 26.8 mm).

Left mandibular fragment with p3–m1 and m3 DTK-275. The specimen is erroneously mentioned as DTK-235 (de Bonis et al., 1990: fig. 7; appendix 1). It lacks the ventral margin of the mandibular corpus, the ascending ramus and the m2 (Fig. 1D). The p3 bears two cuspids, a large labial (protoconid) and a small lingual one (metaconid); its mesial fovea is small, while the distal one is large. The honing facet is well-developed, extending to the mesial root of the tooth. The p4 is more symmetric than the p3, with a rectangular occlusal outline. It bears two mesial cuspids, a small mesial fovea and a larger distal one. The m3 has the typical form for Mesopithecus, with four cuspids and a small and monocuspid hypoconulid.

DKO-38 has been originally described as M. aff. pentelicus based on some differences with respect to the typical M. pentelicus from Pikermi (comparisons limited to the MNHNP collection) (de Bonis et al., 1990) and this attribution was later reported in all articles referred to the Axios Valley Mesopithecus. The original determination mostly relies on the smaller and higher orbits, the thicker supraorbital torus, the zygomatic bone higher under the orbit, and the slightly deeper canine fossa. In the initial description, the comparison of the dentition was not possible because cranium and mandible were still in connection. Their recent separation and new comparisons with a larger sample of M. pentelicus from Pikermi (collections of the MPGA, MNHNP, NHMUK, BSPM, NHMW and PIW) currently allow a revised identification of this specimen.

Most of the M. pentelicus skulls from Pikermi have a deformed face, the best-preserved specimen being NHMUK-M. 8945. The shape of the orbit of DKO-38 differs from that of M. 8945. In DKO-38, the maximal orbital height and breadth correspond to 24.2 mm and 21.4 mm, respectively, while in the Pikermi sample they range between 19.7–25.5 mm and 23.0–30.0 mm, respectively. However, the difference for the orbital breadth clearly results from the lateral compression and deformation suffered by DKO-38. The thickness of the supraorbital torus of DKO-38 is 4.8 mm, which is well within the range of variation of M. pentelicus fromPikermi (3.6–5.5 mm) and Kalimantsi (4.2–5.0 mm; Koufos et al., 2003). At first glance, DKO-38 seems to have a deeper zygomatic bone than M. pentelicus; however, the distance from the lowermost point of the orbit to the zygomaxillary suture of DKO-38 is 16.8 mm, which is within the range of variation of the M. pentelicus from Pikermi (13.2–18.0 mm; Delson, 1973). The canine fossa (= suborbital fossa) looks slightly deeper than in M. pentelicus, but again this is likely due to cranial deformation. According to de Bonis et al. (1990), the dental size of DKO-38 fits better the estimates of M. pentelicus. Indeed, most dental dimensions of DKO-38 support this interpretation, but some measurements are smaller, even when compared to those of the mean female M. pentelicus (Fig. 3A, B). Such differences could indicate a slight dental size decrease in the late Turolian forms of M. pentelicus.

Principal Component Analysis (PCA) of the upper and lower cheek teeth dimensions of DKO-38 and of the large dental samples of M. pentelicus from Pikermi and Kryopigi (Greece) and Kalimantsi (Bulgaria) provides distinct comparative evidence (Fig. 4). The upper dentition of M. monspessulanus is not included in the PCA because the known material is scarce (see Section 4.2.2). The PC1 for the upper teeth (Fig. 4A), which accounts for 67.9% of the total variance, well separates M. pentelicus males and females, respectively located within its positive and negative spaces. In this comparative context, the upper dentition of DKO-38 is closer to that of the M. pentelicus males, but the dimensions of the upper postcanine teeth are slightly smaller than those typical of this taxon. Concerning the lower teeth, the PC1 (62% of variance) again separates males and females of M. pentelicus (Fig. 4B). However, the PC2 (19.4% of variance) separates M. pentelicus from M. monspessulanus, located at its positive and negative parts, respectively. DKO-38 falls within the convex hull of the M. pentelicus males, suggesting size similarities with this species (Fig. 4B). Besides their morphological similarity with those of the M. pentelicus males, the upper and lower canines of DKO-38 are also included within the male size variation of this species, thus supporting most likely attribution of DKO-38 to a male individual (Fig. 5).

The strong deformation and the extreme dental wear of DIT-21 inhibit some morphological comparisons; accordingly, its comparison with M. pentelicus is mainly based on the dental size. The upper dental dimensions of DIT-21 are close to the mean of the estimates of female M. pentelicus, except those of the canine and the P3 length. The P3 breadth is remarkably smaller than measured in female M. pentelicus individuals, but this is likely due to its extreme wear (Fig. 3A). However, the upper canine morphology and size are similar to those of the male M. pentelicus specimens (Figs. 4A). The lower teeth of DIT-21, except the canine, are generally shorter than those of M. pentelicus, but their breadth fits with those of this species (Fig. 3B). In the PCA diagram for the lower teeth, DIT-21 falls within the convex hull for the male M. pentelicus (Fig. 4B). Moreover, its lower canine has large size and matches the values reported for the male M. pentelicus (Fig. 5B).

The specimen DIT-25 does not display maxillary and dental morphological characters suitable for taxonomic assessment, and its determination is uniquely based on dental dimensions, which are close to the estimates for the male M. pentelicus (Fig. 3A). Its canine morphology and its size (Fig. 5A) suggest that it represents a male individual. In the PCA diagram, DIT-25 plots near the convex hull of the male M. pentelicus individuals (Fig. 4A), which confirms its specific and sexual attribution.

The mandibular fragment DTK-275, possibly erroneously reported as M. monspessulanus by Jablonski (2002), preserves the m3. Except for the m3 breadth, the dental dimensions in this specimen are similar to the mean of those recorded for the male M. pentelicus (Fig. 3B). In the PCA diagram, DTK-275 plots together with M. pentelicus within the PC2 positive space, but its sex attribution is uncertain as it is situated in an intermediate position between M. pentelicus males and females (Fig. 4B). Although DTK-275 has only modestly worn teeth, some of its dental dimensions are smaller than those of M. pentelicus, especially the breadth of the molars (Fig. 3B). The narrow lower molars are the main distinctive character separating M. monspessulanus from M. pentelicus (Delson, 1973), thus indicating that some monspessulanus-like features appeared in the late Turolian forms of M. pentelicus.

In conclusion, the relatively large-sized Dytiko Mesopithecus is similar in size and morphology to M. pentelicus and can be attributed to this species. However, the Dytiko M. pentelicus preserves some characters (e.g., narrow lower molars, slightly smaller dental dimensions) similar to those of M. monspessulanus, thus suggesting relationships between the two taxa.

Male mandible with i1–m2 DIT-22. The specimen (Fig. 6A) has been described earlier (de Bonis et al., 1990: fig. 7B).

Maxillary and associated mandibular fragments DTK-276. The specimen is reported as DTK-240 in the original publication (de Bonis et al., 1990: appendix 1). It belongs to an adult individual and preserves the P4–M3 tooth row (Fig. 6B). The base of the zygomatic arch is well distinct, and its anterior margin is situated above the M1/M2 contact (Fig. 6B₁). The P3 has an almost triangular shape; the mesio-labial corner projects mesially, giving a triangularly-shaped occlusal outline. It bears two main cusps, the labial one being higher than the more worn lingual cusp. The occlusal surface is worn, forming a large dentine pit originating from the connection of the corresponding pits on the cusps and the foveae (Fig. 6B₃). The P4 is also bicuspid, but more symmetric and less worn than the P3, with a more rounded occlusal outline. The M1 is badly preserved, having been partially damaged during the separation of the maxilla from the mandible. The M2 lacks the buccal part of the metacone; the paracone is the largest cusp and the metacone is the smallest one. The M3, lacking the lingual part of the hypocone, is similar in shape to the M2, but smaller.

The mandibular fragment DTK-276 preserves the molars and the p4 root (Fig. 6C). The corpus is shallower than typical in M. pentelicus (Table 2). The molars are well preserved, except the mesio-labial part of the m1. The m1 is worn, bearing a large dentine pit almost covering the entire occlusal surface; the protoconid and hypoconid are more worn than the paraconid and the metaconid. The well-preserved m2 is similar in morphology to the m1. The less worn m3 is elongated and bears a fifth cuspid (hypoconulid) (Fig. 6C₃).

Mesopithecus monspessulanus was originally found in Montpellier, France, in the middle Ruscinian freshwater marls of the foundations of the Palais de Justice (Delson, 1973). The known European material of M. monspessulanus is scarce and fragmentary, one of the best known specimens being the female mandible UM 4043 from the Pliocene site of Montpellier described and figured by Mauche (1906: pl. VII, fig. 5–6), as well as two male fragmentary mandibles (NHMB-VJ 87 and NHMB-VJ 130) and a male left hemi-mandible from Villafranca d’Asti (Italy), dated to the upper Pliocene, MN 16, 3.3–3.0 Ma (Gentili et al., 1998, Hürzeler, 1967, Pradella and Rook, 2007 and Rook et al., 2001). The comparison of DTK-276 and DIT-22 with a cast of UM 4043, housed at the LGPUT, shows that the height and thickness of the mandibular corpus are similar (Table 2). The mandibular dimensions of the Greek material are also similar to those of the Italian specimens of M. monspessulanus (Table 2). The main difference of M. monspessulanus with M. pentelicus is its smaller and narrower lower teeth (Delson, 1973). The Simpson logarithmic ratio diagrams comparing the mean dental dimensions of DTK-276 and DIT-22 show that they are closer to M. monspessulanus, more precisely to the males of this species (Fig. 7B). In the PCA diagram for the lower teeth (Fig. 4B), DTK-276 and DIT-22 are close to the convex hull defined by the male M. monspessulanus representatives. In DIT-22, the presence of the canine tooth confirms its sexual attribution, as it distinctly matches with the male canines of M. monspessulanus (Fig. 5B). Conversely, DTK-276 lacks the canine and it is thus difficult to certify its sex. Nonetheless, its location within the PCA diagram suggests a slightly larger size than typically measured in the Pliocene representatives of this species.

The available European M. monspessulanus remains bearing upper teeth are scanty and consist of a few isolated teeth of unknown sex. In the PRIMO database (Primo, 2019), one male upper canine, as well as one m1, one m2 and two m3s of unknown sex, are included, and we used them in a Simpson diagram comparing the upper dentition (Fig. 7A). The maxillary fragment DTK-276 has generally smaller dental dimensions than measured in M. pentelicus, although some are close to this species. It is also interesting that the M3 size (mean of two specimens one from Dorkovo, Bulgaria, and another from Red Grag, England) are very close to M. pentelicus, which suggests that their attribution may deserve revision.

Based on the currently available morphological and dimensional record, the relatively small-sized Dytiko sample of Mesopithecus (DTK-276 and DIT-22) can be more confidently attributed to M. monspessulanus. However, its absolute dental dimensions are slightly larger than those measured in the typical Pliocene M. monspessulanus representatives.