Restos de Acinonyx pardinensis (Croizet et Jobert) del Villafranquiese Medio de Varshets (Bulgaria) y la historia de los guepardos en Eurasia durante el Plio-Pleistoceno-Acinonyx pardinensis (Croizet et Jobert) remains from the Middle Villafranchian locality of Varshets (Bulgaria) and the Plio-Pleis

The article decribes and discusses remains of late Pliocene (middle Villafranchian) cheetah, Acinonyx pardinensis (Croizet et Jobert) from Varshets, N.-W. Bulgaria. It is accepted that three chronoforms A. pardinensis pardinensis, A. p. pleistocaenicus and A. p. intermedius succeed in Europe from the Early Villafranchian (Etouaires, approx. 2.6 Ma) till the Middle Pleistocene. Remains are scarce and at this level of knowledge we could accept the recently proposed subspecific status of all these forms but their taxonomical relations could be rather more complicate. The remains from Varshets correspond to the earliest form A. pardinensis pardinensis which inhabited Eurasia during the late Early and the Middle Villafranchian and the beginning of the Pleistocene. Other European remains of A. p. pardinensis are known only from Spain, France, Italy and the Azov region of Southern Russia. It seems that then the species occupied only the southern part of the European continent (the Mediterranean-Balkan-Northern peri-Pontic area), and from there it probably spread till Central Asia (Tajikistan). Such a distribution supports the concept of the faunal entity of South Europe in Villafranchian time as well as the theory for the Central Asian influence of this fauna mainly trough the Balkans by two ways: via Bosphorus and via the peri-Pontic area. According to the paleontological data the fossil Acinonyx does not reach Central Europe (Austria and Germany) before the Epivillafranchian, possibly after new waves of dispersal of another, Early Pleistocene form of Central Asia. The relationship of the Middle Villafranchian A. pardinensis with the rather contemporaneous forms from China (Hezheng) “Sivapanthera” linxiaensis and A. kurteni and from Siwaliks is not clear, but the Chinese forms must represent taxa different from A. pardinensis. Acinonyx s. str. is characterized by strongly domed and enlarged frontal area, shortened skull and downwards inclined neurocranium. The new data argue the supposition of its Eurasian origin at the end of the Ruscinian/ beginning of the Villafranchian.


Introduction
The well known Middle Villafranchian locality of Varshets, N-W Bulgaria, has yielded more than 150 vertebrate species: about 20 amphibian and reptile species (M. Böhme, N. Tzankov pers. comm.), 74 bird species (BOEV, 1999(BOEV, , 2000, 37 species Micromammalia (POPOV, 2001) and 21 species Macromammalia (Spassov, 2000 and additional pers. data). This is evidently the richest Pliocene locality of vertebrate fauna in Europe. The mammal megafauna of this locality demonstrates an apparent similarity with localities typical for the late Pliocene -Middle Villafranchian, such as Saint-Vallier and Chillhac (France), La Puebla de Valverde (Spain), etc. Varshets is placed, after its macromammal fauna, in the first half of the NMQ17, between Roccaneyra (possibly close to) and St. Vallier. The appearance of the fauna of Varshets shows great similarity to that of the European-Mediterranean localities but unlike most of the well known Mediterranean faunas, that of Varshets also shows a connection with the faunal associations of Middle-and E-Europe. The relationship with the Middle East is evidently strong, too. A tendency of prevalence of taxa with S-European and Middle East distribution emerges. This, in fact, appears to be quite logical bearing in mind the geographical position of the locality and the clear indications of an influence of Asian elements (SPASSOV, 2000;2003). The distribution of the macromammalian species from Varshets in different habitats shows that the physiognomy of the landscape is determined chiefly by forest-steppe/mosaic biotopes or open forests. The open forests (park-like forests) and shrubs prevail. The analysis of the micromammalian species leads to similar conclusions for a temperate, mild climate and a mosaic landscape with forests and meadows prevailing over typical steppe vegetation (POPOV, 2001). The data from the analysis of the ornithofauna are similar. The birds of the open spaces are 63 % of the number of the forest birds, but dominant are the birds of the mixed (mosaic) environments, especially the phasianids (BOEV, 1999).
Till now Acinonyx pardinensis Croizet et Jobert is only mentioned in the check-list of Varshets (Spassov, 1997;2003). The fossil remains of cheetah in Europe are still rare, that determines the importance of the description of the Acinonyx remains from Varshets.
The remains from Varshets, belonging to at least 4 individuals, make the cheetah the most numerous carnivore species in the taphocoenosis of the locality.

Description and comparison
The canines (FM 849; 851) have the characteristic cheetahlike crown shape: relatively short and thick. The typical felid longitudinal furrows are visible on the labial surface. Normally these furrows are absent on the A. jubatus canines. The aforementioned furrows are also not visible on the specimens from the Epivillafranchian cheetah from Untermassfeld (Hemmer, 2001), but exist, weakly marked on the canines of the specimens from St. Vallier, as well as in the lower canines of A. pardinensis from the MNQ18 of Fonelas (Spain) (Garrido, 2008). Unlike most felids however, those furrows on the Varshets canines are relatively weak and short. Probably this felid character disappears secondary in cheetah and could be more frequently manifested in the larger teeth of the older populations of the genus. The canines correspond in their size to the compared specimens of A. pardinensis from St. Vallier (MNQ17) (table1). Upper P4s are characterized by a strong reduction of the protocone which shapes no point and is very slightly protruding in lingual direction. It looks more like a prolongation of the lingual root of the tooth at the crown level. The root itself is closely pressed to the main body of the tooth. Such morphology is typical for cheetahs (Teilhard De Chardin & Leroy, 1945;Viret, 1954;Sotnikova, 1989). In this respect, the P4 differs clearly from the P4 of the recently described species from the Late Pliocene of China, named Acinonyx kurteni (see below). The size of the varshets P4 is statistically larger than the Recent A. jubatus. The preparastyle of the Varshets specimens is only vestigial, clearly less developed than in the recent form, which explains why the mesial paracone border is more oblique than in the recent species. The protocone of the Recent species is slightly more distally displaced. In the position of the protocone (displaced mesially), the Varshets specimens also correspond to another specialized form that is close to Acinonyx and mainly to Puma -the North American Miracinonyx inexpectatus (= M. studeri) (JWT 563-420664 -cast, col. SI-NMNH). The size and morphology of the Varshets P4s completely correspond to A. pardinensis from the Middle Villafranchian of Saint-Vallier (France). Both the MNQ17 samples, the sample from St.-Vallier and the one from Varshets do not reach the dimensions of the Untermassfeld (Germany) A. p. pleistocaenicus teeth (see Hemmer, 2001). The maximum dimensions of the Late Pliocene N. African A. aicha (Geraads, 1997) are also larger. At the same time, the European cheetahs from St. Vallier and Varshets have slightly thicker upper P4s than the mentioned primitive African species (Table 2).
The calcaneus (FM 878) is typically felid (see Gromova, 1960), its size is rather larger than a jaguar and much smaller than a male lion. The sustentacular (inner astragalus) facet is obviously rounded, rounder than in a jaguar or leopard for example. Its articular surface is in a single plane. Plantary from the sustentaculum there is a very strong furrow which usually is not that deep in the noted species of genus Panthera. In the proximal end, on the lateral surface, there is a pronounced elongated fossa. In proximal direction it reaches the proximal edge of the bone (i. e. the edge of the cuboid facet). This fossa is practically missing in lions and jaguars. The bone is larger than the calcaneus of Megantereon and lacks its strong convexity on the lateral surface of the proximal end. Although of approximately the same total length, the Varshets calcaneus differs strongly from Homotherium: it is much more slender, and its distal portion is quite elongated; it lacks also the navicular facet typical of Homotherium (Ballesio, 1963). Significantly larger than the calcaneus of a Recent cheetah (60-78 mm after Gromova, 1960), the Varshets specimen fully corresponds in shape and size to the calcanei of A. pardinensis from Saint-Vallier stored in the MHNL (Table 3) but does not reach the values of the specimen from Untermassfeld.
The astragalus is typically felid. Compared to Homotherium (with length 48.3 mm and width 45.5 mm -coll. MNHNP, Lab. de Paléontologie) it is insignificantly longer but much narrower, with a narrower trochlea and differently situated caput. The caput of the astragalus looks "twisted" in ventral view, if compared to other large felids like leopard and jaguar: its maximal diameter is obliquely placed, compared to the noted species of genus Panthera. There is a strongly expressed droplike fossa on the lateral surface of the caputs' column, with its broader part ending ventrally on the edge of the articular surface of the caput. The elongated sharp end of the fossa touches the small facet for the calcaneus. A similar fossa is present on the astragalus referred by Kurten & Crusafont (1977)   the figured astragalus) to A. pardinensis from Puebla de Valverde. In all noted characters and in size (table 4.) the Varshets astragalus corresponds to the astragalus of A. pardinensis from St.-Vallier (Coll. MHNL) but it is smaller than the specimen from Untermassfeld.
The Varshets cheetah has morphological and morphometrical differences with the modern species. The postcranial remains are not only of larger size than the recent species, but also have more massive proportions. The fossil cheetah from Varshets was much larger than the Recent species and possibly not so fast. After the visual comparison of the astragalus and the calcaneus with these of different recent large cats with known weight, the body weight could be about 100-110 kg. The average weight of the recent male A. jubatus is somewhat more than 40 kg (Krausmann & Morales, 2005) and could reach no more than 65-70 kg. The upper P4 of the cheetah from Varshets has more oblique and elongated parastyle (as in "Miracinonyx" inexpectatus) than in the recent species. This puts their protocone in an even more anterior position than in recent cheetahalmost at one level with the parastyle. Those differences, despite the individual variations, show an obvious regularity. Having in mind the strong difference in body size (see below), the only statistically larger P4 of the Varshets form represents a rather plesiomorphic stage of the carnassial teeth in relation to the Recent species. Both morphological and morphometrical data show that the discused fossils belong to large cheetahs, fully similar according the existent data to Acinonyx pardinensis (Croizet et Jobert) from the Early Villafranchian to the beginning of the Late Villafranchian of Europe. The morphological and metrical comparison of the fossil cheetahs is difficult because of the scarce material and the body weight comparison is only approximate. If the body mass of the Varshets cheetah (~ 100-110 kg after the postcranial remains) is comparable with the estimated weight of the Villafranchian A. pardinensis (probably males and females) from Les Etouaires (80-120 kg), St. Vallier (60-80 kg, calculated after tooth length, but A. pardinensis has smaller carnassials related to A. jubatus: see above), Olivola and Casa Frata (90-120 kg) and Dmanisi (around 100 kg), it would be smaller than the Epivillafranchian cheetah from Untermassfeld (110-140 kg) and larger than the Middle Pleistocene cheetah from Hundsheim and Mosbach (60-90 kg.) (see Hemmer, 2001;Hemmer et al., 2008;Hemmer et al., 2011).

Discussion. What is a cheetah? Problems of the taxonomy, the origin and the phylogeography of the cheetahs
The time and place of the origin of the genus Acinonyx are still debatable. Africa has been pointed out as one possibility (Sotnikova, 1989). Indeed, there are African finds referred to cheetahs, having a supposed absolute age ca. 4 Ma (Omo Mursi) and more than 3.5 Ma (Laetoli). Those remains, however, are very fragmentary, and their generic belong-ing is not unequivocal (Turner, 1990). Turner (1990) supposed that cheetahs are immigrants into Africa. Recently, a new cheetah species with some plesiomorphic features compared to A. Pardinensis, such as a less convex and less specialized skull, was found in Africa (Geraads, 1997). The age of this form, Acinonyx aicha Geraads is about 2.5 Ma. In Africa, the oldest unequivocal fossil remains of genus Acinonyx s. str. are as old as 3 Ma (early MN16) and have been found both in Africa (Makapansgat 3) (Turner, 1990). In Europe the oldest A. pardinensis remains are from Perrier-Etouaire, from where comes the type specimen (a mandible: see Schaub, 1949) of the species A. pardinensis Croizet et Jobert. The age of Les Etouaires is not very clear but it is referred to MNQ16b (Steininger et al., 1990) and could be at about 2.5 Ma (Guérrin, 2007) or most probably slightly older, close to the age of Montopoli (Aguirre et al., 1997). The oldest known Asian remains (India -Pinjor, China) could be slightly younger. Recently a new species of cheetah with an estimated age of 2.2-2.5 Ma, that demonstrates a typical cheetah skull shape, but a rather plesiomorphic stage of the upper carnassials, was described in Hezheng area, China under the name of A. kurteni (Christiansen & Mazák, 2009). A. kurteni seems to be an Acinonyx s. str., characterized by advanced skull shape but plesiomorphic tooth morphology. Having in mind the features of this middle Villafranchian Asian specie and the co-existence of another species in Europe, we could suppose the presence of two parallel lineages of cheetahs in Eurasia from the Early Villaftranchian: A. kurteni and A. pardinensis 1 , and to imagine an Eurasian origin of the genus in Late Ruscinian/Early Villafranchian time. It is accepted that Acinonyx (and possibly Uncia) is a close relative of Puma (Salles, 1992;Van Valkenburgh et al. 1990) (The close affinities of the snow leopard with the Acinonyx-Puma group is contested by recent DNA investigations (Werdelin et al., 2010)). The question of the place and time of the Puma origin is also still unclear. A form morphologically similar to cheetah, the American "cheetah" -Miracinonyx (Van Valkenburgh et al., 1990), that was referred until recently to Acinonyx s. str., possibly shares a common ancestor with Puma from which both, Puma and Miracinonyx diverged, after some opinions, in N. America at about 3.2 Ma ago (Barnett et al., 2005). As Aci- nonyx, Puma is also known in the Early Villafranchian of Eurasia with a form, that merits subgeneric separation as P. (Viretailurus), having some skeletal and cranial differences from the recent form (personal observations). The oldest Puma localities known are Shamar, Mongolia and Kvabebi, Georgia but also Perrier-Etouaires, France (Hemmer et al., 2004;Kurten & Crusafont (1977) and Red Crag, Great Britain, all of them with Early Villafranchian age. Hemmer et al. (2004) refer some African remains with an age of about 3.5-3 Ma to Puma and suppose an African origin of the genus. Except from Varshets European fossil cheetah are known from about sixteen early Villafranchian to middle Pleistocene localities of France, Spain, Italy, Austria and Germany (See Garrido, 2008 andHemmer et al., 2008) as follows: Etouaires, Saint Vallier, Senèze, Le Vallonet, Saint-Estève (France); La Puebla de Valverde, Las Higueruelas, Villarroya, Fonelas (Spain); Montopoli, Villafranca d'Asti, Olivola, Casa Frata (Italy); Untermassfeld and Mosbach (Germany); Hundsheim (Austria). A find from Liventsovka (Priazovian South Russia, NMQ17) noted as Acinonyx sp. (Titov, 2008) must represent the same species.
After the recent concept (Hemmer et al., 2008) the European fossil cheetahs belong to a single and polymorphic large species or "macrospecies" -A. pardinensis (s. lato). The status of Felis arvernensis Cr. et Job., described, as well as A. pardinensis (Cr. et Job.) from Etouaires is still debatable. Some authors express the opinion that this cat has specific differences from A. pardinensis (see Heintz et al., 1974; see also Hemmer et al., 2008). It is difficult to decide from one hemimandible only if this is a separate taxon or an aberrant individual of A. pardinensis from Perrier -Etouaires. Possibly, however, Viret (1954) was right to consider this taxon a synonym of A. pardinensis, and this seems a more parsimonious hypothesis. Indeed, the type mandible has some atypical features such as the elongated distal part of P4, the shape of the paraconid (a specific occlusion) etc. At the same time, a number of characters such as the short massive canine, the short diastema, the rather massive symphysis and the position of P3 indicate cheetah affinities (Spassov, 1999). The history of cheetahs in Europe is still unclear, due to the scarce material. I will follow Hemmer (2001) and Hemmer et al., (2008) who proposed a taxonomic concept according to which the fossil cheetahs of Europe could be dis-tributed in three successive taxa with subspecific rank: A. p. pardinensis (Cr. Et Job.), for the Early Villafranchian till the Pleistocene beginning (Etouaires -Casa Frata), A. p. pleistocaenicus (Zdansky) for the Epivillafranchian form (Untermassfeld): a migrant from Asia) and A. p. intermedius (Thenius) for the Middle Pleistocene cheetah (Hundsheim and Mosbach). At the same time the Cheetah evolution could be much more complicated. Having in mind the considerable time span between the first and the last representative of the species on the continent we could suppose specific differences between the separate successive forms. Besides the probable size differences, some morphological differences support this supposition: Hemmer (2001) and Hemmer et al., (2008) note some morphological differences in the mandibular symphysis between A. pardinensis pardinensis (the Etouaires -Olivola form) on one hand and the Epvillafranchian (Untermassfeld) and the Middle Pleistocene (Hundsheim, Mosbach) forms on another hand. If we judge after the recent carnivores, such kind of features that concern not only size and proportions but also the skull morphology could be regarded as criteria for specific distinction. After this second taxonomic possibility A. pleistocaenicus could be regarded as a separate Asian species, that invaded Europe during the strong aridification at the beginning of the Epivillafranchian.
The Varshets remains fill the knowledge gap on the dispersal of A. pardinensis (especially of "A. p. pardinensis" sensu Hemmer et al., 2008) in the Villafranchian of Europe and show its presence not only in the Western Mediterranean but also in Southeastern Europe. We could conclude now, after the distribution of the known cheetah localities, that in the time interval between the Early Villafranchian and the beginning of the Pleistocene this cheetah occupied only the southern part of the European continent (the Mediterranean-Balkan area). From there, it was probably spread through the peri-Pontic area (South Russia -Liventsovka: Sotnikova, 1989 andDmanisi -Georgia: Hemmer et al., 2011) till Central Asia: Tajikistan (see Sotnikova, 1989) and in case that Felis brachygnathus Lydekker, 1884 (= Sivafelis brachygnathus (Lydekker) sensu Pilgrim, 1932) from Pinjor (Siwalik) represents the same species of cheetah, it could be spread also till India. Such a distribution supports the idea of the specific character of the South European fauna in Middle Villafranchian and of the strong Central Asian influence on this fauna trough the Balkans (Spassov, 2000;2003). According the paleontological data the fossil cheetah does not reach Central Europe (Austria and Germany) before the Epivillafranchian and this is possibly related to a new wave of dispersal from the east, related to the continuation of the aridification.
The assignation of the scarce Indian (Pinjor) material to A. pardinensis is debatable. Qiu et al. (2004) are probably right to separate their new species Sivapanthera linxiaensis from Acinonyx. The skulls of this probably Middle Villafranchian taxon that is again (as A. kurteni) from Longdan, Hezheng area (China), figured in Qiu et al. (2004) differ from Acinonyx mostly by the less domed frontals, the large infraorbital foramen, the strong and elongated sagital crest, the more elongated muzle, the pinched caudal portion of the muzzle (the maxillary bones) on both side of the nasals. With all these features this new taxon shows similarities with Puma and could be regarded as a member of the Acinonyx stem-group. Its relations with Miracinonyx could represent certain interest. At the same time, the close affiliation of the later form "Cynailurus" pleistocaenicus Zdansky with Sivapanthera linxiaensis does not seem to be strongly supported by the paleontological facts and I will follow Hemmer et al. (2008) in the interpretation of this taxon as a trough cheetah, that migrated in the Epivillafranchian in Europe.
The recent fossil material described in Central Asia complicates the problem of the definition of the Acinonyx diagnosis: several forms of Pantherasized cats with (1) short canines and (2) relatively developed frontal sinuses as well as with different degree of expression of the typical cheetah/cheetahlike facial and carnacial features have existed in a very large territory in the Early/ Middle Villafranchian: 1. forms with relatively flattened frontals; medium elongated/shortened skull; large (?) infraorbital foramen, but P4 protocone, with a cheetah-like trend to reduction (Acinonyx aicha, "Sivapanthera" linxiaensis; Miracinonyx); 2. forms with well to strongly domed and enlarged, cheetah-shaped frontal area and shortened skull; but with 2.1.: unreduced P4 protocone (A. kurteni) or 2.2.: reduced protocone (A. pardinensis). We could mention that the reduction of the P4 protocone is a feature that has appeared more than once in the felid evolution and this trend could appear in parallel especially in closely related forms.
Possibly the most peculiar features of the Acinonyx s. str. skull is the visibly bulged frontals, related to the strong development of the frontal sinuses. The development (to a different extent) of the frontal sinuses and the bulging of the frontal area represent synapomorphy of the closely related Puma, Acinionyx and Uncia but the extremal stage this development is reached in Acinonyx. The specific inclined position of the neurocranium of Acinonyx in respect to the palatal plane and the related more oblique position of foramen magnum, as well as the strong vertical elevation of the fronto-orbital region in this felid correspond to the important role of vision and are typical features of the cheetah. These features could be regarded as autapomorphies of Acinonyx as adaptation to a mode of life in open landscapes.
Post scriptum from the author After this paper was in print Deng Tao's publication (2011) came out. This paper revealed the news that the skull (type and single specimen) of this newly described species from China (Christiansen & Mazák, 2009) is a fossil forgery. Deng (2011) especially notes that the zygomatic arch, the occipital part, the sagital crest, the basiooccipital area and parts of the parietal area (above all the posterior part of the skull) bear the traces of artificial modeling. It is possible that the cheetah-like domed shape of the skull roof has not been affected by the human interference (see the answer of Christiansen & Mazák cited in Deng Tao (2011), however we could not be sure in the natural appearance of the skull's contour of this Chinese specimen (Deng Tao, pers. comm.). Neither the age, nor the location of the remains are clear; it could have been from Longdan (2.2-2.5 Ma), and neither is that sure. For this reason the shape of the skull and the name and age of the location cannot be used as firm arguments concerning some of the theses offered in my publication, and in particular: • the existence of "two parallel lineages of cheetahs in Eurasia from the Early Villagranchian: A. kurteni and A. pardinensis ". For this reason the probability, put forth in the paper that the cheetahs rather originated in Eurasia cannot be launched with priority before the version that they are African in their origins. The issue is still awaiting its solution.
The existence in the Middle Villafrachian of forms with strongly domed and enlarged, cheetah-shaped frontal area and shorten skull, but with unreduced P4 protocone cannot be accepted for certain.
It is quite possible that the strongly domed skull was a relatively late apomorphy with cheetahs, dating as late as the Pleistocene.

AKNOWLEDGEMENTS
The comparison of the material is carried out in different museums' collection in the 90s with the support of the Bulgarian Academy of Sciences -Dept. of International Collaboration, the Bulgarian Ministry of Science and Education as well as the Embassy of France in Sofia. The author is very grateful to Zhanxiang Qiu (Inst. of Vertebtrate Paleontology and Paleoanthropology, Beiging) for the helpful discussion, to the late Leonard Ginsburg (MNHN-Paris), Michel Philippe (MHNL) and Steve Jabo (Smithsonian Institution, NMNH, Washington D.C.) for the access to the fossil collections and the useful comparative photos sent. Thanks to Jorge Morales as well as two the anonymous reviewer for the very useful remarks and comments that improved the manuscript.