Reconsideración de la sistemática de Cervavitus (Cervidae, Artiodactyla, Mammalia) del Pleistoceno Inferior-Reconsideration of the systematics of the Early Pleistocene Cervavitus (Cervidae, Artiodactyla, Mammalia)

Cervavitus were usually found from the Late Miocene and Pliocene deposits in East Europe, Middle Asia and North China, but they were found recently in many Early Pleistocene localities in southern China. The latter resulted in the discussion of their systematic status between Cervavitus and Cervus. Here we show the Early Pleistocene forms from southern China are morphometrically more similar to northern China Cervavitus species, and the cladistic analysis shows that the southern China forms are closer to classic Cervavitus species than Cervus and that also proves their systematic status in Cervavitus rather than in Cervus. Cervavitus originated in Moldovan forests of East Europe in the late Vallesian (MN10) from a brachyodont and holometacarpal ancestor with two/three-tined antlers and Palaeomeryx fold and evolved into C. novorossiae. It dispersed into West Europe forests in the earliest Turolian and further west to France in the Ruscinian. It dispersed into northern China forests in the early Turolian and represented by C. shanxius. The great quantity of C. shanxius specimens with brachyodont teeth and complete lateral metacarpals implies the arid Loess Plateau of today was a humid forested region in the Late Miocene. C. shanxius migrated southwards in the Plio-Pleistocene probably due to the drying environment in northern China with uplifting of Himalayas and evolved into C. ultimus and C. fenqii, which survived in southern China until the Early Pleistocene (MNQ18).


Introduction
The first time I met Professor Léonard Ginsburg was in 1986 when I went to France for my doctoral degree in vertebrate palaeontology at the Université de Poitiers. My Chinese colleagues asked me to pass their greetings to French colleagues and I began to know some French colleagues in person further than in literature. When my dissertation materials were focused on cervids, which also fit into Professor Ginsburg's interests (e.g. Ginsburg, 1985Ginsburg, , 1999Ginsburg & Azanza, 1991;Ginsburg & Bulot, 1987;Ginsburg & Crouzel, 1976;Ginsburg & Ukkakimapan, 1983;etc.), I had some discussion with him and I also learned a lot from him by personal communication or by his publications. My first research publication in France (Dong, 1996) was also with his kind help when he was the editorin-chief of the Bulletin du Muséum national d'Histoire naturelle, which has been renamed as Geodiversitas since 1997. Although his Asian field work focused mainly on Thailand (e.g. Ginsburg & Mein, 1987;Ginsburg & Tassy, 1985;Ginsburg & Ukkakimapan, 1983;etc.), he visited China and had some academic exchanges with Chinese colleagues (e.g. Qiu et al., 1985). I am recently working on the Early Pleistocene Gigantopithecus fauna in southern China (Jin et al., 2009;Dong et al., 2010;Dong et al., 2011) where the environments are similar to those in Thailand. I found one of cervid taxa in the fauna, Cervavitus, very problematic which I would discuss with late Professor Ginsburg and I'd like to take this opportunity to communicate with international colleagues who are interested in it.

The first Cervavitus and its follow-up members
During the study of the cervid materials from Tarakliya in Moldova, Khomenko established "Cervavitus tarakliensis" (Khomenko, 1913) for a cervid with two tined antlers, its first or basal tine set high above the burr. Two other cervid species were together named, i.e. "Cervocerus Novorossiae" and "Damacerus Bessarabiae". The former was a cervid with larger antlers than those of "Cervavitus tarakliensis", the three tines of the antlers well developed; while the latter a cervid which had antlers with flattened and somewhat palmate main beam. The three species were included in a new subfamily "Pliocervinae" (Khomenko, 1913). Damacerus was considered as a synonym of Cervocerus by Zdansky (1925), and the view was shared by Simpson (1945). But it was accepted as valid taxon by Teilhard de Chardin and Trassaert (1937). And Cervocerus was later regarded as a junior synonym of Cervavitus (Vislobokova, 1990) because the Cervavitus of Khomenko appears as a juvenile form of Cervocerus of Khomenko. Considering the age and environmental variation of antlers (Heintz, 1970), it is acceptable that both Cervocerus and Damacerus of Khomenko are synonymous of Cervavitus. The dentitions of Cervavitus from Tarakliya are characterized by the evident presence of Palaeomeryx fold on lower molars (Khomenko, 1913). Simpson (1945) included "Pliocervinae" into Cervinae based on close relationship between them and it was followed by many others such as Wang and Wu (1979), Vislobokova (1990), etc. Czyzewska (1968 proposed a new tribe Pliocervini under Cervinae to include primitive Cervavitus and Pliocevus. In addition, Khomenko's "Pliocervinae" is not a valid sub-family name according to arts. 35.3 & 36.1 of International Code of Zoological Nomenclature (http://iczn.org/code). Pliocervini proposed by Czyzewska (1968) was consequently widely accepted (e.g. Vislobokova, 1990;McKenna & Bell, 1997;Petronio et al., 2007). Zdansky (1925) included some cervid materials from Henan Province (previously spelled as Honan) of northern China to "Cervocerus Novorossiae" of Khomenko. Teilhard de Chardin and Trassaert (1937) also included some similar cervid materials from Shanxi Province (previously spelled as Shansi) to this species, but they modified the diagnosis and added three diagnostic characters for Chinese materials: "pedicles prolonged by a ridge on the frontals, Palaeomeryx fold generally missing in the Chinese form, vestigial lateral metacarpals preserved along the whole length of the anterior canon bone" (Teilhard de Chardin & Trassaert, 1937). They also established a new species Cervavitus demissus for some antlers with long pedicle and low brow tine from Baode, Shanxi Province. The present author regards it as a junior synonym of Cervavitus shanxius, because the species was represented by a few specimens that fall well into the age and environmental variations of C. shanxius, and it is furthermore no longer discovered elsewhere since Teilhard de Chardin and Trassaert. The presence of Cervavitus in Shanxi Province was confirmed during the 1950's and 1980's field work in the Yushe Basin (Dong & Hu, 1994;Dong & Ye, 1996), some dozens of skulls and some of which with partial frontal appendages, several hundreds of antlers and dentitions were identified and a sub-specific name, "Cervavitus novorossiae shanxius", was given to Chinese form to distinguish it from its Moldova counterpart (Dong & Hu, 1994). Petronio et al. (2007) proposed further to upgrade the subspecies to Cervavitus shanxius for distinction between eastern European and Chinese Cervavitus. The author shares the same view. Cervavitus shanxius was reported found in Yangjiashan fauna and Shilidun fauna in Linxia Basin, north-western China (Deng, 2009). A new pliocervine, Cervavitus huadeensis Qiu, 1979, was found from Huade in Nei Mongol (Inner Mongolia). It is represented by two fragmental skulls and some antlers, the latter differ from those of C. shanxius by larger and "dagger-like" appearance, slightly lyrated beam and four-tines (Qiu, 1979). The ridge on the frontal formed by prolonged pedicle is less strong in Huade specimens than in Shanxi specimens, and the antlers are similar to those of "Damacerus Bessarabiae" of Khomenko.
Chinese Cervavitus were usually found in northern China (Zdansky, 1925;Teilhard de Chardin & Trassaert, 1937;Qiu, 1979;Dong and Hu, 1994;Deng, 2009) and dated as the Late Miocene (formerly considered as "Pontian"). The first Cervavitus in southern China, C. ultimus (Lin et al. 1978), was reported from the Early Pleistocene Yuanmou Formation at Yuanmou Man Site. Yuanmou specimens are generally similar to those of Shanxi, e.g. pedicles prolonged by a ridge on the frontals, antlers three-tined, Palaeomeryx fold generally missing, but the second and third tines are shorter and the main beam longer in Yuanmou specimens than in Shanxi counterparts. The species is also found at Longgupo Site at Wushan in central China (Huang & Fang, 1991). The second Cervavitus in southern China, Cervavitus fenqii Han, 1987, was reported from the Early Pleistocene cave deposits at the Gigantopithecus Cave in Liucheng, Guangxi (Han, 1987). It is also characterised by pedicles prolonged by a ridge on the frontals, antlers threetined, Palaeomeryx fold generally missing, and it differs from C. ultimus by its smaller antlers, shorter brow tine and main beam, and it differs from C. shanxius by its smaller antlers and shorter brow tine (Han, 1987). C. fenqii was also reported found at Mohui Cave in Bubing Basin, Guangxi (Wang et al., 2007), at Longgudong Cave of Jianshi, Hubei (Chen, 2004), at Renzidong Site of Fanchang, Anhui . And recently, some Cervavitus-like teeth were found from the Early Pleistocene deposits in Sanhe Cavern at Chongzuo, Guangxi (Dong et al., 2011).
Cervavitus was also reported found in the Early Pliocene (MN14) of Montpellier (France) as C. cauvieri (Dong, 1996). It was found from Poksheshty of Moldova in the Late Miocene (MN10), C. variabilis from the Late Miocene to the Early Pliocene (MN12-MN14) and C. flerovi from Esekartkan of Kazakhstan in the Early Pliocene (MN15) (Vislobokova, 1990;. Meanwhile, some "Cervavitus" were included into other cervids, e.g. Cervavitus bessarabiensis to Euprox aff. furcatus, Cervavitus sarmaticus to Euprox sarmaticus (Vislobokova, 2007). For detailed research history on Cervavitus, please refer Petronio et al. (2007) who gave a very good and comprehensive review on the Cervavitus from Europe and some of the Cervavitus in China.

The Early Pleistocene Cervavitus or Cervus?
The Cervavitus materials found in the Late Miocene of northern China have no controversy for their generic attribution. But those found in the Early Pleistocene in southern and central China have two generic attributions: Cervavitus and Cervus.
The three tined antler materials and associated dentitions from Longgudong of Jianshi similar to those of Cervavitus fenqii from Gigantopithecus Cave in Liucheng were considered as a species of Cervus (Chen, 2004), and consequently the Cervavitus ultimus from Yuanmou Man Site and Longgupo, as well as the Cervavitus fenqii from Gigantopithecus Cave were also considered attributable to Cervus (Chen, 2004).
It is a common practice to evaluate first the morphology of antlers in the identification of cervid materials, and for three-tined antlers there are three choices for generic attribution: Cervavitus, Axis and Cervus (Rusa). The Axis have curved and somewhat twisted main beam that is evidently different from that of C. ultimus and C. fenqii. The Early Pleistocene Cervus (Rusa) elegans from Nihewan (formerly spelled as Nihowan) Basin (Teilhard de Chardin & Piveteau, 1930) has three-tined antlers and comparable to C. ultimus and C. fenqii. The materials of Cervavitus shanxius are abundant in the Late Miocene deposits (Zdansky, 1925;Teilhard de Chardin & Trassaert, 1937;Dong and Hu, 1994;Deng, 2009) and the Cervavitus in western Eurasia are found only in the Late Miocene and Early Pliocene (Vislobokova, 1990;Dong, 1996). That gives an impression that Cervavitus is a Late Tertiary genus and both C. ultimus and C. fenqii should be affined to the Early Pleistocene Cervus (Rusa) elegans, and they should be attributed to Cervus or Cervus (Rusa) as indicated by Chen (2004).
Teilhard de Chardin and Trassaert (1937) considered their "Pliocervinae" as a group of "Medium sized Cervidae, connecting in some way the Cervulinae and Cervinae. Pedicles of antlers prolonged by strong frontal ridges. Antlers three-tined in the most typical forms. Lateral metacarpals preserved on their whole length. Palaeomeryx fold sometimes present" (Teilhard de Chardin and Trassaert,1937, p.30). It is noticeable that the pedicles of both C. ultimus and C. fenqii are prolonged on frontal by a strong ridge (Lin et al., 1978;Han, 1987) as in the Late Tertiary Tarakliya forms (Khomenko, 1913) and northern China forms (Zdansky, 1925;Teilhard de Chardin & Trassaert, 1937;Dong & Hu, 1994). The ridge is also present on Longgudong materials (Chen, 2004, fig. 5.79). The ridge is a typical character of most muntiacines and Pliocervini, but it is very weak or absent in Cervini. This character supports the attribution of the Pleistocene C. ultimus and C. fenqii to Cervavitus. Simpson diagrams of cheek teeth also support the similarities of C. ultimus and C. fenqii with previous species of Cervavitus (Figs. 1-2).
The data matrix was edited by NEXUS Data Editor (Version 0.4.8 by Roderic D.M. Page), and then processed by PAUP* (Version 4.0b10 for 32-bit Microsoft Windows by David L. Swofford). 61 best trees were found among 21129 parsimony trees. In considering the importance of the characters as holometacarpal, the state of Palaeomeryx fold, the prolonged pedicle ridge on frontal, etc. the most acceptable tree was selected among these 61 best trees as shown in Fig. 4. The Cervavitus as a whole forms a monophyly that appears in many trees found by Paup*, and they appear at least closer to each other than to the other taxa. It supports the classification of five species into Cervavitus although Palaeomeryx fold is only present in Cervavitus novorossiae and probably in Cervavitus huadeensis. It supports also the inclusion of the Early Pleistocene C. ultimus and C. fenqii into Cervavitus rather than to Cervus. The systematic position of Cervavitus is between muntiacines and cervini.

Biochronological and paleoenvironmental considerations
Much work was done in the last decades for establishing Chinese Land Mammal Age or Terrestrial Biozanation. During the Symposium on Asian Neogene Biostratigraphy and Geochronology held in 2009 in Beijing, many colleagues expressed their impression that it was still premature to form a standard chronological chart based on the existing land mammal sequences. The problem is that no Neogene marine deposits exist in the mainland for large scaled stratigraphic correlation as in Paratethys region. Nevertheless, the European Neogene Biozonation based on mammals has been available for many years and can be used as a reference for chronological distribution of Cervavitus. In addition, the complete merging of European continent with Asian one since the Miocene has undermined the geographic barriers for fauna exchanges between two continents that make the biochronological correlation between two ends of Eurasian continents possible. All recognized species of Cervavitus appeared in publications corresponding to relevant European Neogene (including Quaternary) Mammal Zones are listed in Table 2.
Cervavitus, C. shanxius, was found from Yanjiashan Formation in north-western China (Deng, 2009)  ultimate Cervavitus and so named (Lin et al., 1978). Its earliest appearance was found from Longgupo Site dated earlier than 2Ma (Huang & Fang, 1991), and its last appearance was found from Yuanmou Formation biochronologically equivalent to MNQ17 and probably to MNQ18. The latest Cervavitus known so far is C. fenqii. Its first appearance was found at Renzidong Site in eastern China dated earlier than 2 MA or equivalent to MNQ16 , and its last appearance was found from the Gigantopithecus Cave in southern China (Han, 1987) dated equivalent to MNQ18 (Jin et al., 2009). Based on the chronological distribution of Cervavitus data available, it can be concluded that the genus originated in Moldovan forests of East Europe in the Vallesian from an ancestor with two/three-tined antlers, Palaeomeryx fold, brachyodont cheek teeth and complete lateral metacarpals and evolved into C. novorossiae. It dispersed westwards into West Europe forests (e.g. Kohfidisch, Austria) in the earliest Turolian as Cervavitus sp. (Vislobokova, 2007). It dispersed further west to France (e.g. in Montpellier) and evolved into C. cauvieri in the Early Pliocene. The C. novorossiae in East Europe continued to evolve and C. variabilis was derived in the middle Turolian and survived until the early Ruscinian. On the other side, the C. novorossiae dispersed eastwards into Central Asian and evolved into C. flerovi in the Late Pliocene (Vislobokova, 2007). The C. novorossiae dispersed further eastwards into East Asian forests in the early Turolian and evolved into two forms, C. shanxius which was widespread in northern China in the Late Miocene, and C. huadeensis, which was larger and limited in Inner Mongolia. The extant holometacarpal tragulids (Shen, 1992) and suids are forest dwellers, the complete lateral metacarpals are of great help for balancing the body on irregular slopes. Judged by the presence of brachyodont cheek teeth and complete lateral metacarpals in C. shanxius, the species should be a forest dweller on irregular slopes. The specimens of C. shanxius from the Late Miocene sediments are of great quantity implying large populations of the species. Although the specimens are mostly found from the arid Loess Plateau of today, the areas were quite humid for forest vegetation to feed such large population of C. shanxius in the Late Miocene. C. shanxius disappeared from northern China and dispersed southwards in the Plio-Pleistocene probably due to the drying environment in northern China with uplift-ing of Himalayas and evolved into C. ultimus and C. fenqii, they survived until the middle stage of the Early Pleistocene (MNQ18) and evolved into or replaced by some other larger Cervini such as Axis and Rusa.