Morphological versus genetic diversity
Transcription
Morphological versus genetic diversity
Annals of Anatomy 194 (2012) 88–102 Contents lists available at ScienceDirect Annals of Anatomy journal homepage: www.elsevier.de/aanat Eurasian wild asses in time and space: Morphological versus genetic diversity Eva-Maria Geigl ∗ , Thierry Grange Institut Jacques Monod, UMR7592 CNRS Université Paris Diderot, 15 Rue Hélène Brion, Paris, France a r t i c l e i n f o Article history: Received 13 February 2011 Received in revised form 3 June 2011 Accepted 5 June 2011 Keywords: Equids Hemione Palaeogenetics Phylogeography Conservation biology Taxonomy Speciation s u m m a r y The Equidae have a long evolutionary history that has interested palaeontologists for a long time. Their morphology-based taxonomy, however, is a matter of controversy. Since most equid species are now extinct, the phylogenetic tree based on genetic data can be established only imperfectly via deduction of present day genomes and little is known about the past genetic diversity of these species. Recent studies of ancient DNA preserved in fossil bones have led to a simplification of the phylogenetic tree and the classification system. The situation is still particularly unclear for the wild asses whose geographical distribution in the Pleistocene and the early Holocene stretched from Northern Africa to Eurasia before they became endangered or extinct. Therefore, we performed a phylogeographic study of bone remains of wild asses covering their former geographic range over the past 100,000 years based on the analysis of ancient mitochondrial DNA. Here, we will not show but rather discuss our results calling the morphologybased classification into question and indicating that morphological criteria alone can be an unreliable index in inferring various equid species. Indeed, the diversity of mitochondrial lineages in populations with similar morphology along with genetic signatures shared between morphologically distinct animals reveal a significant morphological plasticity among Equus species. The classification of palaeontological species based on morphological and genetic criteria will be discussed. © 2011 Elsevier GmbH. All rights reserved. 1. Introduction The concept of species has evolved from the static view of Linnaeus that strongly influenced taxonomy, to a dynamic view integrating evolutionary biology and systematics. Ernest Mayr has proposed a very influential definition of species as populations that are reproductively isolated from one another (Mayr, 1963). This biological definition has gained wide acceptance but has also been challenged by numerous alternative definitions in the last 40 years that have emphasized various biological properties (for reviews: de Queiroz, 2005, 2007). One can distinguish biological, ecological, phenetic, evolutionary and phylogenetic concepts that are favoured by different subgroups of biologists. De Queiroz has proposed a unifying, more general concept of species as segments of separately evolving metapopulation lineages where the various biological properties proposed in previous definitions are no longer essential but become contingent (de Queiroz, 2005). In this concept, the various properties so far used to define species have solely the status of cumulative lines of evidence that support the hypothesis that two separating and diverging lineages are distinct species (de Queiroz, 2007). Thus, the secondary species criteria are not relevant to species conceptualization but only to species delimitation (de Queiroz, 2007). Indeed, when two lineages diverge, ∗ Corresponding author. Tel.: +33 157278132. E-mail address: [email protected] (E.-M. Geigl). 0940-9602/$ – see front matter © 2011 Elsevier GmbH. All rights reserved. doi:10.1016/j.aanat.2011.06.002 they will progressively acquire distinctive properties. In particular, they will (i) accumulate specific quantitative and morphological characteristics that will make them phenetically and anatomically distinguishable, (ii) occupy specific niches and adaptive zones that will make them ecologically distinguishable, (iii) accumulate genetic differences that will separate them into monophyletic groups in terms of multiple gene trees, (iv) or develop an intrinsic reproductive barrier due to ethological, anatomical or genetic reasons (de Queiroz, 2005, 2007). Since they will not acquire all of these properties simultaneously or in a defined order, different opinions arise as to whether the two lineages should be considered as distinct species or not. Furthermore, since fluctuations in the environment can, over time, modify the geographical and ecological ranges of the diverging lineages, novel opportunities for gene flow can emerge, thus reversing the divergence underway. The concept of subspecies is sometimes used to characterize metapopulation lineages that have not accumulated all distinctive properties of species. What is considered as distinctive properties for species and subspecies, however, may depend on the school of thought. In the “general metapopulation lineage concept of species” (de Queiroz, 2005), there is no space for subspecies since they correspond solely to segments of lineages unified by secondary criteria. Thus, the term “subspecies” should be considered only as a working hypothesis to facilitate communication among taxonomists. Species classification is particularly challenging when analysing extinct species or lineages based on the morphological analyses of fossils since mostly only anatomical or phenetical characteristics E.-M. Geigl, T. Grange / Annals of Anatomy 194 (2012) 88–102 can be used (tentative palaeoecological reconstructions of the species habitat could be used as well but are rarely used). The definition of new palaeontological species based on minor anatomical differences is a classical methodological trap because of the underestimation of the range of morphological diversity within a species population, or among sexes, or among the various growth stages of a single genus (e.g., Scanella and Horner, 2010). Proper characterization of distinctive features that could be used for classification of distinct species requires identification of a large number of complete, or almost complete, skeletons. This allows assessment of the range of morphological diversity within each putative species population and thus ascertainment that the distinctive features are accurately differentiating two non-overlapping groups. Depending on the situation, the distinctive features could result from either a small or a large number of allelic variations within the genome, and the lower the number of allelic variations involved, the lower the discriminative value of the criterion involved. For example, variations in the expression level of a single dental morphogenic regulator in the mouse, Ectodysplasin, which is a member of the tumor necrosis factor family of proteins, can induce multiple correlated morphological changes of the dentition with simultaneous variations in the number, positions and shapes of cusps, longitudinal crests form, and number of teeth (Kangas et al., 2004). Thus, phylogenetic history based on seemingly independent characteristics can be obscured when the characteristics are correlated to a small number, or even to a single, allelic variation affecting the expression of a developmental regulator (Kangas et al., 2004). Finally, deduction of classifications based on morphological similarities and differences with extant species can also be misleading due to changes in the range of variation of morphological features within a population lineage over time. This is particularly true for domesticated animals whose wild ancestor has disappeared. Indeed, the domesticates cannot be used to infer the variability of the morphology of the wild population because domestication is well known to cause extensive changes in the range of morphological features. Such changes can considerably complicate the task of palaeontologists who rely on comparisons with modern reference collections for their determination. The accuracy of the identification of extinct species would thus benefit considerably from the analyses of other classes of biological properties which are independent of morphology, just as the identification of modern species relies on the accumulation of independent lines of evidence. Palaeogenetic analyses of the genetic information contained in ancient bones of extinct lineages offer the opportunity to extend the range of arguments supporting the identification of extinct species. 2. Asiatic wild asses: the present situation The equid family is a vivid example of a complex lineage with multiple closely related extant and extinct species that are sometimes defined based on a limited number of fossils and that would benefit from clarification of the relationships between lineages using palaeogenetic analyses. The bushy phylogenetic tree of the Equidae spans 60 million years and comprises a multitude of controversially discussed species and subspecies defined by morphological criteria. Identification is a matter of considerable difficulty and debate since the genus Equus comprises many closely related forms, most of which are extinct or rare in the wild (Payne, 1991). E. caballus is extinct in the wild, Equus hydruntinus is completely extinct; E. hemionus survives in Iran and farther east, but the Syrian onager (E. hemionus hemippus) has been extinct since the beginning of the 20th century; African wild asses (Equus africanus) are almost or already extinct in the wild. Zebras are the only wild equids that are still widely distributed, even though only in a single continent: Africa. This situation explains the scarcity of reliable 89 reference collections to which archaeological bone finds could be compared. The Asiatic wild asses, or half-asses, stilt-legged equids of the desert, now survive in small patches of populations that are distributed over a wide area (see Fig. 1 and Table 1). The natural habitat of the Asiatic half-asses is the open semi-arid plains and alluvial valleys characteristic of much of the Middle East from the Mediterranean to the Indus (Meadow and Uerpmann, 1986), Tibet and Mongolia. There are populations that are in contact and hybridize at a low level, producing hybrids of reduced fertility or hybridize at one point of contact but not at another. They can be described as diverging metapopulation lineages that have not accumulated all lines of evidence to be considered as distinct species and are sometimes referred as subspecies, a vague enough concept that leaves plenty of room for controversy (George, 1869; Groves and Mazak, 1967; Groves, 1986; Forsten, 1990; Clark and Duncan, 1992; Denzau and Denzau, 1999; Eisenmann and Mashkour, 1999, 2000). Groves and Mazak ranked the kiang Equus kiang of the Tibetan Plateau as a full species separate from the onager E. hemionus (Groves and Mazak, 1967; Groves, 1986). Indeed, the two are allopatric (their ranges nowhere meet), their biology and morphology are very different, and the various geographic variants of kiang and onager are phenotypically much more different from each other than they are among themselves. Furthermore, Groves and Mazak (1967) and Groves (1986) divide each group in several subspecies, thus defining nine geographical taxa: three subspecies of E. kiang (E. k. kiang in Western Tibet, E. k. polyodon in Southern Tibet, and E. k. holdereri in Eastern Tibet), and six subspecies of E. hemionus: the extinct Hemippe of Syria, E. h. hemippus; the Onager of Persia, E. h. onager; the Kulan of Turkmenistan, E. h. kulan; the Mongolian Kulan, E. h. hemionus in northern Mongolia and adjacent Transbaikalia and Kazakhstan; and the Gobi Kulan or Dziggetai of southern Mongolia and adjacent Gansu in China, E. h. luteus; and finally the Khur of the Rann of Kutch in India, E. h. khur (see Fig. 1 and Table 1). The hemiones fall into three size groups: the two large Mongolian lineages (hemionus and luteus), the three small southerly ones (kulan, onager and khur), and the diminutive Syrian one (hemippus) (Groves and Mazak, 1967). The two Mongolian lineages are generally considered as a single subspecies and E. h. luteus is believed to be a localized sample of a unitary Mongolian lineage living in extreme desert conditions (Groves, 1986). The Mongolian Khulan is the largest remaining hemione population that is currently concentrated mostly in Southern Mongolia and neighbouring Northern China. The exact relationships between all these forms are still a matter of debate. The access to the animals is difficult, some forms have completely disappeared, and many are endangered and parked in reserves or zoos (Denzau and Denzau, 1999). According to Groves (1986), the six subspecies of E. hemionus are quite easy to distinguish by means of their coat colour. Other characteristics, however, do not follow this classification. Therefore, Groves concludes that the differences between the six subspecies are of a mosaic nature, not steps on a unidirectional cline (Groves, 1986). For this author, the easiest way to reconstruct their differentiation is to visualize a formerly continuous population sharing a common gene pool spread over at least the present range, but with differing frequencies of particular characteristics in different parts of the range. Groves concludes that the most parsimonious explanation for the generation of this situation is habitat fragmentation that would have led to well-differentiated isolates within each of which rapid homogenization would have occurred (Groves, 1986). Indeed, the range of the species is known to have been more nearly continuous in the past. It remains unclear, however, whether it was totally continuous and whether the population fragmentation was the consequence of human extermination of intervening populations (Groves, 1986). Analyses of ancient specimens from localities 90 E.-M. Geigl, T. Grange / Annals of Anatomy 194 (2012) 88–102 Fig. 1. Current natural distribution range of the Asiatic wild ass and geographical location of the archaeological sites that were sampled in the present study and the palaeogenetic results yielded: Upper Palaeolithic sites (blue dots), Neolithic sites (green dots); Chalcolithic sites (red dots); Bronze Age sites (orange dots); Iron Age sites (pink dots). Brown clouds indicate the areas where Asiatic wild asses naturally occur at present. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of the article.) where onagers no longer occur are needed to clarify the situation. Most half-ass populations are presently endangered, some are already extinct, like E. h. hemippus of Syria, and several of the few remnant hemione populations may soon become extinct (see Table 1). As pointed out by Meadow and Uerpmann (1986), the situation at the end of the Pleistocene and beginning of the Holocene was very different since several wild equid species roamed the Middle East in large numbers. Better characterization of this situation necessitates the palaeontological analysis of extinct individuals. 3. Palaeontology of the Equidae In order to evaluate the distribution of equids in the past, we have to rely on the fossil record. There is an abundant literature from many authors that is very well covered by the conference proceedings “Equids in the ancient world”, edited by Meadow and Uerpmann (1986). As stated by Eisenmann (1986), the discrimination of the osteological remains of the various equid species, varieties and hybrids is a prerequisite to the understanding of the geographical and chronological distributions. Table 2 summarizes the main morphological features used to discriminate between E. caballus, E. hemionus and E. hydruntinus. Discriminative characteristics are mostly found on limb bones and teeth. Unambiguous species identification seems to be extremely difficult since the different anatomies of modern equids are mosaics of various combinations of a relatively small number of characteristics (Eisenmann, 1986). All kinds of combinations, however, do not seem possible. Teeth of asses, for example, do not exhibit caballine double knots and half-asses do not have robust metapodials (Eisenmann, 1986). How often, however, are teeth and metapodials of the same animal found together at an archaeological site? Thus, even if the whole range of combinations of characteristics in modern species is known, the task remains a challenge when dealing with incomplete and generally fragmented, fossil material. Indeed, whole skeletons would be needed in order to determine their taxonomic status with confidence, but these are extremely rare. Most of the time, palaeontologists and archaeozoologists have to cope with individual and often fragmentary teeth and bones. Fragmentary bones, however, are difficult to identify (Uerpmann, 1986), and so is the determination of equids on single finds from archaeological sites (Von den Driesch, 2000). Groves and Willoughby (1981) conclude that the postcranial skeleton offers the best chances for discrimination especially if different bones are available (the proportions between different bones being better than the shape of any individual bone, the metapodials alone excepted), followed by the cranium, followed by the dentition, which trails badly as a discriminator. This view is not generally shared; others reckon that teeth and complete metapodials are the most reliable indicators (Davis, 1987). Form and length of the protocone are considered a valid distinguishing feature between the various equid species for some authors (e.g., Bökönyi, 1986) but not for others (e.g., Turnbull, 1986). In her thesis, Mashkour summarized the experience of Eisenmann stating that for complex and systematic questions, a global approach to the skeletal material is necessary, if possible entire skulls and, in their absence since they are rare, of metapodials and first phalanxes but that the discriminative characteristics are not necessarily on the same bones or teeth for each species (Mashkour, 2001). One major difficulty in discriminating species stems from poor knowledge of past intraspecies morphological diversity. Palaeontologists seek to analyse as many fossil samples as possible in order to understand the variability of the past populations but generally face the problem that archaeological collections often consist of Table 1 Classification and distribution range of the Asiatic wild ass. Common names Zoological name Synonymous zoological names supposed subspecies Current distribution range Former distribution range Status Palestine to Iraq, including Syria, Israel, Jordan, and Arabian Peninsula Extinct in the wild and in captivity (20th century) Iran (Touran and Bahram-e-Goor Reserves) Endemic to Iran Endangered Protected but isolated in disconnected natural reserves; hybrid population onager/kulan introduced in Israel NW India (Rann of Kutch); gradually moving out and colonizing Greater Rann of Kutch, extending into the neighbouring areas (State of Rajasthan and Jalore in Gujarat, India) Western India, Sindh and Baluchistan, Afghanistan, south-eastern Iran Endangered Protected in the Little Rann of Kutch; suggestions to reintroduce a few individuals into the Thar desert in Rajasthan Turkmenistan, Uzbekistan, Kazakhstan, Tajikistan, Kyrgyzstan Endangered Reintroduced to Uzbekistan and Kazakhstan; introduced to Ukraine; hybrid population onager/kulan introduced in Israel Hemippus Equus hemionus hemippus Onager Persian onager Khur Indian wild ass Equus hemionus khur Turkmenian Kulan Kulan Equus hemionus kulan E. onager kulan Turkmenistan (Badkhys Nature Park) Mongolian wild ass Mongolian Khulan North Mongolian Dziggetaia Equus hemionus hemionusa E. hemionus luteus?a Gobi desert of Southern Formerly possibly Mongolia and N China distributed over NE Mongolia and China. Unclear – possibly extinct if once existed Gobi Asiatic wild ass Gobi Khulan Gobi Dziggetai Equus hemionus luteus E. hemionus hemionus Gobi region of Southern Mongolia, rare in adjacent areas of China Most of Mongolia, small parts of Siberia and Manchuria, W Inner Mongolia and N Xinjiang (China) Unclear – population decline likely Kiang Tibetan wild ass Khyang, Gorkhar Equus kiang E. hemionus kiang 90% China: provinces of Qinghai, Gansu, Xinjiang, and Tibet; 10%: Nepal and Sikkem, India (Ladak), Pakistan and Bhutan Roughly the same as current range Lower risk, but data insufficient Persian wild ass Equus hemionus onager E. onager onager E. k. kiang E. k. holdereri E. k. polyodon E. k. chu E.-M. Geigl, T. Grange / Annals of Anatomy 194 (2012) 88–102 Syrian wild ass Conservation measures China: Kalameili Reserve Based on the status report 2007 of the Large Herbivore Foundation; http://www.largeherbivore.org/assets/pdf/status-report-2007-student-version.pdf) a Unclear whether N Mongolian subspecies ever existed. 91 According to Gromova (1955), Payne (1972b, 1975), Groves (1986), Davis (1980), Bonifay (1991), Burke et al. (2003), Eisenmann et al. (in press) and V. Eisenmann and S. Davis, personal communications. For pictures and schematic drawings of the teeth, see Fig. 3. Both premolars and molars “V” shaped As for E. hemionus 3 (smallest) E. asinus 2 (more gracile) Slenderer Absent Absent Larger, two lobes of more-or-less equal size Both premolars and molars “V” shaped Premolars are like E. hemionus but in the molars there is complete penetration of the external fold between the flexids to the extent that sometimes the external fold touches the internal fold. Short triangular-shaped (like a slipper) Absent Absent Slenderest 3 (smallest) E. hydruntinus 3 (most gracile) Both premolars and molars “V” shaped Shallow, no penetration between the flexids of the external fold in premolars and molars 2 (smaller) E. hemionus 2 (more gracile) Slenderer than horse Absent Absent Larger, two lobes of more-or-less equal size Both premolars and molars “U” shaped Less deep than E. hemionus but in the molars there is slight penetration of the external fold between the flexids and none at all in the premolars Largest, two lobes with the posterior lobe tending to extend some way back (posteriad) Present Present Robust 1 (robust) 1 (biggest) E. caballus Pli caballin Caballine double knots Relative robustness Relative body size Relative robustness index of skeleton Teeth Limb bones and metapodials Whole body Table 2 Summary of the most prominent morphological features on which the morphological discrimination between Eurasian equids is based. Protocone Linguaflexid – lower premolars and molars E.-M. Geigl, T. Grange / Annals of Anatomy 194 (2012) 88–102 Ectoflexid – lower premolars and molars 92 material that is fragmentary and not very abundant. This difficulty is particularly exacerbated for subfossil equid remains because the number of equid bones has generally been rather small in faunal assemblages. This small sample size limits the statistical reliability of many studies (Bökönyi, 1986; Gilbert, 1991). Furthermore, proper assignment of fossil bones is complicated by unsatisfactory modern museum reference collections that poorly sample the morphological variability of the different living wild Equus species. It is difficult to assess the degree of morphological and metrical variation to be expected within the various forms and to identify characteristics that might be restricted to particular populations given the scarcity of modern skeletons of wild true and half-asses and hybrids of any sort. Moreover, very often wild equid specimens in these collections are poorly documented, often come from zoos rather than from the wild, are often juvenile or senile individuals and postcranial skeletons are often completely missing (Payne, 1991). The fact that distinguishing characteristics must be taken from the closest available samples when local samples are not available may clearly introduce a bias (Payne, 1991). This situation has led to contradictory conclusions: Some authors consider true and half-asses no more closely related than are asses and zebras claiming the existence of six groups (Horse, Ass, Hemione, Mountain Zebra, Burchell Zebra, Grevy Zebra) that are separated for a long evolutionary time (Groves and Willoughby, 1981), others found true and half-asses very similar and distinguishable only by a combination of skull and metapodial characteristics (Eisenmann, 1979, 1980). Thus, the classification of equids based on osteological data poses serious problems, which are the consequence of uncertainties of overlapping and variable skeletal criteria (Gilbert, 1991). On the basis of phenotypical characteristics, six different hemione populations or subspecies have been recognized since the end of the Pleistocene, which occupy a geographical area reaching from China to Syria (Eisenmann, 1992b). This phenotypical distinction is reflected also in the post-cranial proportions of the skeleton when the metric measurements are transformed in logarithmic differences with a reference, a method called VSI (Variability Size Index) (Uerpmann, 1986; Meadow, 1999). Following this method, Mashkour (2001) compared the measurements (cumulative frequencies) of different hemiones populations, defined on the basis of their geographical occurrence and their phenotype as discussed before, and found clear size differences between the extinct E. h. hemippus, the Khur from India (E. h. khur), the Kulan of Turkmenistan (E. h. kulan), the Dziggetai of Mongolia (E. h. hemionus) and the Kiang from Tibet (E. h. kiang or E. kiang). These data show a gradient toward larger body size from the Southwest to the Northeast, conforming to Bergmann’s hypothesis linking environmental temperature to body size within a genus (Bergmann, 1847). The Persian Onager and the Turkmenian Kulan, distinguished by very slight morphological differences, occupied close territories in the past, although they are now confined to natural reserves in Touran National Park and Bahramgor Reserve in Iran and in Badkhyz Reserve in Turkmenistan (Schreiber et al., 2000). It is not fully clear whether the differentiation of these two populations has not been exacerbated by recent reduction of their habitats. The existence of half-asses west of the Zagros Mountains is subject to debate (e.g., Ducos, 1986; Uerpmann, 1986). The now extinct “Syrian Onager”, survived in captivity until the first third of the 20th century. Several specimens that supposedly belong to E. hemionus hemippus have been preserved in the Natural History Museums of Vienna, Paris, London, Harvard-Cambridge and Chicago. Their exact geographic provenience, however, is uncertain (Ducos, 1986). The two specimens from the Natural History Museum in Paris, one of which is mounted in the Gallery of Anatomy, were described by Isidore Geoffroy St-Hilaire in 1855: “These two wild horses (“chevaux sauvages”), both females and not yet fully adult, were sent by the Viceroy of Egypt under the name of onagers to Her E.-M. Geigl, T. Grange / Annals of Anatomy 194 (2012) 88–102 Majesty the Empress. Moreover, he wrote in a note about the origin of these animals: “The Viceroy of Egypt had received these animals from Seraskier Izzet-Pacha, governor of Syria, who in turn had taken them from the Arab chief Atha-Bey. They came originally, we are assured, from the desert of Syria between Palmyra and Baghdad. This species had not yet been seen either in Egypt or even in Damascus” (translation by Ducos, 1986). The lack of precision concerning the provenience of the animals is characteristic of all the specimens that were received as living animals and then preserved in the Natural History collections around the world after their death (Ducos, 1986). The fact that their physical appearance resembles one of the equids depicted in the reliefs of the palace of Assurbanipal at Nineveh demonstrates the existence west of the Zagros of a small wild equid with short, horse-like ears and a tail partly devoid of hair, these characteristics being present in the two stuffed and mounted specimens in Paris and Vienna (Ducos, 1986) as well as in the specimen described and photographed by Antonius (1929) in the Schönbrunn Zoo in Vienna. These features point to the Syrian onager as being a half-ass rather than a true ass (Bökönyi, 1986). In 1986, Uerpmann stated that the early Holocene specimens from Syrian sites such as Muraibit and Shams ed-Din could not be called “hemippus” since they were larger than the subrecent hemippus, size being the only criterion of separation of the hemippus from the rest of the hemiones in the case of bone remains (Uerpmann, 1986). Therefore, he attributed the equid bone finds from these archaeological sites to the hemiones and hypothesized that the so-called modern hemippus represent remnants that were dwarfed through isolation and deterioration of its habitat due to human settlement and agriculture (Uerpmann, 1986). An alternative hypothesis is that the modern hemippus were not wild but rather hybrid forms (Ducos, 1986), which would imply that they were not related to the ancient small equids found in Syria and depicted in historical times at Assurbanipal’s palace in Nineveh. Ducos (1986) suggested that these latter were related to wild African asses. This argument does not account satisfactorily for the physical resemblance between the depicted equids and the museum specimens from the 19th and 20th century. Furthermore, the difficulties of sorting out, from bone remains, the presence and distributions of at least three and possibly five forms of equids in the western part of the Middle East at the end of the Pleistocene increase in complexity when the remains of domestic equids begin to appear, at the end of the fourth millennium B.C. In fact, the situation becomes considerably more complicated because of the possibility of encountering the remains of horse-ass, horse-hemione, and ass-hemione hybrids. If characteristics useful for distinguishing horses, asses, and hemiones are hard to work out even for modern specimens of known taxonomic status, the situation when faced with hybrids is that much more troublesome. For example, curious mixed morphological characteristics of some teeth of the Bronze Age site of Godin Tepe, Western Iran, have led the authors to mention mixed zebrine/E. hydruntinus appearance (Gilbert, 1991). Did such mixed characteristic patterns really arise from interbreeding rather than intra- and inter-population variability? 4. The special case of E. hydruntinus A small equid, first described by Regàlia in 1907 (Regàlia, 1907) from the Late Pleistocene site of Grotta di Romanelli (Apulia, Italy) and named E. (asinus) hydruntinus (Stehlin and Graziosi, 1935), shows a mixture between characteristics similar to Equus stenonis, an ancestral European horse from the end of the Pliocene, and hemione characteristics (Eisenmann, 1992b). Its spatiotemporal distribution is still poorly known, as are its phylogenetic affinities (Eisenmann, 1986). The earliest appearance of this small 93 Fig. 2. (A) Rock engravings of equids in the cave of “Les Trois Frères” (MontesquieuAvantes, Ariège, France) attributed to E. hydruntinus. The left drawing is inspired from that represented in Nores Quesada and Liesau von Lettow-Vorbeck (1992). The right drawing was performed by Henri Breuil (1877–1961) and was reproduced in Clot and Duranthon (1990). (B) “Panneau de l’Hémione” in the cave of Lascaux, France. This Magdalenian fresco is situated on the bottom of the wall. On the left side, a depiction of a horse and on the right side, another equid with long ears, a straight back and a long tail suggest that the depicted animal was a hemionus. equid is attested in West Europe around 350,000 years ago at the site of Lunel-Viel (Bonifay, 1991), until the end of the Pleistocene in the Near East in the Levant (Davis, 1980; Clutton-Brock, 1986; Eisenmann, 1992b, 1995; Clutton-Brock, 1999) and in Anatolia until the Neolithic (Payne, 1972a; Uerpmann, 1987; Buitenhuis, 1997; Russel and Martin, 2000) although it does not appear to be present in a systematic manner (Uerpmann, 1986). It was not found in Syria during this period (Meadow, 1986; Vila, 1996), although it seemed to be present earlier (Reynaud Savioz and Morel, 2005). Its presence is also attested in Azerbaidjan (Eisenmann and Mashkour, 1999) and Iran (Mashkour, 2002; Orlando et al., 2006). In addition to the presence of bone material attributed to this animal in many Palaeo-, Meso- and Neolithic sites, its presence is attested to in Europe in Palaeolithic art. Indeed, in the Magdalenian cave of Putois (Haute-Garonne, France), the anterior part of the body of an equid with long ears and slender front legs is engraved with fine, precise lines on a bone fragment, presumably from a reindeer, that has been polished and smoothed (Cleyet-Merle and Madelaine, 1991). Does this engraving represent a hydruntine? This is at least the interpretation of the authors who also correlate the reappearance of this animal in the faunal assemblages of the end of the Middle Magdalenian in the Southwest of France where it had been absent for thousands of years (Cleyet-Merle and Madelaine, 1991). Other art objects possibly representing E. hydruntinus are known from “La Salpétrière” (Pont du Gard, Gard). Various rock art representations of E. hydruntinus are known from “Gabillou” (Sourzac, Dordogne, France), “Bernifal” (Meyrals, Dordogne, France), “Les Combarelles I” (Les Eyzies, Dordogne, France), “Les Cavernes du Volp” including “Les Trois Frères” (MontesquieuAvantes, Ariège, France, Albarracin (Teruel, Spain), and Levanzo (Egedi, Italy) (Cleyet-Merle and Madelaine, 1991). The most convincing depiction, in our eyes, is the one from “Les Trois Frères” (Fig. 2A) since it shows a slender skull, slender legs, a fine tail, long ears, and a moderate convexity of the back. Another interesting example is the Lascaux cave (Montignac, Dordogne, France), where a panel presumably depicts a horse on the left and another equid 94 E.-M. Geigl, T. Grange / Annals of Anatomy 194 (2012) 88–102 appearing to be a hemionus on the right (Fig. 2B). The long ears, the straight back and the tail are characteristic hemionus features. The disappearance of the hydruntine is estimated to be relatively recent, since it was found in numerous Neolithic sites in Eastern Europe (Spassov and Iliev, 2002; Haimovici and Balasescu, 2006), as well as in some Iron Age sites (Wilms, 1989). Based on medieval manuscripts, its disappearance in Portugal was suspected to be very recent (Nores Quesada and Liesau von Lettow-Vorbeck, 1992; Antunes, 2002). The taxonomic position is subject to debate. The species has been included in the subgenus asinus (Stehlin and Graziosi, 1935), hemionus (Azzaroli, 1992), or zebras (Davis, 1980; Bonifay, 1991), or declared the last representative of the Plio-Pleistocene group of the stenonian equids (Gromova, 1949; Forsten, 1986, 1990; Forsten and Ziegler, 1995; Forsten, 1999), or as belonging to a subgenus hydruntinus (Radulesco and Samson, 1965). The main characteristics of hydruntine are its small body and tooth size, its slender leg bones, and the patterns of its dental elaborations (Table 2). Measurement of the length and width of its metapodials allows us to appreciate its slenderness, whereas the limb segment proportions, in particular the relative length of the 3rd metacarpal and metatarsal reveal slightly different proportions relative to the similar E. hemionus (Eisenmann et al., in press). Its relative microdonty compared to various hemiones can be appreciated when relating the size of its upper premolars and molars to the length of its metapodial (Eisenmann et al., in press). The pattern of its dental elaboration is a particular distinctive feature (Fig. 3). Its upper and lower molars are more similar to zebra’s teeth and can be distinguished from horse, hemione and donkey teeth (Fig. 3A). For the upper cheek teeth, the characteristic features are the absence of the pli caballin and a short tri-angular shaped protocone, for the lower cheek teeth, a deep ectoflexid, in particular on the M1 and M2 (Fig. 3A and B) (Davis, 1980; Bonifay, 1991; Eisenmann et al., in press). These features make it similar to Equus stenonis, from which it differs by its slightly longer protocone and simpler enamel pattern (Gromova, 1955). The primitive nature of the distinguishing characteristics led some authors to consider it as a survivor of the Villafranchian E. stenonis, even though common retention of primitive characteristic states does not indicate relationship (Payne, 1972b, 1975). Some of the characteristics, such as the short length of the upper M3 and the particular evolution of the protoconic index, seem to be specific to E. hydruntinus and to give this group a taxonomic identity, in particular as they are combined with the more evolved skeletal characteristics, such as the cursorial proportions and the gracility (Boulbes, 2009). On the basis of the cranial proportions of two almost complete skulls from the Middle Palaeolithic kill and butchering site of Kabazi II (Ukraine), E. hydruntinus was classified as a separate species unrelated to any known Pliocene or Plio-Pleistocene monodactyl equid, to Stenonids, and to zebras, but close to the hemiones (Burke et al., 2003). The aforementioned characteristic dental features allow, however, distinction between hemiones and hydruntines (Fig. 3B). Discrepancies between the characteristic features of various bones and teeth all attributed to E. hydruntinus could be due, however, to the fact that they are not all truly conspecific (Groves, 1986). Gradient-like morphological variations have been observed in this lineage over time and over space. Indeed, its metapodials seem to get more slender over time, judging from Russian and Caucasian material (Groves, 1986). A geographical and chronological gradient for the occurrence of bones attributed to E. hydruntinus has been described, the first occurring from the West to the East with Eastern Europe showing the highest density of bone finds, the latter increasing their numbers in ever earlier archaeological sites reaching up to forty percent of the analysed assemblages, for example from sites dating to the end of the last glaciations in the Mediterranean regions of France (Bonifay, 1991). The closeness with E. hemionus of the dental features, also seen in Late Pleistocene specimens in Southern France (Boulbes, 2009), could be the result of convergent evolution of the two species. Boulbes points out, however, that the enamel structure of all equids shows a high degree of polymorphism, which makes it difficult to determine the degree of extension of the variability of these species and the identification of the limits between them (Boulbes, 2009). Moreover, period-specific size variations are observable in the Pleistocene, the populations of isotope stage 9 to 5 having smaller and those of stage 3 to 1 larger crowns (Boulbes, 2009). Stehlin and Graziosi (1935) argued that, since in the Upper Pleistocene E. hydruntinus was contemporary with hemiones in Asia, it cannot simply be explained away as a hemione with archaic dental characteristics. They speculate that it could have been adapted to a different substrate type rather than to climate. Was it adapted to a particular ecological niche that allowed it to exist in parallel to E. caballus and E. hemionus? E. hydruntinus was a member of the fossil communities that were associated with milder climatic conditions although it was also found in colder and dryer periods if they were not too intense (Prat, 1968; Delpech, 1984; Eisenmann, 1984; Azzaroli, 1990; Bonifay, 1991). Therefore, its geographic distribution varied according to the climatic conditions but it was particularly abundant in the Northern Mediterranean regions and Southeastern Europe. The ecological demands of this species seemed to have forced it either to adapt or to migrate. Population migrations in the second half of the Upper Pleistocene have been identified in well-dated sites in the Southwest of France (Delpech, 2003). Climatic constraints can, when they are tolerated, lead to skeletal adaptations such as size changes, alterations in proportions of the limbs and, in the particular case of equids, of the metapodials (Bignon and Eisenmann, 2002; Bignon, 2003). A certain plasticity of the enamel structure of the teeth has been observed at the classical palaeontological species and subspecies level as a function of the ecological conditions (Eisenmann, 1992a). An increase in the length of the protocone and of the size of the teeth could be an adaptation to a different food source (Gromova, 1949; Guadelli, 1987; Eisenmann, 1991). Thus, climate and nutrition could have favoured morphological changes that might at least partly account for the observed variations. These ecological variations, however, could have also led to population fragmentation and independent evolution. Thus, the populations West of the Alps could have evolved independently of those of the Adriatic region as a consequence of the withdrawal of mammal populations to refuge areas during the Ice Age maxima (Taberlet et al., 1998; Avise, 2000; Hewitt, 2001). Such geographic isolation could have led to genetic differentiation (Michaux et al., 2003; Schmitt, 2007; Provan and Benentt, 2008). In the case of Late Pleistocene horses (Bignon and Eisenmann, 2002; Bignon, 2003) and other taxa (Weinstock, 2000), a regionalization of the populations has been identified on the basis of morphological characteristics such as the size of the metapodials. The analysis of the protocones of E. hydruntinus finds in the Balkans and the South of France suggests a fractionation of the populations that was also explained by migration and differentiation (Boulbes, 2009). Indeed, the scarcity of remains of E. hydruntinus has been interpreted as an indication of small population sizes, which must have undergone differentiation to survive throughout the glacial periods. Moreover, interspecies competition with small-sized caballine (Uerpmann, 2005) or asinine equids (Azzaroli, 1979), E. hemionus (Eisenmann and Mashkour, 1999), and other herbivores (Janis et al., 1994; Grange and Duncan, 2006) could have shaped the ecological and morphological evolution of the hydruntine populations. The coexistence of two equid species is presently attested in Africa (Grevy Zebras and African Wild Ass) and also described for the MiddleUpper Pleistocene site of Binagady (Azerbaidjan) and the Holocene of Iran (Eisenmann and Mashkour, 1999), where these authors also E.-M. Geigl, T. Grange / Annals of Anatomy 194 (2012) 88–102 95 Fig. 3. Characteristic dental elaboration patterns allowing distinction between equid teeth. (A) Photographs and schematic drawings of the patterns of the enamel folds of molar teeth allowing distinction between the various equids. The drawings were kindly provided by Davis (1987) and the pictures by Eisenmann et al. (in press). (B) Comparison of left upper and lower cheek teeth of a representative hemione (Upper: E. h. khur, Lower: E. h. hemionus) and hydruntine (Agios Georgios, SGK 76 and 574, University of Thessaloniki, Laboratory of Geology and Paleontology). The pictures were kindly provided by Eisenmann et al. (in press). propose the co-existence of two “subspecies” of E. hemionus and of E. hydruntinus. The sympatry of two “subspecies”, if correctly assigned, can only be explained if the subspecies/species occupied different ecological niches within the same territory (Eisenmann and Mashkour, 1999). The authors add a cautionary note concerning their taxonomic attribution but consider the possibility that the morphological characteristics of E. hydruntinus, none of which is specific, developed several times in different places and within 96 E.-M. Geigl, T. Grange / Annals of Anatomy 194 (2012) 88–102 different phylogenetic lines. Indeed, the association of primitive characteristics, such as the small size, primitive patterns of upper and lower cheek teeth with evolved characteristics of the skeleton (such as gracility, cursorial proportion as an adaption to dry environment) and teeth (hypsodonty) is also found in the case of the South African E. lylei that differs by having more robust limb bones (Eisenmann and Mashkour, 1999). In contrast, for other authors, E. hemionus and E. hydruntinus clearly occupy different territories and ecological niches in neighbouring areas; for example, in the late Pleistocene in Israel, E. hemionus was restricted to the desertic south and E. hydruntinus was present in the wetter north (Davis, 1980, 1987). It should be noted that the conclusions of each author rely on the taxonomic attributions of the bone finds, which depend on the weights given to various morphological criteria that differ among palaeontologists and archaeozoologists. In conclusion, there seems to be general agreement between many palaeontologists/archaeozoologists that there was a distinctive population of small equids in Europe that could correspond to a species that they all agree to call E. hydruntinus, which would have evolved over time and space, raising the possibility that different equid populations could be associated to the same species by different palaeontologists. Furthermore, there is no general agreement on its phylogenetic relationship to other equids. Genetic analyses should allow us to clarify this situation. 5. Population structure of equids based on modern genetic data The analysis of the hypervariable region of mitochondrial DNA, which is the part of this maternally inherited molecule that mutates at the highest rate, is commonly used to analyse intra- and interspecific relationships. This approach allows a confrontation between the taxonomic classification established from comparative anatomy and a genetic classification. In order to infer properly taxonomic classification from genetic data, it might be necessary to sample a sufficiently large number of loci and individuals and to use probability modelling (Knowles and Carstens, 2007). Furthermore, mitochondrial DNA being exclusively maternally derived can give an erroneous picture when there are sex-specific migration behaviours (e.g., Portnoy et al., 2010). Mitochondrial DNA information is, however, the most readily accessible and can nevertheless provide useful information that enlightens a complex situation, even if it is reasonable to consider it as a starting point that will require more exhaustive analyses to allow genetic data to fully enrich the picture. The Median Joining Network approach is a useful tool to visualize the genetic relationships between DNA sequences (Bandelt et al., 1999). When applied to the mitochondrial hypervariable region from equids, one can easily distinguish 6 clades (clearly separated from each other at roughly equal distances that correspond to the 6 unambiguous equid species, horse, ass, mountain zebra, plain zebra, Grevy’s zebra and half-asses (Fig. 4). The hemione group is separated from all other equids by a 28 bp deletion within the hypervariable region, which certainly resulted from a single evolutionary event, thus precluding homoplasy. This deletion can be used as a “barcode” for the unambiguous genetic identification of the half-asses. In contrast to the horses and the donkeys, there are very few mitochondrial sequences of zebras and half-asses available in Genbank (Benson et al., 2011). Within the half-asses, a phylogeographic pattern could not be deduced, only the relative separation of the Kiangs became apparent (Fig. 4). A similar situation was found when constructing a maximum likelihood tree of the combined mitochondrial hypervariable region and 12S rRNA gene sequences and using the rhinoceros C. simum as an outgroup (Oakenfull et al., 2000). Here as well, the mitochondrial sequences separated two kiang individuals from kulans and onagers. The mitochondrial sequences of two kulans from Turkmenistan clustered with one onager from Iran, while that of two other kulans from Turkmenistan clustered in a different group with those of three other onagers from Iran (Oakenfull et al., 2000). The authors conclude that in these species the speciation events are difficult to resolve, supposedly due to one or more rapid radiation events or to extensive secondary contacts of the metapopulation lineages. The fact that kulan and onager populations share two haplotypes pleads in favour of the secondary contact hypothesis since it suggests that there was recent gene flow between them. Gene flow could have occurred in the ancestral population that only recently split into two populations from which kulans and onagers are derived. Alternatively, kulans and onagers could have separated earlier and mixed recently before separating into their current populations (Oakenfull et al., 2000). A study by Ryder and Chemnick suggested in 1990 that the chromosome numbers of some kulans and onagers deviated from the average chromosome number of 55, which is the same for onagers and kulans (Ryder and Chemnick, 1990). These deviating karyotypes could be the consequence of stable Robertsonian rearrangements, i.e., partial and complete fusion/fission of chromosome pairs 23 and 24, as shown by G- and C-banding studies (Ryder, 1978), indicating an initial difference in chromosome number in the two subspecies followed by limited mixing afterwards (Oakenfull et al., 2000). Alternatively, this situation was explained more parsimoniously with a founder effect, i.e., the separation from the original population of a small population in which the proportion of the chromosome ratios was not the same as in the original population but easily underwent a skew at variance to that of the population from which they were derived, an explanation that is in agreement with the scenario proposed for mtDNA haplotypes (Oakenfull et al., 2000). In this study, DNA from captive animals was analysed. Thus, these founder effects in both mtDNA haplotype and chromosome frequencies could have occurred as the captive populations of onagers and kulans were formed. A more recent study focusing on 31 microsatellite loci and analysing two other captive kulan and onager populations (7 and 11 individuals, respectively) could show some separation between them although supported by low bootstrap values (Kruger et al., 2005). Some individuals in each population were, however, not clearly differentiated from the other population. Because different genetic markers and different captive populations were studied, it remains difficult to ascertain the exact extent of genetic differentiation between these two metapopulation lineages. Should onagers and kulans therefore be classified as evolutionarily significant populations even though their differences could be due to recent human activities causing the isolation of the populations? The skull morphology between these subspecies was shown to have a large overlap in their variation (Eisenmann and Shah, 1996). Thus, the differentiation of morphological characteristics has not been successful in interpreting the evolution of these subspecies. To shed light on this question, it is necessary to obtain data that elucidate not only the dispersal of the populations in extent but also in time. This prompted us to perform a palaeogenetic analysis of individuals belonging to ancient, now extinct populations. 6. Discussion of our palaeogenetic analysis of ancient hemione and hydruntine populations In an attempt to characterize the population structure of half-asses in the past, we sampled roughly 200 ancient E. hemionus and E. hydruntinus specimens spanning the area from Western France to the Caucasus and roughly 100,000 years. We subjected them to palaeogenetic analyses of the hypervariable region of E.-M. Geigl, T. Grange / Annals of Anatomy 194 (2012) 88–102 97 Fig. 4. Median Joining network (Bandelt et al., 1999) depicting the relationships between extant equids based on 408 bp of the mitochondrial hypervariable region. mitochondrial DNA. Here we will describe and discuss the results but the original data will be published elsewhere (Champlot et al., in preparation). The bone and teeth samples came from wellknown archaeological contexts in Spain, France, Belgium, Italy, Slovakia, Serbia, Romania, Turkey, Georgia, Dagestan, Armenia, Syria and Iran, and had been analysed by archaeozoologists and palaeontologists prior to their palaeogenetic analysis. One Iberian sample was roughly 7000 years old and one came from an Iberian Bronze Age context. One sample from France came from a burial context that belongs to isotope stage 5 (roughly 100,000 years ago), one from the Middle Pleistocene and one from a Solutrean context. The Belgian samples belonged to the Middle Pleistocene and one of them was dated to 34,600 years BP. The Slovakian sample was 14 C dated to 3000 BP. The Serbian samples were archaeologically dated to c. 5–4000 BC. The Romanian samples come from a Chalcolithic context dated to c. 4700 BC. The Turkish samples came from different archaeological contexts from the PPNA to the early PPNB (c. 9300–8200 BC), the Late Neolithic (c. 6200–5500 BC), to the Bronze Age (c. 4200–3000 BP). The samples from the Caucasus were archaeologically dated to the 5th and 4th millennia. Finally, the samples from Iran were dated to the period between 6000 and 1000 BC, and those from Syria to the late Bronze Age. The identification of bones as belonging to E. hydruntinus turned out to be a difficult issue that often did not lead to a consensus between eminent archaeozoologists. For all bone and teeth samples that we obtained originally as E. hydruntinus samples and for which we had obtained the entire skeletal fragment, i.e., for 23 samples, the palaeontological determination was performed by several palaeontologists/archaeozoologists. Surprisingly, in almost all cases, there was a divergence in their classification. In the most spectacular cases, five different palaeontologists/archaeozoologists each came up with a different determination of the same sample. Some of them remained cautious and characterized the species only as “little equids” or did not classify them. Often they included the supposed period and geographical origin in their determination, implying that they would have come up with a different taxonomic status if the sample came from a different archaeological context. This shows the difficulty of species classification of the equids based only on morphology and reflects the difficulties discussed above. We established a reliability factor reaching from 0.125 to 1, which expresses the proportion of palaeontological determination in agreement with the determination as E. hydruntinus. Three E. hydruntinus samples had a reliability factor of one (all palaeontologists agreed), two of 0.75 (three out of four palaeontological determinations yielded E. hydruntinus), one of 0.66 (two hydruntinus determinations, one against hydruntinus and one excluding the horse), three of 0.5 and five of 0.33, both with various combinations of determinations, eight of 0.25 (one out of four determinations) and one 0.125. For another 15 samples, we could not establish the reliability factor for the morphological determination since we only obtained a small, uncharacteristic portion of the samples. Of these, 13 samples had been determined by only one archaeozoologist. The bone and tooth specimens were then subject to palaeogenetic analysis in a high-containment laboratory dedicated to ancient DNA research under rigorous experimental conditions as described previously (Charruau et al., 2011). The mitochondrial hypervariable region was analysed using UQPCR that eliminates 99.99% of potential carry-over PCR products (Pruvost et al., 2005). 98 E.-M. Geigl, T. Grange / Annals of Anatomy 194 (2012) 88–102 The initial quantity of target DNA molecules was determined via an internal, homologous reference dilution series as described (Pruvost et al., 2007). The PCR products of overlapping DNA fragments were sequenced, aligned and the 73 obtained and authenticated sequences of various lengths as well as one sequence from an E. hemippus museum specimen from the 19th century, one modern E. h. onager and one modern E. h. luteus sequence, three modern kulan and six modern kiang sequences were analysed by median joining network analysis (Bandelt et al., 1999). Two types of sequences were obtained from the bones assigned as “E. hydruntinus” with various reliability indices, a caballinetype and a hemionus-type. Among the European samples, a 100,000-year-old bone from France that had been determined unambiguously as E. hydruntinus by several archaeozoologists and the four Romanian samples all yielded hemionus-type sequences including the hemionus-characteristic 28 bp-deletion. All the other 14 European samples, however, even those determined by at least one eminent archaeozoologist as being E. hydruntinus, yielded caballine sequences. In contrast, most samples (39) from Southwest Asia, determined as being either E. hydruntinus with various reliability indexes, or as being E. hemionus or “small equids”, yielded hemionus-type sequences. There were, however, exceptions: (i) two samples from the Caucasus and one sample from Turkey that had been determined by eminent archaeozoologists as E. hydruntinus or at least as ass-like types and definitely not caballine-types yielded typical caballine sequences; (ii) for two other samples, one from Iran and one from Turkey, the archaeozoological determination was contradictory, but the sequences that were obtained were clearly of caballine-type. The 28 ancient caballine sequences all cluster with modern caballine sequences, i.e., they are unambiguously contained within the mitochondrial diversity of modern horses. Seven of those came from specimens that were palaeontologically determined as hydruntinus-like and definitely not caballine-like. The two sequences obtained from the Caucasian E. hydruntinus samples represent caballine haplotypes that are now extinct (or were never sampled) and so do some of the samples that had been determined palaeontologically as E. caballus. Since all these sequences were obtained more than once from PCR products produced under rigorous experimental conditions in a high-containment laboratory dedicated to ancient DNA research and in addition using UQPCR that eliminates carry over contamination prior to PCR, we consider it highly unlikely that they are due to laboratory contamination. Moreover, the substitutions are not due to chemical transformations during bone diagenesis either and the sequences can therefore be considered authentic. For one of the Serbian and one of the Caucasian mitochondrially caballine samples, we also obtained a characteristic caballine Y chromosome-specific sequence so that we could exclude the possibility that the corresponding animal was a hybrid. The network of the hemione-type sequences clearly shows a strong population structure of seven well-separated haplogroups or clades (data not shown, Champlot et al., in preparation). This population structure became apparent solely through the ancient sequences, in our study they constitute the majority of available hemionus sequences to date. Only two haplogroups correspond to purely modern populations that are geographically distinct and not covered by our study due to a lack of ancient samples: the kiangs in Tibet and the Mongolian kulans or dziggetais. Moreover, the ancient sequences form three more haplogroups that are void of any modern sequence and characterize populations that are extinct today: there is one haplogroup containing sequences from ancient samples from the Caucasus and Iran, one cluster containing sequences from ancient samples from Turkey and Romania as well as the Pleistocene sample from France, and one cluster containing ancient sequences from E. hemionus samples from Syria and the 19th cen- tury hemippus. Finally, the three modern kulans cluster in two different haplogroups, each one together with ancient sequences from Iran that play a key role in defining the clusters without ambiguity. One of the three modern kulan sequences from Turkmenistan cluster in one haplogroup together with the modern Iranian onager and the other two modern Turkmen kulans belong to the other haplogroup. The 19th century hemippus from the Natural History Museum in Paris has the same haplotype as the four prehistoric samples from Syria. They are well separated from all other haplogroups but distantly linked to one haplogroup constituted by modern Iranian and Caucasian samples. The ancient Syrian E. hemionus were of larger size than the hemippe, and thus the very close genetic relationships between these two groups indicate there was a size reduction in the Syrian hemiones that preceded their extinction in the 20th century, confirming the hypothesis of Uerpmann (1986) that the modern hemippes represent remnants that have been dwarved through isolation and deterioration of their habitat due to human settlement and agriculture. One of the haplogroup clusters all samples from Turkey and Romania for which the archaeozoological determination as E. hydruntinus had the highest support in the ranking list of reliability (three bone samples were determined as E. hydruntinus with a reliability factor of 1, two of 0.75, one of 0.66, one of 0.33, one of 0.5 and four of 0.25). This group also contains the French specimen from isotope stage 5, a perfectly preserved complete 3rd metacarpal that was unanimously determined as E. hydruntinus. One can therefore conclude that this clade shows the best correspondence to what is thought to have been E. hydruntinus. This hemionus-type clade seems to have been present during the Middle Pleistocene in Western Europe and survived in Southeastern Balkans and in Turkey until at least the Late Neolithic. 7. Discussion 7.1. Ambiguities, biases and uncertainties in palaeontological determination of equid bones The phylogenetic tree of the Equidae based on morphological criteria established on a rich fossil record comprises a multitude of species and subspecies that are controversially discussed. Indeed, as discussed above, identification is a matter of considerable difficulty and debate (Payne, 1991). Recent advances in ancient DNA technology have made the reanalysis of a number of ancient Equus samples possible, which has led to proposals for taxonomic revisions at the generic, subgeneric, and species levels (Orlando et al., 2003, 2006, 2009; Weinstock et al., 2005). In particular, while South American hippidions were considered to be descendants of the Pliohippines based on their distinct nasal morphology, the mitochondrial DNA of several hippidion specimens from Argentina formed a tight haplogroup within a larger paraphyletic group of Equus, suggesting either that hippidions and living Equus belong to the same genus or that living equids should be split into several genera (Orlando et al., 2003, 2006, 2009; Weinstock et al., 2005). Similarly, the South American subgenus Amerhippus that was considered on morphological grounds to be rather Equus-like (Azzaroli, 1998), had a mitochondrial haplotype that clustered with the caballine horses (Orlando et al., 2008). Moreover, the group of the stilt-legged equids of North America, that were morphologically similar to the Asiatic wild asses, were shown through mitochondrial aDNA analysis to be endemic to the New World and to harbour fewer species than deduced from morphological criteria (Weinstock et al., 2005). Finally, it was proposed recently on the basis of ancient mitochondrial DNA data to synonymize Cape zebras, quaggas and plain zebras, on the one hand, and E. hydruntinus and E. hemionus, on E.-M. Geigl, T. Grange / Annals of Anatomy 194 (2012) 88–102 the other hand (Orlando et al., 2009). There are thus precedents for inconsistency between palaeontological determination of species based on bone morphology and palaeogenetic data. Although we could clearly establish that the samples that were determined by most archaeozoologists to be E. hydruntinus all carry a characteristic hemione-type sequence, in agreement with the observation of Orlando and colleagues (Orlando et al., 2006, 2009), we observed a large proportion of samples morphologically determined by some archeozoologists as being E. hydruntinus that turned out to carry a typical caballine mitochondrial sequence, and in two cases even a typical caballine Y chromosomal sequence ruling out the possibility of hybrids. This polyphyly is clearly incompatible with the determination of all these bones as a unique species. Most of the caballine-type samples originated from Europe showing that for this geographical area, there is a tendency to attribute horse bones to E. hydruntinus. The difficulties in determining ancient bone remains at the species level are best illustrated by the experimental strategy that we adopted in order to obtain a reliability index for the determination: we asked several palaeontologists/archaeozoologists to determine the same bones. For most of the bones, this procedure led to their attribution to more than one species and for several bones a consensus could not be reached, i.e., each palaeontologist/archaeozoologist attributed a given bone to another species. This suggests that several palaeontologists/archeozoologists, including very experienced and renowned ones, tend to underestimate the morphological variability of caballine equids in this period and area and to overuse the ill-defined E. hydruntinus in Europe to determine bones that seem to be undeterminable. Furthermore, we could also observe that some of the misattributions could be linked to erroneous dating of the bones: for two bones, even though the corresponding stratigraphic layer was archaeologically dated to the Neolithic, 14 C dating showed that the bones were from Late Bronze Age/Early Iron Age, thus biasing the determination that was based on expectations about the size range of wild horses during the Neolithic. Orlando and colleagues previously performed a mitochondrial DNA analysis of four ancient bone samples that had been determined palaeontologically to belong to E. hydruntinus and that were found to carry the 28 bp deletion that is characteristic of the Asiatic wild asses (kiangs, onagers, kulans) (Orlando et al., 2006, 2009). Thus, despite the fact that the average relative dimensions of the metapodials and upper-cheek teeth parameters of E. hydruntinus differed in the specimens that they analysed, the authors conclude, on the basis of the mitochondrial DNA data obtained, that E. hydruntinus was a subspecies of E. hemionus (Orlando et al., 2009). Three other presumed E. hydruntinus specimens from Southwest Siberia, however, formed a new monophyletic group lacking the hemione-specific 28 bp deletion and any specific affinity with other non-caballine horses (Orlando et al., 2009). Therefore, these authors suggest that these latter constitute an extinct group with no extant relatives, a proposal that has been supported by a revisited morphological analysis of these bones (Eisenmann, 2010). The morphology of the lower cheek teeth as well as the size of the third metapodials of these specimens from Proskuryakova and Akhalkalaki were compared to those of E. altidens from Süssenborn, and E. hydruntinus from Lunel-Viel, Staroselie and Dorog. This analysis showed only poor affinities between the Siberian specimens and hemiones and hydruntines but rather strong affinities with the archaic Middle Pleistocene Süssemiones from Süssenborn, Germany (Eisenmann, 2006; Orlando et al., 2009; Eisenmann, 2010). Based on the age and the mitochondrial sequences of the analysed samples, the age of the emergence of this new species was estimated to have occurred between 0.4 and 2.3 MYA (Orlando et al., 2009). Süssemiones supposedly became extinct well before the beginning of the Late Pleistocene (Eisenmann, 2006), but the 14 C dates of the specimens 99 analysed by Orlando et al. (2009), indicate that the Siberian species survived at least until c. 45–50,000. Our identification of the most reliable E. hydruntinus bones to the hemione-type mitochondrial groups is in agreement with the proposal that the hydruntine population is closely linked to E. hemionus, and thus would favour their interpretation that the other mitochondrial lineage identified can clearly be assigned to another lineage, presumably süssemiones as proposed (Orlando et al., 2009; Eisenmann, 2010). As discussed before, the taxonomic position of E. hydruntinus based on morphological criteria was subject to debate and for some authors it was believed to be the last representative of the Plio-Pleistocene group of the stenionans (Gromova, 1949; Forsten, 1986, 1990; Forsten and Ziegler, 1995; Forsten, 1999). Presumably, this attribution was based on bones that belonged to süssemiones rather than to hydruntines. Thus, our results as well as the results of Orlando et al. indicate that the erroneous attribution of bones from another lineage to E. hydruntinus affects at least two equid lineages, süssemiones and caballines, and these erroneous attributions may explain the past discrepancies of the phylogenetic relationships between E. hydruntinus and other equids. Here, the equids may not only suffer taxonomic oversplitting as proposed (Orlando et al., 2009), but also taxonomic undersplitting, several lineages being fused into a single species. The discrepancies between the phylogenetic trees of equids based either on mitochondrial DNA data or morphological characteristics require reconsidering the strength of each line of evidence. We suggest that the morphological variability of ancient equids is underestimated. In particular, temporal and regional body-size variations among Late Pleistocene and Holocene equids might have been more pronounced than generally believed, which could introduce biases in morphological taxonomy. One should keep in mind that morphological characteristics are not as independent as they may seem at first glance. They may rather change when the expression level of a small number of regulatory factors changes due to minor genetic or even epigenetic modifications (e.g., Kangas et al., 2004). Morphological variability within a species could be more widespread than currently assumed. Thus, the use of a few “characteristic” morphological traits to reconstruct phylogeny can be misleading and sometimes provide inconsistencies with phylogenies based on a few genetic markers not related to the morphological traits. In contrast, morphological traits that are based on a large number of genetic variants distributed throughout the genome, as seen for example in the case of size differences in chicken (Johansson et al., 2010), could provide more reliable phylogenies. It is probably presently still premature to rely on morphological traits the genetic basis of which is unknown. Several equid populations might have lived and interacted in the same area at the same time, possibly thriving in different ecological niches, such as different species of zebras do in Africa today. Indeed, natural selection can lead to parapatric occupation of different ecological niches through morphologically distinct populations between which gene flow occurs (Smith et al., 1997; Schneider et al., 1999). If this phenomenon lasts for periods long enough to be detectable in the fossil record but not long enough for complete divergence of the metapopulation lineages, then a situation could arise such as that concerning E. hydruntinus revealed in the course of our study. 7.2. E. hydruntinus as a separate species or an E. hemionus subspecies? Orlando et al., emphasize that extant kiangs and onagers-kulans are classified as separate species based on differences in coat colour, morphology, geographic distribution, and the number of chromosomes although they show only poor mitochondrial differentiation. Similarly, they allow for the possibility that E. hydruntinus is a 100 E.-M. Geigl, T. Grange / Annals of Anatomy 194 (2012) 88–102 separate species, which might have experienced multiple introgression events from different onager or kulan stocks (Orlando et al., 2009). Our identification of the most reliable E. hydruntinus bones to the hemione-type mitochondrial groups is in agreement with the proposal of these authors that the hydruntine population is closely linked to E. hemionus. However, our mitochondrial DNA analysis involving a larger part of the hypervariable region and a large sample size of ancient E. hemionus and E. hydruntinus bones has allowed us to identify substructures within the hemionus sequences that substantially change the evidence supporting the taxonomic status of E. hydruntinus. Indeed, we could identify seven distinct haplogroups within Asiatic wild asses. This in turn allowed us to identify one haplogroup for which independent palaeontological determinations as E. hydruntinus were established with the highest consensus among all the bone remains that we studied. This hydruntine-like clade was present during the first half of the Holocene in the Southwest of the Balkans and in Turkey. The haplotype of the 100,000-year-old specimen from Southwest France clusters with the hydruntine-like haplogroup, but it is not found at a position that connects this haplogroup to the other haplogroups, but rather at the other end. Thus, this sequence, although belonging to a bone that is more ancient than those from the Balkans and Turkey, derives from the Balkan hydruntine population. This suggests that the West European hydruntine corresponded to a subpopulation that evolved from an ancestral population originating in Eastern Europe and Turkey. The hydruntine haplogroup is as distant from the other haplogroups as are the kiangs from the onagers. Thus, if kiangs are considered a distinct species, on the basis of several lines of evidence (phenotypical and genotypical), then the level of genetic differentiation between E. hemionus and E. hydruntinus would be sufficient to consider them as different species as well. A counter argument, however, is that such a genetic distance can also be found separating the two haplogroups that exist within the extant Turkmen kulan and the Iranian onager populations. These two haplogroups comprise ancient Iranian hemiones found near the Caspian Sea from the 6th to the 2nd millennium BC the bone morphology of which does not allow for classification between kulans and onagers. A Bayesian analysis of our mitochondrial DNA sequence data showed that the two groups separated a long time ago (estimated 59 kyrs), at about the same time as they separated from kiangs (Champlot et al., in preparation). The fact that these two haplogroups can be found in the extant Turkmen kulan population, as well as in the Iranian onager population as suggested previously (Oakenfull et al., 2000), indicate that there was gene flow between these two populations at some point after their separation. This could have occurred recently, possibly as a consequence of human-driven habitat disruption and population rearrangements, or could have occurred earlier, with a mixed hemione population that separated only recently in the current onager and kulan population, an hypothesis that was favoured by Oakenfull et al. (2000) because it was the most parsimonious interpretation. Our observation that the two haplogroups had already overlapped in the same geographical region several thousand years ago argues in favour of the recent differentiation of a unique population comprising several distinct haplogroups. Thus, this level of genetic differentiation can accompany a speciation event, as seen for the Kiangs, but does not convincingly demonstrate speciation. Finally, a specimen from Portugal dated to the 17th century was analysed by Orlando et al. (2009) that was reported to be the last specimen of the so-called zebro, a striped equid described in the Medieval Iberian literature and supposed to be at the origin of the name of the African equids, the zebras (Nores Quesada and Liesau von Lettow-Vorbeck, 1992; Antunes, 2002). Cranial morphology and mitochondrial DNA sequence, however, indicate that this animal was a donkey (Orlando et al., 2009). This suggests that E. hydruntinus disappeared earlier than assumed based on the determination of ancient bones, and Orlando and colleagues proposed pushing back the upper limits of hydruntine survival from the Middle Age to the Iron Age. Our study reveals that all the Western European bones that were assigned to E. hydruntinus dating from 35,000 to 6000 BP were erroneous attributions of horse bones. This suggests that the disappearance of E. hydruntinus may have occurred much earlier than believed, or at least, that there was a severe reduction in the size of this population long before the Iron Age. 8. Conclusion Palaeogenetic studies like the one described here can highlight and help to correct systematic biases of the morphological characterization of ancient bones and be useful guiding tools to better define ancient species and their phylogenetic relationships. One should, however, keep in mind, that the present study, as well as those that have preceded it (Orlando et al., 2003, 2006, 2008; Weinstock et al., 2005), have analysed only the maternally inherited, non-recombining mitochondrial DNA that does not take into account certain etiological factors such as territorial and mating behaviour, which is known to be different in horses and wild asses (Burke, 2002). Mitochondrial DNA is a characteristic independent of anatomical traits the analysis of which delivers a certain type of information, different from that which can be obtained from the analysis of the dentition and bone morphology, and can usefully complement the anatomist’s tools. Both the palaeontological and the palaeogenetic and genetic approach analyse secondary species criteria that should be taken as lines of evidence helping to define the species boundary between separately evolving metapopulation lineages. The complexity of speciation events and the scarcity of data from ancient populations mean that caution is necessary before assigning metapopulation lineages as species, whatever the methods used to collect the lines of evidence, whether they are based on genetics or anatomy. Acknowledgments We thank Sophie Champlot and E. Andrew Bennett for the production of the palaeogenetic data and Mathieu Gautier for Bayesian analysis; Véra Eisenmann, Hans-Peter Uerpmann, and Simon Davis for help with palaeontological determinations; Véra Eisenmann, Hans-Peter Uerpmann, Simon Davis, Mietje Germonpré, Marjan Mashkour, Arturo Morales Muniz and Joris Peters for helpful discussions; Benjamin Arbuckle, Adrian Balasescu, Marie-Françoise Bonifay, Jean-Philippe Brugal, Jean-Jacques Cleyet-Merle, Lydia Gamberi, Mietje Germonpré, Christophe Griggo, Stéphane Hinguant, Stéphane Madelaine, Gabriella Mangano, Marjan Mashkour, Arturo Morales Muniz, Pierre-Elie Moullé, Marylène Patou-Mathis, Joris Peters, Maryline Rillardon, Jean-François Tournepiche, and Hans-Peter Uerpmann as well as the Muséum National d’Histoire Naturelle of Paris for providing samples; Véra Eisenmann for providing the photos and validating Fig. 3, and help with Table 2; Simon Davis for critical reading of the manuscript, valuable suggestions, correction of Table 2, providing the drawings of Fig. 3 and correction of the English language; an anonymous reviewer for valuable suggestions; E. Andrew Bennett for critical reading of the final manuscript; The palaeogenetic analyses were funded by the French National Research Centre CNRS. E.-M. Geigl, T. Grange / Annals of Anatomy 194 (2012) 88–102 References Antonius, O., 1929. Beobachtungen an Einhufern in Schönbrunn 1: Der Syrische Halbesel (Equus hemionus hemippus J.Geoffr.). Der Zoologische Garten (NF) 1, 19–25. Antunes, M.T., 2002. The Zebro (Equidae) and its extinction in Portugal, with an Appendix on the noun zebro and the modern “zebra”. In: Mashkour, M. (Ed.), Equids in Time and Space (Proceedings of the 9th ICAZ Conference). Oxbow Books, Durham, pp. 210–235. Avise, J.C., 2000. Phylogeography. Harvard University Press, Cambridge, MA. Azzaroli, A., 1979. On a Late Pleistocene Ass from Tuscany with notes on the history of Ass. Palaeontol. Ital. 71, 27–47. Azzaroli, A., 1990. The genus Equus in Europe. In: Lindsay, E.H. (Ed.), European Neogene Mammal Chronology. Plenum Press, New York, pp. 339–356. Azzaroli, A., 1992. Ascent and decline of monodactyl equids: a case for prehistoric overkill. Ann. Zool. Fennici 28, 151–163. Azzaroli, A., 1998. The genus Equus in North America – the Pleistocene species. Palaeont. Ital. 85, 1–60. Bandelt, H.J., Forster, P., Röhl, A., 1999. Median-joining networks for inferring intraspecific phylogenies. Mol. Biol. Evol. 16, 37–48. Benson, D.A., Karsch-Mizrachi, I., Lipman, D.J., Ostell, J., Sayers, E.W., 2011. GenBank. Nucleic Acids Res. 39, D32–D37. Bergmann, C., 1847. Über die Verhältnisse der Wärmeökonomie der Thiere zu ihrer Grösse. Göttinger Studien 3, 570–595. Bignon, O., Eisenmann, V., 2002. Western European Late Glacial horses diversity and its ecological implications. In: Mashkour, M. (Ed.), Equids in Time and Space (Proceedings of the 9th ICAZ Conference). Oxbow Books, Durham, pp. 161–171. Bignon, O., 2003. Diversité et exploitation des équidés au Tardiglaciaire en Europe occidentale. Implications pour les stratégies de subsistance et les modes de vie au Magdalénien et à l’Azilien ancien du Bassin de Paris. Université Paris X – Nanterre, Paris-Nanterre, p. 856. Bökönyi, S., 1986. The Equids of Umm Dabaghiyah, Iraq. In: Meadow, R.H., Uerpmann, H.P. (Eds.), Equids in the Ancient World (Beihefte zum Tübinger Atlas des Vorderen Orients, Reihe A, Nr 19/1). Dr. Ludwig Reichert Verlag, Wiesbaden, pp. 302–317. Bonifay, M.-F., 1991. Equus hydruntinus Regalia minor n.ssp. from the Caves of Lunel-Viel (Herault, France). In: Meadow, R.H., Uerpmann, H.P. (Eds.), Equids in the Ancient World II (Beihefte zum Tübinger Atlas des Vorderen Orients, Reihe A, Nr 19/2). Dr. Ludwig Reichert Verlag, Wiesbaden, pp. 178–216. Boulbes, N., 2009. Etude comparée de la denture d’Equus hydruntinus (Mammalia, Perissodactyla) dans le Sud-Est de la France. Implications biogéographiques et biostratigraphiques. Quaternaire 20, 449–465. Buitenhuis, H., 1997. Asikli Hüyük: a “proto-domestication” site. Anthropozoologica 25–26, 655–662. Burke, A., 2002. Palaeoethiology as an archaeological tool: a model for the social and spatial behaviour of E. hydruntinus. In: Mashkour, M. (Ed.), Equids in Time and Space (Proceedings of the 9th ICAZ Conference). Oxbow Books, Durham, pp. 62–69. Burke, A., Eisenmann, V., Ambler, G., 2003. The systematic position of Equus hydruntinus, an extinct species of Pleistocene equid. Quat Res 59, 459–469. Champlot, S., Bennett, E.A., Gautier, M., Arbuckle, B., Balasescu, A., Davis, S., Eisenmann, V., Germonpré, M., Mashkour, M., Morales Muniz, A., Peters, J., Tournepiche, J.-F., Uerpmann, H.-P., Grange, T., Geigl, E.-M., Phylogeography of the asiatic wild ass, in preparation. Charruau, P., Fernandes, C., Orozco-Terwengel, P., Peters, J., Hunter, L., Ziaie, H., Jourabchian, A., Jowkar, H., Schaller, G., Ostrowski, S., et al., 2011. Phylogeography, genetic structure and population divergence time of cheetahs in Africa and Asia: evidence for long-term geographic isolates. Mol. Ecol. 20, 706–724. Clark, B., Duncan, P., 1992. Asian wild asses – hemiones and kiangs (E. hemionus pallas and E. kiang moorcroft). In: Duncan, P. (Ed.), Zebras, Asses, and Horses: An Action Plan for the Conservation of Wild Equids. , pp. 17–21. Cleyet-Merle, J.-J, Madelaine, S., 1991. La pendeloque magdalénienne gravée d’un “Equus hydruntinus” de la Grotte du Putois II, commune de Montmaurin (HauteGaronne). Paléo, pp. 119–129. Clot, A., Duranthon, F., 1990. Les Mammifères fossiles du Quaternaire dans les Pyrénées (exposition, 1990). Muséum d’histoire naturelle de Toulouse, Toulouse. Clutton-Brock, J., 1986. Osteology of the Equids from Sumer. In: Meadow, R.H., Uerpmann, H.P. (Eds.), Equids in the Ancient World (Beihefte zum Tübinger Atlas des Vorderen Orients, Reihe A, Nr 19/1). Dr. Ludwig Reichert Verlag, Wiesbaden, pp. 207–225. Clutton-Brock, J., 1999. A Natural History of Domesticated Mammals, 2d edition. Cambridge University Press, Cambridge, UK. Davis, S.J.M., 1980. Late pleistocene and holocene equid remains from Israel. Zool. J. Linn. Soc. 70, 289–312. Davis, S.J.M., 1987. The Archaeology of Animals. Batsford, London. de Queiroz, K., 2005. Ernst Mayr and the modern concept of species. Proc. Natl. Acad. Sci. U.S.A. 102 (Suppl. 1), 6600–6607. de Queiroz, K., 2007. Species concepts and species delimitation. Syst. Biol. 56, 879–886. Delpech, F., 1984. Les Ongulés en Périgord et Nord-Ouest du Quercy durant le Würm III. Géobios 17, 531–548. Delpech, F., 2003. L’environnement animal des européens au Paléolithique supérieur. In: Desbrosse, R., Thévenin, A. (Eds.), L’Europe préhistorique (Collection: Cahiers de Paléontologie). C.N.R.S. Edition, Paris, p. 186. Denzau, G., Denzau, H., 1999. Wildesel. Jan Thorbecke Verlag GmbH & Co, Stuttgart. 101 Ducos, P., 1986. The Equid of Tell Muraibit, Syria. In: Meadow, R.H., Uerpmann, H.P. (Eds.), Equids in the Ancient World (Beihefte zum Tübinger Atlas des Vorderen Orients, Reihe A, Nr 19/1). Dr. Ludwig Reichert Verlag, Wiesbaden, pp. 238–242. Eisenmann, V., 1979. Les métapodes d’Equus sensu lato (Mammalia, Perissodactyla). Geobios 12, 863–886. Eisenmann, V., 1980. Les chevaux (Equus sensu Lato) fossiles et actuels: crâne et dents jugales supérieures (Collection: Cahiers de paléontologie). CNRS Edition, Paris. Eisenmann, V., 1984. Sur quelques caractères adaptatifs du squelette d’Equus et leurs implications paléoécologiques. Bull. MNHN 6, 185–195. Eisenmann, V., 1986. Comparative Osteology of Modern and Fossil Horses, Halfasses, and Asses. In: Meadow, R.H., Uerpmann, H.P. (Eds.), Equids in the Ancient World (Beihefte zum Tübinger Atlas des Vorderen Orients, Reihe A, Nr 19/1). Dr. Ludwig Reichert Verlag, Wiesbaden, pp. 67–116. Eisenmann, V., 1991. Les chevaux quaternaires européens (Mammalia, Perissodactyla). Taille, typologie, biostratigraphie et taxonomie. Géobios 24, 747– 759. Eisenmann, V., 1992a. Systematic and biostratigraphical interpretation of the equids of Qafzeh, Tabun, Shkul and Kebara (Acheulo-Yabrudian to Upper Palaeolithic of Israel). ArchaeoZoologia V, 43–62. Eisenmann, V., 1992b. Origins, dispersals, and migrations of Equus (Mammalia, Perissodactyla). Courier Forsch-Inst Senckenberg 153, 161–170. Eisenmann, V., 1995. What metapodial morphometry has to say about some miocene hipparions. In: Vrba, E.S., Denton, G.H., Partridge, T.C., Burckle, L.H. (Eds.), Paleoclimate and Evolution, with Emphasis on Human Origins. Yale University Press, New Haven and London. Eisenmann, V., Shah, N., 1996. Some craniological observations on the Iranian, Transcaspian, Mongolian and Indian Hemiones. In: Rietker, F., Brouwer, K., Smits, S. (Eds.), European Endangered Species Program Yearbook 1995/1996 including the Proceedings of the 13th EEP Conference Saumur 20–24 June 1996. EAZA Executive Office, Amsterdam, pp. 396–402. Eisenmann, V., Mashkour, M., 1999. The small equids (Perissodactyla, Mammalia) of the Pleistocene of Binagady (Azerbaijan) and Quazvin (Iran): E. hemionus binagadensis nov. subsp. and E. hydruntinus. Geobios 32, 105–122. Eisenmann, V., Mashkour, M., 2000. Data base for teeth and limb bones of modern hemiones. In: Desse, J., Desse-Berset, N. (Eds.), Fiche d’Ostéologie Animale pour l’Archéologie Série B: Mammifères. Centre de Recherches Archéologiques du CNRS. Editions APDCA, Antibes. Eisenmann, V., 2006. Plioene and pleistocene equids: Palaeontology versus molecular biology. In: Kahlke, R.D., Maul, L.C., Mazza, P. (Eds.), Late Neogene and Quaternary Biodiversity and Evolution: Regional Developments and Interregional Correlations Proceedings Volume of the 18th International Senckenberg Conference (VI International Palaeontological Colloquium in Weimar). 25th–20th April 2004. Cour. Forsch. -Inst. Senckenberg, Frankfurt a. M, pp. 71–89. Eisenmann, V., 2010. Sussemionus, a new subgenus of Equus (Perissodactyla, Mammalia). C. R. Biol. 333, 235–240. Eisenmann, V., Howe, J., Pichardo, M. Old world hemiones and new word slender species (Mammalia, Equidae). Paleovertebrata, in press. Forsten, A., 1986. A review of the Süssenborn Horses and the origin of Equus hydruntinus Regalia. Quartärpaläontologie 6, 43–52. Forsten, A., 1990. Old world “Asses”. Quartärpaläontologie 8, 71–78. Forsten, A., Ziegler, R., 1995. The horses (Mammalia, Equidae) from the early Würmian of Villa Seckendorff, Stuttgart-Bad Cannstatt, Germany. Stuttgarter Beiträge zur Naturkunde Series B (Geologie und Paläontologie) 224, 1–22. Forsten, A., 1999. A review of Equus stenonis Cocchi (Perissodactyla, Equidae) and related forms. Q. Sci. Rev. 18, 1373–1408. George, M., 1869. Etudes zoologiques sur les Hémiones et quelques autres espèces chevalines. Annales des Sciences Naturelles, Zoologie et Paléontologie 5ème série, 1–47. Gilbert, A.S., 1991. Equid Remains from Godin Tepe, Western Iran: an interim summary and interpretation, with notes on the introduction of the horse into Southwest Asia. In: Meadow, R.H., Uerpmann, H.P. (Eds.), Equids in the Ancient World II (Beihefte zum Tübinger Atlas des Vorderen Orients, Reihe A, Nr 19/2). Dr. Ludwig Reichert Verlag, Wiesbaden, pp. 75–122. Grange, S., Duncan, P., 2006. Bottom-up and top-down processes in African ungulate communities: resources and predation acting on the relative abundance of zebra and grazing bovids. Ecography 29, 899–907. Gromova, V., 1949. Histoire des chevaux (genre Equus) de l’ancien monde (Première partie): Revue et description des formes, Vol 17. Travaux de l’Institut Paléontologique, Académie des Sciences de l’URSS. t.17 n(1 Traduction Pietresson de Saint-Aubin, 1955, Annales du Centre d’études et de Documentation paléontologique, no. 13. Gromova, V., 1955. Le genre Equus. Annales du Centre d’études et de Documentation paléontologique, no. 13. Groves, C.P., Mazak, V., 1967. On some taxonomic problems of Asiatic wild asses; with the description of a new subspecies (Perissodactyla; Equidae). Zeitschrift für Säugetierkunde 32, 321–355. Groves, C.P., Willoughby, D.P., 1981. Studies on the taxinomy and phylogeny of the genus Equus, 1. Subgeneric classification of the recent species. Mammalia 45, 321–384. Groves, C.P., 1986. The taxonomy, distribution, and adaptations of recent equids. In: Meadow, R.H., Uerpmann, H.P. (Eds.), Equids in the Ancient World (Beihefte zum Tübinger Atlas des Vorderen Orients, Reihe A, Nr 19/1). Dr. Ludwig Reichert Verlag, Wiesbaden, pp. 11–47. 102 E.-M. Geigl, T. Grange / Annals of Anatomy 194 (2012) 88–102 Guadelli, J.L., 1987. Contribution à l’étude des zoocénoses préhistoriques en Aquitaine (Würm ancien et interstade würmien). Université Bordeaux I, Bordeaux, p. 568. Haimovici, S., Balasescu, A., 2006. Zooarchaeological study of the faunal remains from Techirghiol (Hamangia Culture, Dobrogea, Romania). Cercetari Archeologice 13, 371–391. Hewitt, G., 2001. Speciation, hybrid zones and phylogegography – or seeing genes in space and time. Mol. Ecol. 10, 537–549. Janis, C.M., Gordon, I., Illius, A., 1994. Modelling equid/ruminant competition in the fossil record. Hist. Biol. 8, 15–29. Johansson, A.M., Pettersson, M.E., Siegel, P.B., Carlborg, O., 2010. Genome-wide effects of long-term divergent selection. PLoS Genet. 6, e1001188. Kangas, A.T., Evans, A.R., Thesleff, I., Jernvall, J., 2004. Nonindependence of mammalian dental characters. Nature 432, 211–214. Knowles, L.L., Carstens, B.C., 2007. Delimiting species without monophyletic gene trees. Syst. Biol. 56, 887–895. Kruger, K., Gaillard, C., Stranzinger, G., Rieder, S., 2005. Phylogenetic analysis and species allocation of individual equids using microsatellite data. J. Anim. Breed Genet. 122 (Suppl. 1), 78–86. Mashkour, M., 2001. Chasse et élevage du Néolithique à l’äge du Fer dans la plaine de Qazvin (Iran). Etude archéozoologique des sites Zagheh, Qabrestan et Sagzabad. In Histoire de l’Art et d’ArchéologieUniversité de Paris I/Panthéon-Sorbonne, Paris. Mashkour, M., 2002. Chasse et élevage au nord du Plateau central iranien entre le Néolithique et l’âge du Fer. Paléorient 28, 27–42. Mayr, E., 1963. The new versus the classical in science. Science 141, 765. Meadow, R.H., 1986. Some Equid Remains from Cayönü, Southeastern Iran: Faunal Remains from Tepe Yahya and Tepe Gaz Tavila-R37, 5500–3000 BC. Anthropology Harvard University, Cambridge, MA. Meadow, R.H., Uerpmann, H.P., 1986. Equids in the Ancient World I. Dr. Ludwig Riechert Verlag, Wiesbaden. Meadow, R.H., 1999. The use of size index scaling techniques for research on archaezoological collections from the Middle East. In: Historia Animalium ex Ossibus Festschrift für Angela von den Driesch zum 65 Geburtstag. Verlag Marie Leidorf, Rahden, Westf, pp. 285–300. Michaux, J.R., Magnanou, E., Paradis, E., Nieberding, C., Libois, R., 2003. Phylogeography of the Woodmouse (Apodemus sylvaticus) in the Western Palearctic region. Mol. Ecol. 12, 685–697. Nores Quesada, C., Liesau von Lettow-Vorbeck, C., 1992. La zoologia historica como complemento de la arqueozoologia. El caso del zebro. Archaeofauna 1, 61–71. Oakenfull, E.A., Lim, H.N., Ryder, O.A., 2000. A survey of equid mitochondrial DNA: implications for the evolution, genetic diversity and conservation of Equus. Cons. Genet. 1, 345–355. Orlando, L., Eisenmann, V., Reynier, F., Sondaar, P., Hänni, C., 2003. Morphological convergence in Hippidion and Equus (Amerhippus) South American equids elucidated by ancient DNA analysis. J. Mol. Evol. 57 (Suppl. 1), S29–S40. Orlando, L., Mashkour, M., Burke, A., Douady, C.J., Eisenmann, V., Hanni, C., 2006. Geographic distribution of an extinct equid (Equus hydruntinus: Mammalia, Equidae) revealed by morphological and genetical analyses of fossils. Mol. Ecol. 15, 2083–2093. Orlando, L., Male, D., Alberdi, M.T., Prado, J.L., Prieto, A., Cooper, A., Hanni, C., 2008. Ancient DNA clarifies the evolutionary history of American Late Pleistocene equids. J. Mol. Evol. 66, 533–538. Orlando, L., Metcalf, J.L., Alberdi, M.T., Telles-Antunes, M., Bonjean, D., Otte, M., Martin, F., Eisenmann, V., Mashkour, M., Morello, F., et al., 2009. Revising the recent evolutionary history of equids using ancient DNA. Proc. Natl. Acad. Sci. U.S.A. 106, 21754–21759. Payne, S., 1972a. Can Hasan III, the Anatolian Aceramic, and the Greek Neolithic. In: Higgs, E.S. (Ed.), Papers in Economic Prehistory. Cambridge University Press, Cambridge, pp. 191–194. Payne, S., 1972b. On the interpretation of bone samples from archaeological sites. In: Higgs, E.S. (Ed.), Papers in Economic Prehistory. Cambridge University Press, Cambridge, pp. 65–81. Payne, S., 1975. Partial recovery and sample bias. In: Clason, A.T. (Ed.), Archaeozoological Studies. North-Holland Publishing Company, Amsterdam, pp. 7–17. Payne, S., 1991. Early Holocene Equids from Tall-i-Mushki (Iran) and Can Hasan III (Turkey). In: Meadow, R.H., Uerpmann, H.P. (Eds.), Equids in the Ancient World II (Beihefte zum Tübinger Atlas des Vorderen Orients, Reihe A, Nr 19/2). Dr. Ludwig Reichert Verlag, Wiesbaden, pp. 132–165. Portnoy, D.S., McDowell, J.R., Heist, E.J., Musick, J.A., Graves, J.E., 2010. World phylogeography and male-mediated gene flow in the sandbar shark, Carcharhinus plumbeus. Mol. Ecol. 19, 1994–2010. Prat, F., 1968. Recherche sur les Equidés pléistocènes en FranceUniversité de Bordeaux I. Bordeaux France, p. 696. Provan, J., Benentt, K.D., 2008. Phylogeographic insights into cryptic glacial refugia. Trends Ecol. Evol. 23, 564–571. Pruvost, M., Grange, T., Geigl, E.-M., 2005. Minimizing DNA contamination by using UNG-coupled quantitative real-time PCR on degraded DNA samples: application to ancient DNA studies. Biotechniques 38, 569–575. Pruvost, M., Schwarz, R., Bessa Correia, V., Champlot, S., Braguier, S., Morel, N., Fernandez-Jalvo, Y., Grange, T., Geigl, E.-M., 2007. Freshly excavated fossil bones are best for amplification of ancient DNA. Proc. Natl. Acad. Sci. U.S.A. 104, 739–744. Radulesco, C., Samson, P., 1965. Sur la présence de Hydruntinus hydruntinus (Regalia) en Roumanie. Quaternaria 7, 219–234. Regàlia, E., 1907. Sull’Equus (Asinus) hydruntinus Regalia della grotta di Romanelli (Castro, Lecce). Archivio per l’Antropologia e l’Etnologia Firenze 37, 375–390. Reynaud Savioz, N., Morel, P., 2005. La faune de Nadaouiyeh Aïn Askar (Syrie centrale, Pléistocène moyen): aperçu et perspectives. Rev. Paléobiol. 10, 31–35. Russel, N., Martin, L., 2000. Neolithic Catalhöyük: Preliminary Zooarchaeological Results from the Renewed Excavations. In: Mashkour, M., Buitenhuis, H., Choyke, A.M., Poplin, F. (Eds.), Archaeozoology of the Near East, IV. ARC Publicatie 32, Groningen, The Netherlands, pp. 163–169. Ryder, O.A., 1978. Chromosomal polymorphism in Equus hemionus. Cytogenet. Cell Genet. 21, 177–183. Ryder, O.A., Chemnick, L.G., 1990. Chromosomal and molecular evolution in Asiatic wild asses. Genetica 83, 67–72. Scanella, J.B., Horner, J.R., 2010. Torosaurus Marsh, 1891, is Triceratops Marsh, 1889 (Ceratopsidae: Chasmosaurinae): synonymy through ontogeny. J. Vertebr. Paleontol. 30, 1157–1168. Schmitt, T., 2007. Molecular biogeography of Europe: pleistocene cycles and postglacial trends. Front. Zool. 4, 11. Schneider, C.J., Smith, T.B., Larison, B., Moritz, C., 1999. A test of alternative models of diversification in tropical rainforests: ecological gradients vs. rainforest refugia. Proc. Natl. Acad. Sci. U.S.A. 96, 13869–13873. Schreiber, A., Eisenmann, V., Zimmermann, W., 2000. Hemiones: pluridisciplinary quest of their identities and relationships. In: Zimmermann, W. (Ed.), EEP, Asiatic Equids, Husbandry Guidelines. Zoologischer Garten, Köln. Smith, T.B., Wayne, R.K., Girman, D.J., Bruford, M.W., 1997. A role for ecotones in generating rainforest biodiversity. Science 276, 1855–1857. Spassov, N., Iliev, N., 2002. The animal bones from the prehistoric necropolis near Durankulak (NE Bulgaria) and the latest record of Equus hudruntinus Regalia. In: Todorova, H. (Ed.), Durankulak, Band II: Die Prähistorischen Gräberfelder von Durankulak. Deutsches Archäologisches Institut, Berlin, pp. 313–324. Stehlin, H.G., Graziosi, P., 1935. Ricerche sugli Asinidi fossili d’Europea. Mémoires de la Société Paléontologique Suisse 56, 1–73. Taberlet, P., Fumagalli, L., Wust-Saucy, A.-G., Cossons, J.-F., 1998. Comparative phylogeography and postglacial colonization routes in Europe. Mol. Ecol. 7, 453–464. Turnbull, P.F., 1986. Measurements of Equus hemionus from Palegawra Cave (Zarzian, Iraq). In: Meadow, R.H., Uerpmann, H.P. (Eds.), Equids in the Ancient World (Beihefte zum Tübinger Atlas des Vorderen Orients, Reihe A, Nr 19/1). Dr. Ludwig Reichert Verlag, Wiesbaden, pp. 319–326. Uerpmann, H.P., 1986. Halafian Equid Remains form Shams ed-Din Tannira in Northern Syria. In: Meadow, R.H., Uerpmann, H.P. (Eds.), Equids in the Ancient World (Beihefte zum Tübinger Atlas des Vorderen Orients, Reihe A, Nr 19/1). Dr. Ludwig Reichert Verlag, Wiesbaden, pp. 246–265. Uerpmann, H.P., 1987. The Ancient Distribution of Ungulate Mammals in the Middle East. Dr. Ludwig Reichert Verlag, Wiesbaden. Uerpmann, H.P., 2005. Betrachtungen zum Verhältnis zwischen Wildpferd (Equus ferus) und Hydruntinus (Equus hydruntinus) im Jungpleistozän und Holozän auf der Iberischen Halbinsel. In: Homenaje a Jesus Altuna, Tomo I: paleontologia y arqueozoologia. Sociedad de Ciencias Aranzadi, San Sebastian, pp. 351–358. Vila, E., 1996. L’exploitation des animaux en Mésopotamie aux IVe et IIIe millénnaire av. JC. CNRS Editions, Paris. Von den Driesch, A., 2000. Revision zum Vorkommen des Equus (Asinus) hydruntinus (Regalia 1907) im Chalkolithikum der Iberischen Halbinsel. Archaeofauna 9, 35–38. Weinstock, J., 2000. Late Pleistocene reindeer populations in middle and Western Europe. An Osteometrical Study of Rangifer Tarandus, vol. 3. Mo Vince Verlag, Tübingen. Weinstock, J., Willerslev, E., Sher, A., Tong, W., Ho, S.Y., Rubenstein, D., Storer, J., Burns, J., Martin, L., Bravi, C., et al., 2005. Evolution, systematics, and phylogeography of pleistocene horses in the new world: a molecular perspective. PLoS Biol. 3, e241. Wilms, C., 1989. Zum Aussterben des europäischen Wildesels. Germania 67, 143–148.