A fragment of mandible and two metapodials complete unearthed from the fossiliferous aera of Kossom Bougoudi (KB3 and KB26), northern Chad are described. A comparative study allows to assign these specimens to Paracamelus gigas. The evolutionnary degree is compatible with an age around the Mio-Pliocene boundary (ca 5 Ma). Then, the Chadian remains are the oldest adequately dated record of this family in Africa. They are contemporaneous with the oldest known evidence of the genus Paracamelus from the late Miocene of Asia and Europe.
Introduction. – During several field seasons in northern Chad, the “Mission Paléoanthropologique Franco-Tchadienne” (M. P. F. T) discovered new sites in the Kossom Bougoudi (KB) fossiliferous area, west of australopithecine sites [Brunet et al., 1995, 1997; Brunet and M.P.F.T, 2000]. These sites yielded a rich vertebrate fauna (fish, reptiles, birds and mammals), and have been biochronologically dated at around 5 Ma old, close to the Mio-Pliocene boundary [Brunet and M. P. F.T, 2000]. Among the mammal fauna, some remains of Camelidae provide the earliest evidence of this group in Africa, which was previously thought to be younger than 4 Ma, at Laetoli [Harris, 1987] and Koobi Fora [Harris, 1991]. Specimens from sites KB3 and KB26 are described here.
Material : KB3.97.316 : right mandible fragment with p3, p4-m1 roots and m2-m3 teeth; KB3.99.03 : right metatarsus; KB26.97.03 : right metatarsus
The mandible is rather robust with a high horizontal ramus. The mental foramen is located below m1. The p3 alveolus and p4 roots attest elongated premolars. The lingual face of the molars is flat. The third lobe of m3 is less labially shifted than in the living camels. There is no cement, nor cingulum.
The metatarsals are long and robust (tab. III), and show a deep groove on the proximal anterior and posterior faces. The distal condyles are divergent and separated by a deep interarticular notch. They are symmetrical and of the same size differences, in contrast with the extant species where the external condyle is more slender than the internal one.
Comparison. – The mandible (KB3.97.316) differs from the Camelus species mandible by having (1) a robust and deeper horizontal ramus, (2) a well developed p3, (3) a third lobe of m3 less labially shifted (4) Chadian metatarsals are morphologically different from those the living camels and being extremely long (tab. II). All characters of the Chadian specimens are congruent with Zdansky’s  and Teilhard and Trassaert’s  descriptions of genus Paracamelus.
The KB horizontal ramus is deeper than that of P. alutensis (tab. I) from the early Pleistocene of Oltet Valley, Romania [Stefanescu, 1910]. The premolar row is longer. Unfortunately, a detailed comparison with P. aguirrei from the late Miocene (MN13) of Venta del Moro and Librilla, Spain is impossible because this species was defined on skeletal elements (upper molars, calcaneum, phalanxes) not yet recovered from Chad. However, the estimated alveolar length of p3 (20 mm) is similar to those of P. aguirrei (18,8 – 21,6 mm according to Morales ). Lengths of KB tooth row (tab. I) and metatarsals (tab. II) fit into the range of variation recorded by Zdansky  and by Teilhard and Trassaert  for P. gigas from the late Miocene of China. The Chadian material cannot be assigned to the species P. alexejevi from the Pliocene (MN15) of Ukraine, because this species is smaller than P. aguirrei and P. gigas [Morales, 1984].
In conclusion, specimens from Chad do not display any important difference with Chinese species P. gigas and can tentatively be referred to this species.
Biochronology and paleobiogeography. – The earliest known Old World camel correspond to P. aguirrei from the late Miocene (MN13) of Venta del Moro and Librilla in Spain [Morales et al., 1980; Made and Morales, 1999]. After Made and Morales , this species is probably the ancestor of P. alexejevi from of Odessa Catacombs (MN15), Ukraine. In Europe, the chronological range of P. alutensis covers the Plio-Pleistocene. This species is present in the lower Pleistocene of Oltet Valley, Romania [Stefanescu, 1910] and in the early and Middle Pliocene (MN16) of Russia [Baigusheva, 1971]. It is also present in the late Pliocene of Sarikol Tepe, Turkey [Kostopoulos and Sen, 1999]. In China, the earliest record of P. gigas is about 5.5 Ma [Flynn, 1997; Made and Morales, 1999].
In conclusion the chronological range of Paracamelus is from the late Miocene to the Pleistocene. However, the Chadian specimens size is close to P. gigas (first occurrence in China around 5.5 Ma) and P. aguirrei from late Miocene (MN13) of Europe. The occurrence of Paracamelus at KB and its absence from the younger Chadian sites (3-4 Ma) of Koro-Toro and Kollé [Brunet et al., 1995; 1996] as well as in the Plio-Pleistocene localities of Africa, are congruent with an age close to the Mio-Pliocene boundary for the sites of KB. This interpretation is confirmed by the associated fauna, that indicates ca 5 Ma old for the whole of KB fossiliferous area [Brunet and M.P.F.T, 2000].
The age of the Chadian Paracamelus is close to the Mio-Pliocene boundary, slightly younger than specimens from late Miocene of China [Zdansky, 1926; Flynn, 1997], Spain [Morales et al., 1980] and Turkey [Made et al., 2002]. This demonstrates that the group had a wider distribution than previously thought. It indicates that the Camelidae reach a widespread distribution soon after their arrival from northern America [Webb, 1965; Pickford et al., 1993].
Conclusion. – The Chadian material displays distinctive features which allows to refer it to Paracamelus gigas. This taxon, poorly documented in Eurasia, has not been previously recognised in Africa. It will contribute to deciphering the phylogenetic relationships between various species of Paracamelus and the extant Camelus.