Introduction

The Amphicyonidae is an extinct family of the Carnivora. It is included in the infraorder Arctoidea, which also includes extant families such as the Ursidae, Procyonidae, and Mustelidae (McKenna and Bell, 1997; however, some recent studies place the Amphicyonidae at the base of the Caniformia (Hunt, 2003; Wesley-Hunt and Flynn, 2005)). Fossil records of the family are found from the middle Eocene to the late Miocene of North America, Europe, Africa, and Asia (McKenna and Bell, 1997). Occurrences of most Paleogene amphicyonids are in Europe or in North America. There are none in Africa, where they do not appear until the early Miocene, and they are rarely found in Asia.

All Eocene Asian amphicyonid materials, except Guanxicyon from the late Eocene Naduo Formation of southern China (Zhai et al., 2003), have been assigned to Cynodictis, a genus commonly known from Europe (Kotsakis, 1980; Russell and Zhai, 1987 and references cited therein). Some of these are from localities of the Irdinmanhan Asian Land Mammal Age (middle Eocene) and are qualified as the oldest amphicyonid occurrences in Asia (Chow et al., 1973; Manning, pers. comm. 1977 cited in Russell and Zhai, 1987; Tsubamoto et al., 2004).

In 2004, the Hayashibara Museum of Natural Sciences and Mongolian Paleontological Center Joint Paleontological Expedition collected a carnivoran upper molar from the upper Eocene Ergilin Dzo Formation of southeastern Mongolia. Morphology of the upper molar led us to assign the tooth to the Amphicyonidae. Also, the morphology of the upper molar differs from that of the genus Cynodictis, to which Mongolian and Chinese Eocene amphicyonid materials were previously assigned. This study provides a description of the new upper molar material and discusses the systematic position of the previously known amphicyonid materials from the Paleogene of Asia.

Systematic Paleontology

Figure 1.

Amphicyonidae gen. et sp. indet. from the Ergilin Dzo Formation (late Eocene; Mongolia), MPC-M 30/63, a right M2, occlusal view: A. Stereograph. B. Schematic figure.

Comparison

There are a number of Paleogene carnivorans: Viverravidae, Miacidae, Nimravidae, Stenoplesictidae, Paleogale (whose taxonomic position is controversial between the Viverravidae and the basal Aeluroidea), Amphicyonidae, basal Ursida (e.g., Amphicyonodon, Cephalogale, Pachycynodon, basal Mustelida, and Canidae). All these taxa except canids, which never left North America until the middle Miocene (Wang and Tedford, 2008), have fossil records in the Paleogene of East Asia (Russell and Zhai, 1987; Tsubamoto et al, 2004). MPC-M 30/63 can be easily distinguished from Ml and M2 of the above-listed taxa except canids and amphicyonids, in having a combination of unreduced main cusps (the paracone, metacone, and protocone), the reduced parastyle, and the development of a thick lingual cingulum. Hesperocyonines, the earliest canids, differ from the present specimen in the presence of the distinct parastyle, the posteriorly limited lingual cingulum, and the absence of the metaconule (Wang, 1994). Although borophagine canids are endemic to the Oligocene to Pliocene of North America (Wang et al, 1999), the present specimen resembles Ml and/or M2 of some borophagines in the morphology of the trigon cusps and the parastyle; however, two characteristics of the present specimen, the buccal cingulum wider in the posterior part and the lingual cingulum thick on both anterior and posterior sides, are unusual for borophagine canids. In sum, we concluded that the combination of characteristics in the present specimen only occurs in the Amphicyonidae.

The size of MPC-M 30/63 indicates that this material belongs to a small amphicyonid. It is slightly larger than the smallest amphicyonids such as European Cynodictis (based on C. l. lacustris and C. l. intermedius specimens from La Debruge and Quercy, respectively) and North American Paradaphoenus (Hunt, 1998a, 2001).

Compared with other amphicyonids, MPC-M 30/63 is unique in having an extremely thick lingual cingulum that bulges posterolingually and a parastyle that positions anterior to the paracone. Cynodictis has been considered as an ancestral genus of other amphicyonids (Hough, 1948; De Beaumont, 1964) or of the European amphicyonids (Hunt, 1998a). MPC-M 30/63 is similar to Cynodictis in relative size of the main cusps and smooth, posterolingually projected lingual cingulum; however, MPC-M 30/63 differs from Cynodictis in lacking an anterobuccally located and moderately developed parastyle, an anterobuccally oriented preparacrista, an ectoflexus, a strong metaconule, and a moderately thick lingual cingulum. Amphicyonids from the late middle Eocene to early Oligocene of North America have been included in the subfamily Daphoeninae (Hunt, 1998a). MPC-M 30/63 is similar to Eocene daphoenines such as Daphoenus and Paradaphoenus in the relative position of the main cusps and smooth and thick lingual cingulum. Eocene daphoenines differ from the present specimen in having a poorly developed ectoflexus, narrower lingual cingulum that projects more lingually, and lingual width clearly narrower than the buccal border. Some latest Eocene European amphicyonids such as Pseudocyonopsis, Brachycyon, and Cynelos have been included in the subfamily Amphicyoninae (Ginsburg, 1966; Springhorn, 1977). Early amphicyonines are similar to MPC-M 30/63 in absence of the ectoflexus and anteroposteriorly wide lingual border. The formers differ from the latter in clear preparacrista and postmetacrista, the absence of the parastyle, the metastyle compressed to parastyle, and a thin lingual cingulum. Some other late Eocene and Oligocene European amphicyonids have been included in the subfamily Haplocyoninae (Springhorn, 1977). The outline of M2 of Haplocyon is similar to that of MPCM 30/63; however, in Haplocyon, the parastyle is absent, the metacone is close to the paracone, the protocone is located more lingually, and the lingual cingulum is narrow. Simamphicyon is a specialized form, and its M2 is oval with reduced protocone and lingual cingulum; thus, it is not similar to either MPC-M 30/63 or other amphicyonids.

Based on these comparisons, the present material likely belongs to a new genus, although more materials are needed to establish a new genus and to discuss its systematic relationships to other amphicyonids.

Amphicyonids from the Paleogene of East Asia

Russell and Zhai (1987) listed several occurrences of amphicyonids from the Paleogene of East Asia: cf. “Cynodictis” from the Ulan Shireh Formation (Irdinmanhan=middle Eocene) (Manning, pers. comm. 1977 cited in Russell and Zhai, 1987); Cynodictis sp. from the Upper Lushi Formation (Irdinmanhan=middle Eocene) (Chow et al, 1973); Cynodictis? (=cf. Cynodictis) from the Ergilin Dzo Formation (Ergilian=late Eocene) at the Ergilin Dzo locality (Matthew and Granger, 1923); and Cynodictis? elegans from an unnamed formation of the Hsandagolian Asian Land Mammal Age (early Oligocene) in Ulantatal (Huang, 1982, 1993). Recently, a new amphicyonid, Guanxicyon, was described from the late Eocene Naduo Formation (Zhai et al, 2003). Its locality is much more southern than the localities of the other East Asian Paleogene amphicyonids, and this genus is currently endemic to the Naduo fauna. Occurrences of Amphicyon, a Miocene amphicyonid, have been reported from two Oligocene localities: Amphicyon sp. from the Upper Wulanbulage Formation in Qianlishan (Wang et al, 1981) and Amphicyon? sp. from the Saint-Jacques fauna (Teilhard de Chardin, 1926). Both the Upper Wulanbulage fauna and the Saint-Jacques fauna were considered to be Hsandagolian (early Oligocene) by Meng and McKenna (1998), but some part of the Saint-Jacques fauna may belong to the late Oligocene (Wang and Qiu, 2003). Amphicyon and Guanxicyon are much larger than the present specimen, and can be easily distinguished. Thus, all East Asian Paleogene amphicyonids, except two, have been identified as the genus Cynodictis.

The present study is the first description of the Amphicyonidae from the Ergilin Dzo Formation. Although Matthew and Granger (1923) mentioned that an anterior part of a lower jaw with c-p2 from Ardyn Obo (=Ergilin Dzo locality) could be referable to the genus Cynodictis, they did not provide any other information concerning the specimen such as a precise description, figures, or a specimen number. There is a pos- sibility that Cynodictis? sp. (and cf. Cynodictis) listed in the Ergilin Dzo locality by Matthew and Granger (1923, p. 1–2), Dashzeveg (1974, p. 77, fig. 1), and Russell and Zhai (1987, p. 273) might belong to the same species as MPC-M 30/63.

Cynodictis is an amphicyonid mainly known from the late Eocene and early Oligocene of Europe (Kotsakis, 1980). Remains of Cynodictis, C. minor and aff. Cynodictis sp., have been also reported from two middle Oligocene localities in Kazakhstan (Yanovskaya, 1970; Gabunia, pers. comm., 1981 cited in Russell and Zhai, 1987). Kotsakis (1980) made a systematic revision of the genus so that it consists only of the European species, and suggested that the excluded Asian forms are possible immigrants of primitive carnivorans from North America.

Most of the Asian Cynodictis materials have only appeared in faunal lists; thus, it is difficult to discuss their taxonomic validity. Huang (1993) provided a figure and photos of lower premolars of Cynodictis? elegans. Judging from the absence of the posterior accessory cusp on p3 and of a diastema between p2 and p3, these specimens seem more similar to Amphicticeps. This genus is not an amphicyonid but a genus of the basal Ursida (Wang et al., 2005). Cynodictis mongoliensis described from the Tatal Gol Member of the Hsand Gol Formation (Yanovskaya, 1970) has been synonymized with Amphicynodon teilhardi (Russell and Zhai, 1987), thereby removing it from the Amphicyonidae to the basal Ursida (Wang et al., 2005). These two cases agree with the suggestion by Kotsakis (1980) that some of the East Asian Cynodictis materials may belong to neither Cynodictis nor amphicyonids but rather to basal arctoids. Matthew and Granger (1924) reported Cynodictis? elegans and C.? constans from the Tatal Gol Member of the Hsand Gol (=Shand Gol) Formation (Hsandagolian=early Oligocene); however, Hunt (1998b) identified these specimens as belonging to aeluroid carnivorans, and established a new genus, Shandgolictis, for C.? constans.

At present, the new material, which has amphicyonid upper molar characteristics, confirms the presence of an amphicyonid in the Eocene of Asia, and its morphological differences from other amphicyonids suggest the presence of a new amphicyonid. In addition, this opens questions on the previously reported presence and migration of Cynodictis in the Paleogene of Asia.