A new ape from Türkiye and the radiation of late Miocene hominines Author Sevim-Erol, Ayla Ankara University, Faculty of Languages History and Geography, Department of Anthropology, Ankara, Türkiye. aylasevimerol@gmail.com Author Begun, D. R. Department of Anthropology, University of Toronto, Toronto, ON, Canada. david.begun@utoronto.ca Author Sözer, Ç. Sönmez Ankara University, Faculty of Languages History and Geography, Department of Anthropology, Ankara, Türkiye. Author Mayda, S. Ege University Fakulty of Science, Department of Biology, İzmir, Türkiye. Author van den Hoek Ostende, L. W. Naturalis Biodiversity Center, Leiden, The Netherlands. Author Martin, R. M. G. Department of Anthropology, University of Toronto, Toronto, ON, Canada. Author Alçiçek, M. Cihat Pamukkale University, Department of Geology, 20070 Denizli, Türkiye. text Communications Biology 2023 2023-08-23 6 1 842 1 10 http://dx.doi.org/10.1038/s42003-023-05210-5 journal article 10.1038/s42003-023-05210-5 8279657 Anadoluvius gen. nov. Synonomy . Ouranopithecus Bonis and Melentis : Güleç et al. 28 Type species . Anadoluvius turkae comb. nov. Sevim Erol et al. 2023. Etymology . Anadolu is the modern Turkish word for Anatolia and Anatolian. Holotype . CO-205, a fragmented but largely complete male palate with LI1-M3 and RC-M2 (Supplementary Figs. 1 , 2 ). Paratypes . CO-300 (RM 2 ); CO-305 ( male mandibular fragment with RC-M 1 ); CO-710 ( female mandibular fragment with RP 3 - M 2 ); CO-2100 (RI 1 ); CO-2800 (female partial cranium with RC- M 2 , portions of the right maxilla, maxillary frontal processes, frontal maxillary processes and most of the frontal bone) ( Fig. 1 ; Supplementary Figs. 3–5 ) Detailed specimen descriptions and a revised diagnosis for this new taxon appear in the Supplementary Notes 2, 3. The hypodigm is curated in the Department of Anthropology, Ankara University. All samples used in this analysis are listed in Supplementary Tables 1 , 2 . Measurements are provided in Supplementary Tables 3, 4. Supplementary Note 4 presents the results of a comprehensive quantitative analysis of the Anadoluvius hypodigm. Supplementary Note 5 provides details of the phylogenetic analysis. Supplementary Note 6 provides historical, geological and biochronologic background. Table 2 Hominine and pongine synapomorphies (18 taxa, ordered).

Hominine	Pongine
Supraorbital torus	Supraorbital margin
Frontal sinus/glabella	Interorbital space
Ethmoidal sinus	Lateral orbital pillar surface
I 1 marginal ridge shape	Nasal bone length
C implantation	Nasal bone breath
Upper premolar length	Clivus length
Lower molar buccal cristids	Clivus orientation
Supraorbital torus	C inclination
Incisive canal
Incisive foramen size
Upper premolar crests
P 3 hypoproto-postprotocristid

Table 1 Hominine and pongine synapomorphies (18 taxa, unordered).

Hominine	Pongine
Supraorbital torus	Supraorbital margin
Frontal sinus/glabella	Lateral orbital pillar surface
Ethmoidal sinus	Nasal bone length
I 1 marginal ridge shape	Nasal bone breath
C implantation	Clivus orientation
Upper premolar length	C inclination
Lower molar buccal cristids	Upper premolar crests
P 3 hypoproto-postprotocristid

Fig. 1 CO 2100/2800. A female partial cranium. From left to right, palatal, right lateral and anterior views. Fig. 2 Cross sectional anatomy of the palate in Anadoluvius and other hominids (not to scale). Ekembo and extant hominids redrawn from 31 Rudapithecus modified from 40 . Ouranopithecus redrawn from 41 based on a ct scan. The Ardipithecus specimen, modified from 42 , is a surface rendering derived from ct scans and does not show the cross section but the lateral aspect. The Ekembo specimen is based on BMNH 16664, the holotype of Ekembo nyanzae . The Rudapithecus specimens are RUD 12, a female, and RUD 44, a male. The photographs to the right of the line drawings of Rudapithecus are the original specimens. The Anadoluvius specimens are CO-2100/2800 (female, left) and CO-205 (male, right), with photographs of casts of the reconstructed specimens (see SI for details of the reconstruction.) Line drawings of Anadoluvius are original to this work. Cranium . The CO-2100/2800 partial cranium ( Fig. 1 , Supplementary Figs. 2 , 3 ) was recovered with some crushing and displacement of several broken pieces (see SM for restoration details and detailed description.) The frontal bone is nearly intact, missing only portions within the temporal fossa and the squama approaching bregma. This distinguishes it from the most complete facial specimen of Ouranopithecus (XIR-1), which is broken just beyond the superior orbital margins and preserves almost nothing of the frontal squama (Supplementary Fig. 2 ). Though damaged, the position and orientation of the premaxilla is better preserved in CO-2100/2800 than in CO-205 and Ouranopithecus (XIR-1 and RPl 128), confirming previous interpretations of a stepped and overlapping morphology in these specimens 1 – 3 , 11 , 13 – 15 ( Fig. 2 ). The premaxilla of Anadoluvius is short and vertical compared with Pan , Pongo , and australopithecines, and is most like Gorilla and dryopithecins, being relatively short in the alveolar portion but expanded nasally to overlap with the palatine process of the maxilla ( Fig. 2 ) The incisor alveoli are positioned along the mesial transverse plane of the canine crowns ( Figs. 1 , 2 ; Supplementary Fig. 5 ) In two specimens of Ouranopithecus (RPl 128 and XIR 1) the upper incisors are well anterior to the canines (Supplementary Fig. 5 ) In NKT 89 the premaxilla is severely damaged, but the posterior edge of the lateral incisor appears to be aligned with the anterior transverse plane of the canines, a position most like Anadoluvius . The frontal bone of Anadoluvius differs strongly from that of Ouranopithecus in the smooth biconvex squama of the former, contrasting with a broad concavity above glabella in the latter. The superior orbital margins of Ouranopithecus are broad, rounded and slightly projecting while they are sharp and flat in Anadoluvius . Mandible . A principal components analysis based on mandibular measurements available for Graecopithecus , Ouranopithecus and Anadoluvius is presented in Supplementary Fig. 6 and Supplementary Data 1). The Çorakyerler, Nikiti 1 and Graecopithecus mandibles are separated from each other, especially along PC 2, and from Ouranopithecus , illustrating the diversity present in these samples. Supplementary Data 1 includes the data matrix, summary statistics, scores, and loadings. Anadoluvius , like Graecopithecus , and NKT 21, has a relatively narrow mandible compared with the combined sex sample of Ouranopithecus (Supplementary Fig. 7a). Supplementary Fig. 7b compares relative mandibular corpus breadth at each tooth position (P 3 -M 2 ) in Anadoluvius , Graecopithecus , and Ouranopithecus . Anadoluvius is similar in mandibular robusticity at the premolar level but at the level of the molars it matches or strongly exceeds the maximum value in the other taxa (Supplementary Fig. 7b). There is diversity in mandibular robusticity (breadth relative to height) and in dental size ratios as well among the samples of eastern Mediterranean apes (Supplementary Fig. 8a–e). Anadoluvius is distinct from Graecopithecus in all mandibular and dental ratios. Anadoluvius falls beyond the range of variation of Ouranopithecus in relative corpus breadth at M 1 -M 2 (Supplementary Fig. 8a) and M 2 size (Supplementary Fig. 8d). Interestingly, NKT 21 falls outside the Ouranopithecus range in relative P 4 length and M 2 size (Supplementary Fig. 8d–f). It has a relatively short symphyseal-molar distance, at the 25% quartile for the Ravin sample, and a relatively robust mandible at M 1 -M 2 , at the 75% quartile for Ouranopithecus (Supplementary Fig. 8a, c). In Graecopithecus the symphysis is positioned closest to the molars, just barely in the range of the Ravin sample. In summary, in most quantitative comparisons Anadoluvius is distinguished from Ouranopithecus and Graecopithecus . Tooth roots and enamel thickness . Figure 3 shows the root and root canal morphology of CO-300, the male mandible of Anadoluvius (see Methods for segmentation details). Unlike Ouranopithecus , the distal roots of P 3 to M 1 in Anadoluvius and Graecopithecus 11 are single fused roots with two root canals ( Fig. 3 and Supplementary Table 5). Like Ouranopithecus , Anadoluvius has thick enamel. Supplementary Fig. 9 illustrates ranges of variation in relative enamel thickness (RET) in the M 2 of Miocene, Plio-Pleistocene and living hominoids. Anadoluvius has thicker enamel than most Miocene apes, falling at the upper end of the range in Afropithecus 30 Its RET is greater than RPl 641, an M 3 of Ouranopithecus . The relationship between M 2 and M 3 RET is variable in hominoids 30 , but their ranges of variation always overlap. Anadoluvius falls well above the ranges in extant hominids and within the A. afarensis and A. africanus 75% quartiles. Canine size . The results of an ANOVA examining lower canine relative size is presented in Supplementary Table 7. In canine size relative to the geometric mean Ouranopithecus is significantly different from extant African apes in having relatively small canines. The relative size of the CO-305 mandibular canine (0.53) is equal to the mean of Ouranopithecus males and at the low end of the range of variation in Pan males and females. Fig. 3 3-D reconstruction of the left P3 to M1 of CO 300, showing the root, root canal and pulp chamber configurations. Supplementary Table 5 for a comparison of root formulae. Scale =10 mm. Canines are small in the Balkan/Anatolian sample compared with other fossil and extant apes including Ardipithecus , being more consistent with the ranges in Australopithecus . The ratio of lower canine to M 2 size (canine maximum ln x bd/M 2 maximum ln x bd) in Anadoluvius is lower than in any male and most females except Australopithecus (Supplementary Fig. 10a). The Balkan/Anatolian specimens have relatively small canines compared with African apes when scaled to the individual geometric means (Supplementary Fig. 10b and Supplementary Tables 6, 7; geometric mean of 12 variables). Scaled M 2 size is large and beyond the range of variation of African apes in Graecopithecus , Anadoluvius , and Nikiti 1while Ravin de la Pluie Ouranopithecus is intermediate between the other fossil apes and Gorilla (Supplementary Fig. 10c). Canine size is compared to each tooth position in Supplementary Fig. 11a, b. The results of a cladistic analysis using a data matrix of 112 characters and between 18 and 23 taxa are presented in Fig. 4 , Tables 1 , 2 and in the supplementary materials (Supplementary Figs. 12–14; Supplementary Data 2, 3; Supplementary Note 5). All four cladograms are strict consensus. The analysis was run both with all characters unordered and 21 of the 112 characters ordered (see Methods and Supplementary Note 5 for character matrix assumptions). Both analyses yield similar results. The tree topologies are identical, and predictably the tree values are lower in the analyses using ordered characters. All but the cladograms that include the taxa with more than 80% missing data recover a clade that includes Eurasian apes and hominines to the exclusion of pongines and stem hominids (Fig. 4). The potential problem of using data from published sources as opposed to direct observation is illustrated in the results for Sahelanthropus . This taxon, universally attributed to the Homininae and most commonly to the Hominini, is never recovered in these positions in these analyses, being consistently a stem hominid. Sahelanthropus could be coded for 72% of the characters. The potential for coding inconsistent with the criteria used to code other taxa is therefore larger than for Orrorin , which is also coded from the literature, but for which only 29% of the characters could be coded. We mapped synapomorphies and Bremer support values onto two consensus cladograms (Supplementary Figs. 12, 13). Table 1 lists the hominine synapomorphies of the cladogram (ordered and unordered) with the fewest missing data (18 OTUs). A phylogeny consistent with a large majority of the cladograms presented here appears in Fig. 5. Supplementary Fig. 14 shows the results of the analyses of all four taxon sets with all character states ordered. Unlike the unordered and partly ordered analyses, the cladograms with differing OTUs vary widely. The 19 OTU (including Sahelanthropus ) fully ordered analysis is consistent with previous ones while the 18 and 20 OTU analyses result in a pongine clade including all European and Anatolian taxa, which contrasts with all previous cladistic results from analyses with large data sets and numerical cladistic methods (e.g. TNT, PAUP, etc). The 20 OTU analysis includes several other unconventional results such as an Orrorin - Anadoluvius clade that, along with Ouranopithecus as the sister to pongines, and a pongine clade that includes all European and Anatolian taxa. The 23 OTU analysis remains highly unresolved. Given the diversity of results of these all ordered analyses and the broader consistency of the analyses in which the data matrices were either unordered or partly ordered, we consider the latter to be more reliable (see Methods for a discussion of our rationale on ordering character states in this analysis). The consensus cladogram with the three taxa represented by <20% of the character matrix (23 OTUs) is uninformative, given the low level of resolution. The cladograms resulting from the unordered or partly ordered analyses, which exclude these taxa ( Chororapithecus , Samburupithecus and Graecopithecus ) all recover a clade that includes Anadoluvius and Ouranopithecus as sister taxa, which in turn is either the sister taxon to the dryopithecins or in an unresolved polychotomy with the dryopithecins and the crown hominines. Nakalipithecus , which has been interpreted as a potential ancestor of Ouranopithecus , is outside the crown hominids in this analysis, as suggested elsewhere 1 , 31 Ankarapithecus is identified as a pongine and the widely accepted Pan -hominin clade is supported as well. Ouranopithecus , Graecopithecus and Anadoluvius share a suite of derived characters of the jaws and dentition that support their status as a distinct clade. Although Graecopithecus could not be included in the analyses that yielded well resolved phylogenies, due to its many missing data (90%), all previous analyses of Graecopithecus associate it phylogenetically with Ouranopithecus 1 – 3 , 5 , 11 , 13 , 15 , 32 . The core attributes of the Balkan/ Anatolian late Miocene apes are large, thickly enameled molars, transversely robust mandibles, small canines, and large size. However, among these taxa there is diversity. Graecopithecus is distinguished from Ouranopithecus ( contra 35) in its relatively large M 2 compared with both the M 1 and corpus breadth, its more vertical mandibular symphysis and in details of root morphology 1 , 11 , 15 Anadoluvius has the same lower dental root formula (P 3 to M 1 ) as Graecopithecus and both differ from Ouranopithecus . The frontal bone between the superior orbital margin and the anterior temporal line is preserved in XIR-1 ( Ouranopithecus ), which is sufficient to show that it was more vertically oriented than in CO-2100/2800. Anadoluvius is further distinguished from Ouranopithecus and other non-hominin hominines in having mesiodistally shorter canines (as was probably the case in Graecopithecus based on canine root size and shape) that lack mesial grooves and lingual cingula (unknown for Graecopithecus ). Fig. 4 Strict consensus cladograms. The four taxon sets each produced cladograms with the same topology whether character states were left unordered or a subset were ordered (see Methods and Supplementary Note 5 for details). a 18 OTUs. The four taxa with the fewest codable character states ( Graecopithecus , 10%, Chororapithecus ,13%, Samburupithecus , 18%, and Orrorin , 29%) were excluded, as was Sahelanthropus . Both Orrorin and Sahelanthropus were coded from published descriptions, which introduces uncertainty (DRB, who coded all characters in this analysis, was unable to code characters from these taxa through direct observation). b 19 OTUs, with Sahelanthropus added. c 20 OTUs with Orrorin . There is a decrease in resolution with the inclusion of Sahelanthropus and Orrorin but the tree topologies are otherwise consistent. Sahelantthropus is always recovered as a stem hominid and Orrorin as a hominin. The first three cladograms all recover a hominine clade that includes the thickly enameled Balkan taxa and the dryopithecins. d 23 OTUs, including all taxa. Little resolution remains among hominids, with recognized clades (pongines) unresolved. This cladogram also fails to recover Ouranopithecus as a hominine, which is otherwise a common result in previous analyses. Bremer support values, character states, character definitions and the character matrix (nexus) are all included in Supplementary Note 5 and Supplementary Data 3. In quantitative attributes, the PCA (Supplementary Fig. 6) illustrates the overall distinctiveness of Anadoluvius compared with Balkan apes. The male Anadoluvius mandible CO-300/305 is distinguished from male Ouranopithecus in many metric comparisons (canine-M 2 ratio, canine/geometric mean, relative mandibular breadth, symphyseal-molar distance, relative P 4 length and size, and relative P 4 , M 1 and M 2 size (Supplementary Fig. 8a–f). Anadoluvius is distinguished from male Ouranopithecus in canine size relative to postcanine tooth size at every tooth position for both upper and lower tooth rows (Supplementary Fig. 11). Anadoluvius is distinguished from Graecopithecus in relative mandibular breadth, symphyseal-molar distance, P 4 length, M 2 relative size and M 2 size relative to corpus breadth (Supplementary Fig. 8). In addition, the mandibular arch is wider and relative mandibular breadth larger at every dental position in Anadoluvius (Supplementary Fig. 7). Unfortunately, Nakalipithecus is insufficiently preserved to be included in most of these quantitative analyses, which are scaled using the geometric mean. Nakalipithecus lacks one or more of the quantitative attributes needed to generate the geometric mean used in this analysis. We ran an analysis of M 2 size relative to mandibular corpus breadth at the level of mid M 1 , which are data that have been published for Nakalipithecus . In this ratio the Nakalipithecus specimen, which is probably male, falls within the range of gorilla females, Pan of both sexes, and female Ouranopithecus , outside the range of gorilla males, Ouranopithecus males, Nikiti, Graecopithecus , and Anadoluvius (Supplementary Fig. 8f). Ouranopithecus and Anadoluvius lack shared derived characters of the pongines (greatly elongated premaxilla substantially or completely overlapping the maxillary palatine process, expanded zygoma, tall orbits, narrow interorbital space, reduced or absent ethmoidal frontal sinus, circumorbital costae.) There is no evidence for their inclusion in Ponginae. Ouranopithecus and Anadoluvius share with dryopithecins (European middle and late Miocene apes with affinities to Dryopithecus ) a series of characters found among hominines 1 – 3 These include a ventrally rotated palate, a stepped subnasal fossa, broad, flat nasal aperture base, short nasal bones, nasal aperture apex superior to the infraorbital margins, robust lateral orbital pillars, frontal sinus expanded below nasion, incipient supraorbital torus, more horizontal frontal squama 1 – 3 The phylogenetic significance of some of these shared attributes is disputed, particularly concerning the dryopithecins 9 , 24 . However, there is broad agreement that Ouranopithecus shares enough derived characters with hominines to warrant inclusion in that taxon 1 – 3 , 8 – 20 ( Table 1 ). The phylogenetic results presented here regarding Ouranopithecus and its sister, Anadoluvius , are consistent with many previous analyses 1 – 3 , 8 – 20 They are robust in terms of the number of synapomorphies and Bremer support values for many clades (Supplementary Figs. 12, 13 ). The results also recover the widely accepted relations among crown hominids and hominines and relations within all fossil clades. The Bremer support values for the hominid clade as defined here (15-16) are extremely strong. The hominine clade as defined here is moderate to strong (2-3), and very strong (4 to 8) for the pongine clade (Supplementary Figs. 12, 13 ). European and eastern Mediterranean apes are classified as hominine in three of four consensus cladograms. It fails to be resolved only in the analysis of all 23 taxa, including Samburupithecus , Chororapithecus and Graecopithecus , missing 82%, 87% and 90% of the data respectively. The latter cladogram also fails to distinguish between pongines and hominines. Fig. 5 A phylogeny of the taxa included in this analysis consistent with most of the cladograms presented here. Taxa are positioned in chronological order without regard to geography,with most taxa only known from a limited time span. Exceptions are Ekembo and Sivapithecus , with longer time ranges,which are positioned roughly when they are most abundant, in both cases about mid-way in their known time ranges. The different colored “ puddles ” represent hominid clades and/or stages of evolution. These can also be imagined as pools of related species in somewhat delimited space and time with broad ancestordescendant relationships. The lines are disconnected to reflect the difficulty in identifying actual ancestor-descendant relationships, but that these relations can be estimated between “ puddles ” . Blue puddle taxa are stem hominids and are all confined to Africa. Among these taxa the relations of Samburupithecus and Chororapithecus are unresolved in the cladograms except in so far as they are excluded from the clade that includes all Eurasian taxa and crown hominins. Other lines of evidence suggest that these taxa are members of the early or middle Miocene radiation of early apes (see text). The orange puddles are the pongines, which probably have their origin within the middle Miocene puddle, although not necessarily any of the taxa included here (another taxon, Griphopithecus , known from Europe and Türkiye, would be a member of the blue puddle but was not included in the cladistic analyses).While both are pongines, Ankarapithecus lacks derived features shared by Sivapithecus and Pongo , so the line representing the relationship between the latter two bi-passes Ankarapithecus . The three green-shade puddles represent the hominine clades as defined here.Bright green are the dryopithecins,with the younger taxa Rudapithecus and Hispanopithecus depicted as closely related and descendant from any of the older taxa or an unknown taxon sharing attributes with these three. The light green puddle includes the Balkan and Anatolian taxa, likely to have descended from somewhere in the dryopithecin puddle.Possible ancestor-descendant relationships are depicted in this puddle. The darker green puddle represents the crown hominines. The various lineages diverge from unknown ancestors, but probably a member of either of the older green shade puddles.Gorillas diverge first, followed by chimpanzees and humans. Orrorin and Ardipithecus are depicted in a manner consistent with their sister clade status, without implying a direct ancestor-descendant relationship. The relations among the dryopithecins are consistent with most detailed analyses focused on this group, as are the relations within the Asian clade ( Ankarapithecus , Sivapithecus and Pongo ) and the crown hominines 1 – 3 , 8 – 20 The consistencies of these cladograms in many details with previous research lend credibility to these results. Recent analyses by (9 and 10) are broadly similar in their results. Among the most parsimonious cladograms reported in ref. 10 at least one also recovers a hominine clade that includes European fossil taxa9. also recovers Ouranopithecus as a hominine though in that analysis the dryopithecins are recovered as stem hominids. Other taxa have been linked with Ouranopithecus or hominines. Three fossil apes from Africa, Nakalipithecus , Samburupithecus and Chororapithecus , broadly overlap in time with Ouranopithecus , Graecopithecus and Anadoluvius 21 – 23 Nakalipithecus has been identified as potentially ancestral to Ouranopithecus 22 . However, Nakalipithecus differs in many details of dental morphology from Anadoluvius and Ouranopithecus (see Supplementary Note 3). As noted, Nakalipithecus , represented in the data matrix by 43% of the total number of characters, consistently falls outside the crown hominids, which fails to support the hypothesis of an ancestral-descendant relationship with Ouranopitheus . It has been hypothesized that both Chororapithecus and Ouranopithecus have phylogenetic affinities with gorillas 8 , 22 , 23 , 33 . However, Chororapithecus is readily distinguished from both Anadoluvius and Our anopithecus (see differential diagnosis), and its affinity with gorillas has been questioned 1 , 34 A close phylogenetic relationship with gorillas is not supported by our results21. claim that Samburupithecus has phylogenetic affinities with African apes and humans, though this conclusion has also been challenged by 1 , 35 , who conclude that Samburupithecus is not a hominid but instead a vestige of the early Miocene proconsuloid radiation. As with Chororapithecus and Nakalipithecus , Samburupithecus falls outside the crown hominid clade. Diversity and paleobiogeography . A comprehensive review of the taxonomy and phylogeny of late Miocene apes is needed, given recent discoveries and reinterpretations. Here we focus on the diversity and paleobiogeographic implications of Anadoluvius (Supplementary Fig. 15; Supplementary Table 8). Anadoluvius and Ouranopithecus share attributes with other European middle and late Miocene hominids that distinguish them from late Miocene ape fossils from Africa and the broadly contemporaneous pongines Sivapithecus , Ankarapithecus , and Khoratpithecus (Supplementary Note 3). Ouranopithecus from Ravin de la Pluie and Xirochori are dated to 9.6 and 9.3 Ma respectively 6 , 7 . Ouranopithecus from Nikiti 1 is dated to 8.9 Ma 6 , 7 . Çorakyerler is dated to 8.7 Ma (Supplementary Figs. 16, 17; Supplementary Table 9 and Supplementary Note 6). The Nikiti mandible and maxilla are distinct from Ouranopithecus from the more northern Macedonian sites and may represent a different taxon 13 . This possibility, which needs further study, is interesting in terms of regional evolution as Nikiti 1 is likely to be slightly older than Çorakyerler 6 , 7 , 29 . Nikiti, Anadoluvius and Graecopithecus are distinguished from Ouranopithecus in our PCA and in having a greater degree of canine reduction, elongated P 4 and large molars relative to mandibular corpus size. Anadoluvius and Graecopithecus are distinguished from Ouranopithecus in root morphology (unknown in NKT 21.) NKT 21 and Graecopithecus have inferior transverse tori positioned posterior to the mesial edge of the M 1 (unknown in Anadoluvius ), while the torus is anterior to the M 1 in Ouranopithecus . Graecopithecus , which is considerably younger (~7.2 Ma) 4 , has been shown to differ from Ouranopithecus in morphology that replicates differences between late Miocene apes and early hominins, such as reduced relative canine size and premolar root morphology 11 However, the Nikiti specimens have not previously been included in these comparisons. Quantitative and qualitative comparisons reveal many differences among Balkan/Anatolian taxa, indicative of a greater diversity of late Miocene eastern Mediterranean hominids than previously recognized. Hominines appear to have been present and diverse for millions of years in the late Miocene of Europe. Based on the large number of qualitative and quantitative differences among the samples from Macedonia, Attica and Anatolia, we conclude that they represent at least three hominine genera, Ouranopithecus , Graecopithecus , and Anadoluvius . The diversity of hominines in the eastern Mediterranean mirrors that among australopithecines in the Plio-Pleistocene hominin record in Africa. In the phylogenetic analysis presented here the Balkan/ Anatolian taxa are in the sister clade of crown hominines. The fact that dryopithecins are also classified as hominines in this analysis suggests that there was an in situ evolution of thickly enameled late Miocene eastern Mediterranean hominines from more thinly enameled precursors in central and western Europe, though this conclusion has been challenged 9 , 24 . Pierolapithecus, Anoiapithecus , Dryopithecus , Hispanopithecus and Rudapithecus all share attributes with extant hominines and are distinguished from pongines such as Sivapithecus and Ankarapithecus ( Tables 1 , 2 ) 1 – 3 , 36 , 37 , (but see refs. 9 , 24 for alternative views). A clade that includes both thinly and thickly enameled taxa, in this case the dryopithecins and the Balkan/Anatolian apes, has a parallel in Africa with Ardipithecus and australopithecines and with Pan and hominins. The parallel evolution of thickly and thinly enameled members of a clade in Africa and Europe is not proof that the late Miocene European apes are all hominines, but it does make this hypothesis, supported by the results of our cladistic analysis, plausible. It is possible that the generally more thinly enameled dryopithecins and the later occurring thickly enameled Balkan/ Anatolian hominines do not share an ancestor-descendant relationship and represent separate dispersal events into Europe from Africa (e.g. ref. 22 ), though this is less parsimonious biogeographically and contrasts with the results of the phylogenetic analysis presented here. While independent dispersal events are possible, we regard the in situ European hypothesis as more likely and more parsimonious given the current evidence. Other independent lines of evidence are also consistent with the widespread presence of hominines in Europe 38 .