taxonID	type	description	language	source
039787DCFFE99813B81F11F726D7F8A0.taxon	materials_examined	Type species: Indarctos salmontanus Pilgrim, 1913: 290; Hasnot, Pakistan, late Miocene.	en	Geraads, Denis, Kaya, Tanju, Mayda, Serdar (2005): Late Miocene large mammals from Yulafli, Thrace region, Turkey, and their biogeographic implications. Acta Palaeontologica Polonica 50 (3): 523-544, DOI: 10.5281/zenodo.13620702
039787DCFFE99810B81F106927E3FB1E.taxon	materials_examined	Type locality: Montredon, France. Material from Yulafl I. — TTMEU−CY− 46, an almost complete left dentary, lacking only the incisors, the tip of the canine, p 1, m 3, and part of the coronoid process (Fig. 2). Description. — The corpus is robust, but not extremely so, with a uniform depth from p 4 to m 2. The ventral border forms a rather regular curve, being only slightly more convex posteriorly. The anterior border of the coronoid process is slightly inclined backwards. The canine is short and stout, and strongly curved. It has a weak posterior keel, and a stronger mesio−lingual one, with an enamel notch just in front of it. The missing first premolar had a single, rather large root, strongly inclined forwards. It is separated by a short diastema from p 2, which has two roots, almost fused but well distinct labially. Its crown is long, and consists of a single low cuspid, whose apex is more anterior than mid−length, so that the longitudinal cristid has a steeper mesial slope. There is a lingual cingulid, especially marked in the distal half, where it increases tooth width. A longer diastema separates p 2 from p 3, which has two well−distinct roots and is morphologically intermediate between p 2 and p 4. The main cuspid is higher and more posterior than on p 3, the disto−lingual cingulid is weaker, but there is an incipient distal cingulid. In labial view, both edges of the main cuspid are convex. A short diastema separates this tooth from p 4, which is much larger, but not disproportionately so. Its main cuspid is relatively still higher than on p 3; there is a hint of a mesial accessory cuspid where the base of the main cristid turns inwards, and the distal cingulid is somewhat enlarged, forming a narrow shelf. The tooth is only slightly broader above the distal root. The m 1 has the usual morphology for Indarctos, with a very open trigonid, much narrower than the talonid, a reduced metaconid, and an entoconid larger and higher than the entoconulid (see Beaumont 1982). Although the teeth are slightly worn, it is clear that there was no cuspid between protoconid and hypoconid. The m 2 is a large tooth, broader than m 1, without paraconid, but with a transverse cristid between protoconid and metaconid. Comparisons. — Several species of Indarctos have been named, but only a few are now currently recognised in Eurasia. Indarctos atticus (Weithofer, 1888, ex Dames) is mostly known in the classic Turolian localities of Pikermi, Samos, and Maragha. Indarctos atticus probably includes I. salmontanus Pilgrim, 1913, the type species, probably from the Dhok Pathan zone of the Siwaliks, I. lagrelii Zdansky, 1924 and I. sinensis Zdansky, 1924, from Loc. 31 and Loc. 30, respectively, in China, and I. bakalovi Kovachev, 1988, from Kalimantsi in Bulgaria. The recently described I. zdanskyi Qiu and Tedford, 2003, from Baode, China, is similar, but is more derived in several features. All these localities are also of MN 12 / 13 − equivalent age. Indarctos arctoides (Depéret, 1895) is best known from Montredon (MN 10; Depéret and Gomez−Llueca 1928; Beaumont 1988), Westhofen (MN 9?; Tobien 1955), Pfaffstetten (MN 11?; Thenius 1959), and Küçükçekmece (Petter and Thomas 1986), while I. vireti Villalta and Crusafont, 1943, is mostly known from the Vallesian of Spain, chiefly from Can Llobateres (MN 9; Crusafont and Kurtén 1976), but has also been reported from Sinap (Viranta and Werdelin 2003). The status of I. anthracitis Weithofer, 1888, from Monte Bamboli, is disputed. The dentary from Yulafli compares best with a dentary of I. atticus from Samos in the NHMW (Thenius 1959: fig. 7), except that the depth of the latter, as in all I. atticus, increases caudally. Robustness is variable in other specimens of I. atticus but some of them, such as those from Crevillente− 2 (Montoya et al. 2001: pl. 2: 1) and Kalimantsi (NMNHA) have an extremely convex lower border of the corpus, and an anterior border of the ramus which is slightly inclined forwards. This is of course a consequence of the shortening of the cranial basis in this species. The dentaries from the Vallesian of Spain seem to have a much less upright ramus that is, however, largely reconstructed. The dentary from Küçükçekmece (Petter and Thomas 1986: fig. 5) is extremely slender, and perhaps even pathological. The lower canine is poorly known. The description of those of I. vireti by Crusafont and Kurtén (1976) perfectly matches that of our specimen, except that they are smaller. The relatively large p 2 and p 3, which are both doublerooted, contrast with the sharp reduction of these teeth in typical I. atticus (and still more with the loss of these teeth in I. zdanskyi). On the contrary, p 4 is enlarged in I. atticus, so that there is a sharp difference between p 3 and p 4, whereas size harmoniously increases from p 2 to p 4 in our specimen. Molar morphology does not provide many discriminating features. The m 1 of I. atticus often has a labial tubercle behind the protoconid; it is absent in TTMEU−CY− 46. Discussion. — The specific distinction between the middle and late Turolian I. atticus and the early Vallesian form (whether it is called I. vireti or I. arctoides vireti) is widely acknowledged. The former differs from the latter by (1) its larger size; (2) its shortened dentary, and cranial base; (3) its much smaller p 2 and p 3, with only one root; (4) its enlarged and broadened p 4. Other features of the molars do not prove very discriminant. The trigonid / talonid ratio of m 2, used by Petter and Thomas (1986), although potentially meaningful, is too hard to estimate precisely. It is nonetheless clear that species demarcation is blurred when chronologically intermediate forms, many of them called I. arctoides, are taken into consideration. The size of m 1, taken as an indicator of overall size, forms an almost perfect continuum (Fig. 3; the apparent gap in Montoya et al. 2001: fig. 4 results from the non−inclusion of two intermediate specimens, the type of I. lagrelii and the Samos specimen in NHMW). The Lm 2 / Lm 1 ratio, stated by Montoya et al. (2001) to be smaller in I. arctoides than in both other species, is in fact variable. It is always large in I. atticus, but may be small (Küçükçekmece) or large (Pfaffstetten) in I. arctoides, as well as in I. vireti (respectively, Can Purull and Can Llobateres). Morphological characters are also intermediate. For instance, at Montredon, the lower border of the dentary is strongly curved, as in several I. atticus, and the p 3 has its roots “ étroitement soudées ” (Depéret and Gomez−Llueca 1928). At two localities, both usually included in biozone MN 11, Dorn−Dürkheim (Roth and Morlo 1997) and Crevillente 2 (Montoya et al. 2001), I. atticus has been reported to coexist with a more primitive form, I. arctoides in the former site, and with I. cf. vireti in the second. In both cases, this second species is documented by rather poor or fragmentary material. At Crevillente− 2, the identification of two species rests mostly upon size, but the differences certainly do not exceed what can be expected in a single population. Even the “ I. atticus ” there has primitive features: it is rather small, its p 2 and p 3 are bi−radiculated, the parastyle of its P 4 is stronger than in I. vireti, but clearly smaller than in typical I. atticus (Montoya et al. 2001: fig. 5). At Dorn−Dürkheim, an M 2 referred to I. atticus is almost identical in size to a tooth from Montredon, type−locality of I. arctoides. Two M 3 s, each referred to a different species, are little different in size (18.8 × 15.2 and 21.4 × 17.85). It is far more likely that, in both localities, we are dealing with a single species, intermediate between the Vallesian and middle Turolian forms. This is not unexpected in early Turolian sites. Features of these taxa appear to have evolved mosaically, with intermediate forms displaying a mixture of primitive and derived traits. On the whole, not a single trait forbids hypothesising an anagenetic evolution from I. vireti to I. atticus, through I. arctoides. The Indarctos from Yulafli clearly belongs to this intermediate stage. Its m 1 is smaller than those of all I. atticus, and close in size to I. arctoides from Montredon, but its m 2 is large, near the lower range of I. atticus (however, an m 2 from Can Llobateres is almost as large). It has no cuspid between protoconid and hypoconid on m 1, in contrast to I. atticus. The clearest primitive features of the Yulafli specimen are found in the premolars. The harmonious increase in size from p 2 to p 4 is quite unlike I. atticus, where p 4 is much enlarged in respect to the reduced p 3. Correlatively, p 2 and p 3 are still two−rooted, while the former is already one−rooted at Pfaffstetten and Küçükçekmece, and the latter may also be one−rooted in I. atticus.	en	Geraads, Denis, Kaya, Tanju, Mayda, Serdar (2005): Late Miocene large mammals from Yulafli, Thrace region, Turkey, and their biogeographic implications. Acta Palaeontologica Polonica 50 (3): 523-544, DOI: 10.5281/zenodo.13620702
039787DCFFEA9810BB5513D2245CFAED.taxon	materials_examined	Type species: Deinotherium giganteum Kaup, 1829; Eppelsheim, Germany, Vallesian, Miocene.	en	Geraads, Denis, Kaya, Tanju, Mayda, Serdar (2005): Late Miocene large mammals from Yulafli, Thrace region, Turkey, and their biogeographic implications. Acta Palaeontologica Polonica 50 (3): 523-544, DOI: 10.5281/zenodo.13620702
039787DCFFEA9816BB5512A227F2FEF2.taxon	materials_examined	Material from Yulafl I. — TTMEU−CY− 30, right P 4 (Fig. 4 G); TTMEU−CY− 31, left M 3; TTMEU−CY− 109, incomplete toothless dentary. Description. — Both teeth are large (P 4: 71 × 90; M 3: 94 × w 1 = 105 × w 2 = 91). The P 4 is rectangular in outline, being wider than long (Fig. 4 G). The protoloph is complete and united with the paracone, but the metaloph is only a half−loph. The ectoloph is well developed, with a labial ectoflexus. The mesostyle (Harris 1973: fig. 7) is low and located on the mesio−lingual surface of the hypocone. The median valley is wide and opens lingually only. The anterior cingulum forms a ridge along the mesial side, but the distal cingulum is weaker, and there is no labial cingulum. There are three roots; two of them are located under the lophs, the third one is lingual. The M 3 is rectangular in outline, with the protoloph wider than the metaloph. The median valley opens lingually and labially. The anterior cingulum forms a ridge along the anterior side, but the posterior cingulum is weak. There are no lingual or labial cingula, except a small labial cingulum at the opening of the median valley. There are three roots. The anterior root corresponds to the mesio−lingual part of the protoloph, the second root is along the metaloph, and the third root extends along the labial side. The lingual ornamentation of the protoloph is stronger than the labial one. The postmetaloph ornamentation is weak, and forms a small tubercle. The enamel of the tooth is finely wrinkled along the lophs and the lingual and labial surfaces. The dentary TTMEU−CY− 109 has lost its teeth, and the rostral part of the tusk sheaths is also broken away, revealing two parallel alveoli, only slightly decreasing in diameter ventrally, and separated by a narrow septum. The maximum width across the sheaths is 240 mm. Comparisons. — In Eurasia, deinotheres are known in early Miocene to middle Pliocene localities (Bergounioux and Crouzel 1962; Tobien 1988; Huttunen 2002 a), but their taxonomy has long been debated. In this study, following Harris (1973), the name Deinotherium is used for a large−sized deinotheres, which have been recorded from many localities in Europe (review in Huttunen 2002 a). In Turkey, it is known from Tire (Ozansoy 1961), Paşalar (Tobien 1990), Kayadibi (Gaziry 1976), Çandir (Gaziry 1976; Geraads and Güleç 2003), Sinap (Sanders 2003), Küçükçekmece (Malik and Nafiz 1933), and Düzyayla (Kaya and Forstén 1999). The large−size, the presence of the mesostyle, and the reduced postmetaloph ornamentation are diagnostic characters for Deinotherium that distinguish it from Prodeinotherium (Harris 1973). By the presence of a strong anterior cingulum, of a mesostyle, and of incomplete lophs, the P 4 from Yulafli resembles those of D. giganteum from various localities in Austria described by Huttunen (2002 b). There is a clear general trend for size increase in Deinotherium in the Miocene; e. g., the teeth from Yulafli are much larger than those of Prodeinotherium, and also than those of D. aff. levius from the middle Miocene of Paşalar (Tobien 1990) and than the unpublished teeth of P. bavaricum from the middle Miocene of Tire (Figs. 5, 6). However, late Miocene forms exhibit a great size−range not obviously linked with age or geography. The teeth from Yulafli are larger than those of D. giganteum from many European sites, and close to the maximum recorded size for specimens from Vallesian sites such as Montredon, Mannersdorf, Kohfidisch, Eppelsheim or from some Hungarian finds (Gräf 1957; Tobien 1988; Huttunen 2002 b; Mazo and Montoya 2003), but they are only slightly smaller than specimens referred to D. gigantissimum, so that it is hard to draw biochronological conclusions from them.	en	Geraads, Denis, Kaya, Tanju, Mayda, Serdar (2005): Late Miocene large mammals from Yulafli, Thrace region, Turkey, and their biogeographic implications. Acta Palaeontologica Polonica 50 (3): 523-544, DOI: 10.5281/zenodo.13620702
039787DCFFEC9816BB5516A6248DFDA6.taxon	materials_examined	Type species: Mastodon pentelicus Gaudry and Lartet, 1856; Pikermi, Greece, middle Turolian, Miocene.	en	Geraads, Denis, Kaya, Tanju, Mayda, Serdar (2005): Late Miocene large mammals from Yulafli, Thrace region, Turkey, and their biogeographic implications. Acta Palaeontologica Polonica 50 (3): 523-544, DOI: 10.5281/zenodo.13620702
039787DCFFEC9815BB551565200DFA50.taxon	materials_examined	Type locality and age: Yassiören, Turkey, Vallesian, Miocene. Material from Yulafl I. — TTMEU−CY− 22, right dentary with m 3 and alveolus of m 2 (Fig. 4 C); TTMEU−CY− 201, left dentary with m 3; TTMEU−CY− 32, isolated m 3 (Fig. 4 D). Description. — Both dentaries are morphologically similar and certainly belong to the same individual. The corpus is slender relative to that of Tetralophodon of the same locality (Table 1); its depth slightly decreases posteriorly. The ventral border of the corpus is almost straight. The interalveolar crest is curved outward and slopes ventro−mesially, with an angle of about 25 ° relative to the alveolar border. The symphysis is broken, but extends distally as far as the mesial part of m 2. The mandibular foramen, 25 mm in diameter, is located below the middle of m 2. The mandibular canal lies along the corpus, small and square in section anteriorly, becoming broader and triangular posteriorly. Heavy wear on TTMEU−CY− 22 and TTMEU−CY− 201 does not allow description of the structure of the conelets and accessory conules. Both m 3 s have five lophids and a small posterior cingulum with small conelets. The half−lophids of the third and fourth lophids are nearly fused to each other. The mesoconelets are located anterior to the main cones, contributing to an anteriorly−pointing “ V ” shape of the lophids in occlusal view, or “ chevroning ”. The fifth lophid is reduced in width and is composed of a main cusp and a small pretrite mesoconelet. The m 3 TTMEU−CY− 32 has four lophids and a long talonid, which is composed of an irregular arrangement of four small conelets. Wear gradient is strong. It is certainly from a male individual (measurements: Table 1 and Fig. 7). On the pretrite side, the first two half−lophids are accompanied by posterior accessory conules, as on the specimen AS 92.605 from Sinap (Sanders 2003). The fourth lophid has a double mesoconelet. The cement is weak at the base of the interlophids and well developed on the lingual and labial walls of the crown. The chevrons are well developed in the last three lophids. The teeth of Yulafli have well developed ptychodonty, but by contrast weak choerodonty. Comparisons. — During the late Miocene, Choerolophodon was abundant in Turkey, contrasting with its less common occurrences in other Eurasian faunas. It has been recorded, among other sites, at Pikermi, Samos, Ravin de la Pluie in Greece, Veles in FYROM, Maragha in Iran, Ezerovo and Ahmatovo in Bulgaria (Tassy 1983, 1989; Bakalov and Nikolov 1962, as Trilophodon angustidens). The contemporaneous C. corrugatus is documented from Dhok Pathan and Nagri formations of the Siwalik (Tassy 1983), while choerolophodonts span the middle to late Miocene in Kenya (Tassy 1986; Pickford 2001). In Turkey, late Miocene Choerolophodon are known from Çorakyerler, Garkin, Kinik, Gülpinar, Gökdere, Kayadibi, upper and lower levels of Kemiklitepe, Eşme−Akçaköy, Ramiz in Istanbul, and Sinap (Şenyürek 1952; Viret 1953; Ozansoy 1965; Gaziry 1976; Tassy 1994; Tassy et al. 1989; Sanders 2003). Moreover, a few middle Miocene Choerolophodon have been recorded in Turkey (e. g., Sofça) and Chios (Gaziry 1976; Tobien 1980).	en	Geraads, Denis, Kaya, Tanju, Mayda, Serdar (2005): Late Miocene large mammals from Yulafli, Thrace region, Turkey, and their biogeographic implications. Acta Palaeontologica Polonica 50 (3): 523-544, DOI: 10.5281/zenodo.13620702
039787DCFFEC9815BB551565200DFA50.taxon	description	The m 3 s from Yulafli fall within the size ranges of C. pentelici, C. anatolicus, and also C. corrugatus (Fig. 7), but the size of the cheek teeth appears to be less important than occlusal morphology, since there is much size overlap between species, partly because of sexual dimorphism (Sanders 2003). The morphological characters of the Yulafli sample suggest an intermediate form showing more resemblance to C. anatolicus than to C. pentelici sensu stricto.	en	Geraads, Denis, Kaya, Tanju, Mayda, Serdar (2005): Late Miocene large mammals from Yulafli, Thrace region, Turkey, and their biogeographic implications. Acta Palaeontologica Polonica 50 (3): 523-544, DOI: 10.5281/zenodo.13620702
039787DCFFEF9815B81F12C1211AF9D5.taxon	materials_examined	Type species: Mastodon longirostris Kaup, 1832; Eppelsheim, Germany, Vallesian, Miocene.	en	Geraads, Denis, Kaya, Tanju, Mayda, Serdar (2005): Late Miocene large mammals from Yulafli, Thrace region, Turkey, and their biogeographic implications. Acta Palaeontologica Polonica 50 (3): 523-544, DOI: 10.5281/zenodo.13620702
039787DCFFEF9814B81F117426F3FC32.taxon	materials_examined	Material from Yulafl I. — TTMEU−CY− 23, left dentary, with m 2 – m 3 and alveoli of m 1; TTMEU−CY− 202 a, right dentary with m 2 and alveoli of m 1 (Fig. 4 A); TTMEU−CY− 202 b, isolated right m 3 (Fig. 4 B); TTMEU−CY− 111, right dentary with heavily worn m 1; TTMEU−CY− 112, right dentary with heavily worn m 2, and anterior part of m 3; TTMEU−CY− 48 isolated left M 2. Description. — TTMEU−CY− 23 and TTMEU−CY− 202 belong to adult individuals. The corpus of each specimen is robust and broadens posteriorly. The interalveolar crest and the mandibular foramina are missing in TTMEU−CY− 23, but the interalveolar crest of TTMEU−CY− 202 is straight, long and mesio−ventrally sloping. It is deflected downward at ~ 45 ° relative to the alveolar border, while the ventral margin of the symphysis exhibits a downward deflection of 15 °. These measurements for TTMEU−CY− 112 are 40 ° and 17 °, respectively. There are four mandibular foramina on TTMEU− CY− 202. The largest one is positioned below the alveoli of m 1, the others are positioned anteriorly on the symphysis. TTMEU−CY− 111 belongs to a juvenile individual with a slender corpus exhibiting three mandibular foramina. The m 2 of TTMEU−CY− 202 has four lophids, plus a small and low distal cingulum. The anterior margin of the tooth is broken. The pretrite half−lophids 2 and 3 preserve large posterior accessory conules while the fourth one has a smaller one. All other intermediate molars have four loph (id) s, but due to the wear stage in all specimens, it is impossible to determine their precise composition. All m 3 s are morphologically similar. They each have five lophids and a bituberculate talonid, which could also be considered as a lophid. Each half−lophid is composed of a large main cone, a smaller mesoconelet and anterior and posterior accessory conules on the pretrite half−lophids and a posterior accessory conule on the first posttrite half−lophid. The pretrite half−lophids 4 and 5 of TTMEU−CY− 23 do not exhibit any accessory conules. The anterior cingulum is well developed in all teeth. There are no traces of labial and lingual cingula except a thick basal one on the labial side of the last lophid of TTMEU−CY− 23. There is a trace of cement at the base of the interlophids. Comparisons. — Tetralophodont records are scarcer in Turkey than those of choerolophodonts, but they are more numerous in other Eurasian localities. Tetralophodon longirostris ranges from the late middle Miocene to the late Miocene. It is well known from Eppelsheim, Dorn−Dürkheim 1, Belvedere in Austria, and Nombrevilla in Spain (Tobien 1978; Gaziry 1997; Göhlich 1999). The genus Tetralophodon has also been recorded in Istanbul (Viret 1953). The materials from Yulafli are very similar in size and morphology to those of Tetralophodon longirostris from Eppelsheim (Tobien 1978: pl. 10: 1) and Dorn−Dürkheim 1 (Gaziry 1997: pl. 1: 2). The m 3 s have five lophids and a talonid, a simple crown pattern without anancoidy, a slight cement cover in the posterior interlophids, and the symphysis is down−turned. The fossils from Yulafli are distinguished from other tetralophodonts (“ Mastodon ” longirostris forma gigantorostris and “ Mastodon ” grandincisivus) (Tobien 1978; Mazo and Montoya 2003) by having a simple crown pattern, a weak cement cover, and smaller cheek teeth. The material from Yulafli falls within the size range of T. longirostris from late Miocene localities (Eppelsheim, Esselborn, Mannersdorf), which is clearly distinct from that of “ T. longirostris − grandincisivoid form ” – “ Mastodon ” grandincisivus (Fig. 8).	en	Geraads, Denis, Kaya, Tanju, Mayda, Serdar (2005): Late Miocene large mammals from Yulafli, Thrace region, Turkey, and their biogeographic implications. Acta Palaeontologica Polonica 50 (3): 523-544, DOI: 10.5281/zenodo.13620702
039787DCFFEE981BB81F14E821BFF985.taxon	materials_examined	Type species: Amebelodon fricki Barbour, 1927; Cambridge quarry, Nebraska, middle Pliocene.	en	Geraads, Denis, Kaya, Tanju, Mayda, Serdar (2005): Late Miocene large mammals from Yulafli, Thrace region, Turkey, and their biogeographic implications. Acta Palaeontologica Polonica 50 (3): 523-544, DOI: 10.5281/zenodo.13620702
039787DCFFEE981BB81F14E821BFF985.taxon	materials_examined	Type locality and age: Maragheh, Iran, Turolian, Miocene. Material from Yulafl I. — TTMEU−CY− 143, left i 2 (Fig. 4 F); TTMEU−CY− 203, isolated left M 2 (Fig. 4 E). Length (mm) Fig. 8. Length versus width plot of Tetralophodon and “ Mastodon ” grandincisivus m 3 s. Description. — The incisor is slightly curved outwards (Fig. 4 F 2). It has a rather flattened transverse section (145 × 65 mm). There is no trace of enamel. On the dorsal side there are two grooves; the labial one being deeper than the lingual one. On the ventral side, there are five shallow grooves. The tip of the incisor has a thin cement layer, underneath follows a concentric laminated dentine layer (10 mm), then a uniform core dentine (max. 41 mm). The basal part has a regular dentine layer (10 mm), but the core of the tooth (max 44 mm) consists of rod−cone structures, or tubular dentine (Tassy 1999). TTMEU−CY− 203 is a nearly complete, heavily worn left M 2 with four lophs, a weak anterior cingulum and a well−developed double talon. The tooth is rectangular. The pretrite half−lophs are composed of a main cone, a smaller mesoconelet and anterior and posterior accessory conules that exhibit trefoil wear pattern in the first two half−lophs. On the posttrite side, the posterior accessory conules are present on the first and third half−lophs; and an incipient anterior accessory conule on the second half−loph. The last two half−lophs show an incipient anancoidy with the pretrite side positioned anterior to the posttrite side. There is a trace of cement at the base of the interlophs. Comparisons. — The systematic position of “ Mastodon ” grandincisivus among late Neogene elephantoids has long been debated. It was included in Stegotetrabelodon by Tobien (1978), and more recently in the amebelodonts (Tassy 1999). We follow this latter opinion, since the dorso−ventral compression and tubular dentine structure of the holotype (Schlesinger 1922) clearly demonstrate affinities with this group. Similar incisors are known from some (but not many) localities in Eurasia, Africa, and North America, of middle to late Miocene age, but the differences in outline of the cross−section point to the occurrence of two or more different species (and / or sexual dimorphism) in the Old World. The structure of the rod cones from Yulafli is similar to that of an incisor from Arapli, near Tekirdaǧ, referred to Amebelodon (Platybelodon) by Gaziry (1976: pl. 3: 2), but the latter was certainly more dorso−ventrally compressed, and more regularly compressed dorsally. At Kerch in Crimea (Pavlov 1904) and Sahabi in Libya (Gaziry 1987), the section is almost rectangular, whereas it is more irregular, and thus more similar to that of Yulafli, at Maragha and Pestszentlörincz in Hungary (Schlesinger 1917; 1922; comparisons in Tassy 1999: fig. 18.8), and perhaps also in Gansu, China (Tobien et al. 1986: fig. 24). However, the incisor from Yulafli is smaller and more convex ventrally than the teeth from Maragha and Pestszentlörincz (Table 2 and Fig. 9). Perhaps the specimen most similar to the tooth from Yulafli is from Orjachovo in Bulgaria (Bakalov and Nikolov 1962: pl. 66). The M 2 TTMEU−CY− 203 differs from that of T. longirostris from Yulafli, which has a simpler lophid structure and a clearly smaller size. It resembles the Mannersdorf sample of “ Mastodon ” grandincisivus (Schlesinger 1917: pl. 15: 1). They share the pretrite trefoil pattern in the first two half−lophs, posterior accessory conules on the posttrite side, and secondary conules blocking the interlophs, but the teeth from Mannersdorf are larger. The mastodont from Arapli is trilophodont, and this is a further difference, if in both sites the molars are correctly referred to the same taxon as the incisors. Thus, there is little doubt that these two teeth should be referred to what is often called “ Mastodon ” grandincisivus, but what should more correctly, even if provisionally, be called Amebelodon. Its smaller size than the similar form from Kerch and Maragha suggests that it might be earlier, but given the great size variability in proboscideans, this conclusion is, admittedly, quite weak.	en	Geraads, Denis, Kaya, Tanju, Mayda, Serdar (2005): Late Miocene large mammals from Yulafli, Thrace region, Turkey, and their biogeographic implications. Acta Palaeontologica Polonica 50 (3): 523-544, DOI: 10.5281/zenodo.13620702
039787DCFFE19818B81F10A024B0FAB7.taxon	materials_examined	Type species: Equus primigenius von Meyer, 1829; Eppelsheim, Germany, Vallesian, Miocene.	en	Geraads, Denis, Kaya, Tanju, Mayda, Serdar (2005): Late Miocene large mammals from Yulafli, Thrace region, Turkey, and their biogeographic implications. Acta Palaeontologica Polonica 50 (3): 523-544, DOI: 10.5281/zenodo.13620702
039787DCFFE19818B81F10A024B0FAB7.taxon	description	the deeper hypoconal groove, the confluent distal wall of the postfossette, and stouter metacarpals (Fig. 11). The Mc−IIIs from Yulafli are also smaller and less massive than the Mc−IIIs from Eppelsheim and Rudabánya. The Yulafli hipparion differs from both Cormohipparion sinapensis and “ H ”. ankyranum from Sinap Tepe (Ozansoy 1965; Bernor et al. 2003) in the following characters: the maxillary cheek teeth have highly ornamented pre−and postfossette, the protostylid is present in the lower teeth, and the metacarpal III is lightly built (Bernor et al. 2003: fig. 11.8). They share a lingually flattened protocone, a deep distal hypoconal groove, a complex or bifid pli caballin, and a distinct posterior wall of the postfossette. However, the Mc−IIIs from Sinap Tepe have very diverse proportions, and some specimens from Loc. 12 (early MN 10), conform well with our specimens in being short and slender (Fig. 11); they were referred by Bernor et al. (2003) to “ Hipparion ” sp. 1, while slightly stouter specimens from Sinap Loc. S 01 were identified as aff. “ Hipparion ” kecigibi. Our material is also distinct from the H. aff. depereti from Pentalophos (Koufos 2000), which has teeth with moderate enamel plications (19 for M 1 – 2) and short massive metacarpals (the index 11 / 1 = 170). The teeth from Yulafli certainly differ from the Turolian hipparions of Gülpinar and Kemiklitepe (TTMEU), where the enamel plication is simple, the protocone is rounded, and the pli caballin is simple. The keel index of Mc−IIIs is 111 – 113 indicating Vallesian forms (Sen et al. 1978; Staesche and Sondaar 1979) from that of the Turolian forms. Similarities in dental morphology indicate that the Yulafli hipparion is closer to Vallesian forms than to Turolian ones, but the Mc−IIIs are smaller and less massive. These characters suggest that it is more advanced than early Vallesian forms of H. primigenium.	en	Geraads, Denis, Kaya, Tanju, Mayda, Serdar (2005): Late Miocene large mammals from Yulafli, Thrace region, Turkey, and their biogeographic implications. Acta Palaeontologica Polonica 50 (3): 523-544, DOI: 10.5281/zenodo.13620702
039787DCFFE29818BB55128E258BF98C.taxon	materials_examined	Type species: Sus antiquus Kaup, 1832; Eppelsheim, Germany, Vallesian, Miocene.	en	Geraads, Denis, Kaya, Tanju, Mayda, Serdar (2005): Late Miocene large mammals from Yulafli, Thrace region, Turkey, and their biogeographic implications. Acta Palaeontologica Polonica 50 (3): 523-544, DOI: 10.5281/zenodo.13620702
039787DCFFE2981EBB55111E2580FE32.taxon	materials_examined	Material from Yulafl I. — TTMEU−CY− 45, symphysis with i 1 – i 2 on both sides, bases of canines and right p 1 (Fig. 10 J); TTMEU−CY− 49, right mandibular ramus with root of canine, and p 3 – m 2 (Fig. 10 K). Measurements are given in Table 4. Description. — A slight overlap in the preserved parts of the dentaries show that the two specimens are not from the same individual. Although the posterior border of TTMEU−CY− 45 is missing, it is clear that the symphysis was short and stout. The incisors are inserted along a rather shallow arch, the line joining the alveoli of i 1 and i 3 being inclined at about 50 ° in respect to the sagittal line. There is a minute diastema between i 3 and the canine, itself separated from p 1, which is present on the right side only, by a very short diastema. The incisors are robust but quite short, although they are only slightly worn, at their tips and along the lingual (dorsal) ridge of i 1. The i 2 is much broader than i 1, and its flange, laterally offset, overlaps the labial face. The third incisor is missing on both sides, but was intermediate in diameters between i 1 and i 2. The length and relative position of the incisors, as well as the lack of diastemata, clearly point to a suid with a shortened muzzle. The canine, rather vertically inserted, is imperfectly preserved, but it has an oval cross−section and a clear demarcation between crown and root. The latter, not being visible at the break just behind p 1, must have been quite short, in sharp contrast to that of the other specimen (TTMEU−CY− 49), and TTMEU−CY− 45 is likely from a female individual. The p 1 is small but not vestigial (Fig. 10 J). It is strongly compressed transversally, with a main tubercle, plus an anterior cuspid and several small ones along the main cristid. There are two fused roots. TTMEU−CY− 49: this specimen is broken in front of p 2. Posteriorly, the symphysis reached at least the level of p 3, and perhaps even that of p 4. Although there is no direct evidence of it, the shape of the dentaries TTMEU−CY− 45 + TTMEU− CY− 49 suggests that the diastema between p 1 and p 2 was short. The canine, of which a part is preserved inside the bone, is much larger than that of TTMEU−CY− 45, and therefore likely from a male individual. The cross−section is of verrucosus − type, with the following approximate widths (in mm) of the three sides: lingual = 20; antero−labial = 15; postero−labial = 15. The former two sides are covered with enamel. Only the posterior root of p 2 is preserved; all that can be said is that this tooth was rather large. The other cheek−teeth are in medium wear, except m 1, which is in late wear. The p 3 is a large and robust tooth, being even slightly longer than p 4. Its morphology is simple, without any evidence of division of the main cuspid, which is inflated, especially labially; the talonid is expanded disto−labially into a strong vertical buttress. In lateral view; the steep slope of the anterior wear facet shows that the anterior accessory cusp was low, but it is also buttressed on both the lingual and labial sides. The p 4 is stout and broad. As on p 3, the anterior accessory cuspid is broadened. Comparisons. — The most common suid of the Mediterranean late Miocene is Microstonyx, whose systematics have long been debated. In the Turolian, in spite of the wealth of the material, recent reviews (Bonis and Bouvrain 1996; Kostopoulos et al. 2001) have highlighted the difficulty to recognise two or more taxonomic entities. The variation range of the Pikermi m 3 s encompasses those of most of the other samples, except some Greek ones (Vathylakkos, Kerassia, Perivolaki), and no clear metric trend through time is evident. Therefore, we will include all of them in Microstonyx major. This species differs from the one present at Yulafli by a number of features: – the anterior part of the dentary is much more elongated so that, even though the symphysis is much longer, it does not reach farther posteriorly than the level of p 2, and usually remains more anterior. The i 3 is more posterior relative to i 1 – i 2; there is a diastema between i 3 and the canine, and a very long one between the latter and p 2. – i 1 and i 2 are much longer, adding to the long slender as− pect of the symphysial area, which much contrasts with that of the Yulafli specimen. This difference has also been mentioned by Made (2003). The i 2 is not so broad relative to i 1. – the canine is much smaller (it may even be missing), the difference being more marked in the male. – p 1 is always missing. – p 2 is smaller. – the premolars are smaller, and p 3 is shorter relative to p 4 (Fig. 12), but its anterior accessory cuspid is higher and narrower. – p 4 is narrower, especially anteriorly, although some specimens approach the condition seen at Yulafli. This tooth, which is rather variable in M. major, especially in the de− velopment of the “ Innenhügel ” does not significantly differ in other morphological features. Therefore, the Yulafli suid cannot be referred to M. major. It is much closer to the earlier species antiquus, often included in the same genus, or in Hippopotamodon Lydekker, or in Limnostonyx Ginsburg, of which it is the type−species. Following most recent authors, we will regard both latter generic names as synonymous, and include antiquus in it. Bonis and Bouvrain (1996) gave a clear account of this species, and we will follow their conclusions here. Besides the type locality, Eppelsheim, well−documented reports of this species are from Montredon in France (Ginsburg 1988) and perhaps from Sophades in Greece (Thenius 1955) although, as noted by Kostopoulos et al (2001) the teeth from Sophades are small. Bonis and Bouvrain (1996) also referred to this species some teeth from Akkirma, a site of unknown age near Ankara, and the anteriorly broadened premolars described by Senyürek (1952) support this identification. Further material of this species in Turkey includes a dentary MTA− 2388 from Bayraktepe, and a dentary MTA− 1964 from “ Uşak ”. The latter locality is very imprecise but, as this dentary is very different from those of Kemiklitepe, one of the main sites close to Uşak, Akçaköy is a more likely provenance. A few more specimens come from Sinap Tepe, near Ankara. MTA− 1955 (or 1953) displays the lower incisors, set in a shallow arch, and without any diastema between them and the canines, which are large. In the MNHNP, the holotype of Dicoryphochoerus meteai Ozansoy, 1965, from Yassiören, is a complete right dentary. The i 2 is much larger than i 1; there are only very short diastemata between i 3 and the canine, and between p 2 and p 1, which is long and bi−rooted. The p 3 is long, and p 4 is broad. All these specimens, in contrast to M. major, share the features observed in the Yulafli specimens. These differences between H. antiquus (including the Yulafli suid) and M. major far exceed those between any two Turolian samples of Microstonyx. Even if there is only one species of this genus in the Turolian, we find it difficult to include antiquus in the same genus, as the differences between the two species would be far greater than between two living suid species (e. g., Sus scrofa / S. barbatus, Phacochoerus aethiopicus / P. africanus), and we prefer to use Hippopotamodon. Indeed, pending detailed phyletic analysis, there is no evidence that H. antiquus and M. major form a monophyletic group. We agree with Bernor and Fessaha (2000) that “ There is little data supporting its [Microstonyx] transition in MN 10 from Hippopotamodon antiquus. ”. In sharp contrast to Indarctos, for instance, no intermediate form is known, and Microstonyx is more likely to be a Turolian immigrant into Europe and the Eastern Mediterranean. In any case, there is a clear chronological distinction between both genera, Microstonyx being known only in Turolian−equivalent sites, while all sites with H. antiquus are earlier.	en	Geraads, Denis, Kaya, Tanju, Mayda, Serdar (2005): Late Miocene large mammals from Yulafli, Thrace region, Turkey, and their biogeographic implications. Acta Palaeontologica Polonica 50 (3): 523-544, DOI: 10.5281/zenodo.13620702
039787DCFFE4981EBB55157921BFFC52.taxon	materials_examined	Type locality and age: Lyon Croix−Rousse, Vallesian, Miocene. Material from Yulafl I. — TTMEU−CY− 139, dentary with slightly worn m 3; TTMEU−CY− 140, dentary with p 4 − m 3, in middle wear; TTMEU−CY− 141, dentary with p 4 − m 1; TTMEU−CY− 141, metatarsals III−IV, lacking distal ends. Description. — TTMEU−CY− 140 has the most complete tooth series, as it includes also the alveolus for p 3, which was about as long as p 4. The latter is a long narrow tooth, not very much broader across the talonid than across its anterior lobe, which is also relatively long. The morphology of m 3 is best seen on TTMEU−CY− 139 (Fig. 10 H). The labial crescents are fully selenodont, and the tooth is not very brachyodont. The disto−labial wing of the protoconid meets the hypoconid, while its disto−lingual wing joins the mesial end of the entoconid, which is rather more labial than the disto−labial wing of the metaconid, which curves labially near its distal end. The third lobe is large and forms a complete loop; the lingual wall is thin but reaches the disto−lingual corner of the hypoconid. The ectostylid is weak, as on the other specimens, but it is larger on the m 1 of TTMEU−CY− 141, a dentary with the ascending ramus preserved (Fig. 10 I). The labial cingulum is weak. Measurements are given in Table 5. Comparisons. — As acknowledged by most authors, systematics of Dorcatherium is not easy, because of dental uniformity and overlap of size ranges between species. Furthermore, the genus is not common in the Eurasian upper Miocene, and descriptions are very scarce. The m 3 s are larger than a Vallesian tooth from Sigindere (Alçitepe) near Gelibolu, in the TTMEU (length of m 3 = 17.8), but slightly smaller than an m 3 from Kozbaşi near Çanakkale, a locality which is probably also of Vallesian age, as shown by the occurrence of Hippopotamodon antiquus. They are also slightly larger than the teeth from Eppelsheim, whose length was given as 17 – 19 mm by Kaup (1832). Their dimensions are almost identical with those from the type locality of the Vallesian, Los Valles de Fuentidueña, which were given by Morales and Soria (1981) as 12.1 × 5.4 and 20.1 × 8.8 for p 4 and m 3 respectively. However, the p 4 s from Yulafli lack the disto−labial expansion of the hypoconid present in the illustrated tooth from Spain (Morales and Soria 1981: fig. 1 C). Both the m 3 and m 1 – m 3 lengths at Yulafli are similar to those of D. jourdani from Küçükçekmece. The latter measurement is given by Malik and Nafiz (1933) as 41 mm, and the length of the largest m 3, which is similar to the Yulafli teeth, is 20.5 mm (MNHNP).	en	Geraads, Denis, Kaya, Tanju, Mayda, Serdar (2005): Late Miocene large mammals from Yulafli, Thrace region, Turkey, and their biogeographic implications. Acta Palaeontologica Polonica 50 (3): 523-544, DOI: 10.5281/zenodo.13620702
039787DCFFE4981EBB5516E925F4FDD0.taxon	materials_examined	Type species: Dorcatherium naui Kaup, 1836; Eppelsheim, Vallesian, Miocene.	en	Geraads, Denis, Kaya, Tanju, Mayda, Serdar (2005): Late Miocene large mammals from Yulafli, Thrace region, Turkey, and their biogeographic implications. Acta Palaeontologica Polonica 50 (3): 523-544, DOI: 10.5281/zenodo.13620702
