Dorcatherium maliki, Kostopoulos & Sen, 2016

Kostopoulos, Dimitris S. & Sen, Sevket, 2016, Suidae, Tragulidae, Giraffidae, and Bovidae, Geodiversitas 38 (2), pp. 273-298 : 277-282

Dorcatherium maliki n. sp.

( Fig. 3 ; Tables 2, 3)

Dorcatherium puyhauberti – Malik & Nafiz 1933: 61.

Dorcatherium jourdani – Nicolas 1978: 476.

Dorcatherium cf. jourdani – Geraads et al. 2005: 536 .

HOLOTYPE. — TTMEU-CY-141, partial left mandibular ramus with p4-m3, illustrated by Geraads et al. (2005: fig. 10I).

FORMATION AND LOCALITY. — Yulaflı site, 10 km NW of Çorlu in Turkish Thrace (41°12’13”N, 27°49’46”E), channel-filling sand horizons of Ergene Fm ( Geraads et al. 2005).

AGE. — Late Miocene, Vallesian.

ETYMOLOGY. — Dedicated to Ahmet Malik Sayar, discoverer of the Küçükçekmece fossil site.

DIAGNOSIS. — Medium sized Dorcatherium characterized by advanced selenodonty, absolutely and relatively long premolars and M3/m3; tricuspid dp2/(p2) and p3 without bulky (bunodont) protoconulid; protoconid and hypoconid of dp2 and p3 of similar lengths; simple distal fossette on p4; presence of a distinct lingual protocristid on the lower molars; long metatarsals and large astragali.

DIFFERENTIAL DIAGNOSIS. — Dorcatherium maliki n. sp. is larger than D. guntianum von Meyer, 1846 and smaller than D. peneckei (Hofmann, 1893) and D. vindebonense von Meyer, 1846 . It also differs from D. peneckei , D. vindebonense and from the slightly smaller D. crassum (Lartet, 1851) in the more advanced selenodonty. Dorcatherium maliki n. sp. distinguishes from D. crassum by the presence of a well-developed lingual cuspid on P2; the less bulky protoconulid of dp2 and p3; the longer p3 and the slightly longer and relatively narrower p4; the longer hypoconid on dp2 and p3 compared to the protoconid; the shorter Dorcatherium -fold on the lower molars (always much shorter than the postmetacristid); the buccallly shifted third lobe of m3; the distally larger radius with wider and shallower groove for the extensor carpi radialis and much weaker groove for the common extensor tendon of the digits; the absence of fusion between tibia and maleolus lateralis and the much narrower and more plantarily extended medial concavity of the cochlea tibia; the larger astragali with weaker angle between the axes of the proximal and distal trochlea, wider lateral condyle of the caput tali and more prominent distally medial condyle; the calcanei with weaker coracoid process, longer and narrower articular surface for the malleolar, and plantary facing articular surface for the cubonavicular; and the longer metatarsals with much wider frontal groove between the MtIII and MtIV and more unevenly placed proximal surfaces for the cubonavicular and the ectomesocuneiforme. Dorcatherium maliki n. sp. differs from D. jourdani Depéret, 1887 in having broader upper and lower molars, longer m3 compared to m2, and tricuspid dp2. Dorcatherium maliki n. sp. differs from D. puyhauberti Arambourg & Piveteau, 1929 in the longer m3 compared to m2, and the presence of cingulum all around the lingual and distal wall of the posterior lobe of upper molars. Dorcatherium maliki n. sp. resemble more closely D. naui Kaup, 1833 from which, however, it differs in the slightly larger M3/m3, the longer lower premolars, the longer hypoconid on dp2 and p3 compared to the protoconid, the strong protoconulid of dp2, the simple distal fossette of p4, and the frequent presence of a distinct lingual protocristid on the lower molars.

STUDIED MATERIAL. — Küçükçekmece West: right dP4, MNHN.F.TRQ763; left dP4,TRQ762, TRQ764; left P2,TRQ759; right M1, TRQ754, TRQ755; left M2, TRQ751, 752; right M2, TRQ753, TRQ757; right M3, TRQ750; right dp2, TRQ779; right p4, TRQ772; left p4, TRQ768; partial right mandibular ramus with m1-m3, TRQ748; left m2-m3, TRQ749; right m1, TRQ765, TRQ775; right m2, TRQ783, TRQ767, TRQ782; left m2, TRQ766, TRQ771, TRQ no No. (two specimens); right m3, TRQ784, TRQ769; left m3, TRQ785, TRQ770; distal humerus, TRQ727, TRQ728; distal radius, TRQ733; distal metacarpals TRQ746, TRQ747; tibia, TRQ718; distal tibia, TRQ717, 720, 721; astragalus,TRQ706, TRQ707, TRQ708, TRQ710, TRQ711, TRQ713, TRQ714, TRQ716; calcaneum, TRQ703, TRQ704, TRQ705; metatarsal, TRQ741; distal lateral metatarsal TRQ765, Phalanx I, TRQ729, TRQ731.

Küçükçekmece East: left P4, ITU no No.; M1, ITU no No.; M3, ITU379, ITU no No.; left p3-p 4 in situ, ITU371; right p3, ITU371; left p4, ITU no No.; right m1, ITU no No.; left m3, ITU380, ITU no No.; astragalus, ITU352.

MEASUREMENTS. — See Tables 2 and 3.

DESCRIPTION

Only isolated upper teeth are preserved; among them, there are three molariform dP 4 in different wear stages. They show a proportionally stronger and more mesially protruding parastyle than the permanent molars ( Fig. 3G ). The preparacrista ends mesially just in front of the parastyle. Both the paracone and the mesostyle are well pronounced and coniform, whereas the metacone is significantly weaker and the metastyle barely distinct. A weak crest at the distolingual wall of the paracone is present in the two less worn specimens. The postprotocrista is faint and short, weakly distinct from the long neocrista. A strong cingulum runs along the base of the mesolingual wall to the middle of the posthypocrista. A kind of basal knob or pillar develops between the lobes in two of the three available specimens.

The tri-rooted P2 ( Fig. 1C ) is much longer than wide with well-developed lingual (protocone) and buccal (paracone) cuspids and a distinct distolingual cingulum. Mesially, the tooth bears a strong buccal style and a well-developed lingual cuspid.

The single P4 ( Fig. 1D ) is three-rooted and has strong paracone and protocone. The protocone shows well-expressed selenodonty and its height reaches almost 2/3 of the paracone height. The preparacrista is long, leading to a well-developed parastyle that projects mostly mesially. The postparacrista is equally long but less sloping down and leads to a weakly developed metastyle that bears a blunt distobuccal cingulum. The preprotocrista reaches marginally the parastyle. The postprotocrista turns buccally, terminating in between the paracone and the metastyle, and closing the fossette. The distolingual cingulum is strong rising up to the distal surface of the protocone, which is also surrounded by a well-developed cingulum.

The preserved upper molars show advanced selenodonty though in moderate to advanced wear stage, and thus little informative ( Fig. 3J ). The prehypocrista is significantly longer than the neocrista and reaches the lingual wall of the postparacrista. In two molars at least, there is evidence of a very weak postprotocrista. The para- and mesostyle are strong and the paracone has well-convex lingual and buccal ribs. The cingulum is strong all along the lingual wall of the molars and mostly on the mesial lobe ( Fig. 3J ).

Discriminating between isolated dp2, p2 and p3 is quite difficult in case all appear tricuspid. We assume, however, the isolated and tricuspid MNHN.F.TRQ779 ( Fig. 3E ) is a dp2 as its roots are more divergent than in permanent premolars and the talonid is high (Geraads, Aiglstorfer & Rössner pers. comm. 2016). A similar tricuspid dp2 from Küçükçekmece East is also illustrated by Malik & Nafiz (1933: pl. XI, fig. 2). Both the dp2 and p3 ( Fig. 3B, E ) are very similar in morphology: tricuspid, long, and buccolingually compressed, though the p3 width reaches distally the minimum values for the p4. The protoconid is as long but slightly higher than the hypoconid, whereas the protoconulid is shorter than the other two, low, directs mesio-lingually, and though well developed it is not bulky unlike D. crassum (Lartet, 1851) . The prehypocristid of both dp2 and p3 is shorter than the preprotocristid. Two sharp distal cristids branch from both the protoconid and the hypoconid. One is running along the mesio-distal axis of the tooth (postprotocristid and posthypocristid, respectively) and is long and slightly convex. The other is much shorter (about half of the length of the buccal cristids) and directs distolingually. Distally, the posthypocristid joins near the base the weakly developed but clear distolingual cingulum and, together with the distolingual cristid of the hypoconid, they define a subrounded distolingual valley. A weak basal lingual cingulum also connects the protoconulid with the protoconid.

The p4 ( Fig. 3B, H ) is slightly broader than the p3. The protoconid (mesiobuccal conid) dominates at the center of the tooth. The preprotoconulid is undistinguished from the preprotocristid even at initial wear stages. Their contact is marked by the deep and abrupt buccal clefts of the wide and open lingually anterior valley. An incipient preparacristid is present in two out of four specimens.Two cristids branch on the posterior part of the protoconid: the lingual one is almost parallel to the mesiolingual axis of the tooth and reaches the distolingual corner, and the distal one fuses with the hypoconid. The hypoconid is well convex buccally-distobuccally, more so than the protoconid. A narrow posthypocristid directs vertically compared to the mesiodistal axis of the tooth reaching the distolingual corner. The long and rather narrow posterior fossette (fossetid) of the tooth is almost fully enclosed but remains narrowly open at the distolingual corner till advanced wear stage. In one specimen (MNHN.F.TRQ772) an additional isolated tubercle appears in the mesial-most corner of the posterior fossette. Short mesiolingual and distolingual cingula are developed.

The lower molars ( Fig. 3A, F, K ) are rather narrow and relatively high (height represent 70-80% of the maximum length). The length of the lower molar row is 41.8 mm in MNHN.F.TRQ748( Fig. 3A ). The distal lobe of m1 and m2 is slightly broader than the mesial one. The m1 and m2 ( Fig. 3A, F, K ) have well-developed Σ- structure. The Dorcatherium -fold is always short representing about half of the postmetacristid length. The postprotocristid and the Tragulus -fold are equally developed. In some specimens the postmetacristid passes over the postprotocristid, reaching the terminal point of the Tragulus -fold. Zhailimeryx -fold is weak to absent. In most specimens (6 out of 8 in initial to mid wear stage) the postentocristid is double, formed by a long and thin lingual and a very short buccal cristid. The posthypocristid is longer than the postentocristid and turns lingually, closing the posterior fossette. The lingual wall of the protoconid shows an additional crest (called lingual protocristid) invading the anterior fossette and reaching the base of the premetacristid; though always present, its development varies from weak (knob-like) to important (c. half of the preprotocristid length). A small ectostylid is usually present on m2 but very weak to absent on m1. The mesial cingulum is strong, especially mesiolingually; the buccal cingulum is less marked. The morphology of the two mesial lobes of the m3 is identical to that of m1- m2 ( Fig. 3A, I ). The posthypocristid is divided distally into a long internal and a very short external cristid. The former reaches lingually the postentocristid, closing the distal fossette, whereas the latter connects the prehypoconulid-cristid with wear. The posthypoconulid-cristid is rather short leading to a small entoconulid. The prehypoconulid-cristid extends lingually reaching the short postentoconulid-cristid, and closing the back fossette of the m3. The third lobe is oval in occlusal view and directs toward the buccal side, deviating from the main axis of the tooth ( Fig. 3I ). There is no ectostylid but a low cingulum appears buccally between the lobes and especially between the second and third lobe.

REMARKS

Recent advances on European Miocene tragulids partly solved longstanding taxonomic confusion and allowed recognizing two possible lineages with distinct morphometric features ( Hillenbrand et al. 2009; Rössner 2010; Alba et al. 2011, 2014; Sánchez et al. 2011; Morales et al. 2012; Rössner & Heissig 2013; Aiglstorfer et al. 2014). Nevertheless, the SE European record remains little known and still out of the discussed frame.

Malik & Nafiz (1933: pl. XI, fig. 2; pl. XI, fig. 3; pl. XI, figs 4, 5; pl. X, figs 3, 4) described from Küçükçekmece East a right mandible with dp2-dp4, a part of a left mandible with p4-m3, an isolated m2, one m3 and two metatarsals as Dorcatherium puyhauberti Arambourg & Piveteau, 1929 . This material is morphologically and dimensionally identical with the one described here, representing a single species. Geraads et al. (2005) also described from the nearby site of Yulaflı three partially preserved lower toothrows, and a metatarsal III+IV without distal ends, indicating strong similarities between the Yulaflı and the Küçükçekmece tragulids. As Nicolas (1978) before them, Geraads et al. (2005) refer samples from both fossil sites to Dorcatherium jourdani Depéret, 1887 , which taxonomic validity is presently questioned (e.g., Morales et al. 2012; Aiglstorfer et al. 2014).

The occlusal length of the lower molar row is 41 mm in the specimen illustrated by Malik & Nafiz (1933), 42 mm in two samples from Yulaflı ( Geraads et al. 2005; though maximum molar row length may reach 44.40 mm; S. Mayda pers. comm. 2015) and 41.8 mm for the specimen studied here. Similarly with the Küçükçekmece tragulid, the Yulaflı specimens show advanced selenodonty, long p4 with simplified distal valley, large m3 with similar structure of the third lobe, comparable development of Dorcatherium and Tragulus folds, weak buccal postentocristid, weak cingula and basal pillars (entostylids), and presence of a distinct lingual protocristid. Hence, we suggest tragulids from these two neighboring and roughly contemporaneous sites represent a single Dorcatherium species, in agreement with Geraads et al. (2005).The Küçükçekmece-Yulaflı species is twice as large as the living Moschiola meminna (Erxleben, 1777) and about 30% larger than the dentally more similar Hyemoschus aquaticus (Ogilby, 1841) . Using body mass equations proposed by Janis (1990), the average weight of the Küçükçekmece –Yulaflı species, based on the length of the M2, m2 and m1-m3 row, is estimated at 31.4 kg, ranging between 24.8 and 38.5 kg (n = 13). According to the body weight data provided by Alba et al. (2011) the Küçükçekmece-Yulaflı tragulid is comparable in size to D. naui Kaup, 1833 from Can Petit, Ballestar and Los Valles de Fuentidueña, Spain.

Upper and lower molar proportions of the Küçükçekmece and Yulaflı sample occupy an area in between the maximum observed values for Dorcatherium naui and the minimum for D. vindebonense von Meyer, 1846 ( Fig. 4 ). Compared to the studied species, D. guntianum von Meyer, 1846 is too small to be further considered ( Fig. 4 ; metrical comparison based on data provided by Rössner & Heissig 2013; Alba et al. 2011, 2014). The little known D. jourdani from Croix Rousse, France is also smaller than the Küçükçekmece-Yulaflı species, showing slightly shorter and constantly narrower upper and lower molars and shorter m3 ( Fig. 4 ; metrical comparison based on data provided by Rössner & Heissig 2013). According the restricted available data, D. puyhauberti from Vathylakkos, Axios Valley, Greece, is characterized by smaller upper molars and much shorter m3 ( Fig. 4 ), though the rest of its lower molars are placed within the normal range of D. naui and close to those of the studied species. The index “length m3/length m2 %” is 126.9 in MNHN.F.SLQ788 of D. puyhauberti and 135 in CR68 of D. jourdani instead of 146.5-148.7 in Küçükçekmece (n = 2), 142.1-143.9 in Yulaflı (n = 2), 146.1-163.5 in D. naui from Eppelsheim, Germany (n = 3), and 145.8-172.3 in D. crassum from Sansan, France (n = 3) (metrical data from Rössner & Heissig 2013).

By their size, the three dP4 from Küçükçekmece are placed close to the values recorded for D. jourdani from Croix Rousse and D. naui from several West European sites (metrical data from Alba et al. 2011; Morales et al. 2012; Rössner & Heissig 2013; Aiglstorfer et al. 2014) ( Fig. 4 ). The p3 is as long as that of D. vindebonense but proportionally narrower ( Fig. 4 ). The sample of the adequately represented p4 from Küçükçekmece (n = 4) plus Yulaflı (n = 2) shows a great overlapping with that of D. naui from several Central and West European sites, though with a tendency to be longer ( Fig. 4 ). The single preserved dp2 from Küçükçekmece West ( Table 2) is placed within the range of D. crassum from Sansan and Els Casots (L = 10.5-11.6 mm, n = 7; metrical data from Rössner & Heissig 2013 and Alba et al. 2014), being significantly longer than a single known dp2 of D. naui from Gratkorn (L = 8.9 mm; Aiglstorfer et al. 2014).

The advanced hypsodonty and selenodonty of the studied species clearly differentiate it morphologically from both D. crassum and D. vindebonense . Like D. naui and D. jourdani the Küçükçekmece-Yulaflı species is characterized by upper molars with less bulky styles, a short Dorcatherium -fold, a buccally shifted third lobe of m3, and weak mesiobuccal conid on the p3 ( Sánchez et al. 2011; Morales et al. 2012; Aiglstorfer et al. 2014; not all features known for D. jourdani ). In difference, however, from D. naui and D. jourdani ( Depéret 1887: pl. XIII, fig. 38) the Küçükçekmece sample shows a tricuspid dp2 like that of D. crassum ( Morales et al. 2012: figs 29-31), though without bulky protoconulid and with longer hypoconid in comparison with the protoconid. As the p2 and dp2 of European Dorcatherium species show the same condition (namely, tricuspidy or bicuspidity: Morales et al. 2012; Aiglstorfer et al. 2014), the tricuspidy of the Küçükçekmece dp2 leads us to assume that this taxon had a tricuspid p2 as well, in contrast again to D. naui . Unlike D. crassum from Sansan, France and Sandelzhausen, Germany ( Rössner 2010), D. naui from Can Mata, Spain and Gratkorn, Austria ( Alba et al. 2011; Aiglstorfer et al. 2014) and D. jourdani from Croix Rousse, the p3 from Küçükçekmece shows a more mesiodistally developed hypoconid compared to the protoconid. The Küçükçekmece-Yulaflı p4 differs from that of D. crassum in the well-individualized protoconulid and the long posterior fossette, which is simpler than in the specimens of D. naui from Gratkorn and D. guntianum from Wannenwaldtobel 2 ( Sach 1999; Aiglstorfer et al. 2014: fig. 5) and rather closer to the Abocator de Can Mata and Eppelsheim morphotypes of D. naui . A lingual protocristid, occasionally seen in an incipient stage of development in West European populations of D. naui , becomes more frequent and distinct in the Küçükçekmece and Yulaflı lower molars. In a lesser degree, the same feature occurs in three out of the six preserved lower molars of D. puyhauberti from Vathylakkos, Greece (MNHN; pers. obs.). A tendency similar to the supposedly typical character of the latter species (i.e. an extremely long postmetacristid, marginally reaching the Tragulus -fold; Geraads et al. 2005) is also seen on the m1 of MNHN.F.TRQ748 ( Fig. 3A ) and in two additional isolated lower molars (MNHN.F.TRQ765, ITU no No.) from Küçükçekmece. Nevertheless, D. puyhauberti from Vathylakkos has significanty smaller M2, M3 and m3 ( Fig. 4 ), and reduced cingulum on the posterior lobe of the upper molars compared to the Küçükçekmece tragulid. A well preserved half-mandible from Strumyani, Bulgaria referred to D. cf. puyhauberti by Geraads et al. (2011), also shows shorter p3, p4 and m3 compared to the Küçükçekmece-Yulaflı species.

A comparison of the postcranial elements is not really feasible as – apart for D. crassum from Sansan ( Morales et al. 2012) – most species are insufficiently known. Additionally, the best-represented postcranials from Sansan (astragali, n = 36; calcanei, n = 15; metatarsals, n = 9; Morales et al. 2012: table 2) indicate that differences between minimum and maximum measured values may exceed 20%, indicating large intraspecific variation. The distal humeri from Küçükçekmece ( Table 3) are comparable in size to a single specimen of D. crassum from Sansan and close to D. naui from Eppelsheim and Atzelsdorf (Kaup 1832; Hillenbrand et al. 2009; Morales et al. 2012). The distal radius from Küçükçekmece (MNHN.F.TRQ733, Table 3) appears significantly larger than specimens from Sansan (n = 6) and a single specimen of D. jourdani from Croix Rousse reported by Depéret (1887). It additionally differs from D. crassum in the wider and shallower groove for the extensor carpi radialis and the much weaker groove for the common extensor tendon of the digits ( Fig. 3M , compared to Morales et al. 2012: figs 37, 41). In these features it looks similar to the Atzelsdorf distal radius of D. naui ( Hillenbrand et al. 2009: pl. 1, fig.13). The single complete tibia from Küçükçekmece (MNHN.F.TRQ718, Fig. 3P ), which however belongs to a young adult individual, appears similar in length to that from Croix Rousse and quite shorter than that from Sansan ( Morales et al. 2012: tab. 2 but Depéret 1887: 225 reported a tibia of D. crassum with comparable length to the Küçükçekmece one). As in D. naui and unlike D. crassum ( Aiglstorfer et al. 2014) the Küçükçekmece tibia shows no fusion with maleolus lateralis and a much narrower and more plantarily extended medial concavity of the cochlea tibiae ( Fig. 3O, P ). The size range of the Küçükçekmece astragali (n = 7) appears much greater than that of D. crassum from Sansan. The size variability of the astragali (v’ = 31.5; sensu Freudenthal & Cuenca-Bescós 1984) is, however, less than the intaspecific size variability estimated from the body weight of the species based on dental data alone (v’ = 43.1), precluding second thoughts on the homogeneity of the material. Excluding two small astragali (MNHN.F.TRQ710, TRQ714) that possibly represent young individuals, the rest of the studied sample indicates large dimensions ( Fig. 4 ), marginally overlapping with the Sansan size range of D. crassum . MNHN.F.TRQ706, TRQ707, TRQ708 and TRQ711 have similar size to the larger specimens of D. naui from Atzelsdorf and the single astragalus from Gratkorn ( Fig. 4 ), whereas MNHN.F.TRQ716 and TRQ732 are as large as a single astragalus from the Vallesian fauna from Kastellios Hills in Crete Island refered to as cf. Dorcabune anthracotheroides by Made (1996). As the astragali of D. naui from Gratkorn and Atzelsdorf ( Hillenbrand et al. 2009: pl. 1, fig.12; Aiglstorfer et al. 2014: fig. 2s) and in difference from D. crassum from Sansan ( Morales et al. 2012: figs 63-64), the Küçükçekmece astragali ( Fig. 3O ) show weaker angle between the axes of the proximal and distal trochlea, whereas the lateral condyle of the caput tali is wider and the medial condyle more prominent distally in dorsal view. By their size, the two complete calcanei from the Küçükçekmece ( Table 3; Fig. 3L ) are well within the size range of the Sansan sample ( Morales et al. 2012). Nevertheless, the coracoid process is weaker, the articular surface for the maleolar is longer and narrower, and the articular surface for the cubonavicular faces plantary instead of medially in D. crassum ( Morales et al. 2012: figs 58-59). The Küçükçekmece metatarsals (including those described by Malik & Nafiz [1933]) appear slightly more robust and certainly longer than those of D. crassum from Sansan (L <100, n = 9; Morales et al. 2012, instead of>106, n = 3 in Küçükçekmece). They also differ in the much wider frontal groove between the MtIII and MtIV ( Fig. 3N ), and the more unevenly placed proximal surfaces for the cubonavicular and the ectomesocuneiforme. Unfortunately no complete metatarsals of D. naui or D. jourdani are known to us for a further comparison.

To sum up, the Küçükçekmece-Yulaflı tragulid shows great general morphometric resemblance to the West and Central European D. naui , from which, however, it differs in the slightly larger size (M3/m3, astragali), longer lower premolars, and the expression of some dental features. Among them, the most important are: (1) the presence of tricuspid dp2 (and possibly p2); (2) the stronger mesiodistal development of the hypoconid on the dp2 and p3; (3) the rather simple posterior fossette of the p4; and (4) the common presence of a dintinct lingual protocristid on the lower molars. In character (2) the Thracian tragulid recalls D. guntianum from Wannenwald- tober 2, Germany ( Sach 1999) and D. naui from Abocador de Can Mata, Spain ( Alba et al. 2011), both representing earlier members of the European “selenodont phylogenetic lineage” (sensu Rössner & Heissig 2013). Characters (3) and (4) are not unique to the studied form and their variation in different chrono-spatial populations of the genus is still unknown. Character (1) excludes, however, D. naui as a possible affiliation and in combination with the rest of the observed features (i.e. advanced selenodonty, large M3/m3, long premolars, long hypoconid compared to the protoconid on dp2, p3, and common presence of a distinct lingual protoctistid on lower molars) justify, in our opinion, the foundation of a new species, namely Dorcatherium maliki n. sp. The discontinuous geographic distribution between D. naui at the northwest and the Thracian taxon to the south-east (see Geraads et al. 2005: fig. 14) would also support such a decision. Phylogenetic relations between D. maliki n. sp. and the later D. puyhauberti can also be alleged on the basis of some secondary dental features, but more material of both species is necessary to test this hypothesis.

Figures

FIG. 3. — Dorcatherium maliki n. sp. from Küçükçekmece West, Turkey: A, m1-m3 dex, MNHN.F.TRQ748 in lingual (up), occlusal (middle), and buccal (down) views; B, p3-p4 sin, ITU371 in occlusal (reversed-up) and buccal (down) views; C, P2 sin, MNHN.F.TRQ759 in occlusal view; D, P4 sin, ITU no No. in buccal (left), occlusal (middle), and lingual (right) views; E, dp2 dex, MNHN.F.TRQ779 in occlusal (up) and buccal (down) view; F, m2 sin, MNHN.F.TRQ766 in oc- clusal view; G, dP4 dex, MNHN.F.TRQ763 in occlusal view; H, p4 sin, MNHN.F.TRQ768 in occlusal view; I, m3 dex, MNHN.F.TRQ784 in occlusal view; J, M2 sin, MNHN.F.TRQ751 in occlusal view;K, m2 dex,MNHN.F.TRQ783 in occlusal view;L, calcaneum,MNHN.F.TRQ703 in medial view;M, distal radius,MNHN.F.TRQ733 in distal view; N, metatarsal III+IV, MNHN.F.TRQ741 in plantar (left) and dorsal (right) view; O, distal tibia and talus, MNHN.F.TRQ716, TRQ717 in dorsal view; P, tibia, MNHN.F.TRQ718 in dorsal view. Scale bars: 2 cm.

FIG. 1. — Hippopotamodon cf. antiquus from Küçükçekmece West, Turkey: A, i3 dex,MNHN.F.TRQ682 in occlusal (up) and buccal (down) views;B, p3 sin, MNHN.F.TRQ675 in occlusal (up) and lingual (down) views;C, distal part of dp4 dex, MNHN.F.TRQ679 in occlusal view; D, part of?M1 dex, MNHN.F.TRQ678 in occlusal view. Scale bar: 2 cm.

FIG. 4. — Metrical comparison of several European Dorcatherium Kaup, 1833 species.Scatters adopted and modified from Alba et al. (2011: fig. 4A) for DP4 and Alba et al. (2014:fig. 7) for M1, M2, M3, p3, p4, m2, m3, adding data from Rössner (2010),and Aiglstorfer et al. (2014), as well as personal data on D. puyhauberti Arambourg & Piveteau, 1929 from Axios valley, Greece (based on MNHN collection). Data sources for astragali from Morales & Soria (1981), Hillenbrand et al. (2009), Morales et al. (2012) and Aiglstorfer et al. (2014). Coloured areas represent convex hulls of original distributions with outliers per species indicated with the same coloured circles. For astragalus,boxes represent minimum-maximum size range.The number of records per species (N) is given in the similar coloured boxes associated with each distribution. White area, D. guntianum von Meyer,1846; yellow area, D. crassum (Lartet, 1851); green area, D. naui Kaup & Scholl, 1834; blue area, D. vindebonense von Meyer, 1846. Symbols:, D. jourdani Depéret, 1887; •, D. puyhauberti Arambourg & Piveteau, 1929; ê, Küçükçekmece tragulid;, Yulaflı tragulid;, Kastellios tragulid, Crete Island. The dashed box on m3 indicates minimum-maximum range for Küçükçekmece.