The northeast-trending Burdur Basin rests on the Mesozoic basement of allochthonous limestone and ophiolitic units of Lycian nappes and Eocene–Oligocene parallochthonous sedimentary units. The Neogene sedimentation consists of alluvial, fluvial and lacustrine deposits with volcanic intercalations (Fig. 1). Geological mapping and lithostratigraphic subdivision of the basin was completed by Wedding (1966), and later the studies by Karaman (1986) and Price (1989) proposed a Miocene–Pliocene age for the basin-fill succession. More recently, geologic mapping of the basin and its surroundings were compiled by Şenel, 1997 and Şenel, 2002. A summary of the basin stratigraphy was presented by Price and Scott, 1989 and Price and Scott, 1994 suggesting the basin formed in a half-graben setting in the control of a NE-dipping main fault as a result of crustal extension. They proposed an 1100 m thick of basin-fill succession consisting of coarse- to fine-grained clastic deposits and subdivided into Çendik, Akdere and Günalan members in ascending stratigraphic order. The Çendik and Günalan members are dominated by fluvial sediments that laterally pass into alluvial-fan deposits, whereas the Akdere member is restricted to the central basin area and dominated by lacustrine deposits. At the present time, the basin is actively evolving, punctuated by seismic activity along secondary faults confining the present Burdur Lake and causing a shift of its depocentre.

Terrestrial sedimentation in the Burdur Basin started in the Late Miocene in an extensional setting, hosting alluvial-fan, fluvial and palustrine depositional systems (Price and Scott, 1989). The basin has a similar developmental history with a coeval basin array along the Tauride hinterland that has been documented in detail in several basins such as Çameli, Eşen and Çal (Alçiçek, 2007, Alçiçek et al., 2005 and Alçiçek et al., 2012). The initial basin subsidence is attributed to a rifting pulse that affected the entire hinterland of the Lycian Taurides. At the initial stage of sedimentation, the marginal alluvial-fans were prograded from basin-bounded fault escarpments into axial fluvial systems. During the Early Pliocene, the resumption of regional extension caused a new pulse of rifting in SW Anatolia and, accompanied by regional humid climatic conditions, caused more accommodation space for extensive lacustrine basins. The lacustrine environments were gradually shrunk due to denudation and degradation of the surrounding basement rock that caused it to overfill the basin. The fossil locality presented in this paper is surrounded by marginal alluvial-fan to fan-deltaic depositional settings of the uppermost reaches of the basin-fill succession. By the Early Pleistocene, changes in the regional extensional direction were attributed to a third rifting pulse, dissecting the basins by newly generated faults and rearranging basin floors by tilting to establish the present basin configuration and morphology.

The locality was discovered by MCA in 1999 during field reconnaissance in the Burdur Basin, and the first fossil discoveries were shown to Gercek Saraç, who described them as Equidae Hipparion sp. and Giraffidae indet. in his extensive report on the Anatolian mammal localities (Saraç, 2003). Later, the findings were given to SM who tentatively identified them as Paracamelus sp. Then the locality was revisited and new materials were discovered. VVT joined in for further descriptions and he recognized the significance of the specimens for interregional faunal linkage. As Paracamelus is rather rare all over Eurasia, the Burdur material is of crucial importance to contribute to basin stratigraphy and interbasinal correlation. The specimens are currently stored in the collection of the Natural History Museum of Ege University in Izmir, Turkey.

Order PERISSODACTYLA Owen, 1848

Family EQUIDAE Gray, 1821

Genus Equus Linnaeus, 1758

Equus (Allohippus) sp.

Fig. 2d

Material: Left posterior second phalanx of a large monodactyl horse (PV-3504; Fig. 2e).

Description: The bone is wide and relatively short (the ratio of length to mid-shaft width is 0.92; Table 1). Its upper end is slightly narrow than the lower one (the ratio of proximal width to the distal one is 1.1). The width of the diaphysis is noticeably narrower than the proximal and distal ends of the phalanx. The angle of the proximal articular surface relative to its longitudinal axis is 78°. This peculiarity was more typical for E. (A.) stenonis, in which such an angle is generally less than in caballoid horses (Gromova, 1949).

The well-developed ligamental fossae for the attachment of the lateral ligaments of the hoof joint weakly go to the front surface, and are bordered by the ridges from shallow median depression on their dorsal surfaces. Gromova (1949) proposed that the posterior phalanges of stenonid-like horses are characterized by a stronger development of these ligamental fossae compared with caballoid horses.

Remarks: Judging by its size, PV-3504 is smaller than most stenonid-like horses (Table 1), but similar to Equus (A.) pamirensis from Kuruksay (Tadzhikistan; Sharapov, 1986), E. (A.) stenonis senezensis, Equus (A.) sanmeniensis (Tologoy, China), and E. (Equus) nalaikhaensis (Eisenmann and Kuznetsova, 2004, Prat, 1980 and Sharapov, 1986). It falls within the limits of variability of E. (Plesippus) altidens (Alberdi and Palombo, 2012; Fig. 3). It is larger than E. (Hemionus) hemionus, E. (E.) hydruntinus, and E. (E.) gmelini (Kuzmina, 1997), but smaller than E. A.) livenzovensis from Liventsovka and Khapry, E. (A.) bressanus, E. stenonis ssp. from Olivola, E. (A.) verae, and large caballoid horses E. (E.) mosbachensis, E. (E.) uralensis, and E. (E.) latipes. Its measurements fit into the lower limits of E. (A.) stenonis vireti from Saint-Vallier, E. (A.) süssenbornensis, E. (A.) stenonis ssp. from El Rincon and caballoid horses Equus (E.) lenensis (Alberdi and Palombo, 2012, Eisenmann, 2004, Kuzmina, 1997 and Prat, 1980). The find from Burdur differs from phalanges of hipparions by its larger size and the insignificant restriction of the distal epiphysis of the bone.

PV-3504 differs by its massive proportions from most caballoid horses, onagers, E. (Plesippus) altidens, and most stenonid horses. Only phalanges of E. (E.) latipes are notable by more robust construction. By this parameter the Burdur specimen is similar to E. (A.) stenonis vireti, E. (A.) pamirensis, and E. (A.) major from Psekups (Fig. 4). Indices of proximal, distal and diaphysis widths are slightly higher than those of other stenonid horses, but similar to them. Generally, they are lower than those of true horses, and are greater than those of donkeys (Table 1). Based on this comparison, we interpreted PV-3504 as a small E. (Allohippus) sp. with rather massive distal limb bones. The stratigraphic distribution of such forms is Late Pliocene–Early Pleistocene.

Order ARTIODACTYLA Owen, 1848

Family CAMELIDAE Gray, 1821

Genus Paracamelus Schlosser, 1903

Paracamelus cf. gigas Schlosser, 1903

Fig. 2a–c

Material: Fragments of the right radius (PV-3501; Fig. 2a), right tibia (PV-3502; Fig. 2b) and a left astragalus (PV-3500; Fig. 2c) presumably belonging to one individual of a large camelid.

Descriptions: Radius. The distal part of a radial bone of a large animal (Table 2).The medial ridge of the distal articular surface is perpendicular to the width of the bone. The medial facet (to the navicular bone) is not narrowed towards its back. The medial facet (for the semilunar bone) is narrower than the medial and lateral ones.

Tibia. The articular surface of the condylus lateralis is narrowed in the anterior–posterior direction, its anteroposterior diameter is less than its width and less than the anteroposterior diameter of the condylus medialis. The width of the proximal epiphysis is 138.3 mm (Table 3).

Astragalus. The astragalus extends distally. The width of the distal block is greater than the proximal at 10 mm. The inner surface of the bone is bent substantially at its posterior side. The cuboid fossa is rather large. A large malleolar eminence is below the middle of the outer surface of the bone. The tubercle of the inner ridge is located below the malleolar eminence. The proximal block is significantly asymmetrical; its lateral crest is distinctly above the medial one (Table 4). On the lower block a medial crest is well defined on its external part.

Remarks: The available remains do not allow a definite identification of the Burdur camel fossils to either Paracamelus or Camelus. However, the large size and some indices show similarities to large camels like P. gigas, which are known from the Sea of Azov region, Azerbaijan, north-western Mongolia, Kazakhstan, and China (Bajgusheva, 1971, Belajeva, 1937, Teilhard de Chardin and Trassaert, 1937, Titov, 2008b and Zdansky, 1926) (Fig. 5). Taking into consideration the reliable data of the Khapry fauna (Middle Villafranchian; Sea of Azov Region, Russia) the real period of P. gigas existence in eastern Europe is accepted as Late Pliocene (= early Early Pleistocene) (Titov, 2008b). Unfortunately, the ages of other localities of the giant camel very often have no exact stratigraphic position. The discovery of the astragalus in the conglomerates of Bozdağ (Azerbaijan) (Burchak-Abramovich and Akhundov, 1960) is dated as Upper Apheronian, which corresponds with the late Late Villafranchian. The astragalus from locality Oshi (Northwest Mongolia) is dated as Late Neogene (Belajeva, 1937).

Remains of giant camel are known from several localities from the territory of China. The exact age and location of teeth of a giant camel from Honan Province, which are holotypes for the genus and species Paracamelus gigas (Schlosser, 1903), is unknown. More numerous finds of giant camel bones come from SE Shansi (Teilhard de Chardin and Trassaert, 1937), from the province of Honan, Yang-Shao-Tsun (localities 102, B, D; Zdansky, 1926), from Sangan River valley, west of Beijing (Teilhard de Chardin and Piveteau, 1930), and from the Chouk’outien (Zhou-kou-dian) ‘Lower Cave’ site (Young, 1932). The dating of these localities falls in different periods in the Late Pliocene–Early Pleistocene. At various times, the presence of P. gigas was indicated in the early Middle Villafranchian faunas of Youhe and Nihewan (Burchak-Abramovich and Akhundov, 1960 and Haveson, 1954). In recent reports, this camel species is a member of the Xiashagou fauna, which is dated as 0.8 to 2.0 Ma (Deng et al., 2008).

Family CERVIDAE Goldfuss, 1820

Cervidae gen. indet.

Fig. 2d

Material: Radius. Incomplete right radius (PV-3503; Fig. 2d), lacking its proximal part, of a medium-sized deer.

Description: In the distal epiphysis of the bone at the anterior surface there are two well-defined longitudinal ridges with a groove between them. The lower end of the bone is fused with the distal end of the ulna at its lateral posterior surface. The boundary between them at the articular surface is presented in the form of a crack. Ridges on the distal articular surface are splayed relative to the anteroposterior diameter of the distal epiphysis. The width of the facet for the navicular bone is approximately equal to the width of the facet for the lunate bone.

Remarks: The dimensions of the radius are larger than those of Cervus (Rusa) philisi, Croizetocerus ramosus, and are similar to those of the Late Pliocene–Early Pleistocene deer Eucladoceros and Arvernoceros (Heintz, 1970; Table 5). But the Burdur specimen differs by the larger anteroposterior diameter of the distal part of the radius (Fig. 6). The exact determination of the specimen is difficult due to the lack of a developed diagnostic system of the postcranial bones of deer, and the absence of published data about postcranial bones of other Asiatic Plio-Pleistocene deer such as Praemegaceros, Sinomegaceros, Axis, and Elaphurus.

Stratigraphy and palaeoenvironmental changes in the Burdur Basin show an overall tectono-sedimentary interaction from the Late Miocene onward with a close relation with coeval neighboring basins. In the early basin evolution stage, by the Late Tortonian, the basin was subsided to attract a transition of coarse-clastic alluvial-fans into the axial fluvial systems. The Pliocene of the basin was signified by denudation and degradation of surrounding basement rocks and resulted in the deposition of fining upward sequences from alluvial-fan, fluvial into lacustrine deposits, implying widespread humid conditions that were sustained up to the Late Pliocene, comparable with neighbouring lacustrine basins. In the Pleistocene, reactivated extension rearranged the basin configuration. Uplift tilted the former basin-fill that was unconformably overlain by fluvial deposits.

Because fossil camel remains from Turkey are rare, the locality of Burdur is primarily interesting because of the finds of bones of camels. The Anatolian Camelidae localities were reported by Saraç (2003) as Paracamelus cf. alexejevi from Gülyazi (Early Pliocene MN16, Sickenberg et al., 1975), Paracamelus cf. alutensis from Sarıkoltepe (Early Pleistocene MN17, Kostopoulos and Şen, 1999), Paracamelus sp. from Yukarısöğütönü (Late Pliocene MN17, Becker-Platen and Sickenberg, 1968 and Sickenberg et al., 1975) and Camelidae gen. indet. noted from Upper Pleistocene of Sinap Tepe (Sen, 1990). Van der Made et al., 2002 and Van der Made et al., 2003 reported Paracamelus cf. aguirrei from Çobanpınar (Late Miocene MN13, Ozansoy, 1965) as one of the oldest Old World camels. But recently that specimen was withdrawn because the fossil probably came from a nearby archaeological site, but was stored with the Çobanpınar material. The state of fossilization together with its colour and matrix do not fit those of the fossils collected from Çobanpinar (Şen, 2010).

In general, the history of camel evolution and distribution in Eurasia is known (Haveson, 1954). Although Camelidae remains are not very common in the Late Cenozoic of Eurasia, several papers were dedicated to the question of the latest extent of its ancient representatives (Harris et al., 2010, Likius et al., 2003, Pérez-Lorente et al., 2009, Titov, 2008a, Titov and Logvynenko, 2006 and van der Made et al., 2006). As suggested by Honey et al. (1998), Camelidae dispersed through North America during the mid-Eocene and migrated into Eurasia via the Bering land-bridge caused by sea-level fall in the Late Miocene. The earliest Camelidae finds in Eurasia are from the zone MN12-13 localities Pavlodar (Kazakhstan), Sinyavskaya and Novocherkask (Russia), Eupatoria, Odessa and Yabloniya (Ukraine), and MN13 from Venta del Moro and Librilla (Spain) (Morales et al., 1980, Pickford et al., 1995 and Titov and Logvynenko, 2006). These large animals quickly dispersed over Eurasia, occupying ‘savanna’ type biotopes. Apparently, the earliest Eurasian camels were typical members of Late Miocene hipparion communities (Titov, 2008a).

The camel Paracamelus khersonensis (= alexejevi) was typical of the Ruscinian (= Early Pliocene) fauna of the northern Black Sea region. Similar forms are known from Turkey and northern Africa. In western Asia, P. praebactrianus was common during this time. The Middle Pliocene (Early Villafranchian) record of camels is rather fragmentary. Only from Kazakhstan a very large P. longipes is known. In the Late Pliocene, camels were the most varied, and were widely distributed in eastern Europe, Asia and Africa. In the Black Sea region and possibly in the Middle East, the small camel Paracamelus alutensis was common in the Middle and Late Villafranchian faunas. At the same time in different parts of Asia lived larger forms of camels – P. gigas (China, Kazakhstan, Mongolia), and P. trofimovi (Tajikistan). In the Middle Villafranchian of the Sea of Azov fauna, two forms were found, the small camel P. alutensis and the large P. cf. gigas (Bajgusheva, 1971, Bajgusheva et al., 2001 and Titov, 2008b). From the Pinjor beds of the upper Sivaliks from Pakistan and India remains of “Camelus” (= Paracamelus) sivalensis are known (Nanda, 2002).

The late Early Pleistocene history of Eurasian camels is insufficiently known. There are only fragmentary remains from eastern Europe and the Middle East. In the early Middle Pleistocene, the genus Paracamelus was replaced throughout Eurasia by representatives of the genus Camelus.

Monodactyl horses, also immigrants of North American origin, became common in Eurasia in the terminal Late Pliocene–early Early Pleistocene at the Gauss-Matuyama boundary, or even in the upper part of the Gauss chron (beginning of Middle Villafranchian; Agusti et al., 2001 and Lindsay et al., 1980). Although there is evidence that they started to appear in Eurasia sometime earlier (in the Early Villafranchian of France and Greece; Lacombat et al., 2008). During the Late Pliocene–Early Pleistocene monodactyl horses displaced hipparions in Eurasia. But this process occurred gradually and unevenly. During the Middle–Late Villafranchian, stenonid-like horses Allohippus were common.

In contrast to the Miocene, Plio-Pleistocene localities with large mammals are quite rare in Turkey. In limited localities, the systematically studied equid records have come from Sarikoltepe (Early Pleistocene; Kostopoulos and Şen, 1999), Gülyazi (Late Pliocene, Bernor and Lipscomb, 1991), Denizli (Erten et al., 2005) and Çal (Alçiçek et al., 2012) as Equus stenonis, Plesiohipparion huangheese, and E. süssenbornensis and E. hydruntinus, respectively. It is worth to note that a new Equus hydruntinus teeth were discovered by MCA in the alluvial-fan deposits of uppermost reach of the Acigöl basin-fill succession to the west of the Burdur basin, as another important correlation material for Burdur equids. Equus sp. was recorded at the localities of Kamişli, Yukari Söğütönü (Early Pleistocene, Sickenberg et al., 1975) and Dursunlu (early Middle Pleistocene, Howell et al., 1999), but this material has not been studied in detail, and only presented as faunal lists. The limited fossils from Yukari Söğütönü in the collections of MTA (General Directorate of Mineral Research and Exploration, Ankara/Turkey) museum show one type of equid that has close similarities with the small E. altidens forms and the more advanced Dursunlu (early Middle Pleistocene) equids are similar to those of E. mosbachensis and E. cf. altidens (Yiğit, 1998). The few equid fossils from the Çobanisa (late Early Pleistocene, Mayda, 2004) and Kamisli localities are closely comparable with a medium-sized E. stenonis. In conclusion, the currently available limited data on the Turkish Plio-Pleistocene equids show close resemblance with southern European forms in having small to medium-sized forms dominant during the Early Pleistocene and larger forms at the beginning of the Middle Pleistocene.

We should also provide a note on the Elmacık fauna 30 km to the the southwest of Burdur locality (Fig. 1), located at the same stratigraphic level and mostly composed of large mammals that were mainly assigned to “Mastodon” by Kahraman (2009). However, the preliminary reports from this locality surely do not give us an adequate picture of a “Mastodon” type proboscidean. On the contrary, we are now sure that there is also an Elephantinae, most likely a primitive “southern elephant” that inhabited the Burdur Basin during the Late Pliocene–Early Pleistocene (see the mandible on page 98 of Kahraman, 2009). These changes the age considerably from that given in preliminary reports of this fauna as “Mio-Pliocene” and in accordance with the faunal age of the Burdur locality introduced in this paper. Further to the southwest, in the village of Hasanpaşa near Tefenni, another well preserved lower molar of a “southern elephant” Mammuthus meridionalis together with a Gazella sp. were found in coal-bearing fluvio-lacustrine deposits (Sickenberg et al., 1975). The age of the fauna (MN16-17) and stratigraphic position of the locality are in concordance with both further Burdur and Elmacık localities.

Recent discoveries of Equus (Allohippus) sp. and Paracamelus cf. gigas remains allow the confirmation of a Late Pliocene–Early Pleistocene age for the uppermost reaches of the Burdur Basin succession. The first remains of a large camelid from this layer on the territory of Anatolia expand the known distribution of that form. The combination of Paracamelus, Equus (Allohippus) and a medium-sized deer show the typical Middle–Late Villafranchian association, that typical of eastern Europe, and western and central Asia. Taking into consideration the recent data on the dating of the Chinese localities of the Xiashagou fauna with P. gigas (0.8–2.0 Ma; Deng et al., 2008) and the presence of medium-sized stenonis-like horse with massive distal limb bones, we tend to think that the Burdur association is most likely to be correlated with European localities of the early Late Villafranchian.