Amphicticeps and Amphicynodon (Arctoidea, Carnivora) from Hsanda Gol Formation, Central Mongolia and Phylogeny of Basal Arctoids with Comments on Zoogeography

Home , Arctoidea, Cephalogale, Hsanda Gol Formation, Miacidae, Palaeogale, Plesictis

PUBLISHED BY THE AMERICAN MUSEUM OF NATURAL HISTORY CENTRAL PARK WEST AT 79TH STREET, NEW YORK, NY 10024 Number 3483, 57 pp., 17 ®gures, 8 tables July 25, 2005

XIAOMING WANG,1 MALCOLM C. McKENNA,2 AND DEMBERELYIN DASHZEVEG3

ABSTRACT

Amphicticeps shackelfordi and Amphicynodon teilhardi are two small carnivorans from the early Oligocene Hsanda Gol Formation of central Mongolia, and as basal arctoids (infraorder Arctoidea) in Asia, feature unique combinations of morphologies that offer insights into early diversi®cation and zoogeography of the arctoids. Lack of adequate study of Amphicticeps and incomplete knowledge about Amphicynodon, however, prevented them from being ®gured in the discussions of arctoid relationships. New associated dental and cranial materials collected during recent expeditions in the 1990s substantially enrich our knowledge of the two genera and their stratigraphic positions, and serve as an impetus for a study of their phylogenetic relationships in the broad perspective of basal Arctoidea. Hsanda Gol arctoids are represented by six small- to medium-sized species: Amphicticeps shackelfordi Matthew and Granger 1924, A. dorog, n.sp., A. makhchinus, n.sp., Amphicynodon teilhardi Matthew and Granger 1924,? Cephalogale sp., and Pyctis inamatus Babbitt, 1999. The three species of Amphicticeps apparently form an endemic clade con®ned to central Asia, whose zoogeographic origin is currently unknown. Amphicynodon has a much higher diversity in Europe than in Asia, and phylogenetically the Asian A. teilhardi seems to be nested within

1 Division of Paleontology, American Museum of Natural History; Department of Vertebrate Paleontology, Natural History Museum of Los Angeles County, 900 Exposition Blvd., Los Angeles, California 90007; Institute of Vertebrate Paleontology and Paleoanthropology, Chinese Academy of Sciences ([email protected]). 2 Division of Paleontology, American Museum of Natural History ([email protected]). 3 Geological Institute, Mongolian Academy of Sciences, Ulaanbaatar, Mongolia.

the European congeneric species, indicating an eastward dispersal for this group, linking the European ``Grande Coupure'' and the Asian ``Mongolian Reconstruction'' events. To avoid excessive homoplasies in crown groups, we attempted a phylogenetic reconstruction based mostly on stem arctoids. Twenty genera of primitive arctoids occupying basal positions of nearly all major clades are selected for the analysis. The resulting tree, based on 39 characters, approximates the initial divergence of the arctoids. The traditionally dichotomous Arctoidea, formed by sister clades Ursida and Mustelida, is recovered in our analysis. Mustelida is also largely dichotomous with mustelid-like forms on one side and procyonid- like forms on the other. Despite its rather hypercarnivorous dentition, Amphicticeps is found on the Ursida side of the arctoids, although support for such a topology is relatively weak. Amphicynodon is a stem taxon of the Ursida and is a sister to an ursid±pinniped clade.

INTRODUCTION ticeps shackelfordi found by the Mongolian Academy of Sciences and the American Mu- Among the intriguing discoveries made seum of Natural History joint expeditions during the Central Asiatic Expeditions by the (MAE) during the 1994 ®eld season. The American Museum of Natural History in the new MAE materials collected throughout the 1920s were several small carnivorans found 1990s complement in important ways the in the rich deposits of Shand Gol (preferred original holotype skull by ®lling in gaps in spelling over ``Hsanda Gol'' of most previ- our knowledge of the upper and lower teeth ous literature) and Tatal Gol in the Mongo- missing in the holotype, furnish additional lian People's Republic. The initial discovery materials that indicate the existence of two of fossiliferous localities near Shand Gol (®g. more species, and help to resolve some long- 1) in 1922 led to a number of preliminary standing problems that have plagued pale- reports about a fauna that was almost entirely ontologists for many years. new to science (see Mellett, 1968, for a sum- Also among the newly collected materials mary). Amphicticeps shackelfordi Matthew are important additions to Amphicynodon teil- and Granger 1924, a small badger-sized car- hardi Matthew and Granger 1924, another nivoran, was brie¯y reported as such and be- basal arctoid that has a phylogenetic af®nity came part of a diverse Shand Gol ``fauna'' with some congeneric species from European that serves as a regional standard in the mid- early Oligocene ®ssure ®lls in the Quercy re- Tertiary of continental east Asia (Russell and gion of France. A. teilhardi was originally Zhai, 1987, and references within). based on a lower jaw fragment with m1±2, At the time of its formal announcement, and our new materials combine to include the holotype of Amphicticeps shackelfordi much of the skull and lower jaw. With such was the only specimen ®gured and brie¯y improved knowledge about the anatomy of described, and was represented by a nearly Amphicticeps and Amphicynodon, along with complete skull (Matthew and Granger, 1924). better documentation of the locality and strati- It stood out as one of the best specimens graphic data by the MAE ®eld parties, a sub- among half a dozen small carnivorans ini- stantial contribution is now possible to resolve tially reported from the Shand Gol and Tatal some long-standing systematic issues. Gol localities. Although several mandibles of Matthew and Granger (1924: 4) consid- Amphicticeps-sized individuals were collect- ered Amphicticeps so out of place in the then ed during the 1922 ®eld season (additional existing understanding of primitive carnivor- materials were also collected in 1925), Mat- ans that they originally regarded it as ``a thew and Granger did not list the isolated jaw highly progressive miacid'' rather than be- fragments, mentioning only the morphology longing to any existing family of carnivor- of the lower carnassials, presumably partly ans. Its puzzling array of mixed dental and because of a lack of a secure association of basicranial morphology seems to suggest to these mandibles with the holotype skull. recent authors (Schmidt-Kittler, 1981; Wol- This lack of association is now remedied, san, 1993; Hunt, 1996b; Wang and Qiu, more than half a century later, by naturally 2003a) a relationship of either musteloids or associated upper and lower jaws of Amphic- ursoids, two major clades of carnivorans that 2005 WANG ET AL.: CARNIVORANS FROM CENTRAL MONGOLIA 3

Fig. 1. Map of early Oligocene localities (solid circles) of basal arctoids in Mongolian People's Republic and People's Republic of China. Map locations of fossil localities are based on Russell and Zhai (1987). Amphicticeps and Amphicynodon were reported from the Khatan-Khayrkhan locality in the Altai region of western Mongolia (Russell and Zhai, 1987: 324). However, no description of the materials was given and their occurrence in this locality needs to be con®rmed. include all living families of Arctoidea (Ur- that features key anatomical innovations in sidae, Procyonidae, Mustelidae, Ailuridae, various families of the Arctoidea (e.g., Hunt, plus the Phocoidea [ϭ Pinnipedia] clade). 1974, 1977; Schmidt-Kittler, 1981; Cirot and With its occurrence in the early Oligocene, a Bonis, 1993; Wolsan, 1993; Wang, 1997), is critical period of time when some European still not available. With the new materials basal arctoids began to make their ®rst ap- and revised stratigraphic framework in hand, pearance, Amphicticeps seems to have fallen the time has come to address the various is- somewhere in the initial diversi®cation of the sues outlined above. We aim to accomplish arctoids and is thus relevant in discussions of the following three objectives in this contri- higher level relationships of the Arctoidea. bution: (1) to restudy the holotypes and to However, despite an exceptionally pre- describe new materials of Amphicticeps served holotype skull, Amphicticeps has not shackelfordi and Amphicynodon teilhardi; received more than a cursory mention or a (2) to place on record two new species of few speculative remarks. A detailed descrip- Amphicticeps from the Hsanda Gol Forma- tion of its basicranial morphology, a region tion, as well as more fragmentary materials 4 AMERICAN MUSEUM NOVITATES NO. 3483 of uncertain taxonomy; and (3) to conduct a Although we did not have full access to the phylogenetic analysis of the arctoids from PIN materials, casts of key specimens are Hsanda Gol Formation in relation to major available to us. The Polish±Mongolian Pa- clades of Arctoidea. leontological Expeditions in the 1960s also amassed a collection in Warsaw. Carnivores INSTITUTIONAL ABBREVIATIONS from the ZPAL collection were described by AMNH Department of Vertebrate Paleontol- Lange-BadreÂ and Dashzeveg (1989: 141), ogy, American Museum of Natural and only a few jaw fragments were relevant History, New York in the present study. New materials collected BSP Bayerische Staatssammlung fuÈr Pa- by the Mongolian Academy of Sciences and laÈontologie und historische Geologie, the American Museum of Natural History Munich joint expeditions (MAE) add substantially to F:AM Frick Collection, Department of Ver- various anatomic components to be described tebrate Paleontology, American Mu- below, and also furnish a modern stratigraph- seum of Natural History, New York FSP Collections of FaculteÂ des Sciences, ic context. In 1995±1997, a joint Austrian± Laboratoire de GeÂobiologie, Bio- Mongolian team also made a collection in the chronologie et PaleÂontologie Humai- nearby Taatsiin (Tats) Gol area, particularly ne de l'UniversiteÂ du Poitiers, Poitiers by screen washing for small mammals IVPP Institute of Vertebrate Paleontology (Daxner-HoÈck et al., 1997; Daxner-HoÈck, and Paleoanthropology, Chinese 2000, 2001; Erbajeva and Daxner-HoÈck, Academy of Sciences, Beijing 2001; Daxner-HoÈck and Wu, 2003) but large MAE Collections of the joint Mongolian mammals were also collected (Vislobokova Academy of Sciences±American Mu- and Daxner-HoÈck, 2002). The carnivore ma- seum of Natural History Paleontolog- terials are currently being studied by Nagel ical Expeditions, currently housed in the American Museum of Natural and Morlo (2001, 2003), who have kindly History, New York supplied us with two casts. Various groups MM MuseÂum d'Histoire naturelle de Mon- of mammals from the new MAE collection tauban, Montauban are under study by specialists, a few of which MNHN Institut de PaleÂontologie, MuseÂum have been published (Bryant and McKenna, National d'Histoire Naturelle, Paris 1995; Kellner and McKenna, 1996; Babbitt, NMB Naturhistorisches Museum Basel, Ba- 1999; Wang, 2001; Geisler, 2004). sel The following specimens of North Amer- PIN Institute of Paleontology, Russian ican primitive musteloids, referred to as Oli- Academy of Sciences, Moscow gobuninae (Baskin, 1998a), are examined in PST Paleontology±Stratigraphy Section of Mongolian Academy of Science, this study: Oligobunis crassivultus: AMNH Ulaanbaatar 6903, 6906, Turtle Cove Member of John UCMP University of California Museum of Day Formation, early Arikareean; Promartes Paleontology, Berkeley cf. olcotti: F:AM 27584, 27587, 27589, YPM-PU Princeton University Collection of Marsland Formation, early Hemingfordian; Peabody Museum of Natural History, Zodiolestes daimonelixensis: F:AM 27598, Yale University, New Haven 27599, 27600, Harrison Formation, late Ari- ZPAL Institute of Paleobiology, Polish kareean; Megalictis ferox: F:AM 25430, Academy of Sciences, Warsaw Marsland Formation, early Hemingfordian; AMNH 12880, upper part of Rosebud For- MATERIALS AND METHODS mation, late Arikareean. Hsanda Gol materials collected during the For European primitive musteloids, Wol- American Museum Central Asiatic Expedi- san's (1993) character matrix forms the prin- tion in the 1922 and 1925 seasons form an cipal database for comparison. Where pos- important basis of the present study. The So- sible, his observations are checked against viet Academy of Sciences 1946±1949 Ex- casts or actual specimens of the following peditions obtained a signi®cant collection of taxa available in the AMNH collection and small carnivorans, particularly Amphicyno- elsewhere: Simocyon primigenius: IVPP don (Janovskaja, 1970), housed in the PIN. V12162; Pseudobassaris riggsi: YPM-PU 2005 WANG ET AL.: CARNIVORANS FROM CENTRAL MONGOLIA 5

11455; Mustelavus priscus: AMNH 129168, We rely on Cirot (1992) for a recent synthe- YPM-PU 13775; Broiliana nobilis: BSP sis on Amphicynodon, supplemented by our 1937 II 13524 (cast); Stromeriella franconi- own personal observations of the new FSP ca: BSP 1937 II 13533 (cast); Potamother- collection of specimens of this genus, partic- ium valletoni: AMNH 11003, AMNH 22520, ularly the well-preserved cranial and dental and uncataloged basicranial materials from materials (FSP ITD 876, 312, 60, 569, 356) NMB. For most of the skull and dental char- of A. leptorhynchus from Itardies, as well as acters, Wolsan's observations can also be our own casts of some critical specimens. We veri®ed by published descriptions: Simocyon were able to study a nearly complete skull of primigenius in Beaumont (1964) and person- Pachycynodon boriei in the FSP (uncata- al observations; Angustictis (Plesictis mayri) loged) collection and a right jaw (NMB QB in Dehm (1950); Bavarictis gaimersheimen- 357). We also have access to casts of two sis in MoÈdden (1991); Franconictis (Plesictis mandibles of P. ®lholi (NMB QB 268 and humilidens) in Dehm (1950); Mustelictis piv- 451). We examined UCMP 24106, holotype eteaui in Lange-BadreÂ (1970); Mustelictis and only specimen of Allocyon loganensis. olivieri in Bonis (1997); Plesictis genettoides in Helbing (1930) and Dehm (1950); Para- STRATIGRAPHIC FRAMEWORKS gale huerzeleri in Petter (1967); Plesiogale angustifrons in de Beaumont (1968b) and Little distinction was made in the strati- Helbing (1930). Basicranial morphologies of graphic relationships of the historical AMNH some of these musteloids are gleaned from localities when the geology was being worked works by de Beaumont (1968a), Schmidt- out by C. P. Berkey and colleagues. Nor was Kittler (1981), Cirot and de Bonis (1993), this a major concern at a time when the pri- and Wolsan (1993), as well as from direct mary objective was to secure the best possible observations of the above specimens in the mammalian fossil collections. The concept of AMNH. A recent acquisition of a nearly the Hsanda Gol Formation was originally pro- complete skull of Mustelavus in the AMNH posed by Berkey and Granger (1923; more (still undescribed), complemented by earlier elaborated in Berkey and Morris, 1927) to en- descriptions of its holotype (Scott and Jep- compass strata with Oligocene fossil mam- sen, 1936; Clark, 1937), permits the inclu- mals, but also included rocks as old as Cre- sion of this primitive, presumably very basal taceous due to erroneous extrapolations of musteloid in our analysis. A basal leptarctine ®eld geology. The Hsanda Gol Formation is mustelid, Kinometaxia, was recently found in now known to overlie the Eocene Kholobol- the early Miocene Tabenbuluk area (Wang chi and Elegen formations (Tsagaan Ovoo and Qiu, 2003b; Wang et al., 2004) and its Formation of HoÈck et al., 1999) and underlie inclusion in this analysis permits a better the Oligocene/Miocene Loh Formation (Mc- sense of the Mustelinae clade. Kenna, 1995; HoÈck et al., 1999). Relationships of basal ursoids have not A complete revision of the biostratigraphy been worked out in detail. We use Cephalo- and chronology of the Shand Gol and Tatal gale as a basal form of the Ursidae (e.g., Gol areas is underway (McKenna, 1995; Mc- Beaumont, 1965; Ginsburg and Morales, Kenna et al., MS). Relocation of classic lo- 1995; Hunt, 1998c), and we examined the calities and improvements in the documenta- primitive species C. minor based on the re- tions of new collections help to establish a cent FSP collection from the Pech du Fraysse new stratigraphic framework. The importance locality of the Quercy district in late Oligo- of this contextual information becomes more cene (Paleogene biochronological level MP apparent in light of revelations that more than 28). A loosely de®ned Amphicynodontidae one fossil-bearing horizon can be recognized has been in circulation as a primitive group (see also HoÈck et al., 1999), in contrast to the of ursoids that may have given rise to pin- traditional practice of lumping all specimens nipeds (Tedford et al., 1994; Baskin and Ted- from near Shand Gol and Tatal Gol as rep- ford, 1996; Hunt, 1996b, 1998c; Wang and resenting a single assemblage (Mellett, 1968). Qiu, 2003a), and it generally includes Am- Within the newly de®ned Hsanda Gol For- phicynodon, Pachycynodon, and Allocyon. mation, the reddish brown mudstones are di- 6 AMERICAN MUSEUM NOVITATES NO. 3483 vided by a prominent basaltic lava into lith- 1987; Lange-BadreÂ and Dashzeveg, 1989; ologically recognizable upper and lower Wang, 1992). A broader faunal analysis by parts. The lower Tatal Member includes sed- Meng and McKenna (1998) further proposed iments below the lava whereas the upper the existence of a ``Mongolian Remodeling'' Shand Member includes strata above it corresponding to the European ``Grande (Dashzeveg, 1996). Faunally, two units can Coupure'', although whether this is linked to be differentiated, in contrast to the traditional climatic change is still being debated (e.g., assumption of a single faunal assemblage in Tsubamoto et al., 2004). Such a downwardly the Hsanda Gol Formation, but the two fau- shifted chronological framework is consis- nal units do not exactly correspond to the tent with the above radiometric dates and two lithological members separated by the with the new North American Eocene±Oli- lava. Fossils from beneath the lava and those gocene boundary between the Chadronian a few meters above it belong to the Ulaan and Orellan. The Orellan land mammal age Khongil fauna characterized by rodents such was correlated by Mellett (1968) to the com- as Cricetops dormitor, Karakoromys, and Se- posite fauna collected from the Hsanda Gol lenomys, whereas above it is the Zavlia fau- Formation. In terms of the European conti- na, which is characterized by Yindirtemys nental stratigraphic framework, small carni- and Tachyoryctoides (McKenna, 1995). This vorans from the Hsanda Gol Formation, such scheme roughly corresponds to the rodent as Amphicynodon, Palaeogale, Stenoplesic- Biozones A and B of HoÈck et al. (1999: ®g. tis, and so forth, are comparable to those in 22), although their Biozone B extends to the the early part of the Rupelian after the top of the lava and their Hsanda Gol/Loh for- Grande Coupure event in MP21 (Schmidt- mation boundary is postulated to be time Kittler, 1989; Berggren et al., 1992; LeÂveÃque, transgressive. 1993). Absolute age for the lower of the two The name Shand Gol is the preferred spell- Hsanda Gol faunal levels is constrained by ing over ``Hsanda Gol'' in much of the pre- radioisotopic dates of the basaltic lava within vious literature for a geographic feature. In the Hsanda Gol Formation. The lava yielded the text below, we use Hsanda Gol Forma- a whole-rock potassium±argon date of 31.2± tion for formal reference to an established 32.0 Ma (Evernden et al., 1964: 193), which geologic unit, but use Shand Gol to refer to after correction (see Dalrymple, 1979), the geographic area in the vicinity of the would be 32.0±32.8 Ma. Slightly younger Shand Gol drainage. Our usage of the Hsan- dates of 28 Ϯ 1.1 and 30 Ϯ 1.1 Ma have da Gol area or Hsanda Gol carnivorans refers been obtained by Gabuniya and Rubinshtein to the larger area that includes Shand Gol (1975). Most recently, HoÈck et al. (1999) ob- and Tatal Gol, as well as other localities in tained an 40Ar/39Ar age of around 31.5 Ma the Hsanda Gol Formation. (with a range of 30.4±32.1 Ma) for their Ba- salt I within the Hsanda Gol Formation, and SYSTEMATIC PALEONTOLOGY around 28 Ma (with a range of 27±29 Ma) ORDER CARNIVORA BOWDICH, 1821 for their Basalt II near the bottom of the overlying Loh Formation. SUBORDER CANIFORMIA KRETZOI, 1943 Following a recent shift of the terrestrial INFRAORDER ARCTOIDEA FLOWER, 1869 Eocene-Oligocene boundary in North Amer- ica (Swisher and Prothero, 1990; Prothero PARVORDER URSIDA TEDFORD, 1976 and Swisher, 1992) and elsewhere in the SUPERFAMILY URSOIDEA FISCHER DE world, Ducrocq's (1993) analysis of Paleo- WALDHEIM, 1817 gene Asian faunal turnover and Hunt and Tedford's (1993) study of fossil carnivorans Amphicticeps Matthew and Granger, 1924 suggest that the composite Hsanda Gol fau- TYPE SPECIES: Amphicticeps shackelfordi nas correspond to the Early Oligocene in- Matthew and Granger, 1924. stead of Middle Oligocene as traditionally INCLUDED SPECIES: Amphicticeps shackel- recognized (e.g., Mellett, 1968; Kowalski, fordi Matthew and Granger 1924, A. dorog, 1974; Li and Ting, 1983; Russell and Zhai, n.sp., and A. makhchinus, n.sp. 2005 WANG ET AL.: CARNIVORANS FROM CENTRAL MONGOLIA 7

EMENDED DIAGNOSIS: Amphicticeps pos- damentally arctoid basicranium of Amphic- sesses the following derived characters that ticeps and its musteloid-like molar reduction distinguish it from basal ursoids and muste- (transversely elongated M1) and short ros- loids such as Amphicynodon, Pachycynodon, trum. However, he did not consider it a mus- Cephalogale, Mustelavus, Amphictis, Bavar- teloid because of its shallow suprameatal fos- ictis, Pseudobassaris, Mustelictis, and Bro- sa, a character especially emphasized in his iliana: broad and short rostrum, short infra- analysis of musteloid phylogeny. The form orbital canal, enlarged M1 parastyle, small of its M1 seemed to him to be another ob- angle between labial borders of P4 and M1, stacle to recognizing it as a musteloid. Spe- reduced and lingually positioned M2, re- ci®cally, he regarded the somewhat swollen duced m2, and extremely reduced or lost m3. buccal border and a ``knoblike'' (hoÈckerar- It is primitive compared to Kinometaxia, tige) lingual cingulum of the M1 as atypical Paragale, Plesiogale, and other mustelids in of a musteloid. He therefore regarded Am- its possession of a carnassial notch on P4 and phicticeps as a ``basal arctoid'' prior to the a shallow suprameatal fossa. In contrast to emergence of the musteloid clade. Wolsan the North American oligobunines, Amphic- (1993) compared its lingually located M2 ticeps possesses a postprotocrista on the M1, with those of Potamotherium, but did not lacks of a lingual notch on the m1 entoconid draw de®nite conclusions. Hunt (1996b, crest, and has reduced M2 and m2. 1998c) suggested that Amphicticeps may be DISTRIBUTION AND AGE: Hsanda Gol For- an amphicynodontid possibly ancestral to the mation, Tsagan Nor Basin, eastern Valley of North American Allocyon and Kolponomos, Lakes, central Mongolian People's Republic. a suggestion that was followed by Wang and Early Oligocene (see more comments in Ge- Qiu (2003a). ology and Age under Amphicticeps shackel- Among the small carnivorans from the fordi). Hsanda Gol Formation, Amphicynodon teil- COMMENTS: Ever since its original descrip- hardi (Matthew and Granger, 1924), founded tion, Amphicticeps shackelfordi has remained on a few jaw fragments (see description be- something of an enigma in its phylogenetic low), is the only similar-sized arctoid that relationships. Offering no formal classi®ca- may potentially be confused with Amphicti- tion, Matthew and Granger (1924: 4) initially ceps (other Hsanda Gol carnivorans, such as remarked that ``it has the sharply reduced Stenoplesictis, Palaeogale, and Viverravus, post-carnassial dentition of [stenoplesictoids] are easily distinguished on the basis of their with the short, heavy precarnassial dentition far more trenchant carnassials that are typical of [cynodontoids]. It is not close to any one of feliforms; Hunt, 1998b). With the bene®t genus with which I [sic] have made compar- of the more complete materials for both Am- isons and might be regarded as a highly pro- phicticeps and Amphicynodon, these two gressive miacid rather than as a member of primitive Shand Gol arctoids are contrasted any of the existing families of ®ssiped Car- in table 1 to facilitate identi®cation. nivora.'' Subsequent classi®cations also re- ¯ect this ambiguity; Simpson (1945: 110 and Amphicticeps shackelfordi Matthew and 115) listed it under both ``?Amphicynodon- Granger, 1924 tinae incertae sedis'' and ``?Stenoplesictinae Figures 2±7; Tables 2±4 incertae sedis'', whereas Piveteau (1961: 721) considered it as incertae sedis but com- HOLOTYPE: AMNH 19010, nearly compared it to Cynodon (ϭ Amphicynodon). plete skull with left and right P1±2, P4±M1, Without suggesting a taxonomic position for and alveoli of left and right C1, P3, and M2 the genus, Bonis (1971) commented on its (Matthew and Granger, 1924: ®gs. 4±5). ``parallel'' resemblance to Harpagophagus,a TYPE LOCALITY: Originally designated as genus based on a single left M1 and thought from ``Hsanda Gol formation, Loh'' (Mat- to be an amphicyonid. thew and Granger, 1924: 4), AMNH 19010 In the ®rst substantial discussion of Am- (®eld no. 89) was collected from about 2 mi phicticeps since its original description, southwest of the Loh campsite, in Tsagan Schmidt-Kittler (1981) pointed out the fun- Nor Basin, eastern Valley of Lakes, Obor- 8 AMERICAN MUSEUM NOVITATES NO. 3483

TABLE 1 Contrast of Two Genera of Small Primitive Arctoids in the Hsanda Gol Formation

Amphicynodon Amphicticeps Rostrum not shortened shortened Rostrum narrow broad Distance from postorbital processes to postorbital constriction short long C1 slender long M1 parastyle reduced robust M2 labially positioned enlarged Lower premolars slender lingually positioned m1 hypoconid low crowned high crowned m1 talonid basin crenulated not crenulated m2 protoconid/metaconid widely separated closely spaced m3 always present vestigial/absent

Khangay Province, in north-central Mongo- mentioned in the early Oligocene Khatan- lia. Khayrkhan locality of Altai Province of REFERRED SPECIMENS: AMNH 19017, par- Mongolia by Russell and Zhai (1987: 324). tial left ramus with p2±m1 and alveoli of c1± GEOLOGY AND AGE: The above referred p1 and m2, ®eld no. 69, from Loh; AMNH specimens of Amphicticeps shackelfordi 19127, partial right ramus with p3, m1, and come from three localities (some specimens alveoli of c1±p2, p4, and m2, ®eld no. 84, lack a detailed locality record): (1) general from Loh; AMNH 19128, right ramal frag- vicinity of Loh for AMNH 19017 and 19127; ment with m1 and alveoli of c1±p4 and m2, (2) 2 mi southwest of Loh for AMNH 19010, ®eld no. 92, from 2 mi southwest of Loh; AMNH 19128, MAE BU.91.9187±90, and AMNH 21695, left ramal fragment with m1 MAE M-217; (3) general vicinity or 2 mi and alveoli of c1±p4, ®eld no. 536, from 2 west of the Ulaan Khongil (``Grand Canyon'' mi west of the ``Grand Canyon'' area, which or Tatal Gol) for AMNH 21695 and 83610. is 10 mi west of Loh; AMNH 83610, left While ®eld studies are currently pursued ramal fragment with p4±m2 and m3 alveo- by the on-going joint expeditions of the lus, ®eld no. 531, ``Grand Canyon''; AMNH MAE and formal stratigraphic revisions will 81336, left ramal fragment with m2 and m3 have to wait for that result (see HoÈck et al., alveolus; AMNH 85749, right ramal frag- 1999, for a recent summary), it is relevant to ment with m2±3, ®eld no. 548; MAE note here that the above three Amphicticeps- BU.91.9187±90 (AMNH cast 129686), par- producing localities fall within a more re- tial rostrum with right alveoli of C1±P2, right stricted concept of the Hsanda Gol Forma- P3±M2, and left P3±M1, partial mandible tion close to the level of a discontinuous but with left p2, p4±m2, and alveoli of c1±p1 approximately contemporary basaltic lava and p3, and right p1±m2 and root of m3, (Basalt I of HoÈck et al., 1999). collected by PerleÂ Altangerel in 1994, ®eld Historic collections are largely concentrat- no. M-152, from 2 mi southwest of Loh; ed in the Ulaan Khongil fauna in the lower MAE SG.95.7518, right ramal fragment with part of the Hsanda Gol Formation below the p2, p4, and m1 (broken); MAE SG.95.8919, prominent basaltic lava and immediately posterior skull fragments with top of the in- above. Amphicticeps specimens from near ion and left basicranial region, 45Њ17Ј49ЉN the Loh campsite and 2 mi southwest of Loh 101Њ37Ј13ЉE, collected by Khosbayar on 10 (including the holotype) are darkly stained August 1995; and MAE M-217, isolated left due to the percolation of ground water, and m1, ®eld no. M-217, from 2 mi southwest of they all belong to the Ulaan Khongil fauna. Loh. AMNH 21695 and 83610 from near the DISTRIBUTION: Early Oligocene of north- ``Grand Canyon'' area, on the other hand, are central Mongolia. An undescribed record was light-colored and may belong to the Zavlia 2005 WANG ET AL.: CARNIVORANS FROM CENTRAL MONGOLIA 9

Fig. 2. Skull of Amphicticeps shackelfordi, AMNH 19010, holotype. A, lateral, B, ventral, and C, dorsal views. Scale ϭ 20 mm. 10 AMERICAN MUSEUM NOVITATES NO. 3483

Fig. 3. Partial skull of Amphicticeps shackelfordi, referred specimen, MAE BU.91.9187±90. A, lateral and B, dorsal views. Scale ϭ 10 mm. fauna in the upper part of the Hsanda Gol and Granger (1924: ®g. 5) is mostly accurate Formation above the lava. This presumed in overall proportions except for a more pos- younger age of AMNH 21695 and 83610 rel- teriorly displaced mastoid process (relative to ative to the rest of the A. shackelfordi hy- the nuchal crest) in the dorsal view. podigm is also consistent with the former's For a small carnivoran, the skull is rather wider m1 and more prominent lingual cin- strongly built, with a short and broad ros- gulum on the m1 trigonid, tendencies that in- trum. The incisor-bearing part of the premax- dicate a slightly more advanced stage of evo- illary is broken off and only the posterior lution for the species. processes of the premaxillary between the EMENDED DIAGNOSIS: Amphicticeps shack- nasal and maxillary are preserved; they ex- elfordi is distinguishable from the more de- tend slightly behind the level of the P2. The rived A. makhchinus and A. dorog by its posterior tip of the nasal reaches nearly to smaller size, smaller angle between the labial the level of the postorbital process of the borders of P4 and M1, more enlarged M1 frontal. In keeping with the broad snout, the parastyle, larger M1 metaconule, more re- frontal shield is also wide, that is, there is a duced anterior cingulum of M1, and more long distance between upper rims of the or- lingually located M2. In addition, the P4 pro- bits. There is a small fossa above the antor- tocone of A. shackelfordi is larger than in A. bital rim on the frontal/maxillary suture, for dorog, but is less well developed than in A. the insertion of the levator nasolabialis, and makhchinus. this fossa is more prominent on the right side DESCRIPTION: Matthew and Granger's of the holotype. The postorbital process of (1924) original report of Amphicticeps shack- the frontal is small but rather sharply point- elfordi consisted of a brief diagnosis only. A ed; that on MAE BU.91.9187±90 is more re- full description is furnished here for the ho- duced. The distance between the postorbital lotype and the newly referred materials. constriction and the postorbital process is rel- Skull (®gs. 2±4): The holotype, AMNH atively elongated (the postorbital constriction 19010, is still the only nearly complete skull is disjointed in the type but enough is pre- available, although additional referred cranial served on the right side to indicate this elon- fragments supplement the holotype in a num- gation), and is approximately 12 mm, as is ber of important ways. Its rostral part is also seen in Potamotherium and Paragale. slightly crushed, such that the left cheek re- The temporal crests merge into the sagittal gion is uplifted by approximately 3 mm. The crest slightly behind the postorbital constric- reconstructed skull illustrated by Matthew tion. The braincase is not laterally expanded 2005 WANG ET AL.: CARNIVORANS FROM CENTRAL MONGOLIA 11

Fig. 4. Posterolateral view of the orbital region of Amphicticeps shackelfordi, MAE BU.91.9187± 90, referred specimen. C1 al., partial alveolus of upper canine; Fr., frontal; Inf. C., infraorbital canal; Ju.-Max. s. s., jugal-maxillary suture surface; Lac., lacrimal; Lac. f., lacrimal foramen; Max., maxillary; Pal., palatine; Sph.-pal. f., sphenopalatine foramen. near the postorbital constriction as in Pota- ¯oor that opens into the dorsal wall of the motherium or nearly becoming so in Para- infraorbital canal. A slender process of the gale. Although not very high, the sagittal palatine meets the lacrimal and excludes or- crest is thick and robust; so is the nuchal bital contact of the frontal with the maxillary. crest. The temporal region of the skull has a At the palatine±maxillary±lacrimal junction, rugose surface texture. In lateral view, the there is a small, oval fenestra, probably due skull is somewhat shallow and has a rather to lack of ossi®cation at this stage of the on- ¯at forehead. togeny, although a fossa for inferior oblique The anterior half of the right orbital region muscle in hyaenids has been identi®ed at the is well preserved on MAE BU.91.9187±90 same triple junction (Werdelin and Soloun- (®g. 4). The infraorbital canal is short, about ias, 1991: ®g. 26). The anterior process of 3 mm long, and has a round cross section. the jugal is broken away in both AMNH Immediately above the canal is a small, 19010 and MAE BU.91.9187±90, leaving rounded lacrimal bone forming the inner rim the jugal-maxillary suture surface well ex- of the antorbital rim. The lacrimal foramen posed in both specimens. From these sutures, on the lacrimal bone opens posterodorsally. it can be deduced that the anterior tip of the About 1 mm into the ori®ce for the lacrimal jugal stops just above the infraorbital canal sac, there is a small foramen on the ventral and does not reach the lacrimal bone. 12 AMERICAN MUSEUM NOVITATES NO. 3483 2005 WANG ET AL.: CARNIVORANS FROM CENTRAL MONGOLIA 13

Basicranium (®g. 5): The occipital con- dal direction, and is excavated slightly to- dyles are broken off on both sides of the ho- ward the ventrolateral aspect such that it be- lotype. The remaining basioccipital ¯oor begins to be hidden by a thin bony rim, al- tween the bullae is distinctly widened pos- though the degree of excavation is far less teriorly, such that the lateral edges of the ba- than seen in some procyonids and mustelids. sioccipital form a 25Њ angle, in contrast to Although both bullae are missing, the smaller angles in primitive arctoids such as presence of an ectotympanic bulla is indicat- Amphictis and to nearly parallel (0Њ) edges in ed by a clearly de®ned scar posterior to the canids. A small, rounded process for the at- postglenoid foramen and by a broad, smooth tachment of the rectus capitis ventralis mus- depression on the alisphenoid/squamosal su- cle lies close to the lateral edges of the ba- ture (fused) area just medial to the postglen- sioccipital and is slightly in front of the pos- oid process. Such surface markings leave lit- terior lacerate foramen. Both glenoid fossae tle doubt as to where the anterior crus of the are missing. On the left side, however, the ectotympanic ring was attached (also see the medial segment of the postglenoid process is description under Amphicticeps dorog for the still preserved. Behind this broken process is preserved ectotympanic). A broad facet fac- a small postglenoid foramen, 1 mm in di- ing anteroventrally between the posterior lac- ameter. erate and stylomastoid foramina at the base The mastoid part of the petrosal is in¯ated, of the paroccipital process is apparently the forming a prominent, laterally protruding site of attachment of the posterior crus of the mastoid process. The process has a smooth ectotympanic. The presence of an entotym- and ¯at lateral facet and is connected to the panic, on the other hand, is indicated by a 2- paroccipital process via a posterior ridge and mm-wide rugose area on the ventral surface to the lambdoidal crest via a more prominent of the promontorium immediately lateral to dorsal blade. Such a blade can also be seen the petrosal/basioccipital juncture. It is not in most North American oligobunines. The possible to ascertain whether a bony external mastoid tubercle (processus hyoideus) is auditory meatus was present. However, the formed by the petrosal. The posteriorly di- rather distinct mark of the above mentioned rected paroccipital process is broadly based ectotympanic attachment behind the post- because of its expanded wings on each side, glenoid foramen suggests that the anterior but shows no sign of fusion with the bulla crus of the ectotympanic does not wrap (not preserved) at the base. There is a low, around to superimpose on the squamosal longitudinally oriented ridge on the ventral around the dorsal bony passage of the meatus surface of the paroccipital process. to form a complete ring by the ectotympanic, In front of the mastoid process is an oval- as happens in many arctoids that have a tu- shaped suprameatal fossa; its long axis is bular external bony auditory meatus. transversely oriented. The fossa is not fully The promontorium of the petrosal is prom- enclosed toward the medial side, and is thus inently domed ventrally, particularly near the incompletely rimmed. Approximately 1.5 fenestra cochleae and fenestra vestibuli (oval mm deep, the fossa is excavated into the and round windows). Its ventral surface is squamosal bone, which forms the anterior marked by at least two indistinct grooves wall of the mastoid process. The suprameatal (more clearly shown on the left side) that be- fossa is primarily developed toward the cau- gin posteriorly at a small tubercle near the

← Fig. 5. Ventrolateral view of the right side of the basicranium of Amphicticeps shackelfordi, AMNH 19010, holotype. The region near the postglenoid process is reconstructed from that of the left side. cfn, canal for facial nerve; cica, canal for internal carotid artery; er, epitympanic recess; fc, fenestra cochleae (fenestra rotunda); fo, foramen ovale; fs, fossa for stapedius muscle; ftt, fossa for tensor tympani muscle; fv, fenestra vestibuli (fenestra ovalis); Gf, Glaserian ®ssure; ips, inferior petrosal sinus; mlf, middle lacerate foramen; mp, mastoid process; pf, promontory foramen; pgf, postglenoid foramen; plf, posterior lacerate foramen; pp, paroccipital process; pr, promontorium; sf, suprameatal fossa. 14 AMERICAN MUSEUM NOVITATES NO. 3483

Fig. 6. Amphicticeps shackelfordi? MAE SG.95.8919, braincase and basicranial fragments. A, caudal view; B, lateral view; C, stereophotos of lateroventral aspect; and D, stereophotos of ventral view. cf, condyloid foramen; eam, external auditory meatus; Ec, Eustachian canal; mp, mastoid process; pcf, posterior carotid foramen; plf, posterior lacerate foramen; pp, paroccipital process; sf, suprameatal fossa; smf, stylomastoid foramen; tb, tympanic bulla. 2005 WANG ET AL.: CARNIVORANS FROM CENTRAL MONGOLIA 15 entotympanic/promontorium contact facet. left side, showing pneumatic spaces beneath These grooves make small arches laterally at the bony surface. a level slightly in front of the fenestra co- The fossa for the tensor tympani is deep chleae and then turn medially toward the en- and located anteromedial to the epitympanic totympanic/promontorium suture. Despite recess. There is a thin sheet of bone covering the super®cial resemblance of the course of the canal for the facial nerve; some segments these grooves to the sulcus of the promon- of this sheet are so thin that the bone is rather torial branch of the internal carotid artery transparent along the nerve canal. The epi- and nerve in primitive caniforms (Wang and tympanic recess is shallow, and walled by Tedford, 1994), its occupant is unlikely to be squamosal ventrolaterally and petrosal dor- a promontorial artery, contrary to Cirot solaterally. (1992: ®g. 2), who postulated a promontorial Medial to the Eustachian canal at the level artery in the primitive musteloid Amphictis, of the presumed anterior end of the entotym- and to Schmidt-Kittler (1981), who implied panic, the basisphenoid is deeply excavated the existence of an internal carotid artery on into a large pit just anterior to the median the promontorium of Amphicticeps. In taxa lacerate foramen. This space marks the turn- with a promontorial artery, there is usually a around point of the internal carotid artery stapedial branch leading transversely toward (cica of ®g. 5; Wang and Tedford, 1994). The the oval window. In Amphicticeps there is no inferior petrosal vein is excavated into the such transverse sulcus medial to the oval lateral wall of the basioccipital but remains window. Instead, there is a distinct groove rather thin (approximately 1 mm in diameter) along the ventral rim of the oval window. within a canal formed by the petrosal and Such a longitudinal groove can also be found basioccipital (ips of ®g. 5; best seen near the in Promartes and in living procyonids. In the posterior lacerate foramen). The caliber of case of the latter group, the soft structures the inferior petrosal vein is such that it is unlikely to be able to accommodate a double- that left the grooves are ®ne branches of the looped internal carotid artery hypothesized to caroticotympanic artery and the accompa- be present in many ursoids (Hunt, 1977; nying caroticotympanic nerves (Story, 1951: Hunt and Barnes, 1994). ®g. 83). The caroticotympanic artery, which The area anterior to the Eustachian canal is a minor component in the internal carotid is damaged on both sides of the skull, and it artery, arises from the main internal carotid is not possible to ascertain the status of the artery within the carotid canal, and after alisphenoid canal except by indirect infer- looping across the promontory, anastomoses ences. Schmidt-Kittler (1981: 784) stated, with the tympanic arteries. The inferior tym- without elaboration, that Amphicticeps has an panic artery and the tympanic nerve loop alisphenoid canal on each side of the skull. around the posterior edge of the round win- On AMNH 19010, only the anterodorsal roof dow instead the anterior position as in a of the foramen rotundum (shared with the an- promontory artery. The surface sulci on the terior opening of the alisphenoid canal) is promontorium of Amphicticeps are thus best partially preserved on each side of the skull, reconstructed as left by an arterial and ner- and the ventral ¯oor of the canal (if present) vous con®guration similar to that of extant is missing. Further preparation on the better procyonids, that is, no promontory artery is preserved left side of the holotype reveals a present. The main course of the internal ca- short segment of bone, about 2 mm in length, rotid artery is assumed to be in the typical between the posterior aspect of the foramen arctoid fashion enclosed within the medial rotundum and the foramen ovale. In Canis bullar wall (see the description under A. teil- (Evans and Christensen, 1979) and Ailurus hardi for further evidence of a medially po- (Story, 1951), the maxillary artery enters the sitioned internal carotid artery). posterior opening of the alisphenoid canal At the posteromedial corner of the pro- (caudal alar foramen in Evans and Christen- montorium, there is a distinct posterior pro- sen, 1979) and emerges from the foramen ro- cess protruding toward the posterior lacerate tundum (rostral alar foramen). In Procyon foramen. This process is broken off on the (which lacks the alisphenoid canal), on the 16 AMERICAN MUSEUM NOVITATES NO. 3483 2005 WANG ET AL.: CARNIVORANS FROM CENTRAL MONGOLIA 17

TABLE 2 Cranial Measurements of Amphicticeps shackelfordi and Amphicynodon teilhardi (in mm)

Amphicticeps Amphicynodon shackelfordi teilhardi (AMNH 19010) (PST 17/34) Length of skull, inion to nasal tip 87.0 Ð Breadth of rostrum across C (labial side) 22.6 13.8 Breadth of rostrum across P1 (lingual side) 13.8 9.3 Breadth of palate across left and right P4±M1 35.5 26.3 Breadth of frontal across postorbital processes 29.0 18.3 Breadth of postorbital constriction 13.3 13.1 Distance between postorbital constriction and postorbital process 12.2 7.0 other hand, only a small branch of the inter- posterolateral aspect of the skull, preserving nal maxillary artery, the medial meningeal much of the left bulla as well as the occipital artery, enters the foramen ovale (Story, 1951: condyles and the top of the skull. Although ®g. 82). On AMNH 19010, the bony bridge much of the bony relationships between var- ¯ooring the orbital ®ssure and roo®ng the ious elements are intact, crushing has dis- alisphenoid canal forms a half-pipe structure torted some areas so that full restoration of on the ventral view, possibly because of a their original relationships is no longer pos- broken ventral ¯oor for the alisphenoid ca- sible. nal. A tiny foramen is present on the medial The top of the braincase is well preserved, wall of the alisphenoid canal at the level of including about 25 mm of the posterior-most the presumed posterior entrance of the canal, segment of the sagittal crest and a complete as is also seen in Amphicynodon teilhardi. nuchal crest. The sagittal crest is 7 mm high While structural damages do not permit us to at its deepest point just in front of the nuchal state with certainty the existence of a poste- crest. The posterior segment of the sagittal rior opening of the alisphenoid canal, we crest on the holotype is missing, but based agree with Schmidt-Kittler that an alisphe- on the height of its nuchal crest, seems to be noid canal is likely present. slightly lower than that in MAE SG.95.8919. Bulla of MAE SG.95.8919 (®g. 6): We ten- The temporal foramen at the suture of pari- tatively refer a left bulla and associated pos- etal and supraoccipital is more posteriorly lo- terior-most portion of the skull, MAE cated than in the holotype. The pro®le of the SG.95.8919, to Amphicticeps shackelfordi. nuchal crest, viewed from the caudal end, is Although the bullar size and overall basicra- also different from that of the holotype; in- nial morphology seems to be compatible stead of a rather ¯at top in the holotype, with the holotype, there are a number of dif- MAE SG.95.8919 has a rather pointed inion ferences that prevent us from being certain with a more steeply sloped nuchal crest on of our reference. Furthermore, in the absence either side. The nuchal crest is also slightly of associated dental materials in MAE thinner than in the holotype. SG.95.8919, it is prudent to describe this bul- The most prominent difference between la separately in order to highlight the con- MAE SG.95.8919 and the holotype is in the ¯icting morphologies in the basicranial area size and lateral extrusion of the mastoid pro- from those in the holotype. cesses, although the overall construction of MAE SG.95.8919 consists of a crushed the mastoid process in MAE SG.95.8919 is

← Fig. 7. Amphicticeps shackelfordi, MAE BU.91.9187±90, referred specimen, photographs of polyester cast. A, occlusal view of upper teeth, stereophotos; B, occlusal view of lower teeth, stereophotos; C, medial and D, lateral views of lower jaw. Scales ϭ 10 mm. 18 AMERICAN MUSEUM NOVITATES NO. 3483

TABLE 3 Measurements of Upper Teeth of Amphicticeps (in mm)

shackelfordi dorog makhchinus AMNH MAE AMNH MAE MAE MAE 19010 BU.91.9187-90 85224 SG.9799 SG.9194 93-213 Length of C1 (alveola) 6.8 Ð Ð Ð Ð Ð Width of C1 (alveola) 4.5 Ð Ð Ð Ð Ð Length of P1 2.8 Ð Ð Ð Ð Ð Width of P1 1.6 Ð Ð Ð Ð Ð Length of P2 4.7 Ð Ð Ð Ð Ð Width of P2 2.6 Ð Ð Ð Ð Ð Length of P3 Ð 5.4 6.4 Ð Ð Ð Width of P3 Ð 3.0 4.0 Ð Ð 4.7* Labial length of P4 9.7 9.5 Ð Ð 11.0 12.7 Lingual length of P4 (protocone to metastyle) 10.8 10.7 Ð Ð 11.7 14.4 Anterior breadth of P4 6.9 6.6 Ð Ð 7.7 10.6 Labial length of M1 7.1 6.7 Ð 7.9 8.1 10.3 Max. transverse width M1 10.5 9.7 Ð 10.6 11.4 14.2 Longitudinal length of M2 Ð 2.6 Ð Ð Ð Ð Max. transverse width M2 Ð 4.0 Ð Ð Ð Ð Alveolar distance P1±M2 28.9 26.0 Ð Ð Ð Ð * Indicates an estimate. similar to that in the holotype. A laterally angular. A posteroventral facet for attach- expanded mastoid forms a conspicuous, ment of the obliquus capitis cranialis muscle rounded (in dorsal view) crest continuous (AntoÂn et al., 2004) is the largest surface of from the lambdoidal crest. In lateral view, the the process, and this facet is less posteriorly outline of the mastoid process is roughly tri- oriented than in the holotype. The depth of

TABLE 4 Measurements of Lower Teeth of Amphicticeps shackelfordi (in mm)

MAE MAE MAE AMNH AMNH AMNH AMNH AMNH AMNH AMNH BU.91. SG.95. SG.97. MAE 19017 19127 19128 21695 81336 83610 85749 9787-90 7518 3576 M-217 Length of p1 Ð Ð Ð Ð Ð Ð Ð 2.8 Ð Ð Ð Width of p1 Ð Ð Ð Ð Ð Ð Ð 1.5 Ð Ð Ð Length of p2 4.0 Ð Ð Ð Ð Ð Ð 3.9 4.6 Ð Ð Width of p2 2.2 Ð Ð Ð Ð Ð Ð 2.4 2.4 Ð Ð Length of p3 5.0 4.5 Ð Ð Ð Ð Ð 5.0 Ð Ð Ð Width of p3 2.5 3.0 Ð Ð Ð Ð Ð 2.9 Ð Ð Ð Length of p4 Ð Ð Ð Ð Ð 6.3 Ð 6.1 6.2 Ð Ð Width of p4 Ð Ð Ð Ð Ð 3.7 Ð 3.6 3.0 Ð Ð Length of m1 8.7 8.5 9.2 9.2 Ð 9.2 Ð 9.2 Ð 9.8 9.4 Trigonid length of m1 5.4 5.4 5.6 6.0 Ð 6.2 Ð 6.1 Ð 6.0 6.3 Trigonid width of m1 4.2 4.4 4.8 4.6 Ð 4.5 Ð 4.8 4.4 4.2 5.1 Talonid width of m1 3.7 3.8 4.4 4.2 Ð 4.0 Ð 4.3 Ð 4.5 4.9 Length of m2 Ð Ð Ð Ð 3.9 3.5 4.3 3.8 Ð Ð Ð Width of m2 Ð Ð Ð Ð 3.7 3.1 3.6 3.4 Ð Ð Ð Diameter of m3 alveola Ð Ð Ð Ð Ð Ð 2.0 1.5 Ð Ð Ð p1±m2 (alveolar) 28.7* 27.3 31.5* Ð Ð Ð Ð 30.1 Ð Ð Ð * Indicates an estimate. 2005 WANG ET AL.: CARNIVORANS FROM CENTRAL MONGOLIA 19 the mastoid process in MAE SG.95.8919 is roughened area may be one possible inter- also signi®cantly less than in the holotype, pretation of the rostral entotympanic± resulting in a smaller area of the lateral facet ectotympanic contact, although such an in- of the mastoid for attachment of the sterno- terpretation is highly speculative and other mastoideus muscle. Overall, one gets the im- alternatives are just as likely. The posterior pression that the much enlarged and laterally carotid foramen is located on the anterior rim extruded mastoid process in the holotype is of the large posterior lacerate foreman. mostly related to the increased size and le- Toward the medial aspect of the bulla, the verage of the m. obliquus capitis cranialis, basioccipital±basisphenoid region is frac- and thus presumed more powerful head ro- tured, and anatomic relationships are dif®cult tation (AntoÂn et al., 2004), although onto- to interpret. The nearly vertical medial wall genetic variation may also be responsible for of the bulla is buttressed by a thickened lat- such differences. eral wall, up to 6 mm in depth, of the basi- Associated with its smaller mastoid pro- occipital. The lateral surface of this lateral cess, the paroccipital process in MAE wall of the basioccipital is essentially ¯at and SG.95.8919 is also narrower in ventral hugs the medial wall of the bulla, although viewÐthe broader process in the holotype is there is a narrow gap between these two apparently the result of a proportional lateral walls toward the posterior aspect of their expansion due to its greatly expanded mas- contact. The medial wall of the basioccipital toid process. Otherwise, the paroccipital pro- lacks a prominent invagination for the em- cess is of the same general construction, with bayment of the inferior petrosal sinus seen in a dorsally convex and ventrally ¯at process many ursoids such as ursids, amphicyonids, that is completely posteriorly oriented with- and basal pinnipeds (e.g., Hunt, 1977; Hunt out any hint of a ventral bending toward the and Barnes, 1994). The above mentioned gap bulla. Such a ``free'' paroccipital process, between the lateral wall of the basioccipital without hugging the bulla, is often a primi- and the medial wall of the bulla-petrosal tive condition for all caniforms (e.g., see seems too small to accommodate an enlarged Wang, 1994; Wang and Tedford, 1994; Wang inferior petrosal sinus. On the medial side of et al., 1999). the lateral basioccipital wall a small canal is The bulla is more or less intact with the embedded within the basioccipital bone. This exception of a crack on the ventrolateral as- canal, probably for a nutrient blood vessel, pect. The lateral half of the ectotympanic emerges anteriorly into the braincase slightly ring is slightly caved in by approximately 1 behind the level of the anterior carotid fora- mm along this crack. Other than such a dis- men. tortion, the bulla seems to maintain its orig- The external auditory meatus is quite well inal proportions. The form of the bulla is developed for a basal arctoid. The ventral lip quite in¯ated for a basal arctoid, more so of the meatus is 3±4 mm long, much longer than the modern ursid ``type A'' bulla (Hunt, than those in European basal arctoids such as 1974). The axis along the ventralmost rim of Amphicynodon leptorhynchus (FSP ITD 312) the bulla forms a slight angle with the para- and Amphictis ambiguus (FSP PFRA 28). sagittal axis of the skull, in contrast to the Areas inside the meatus were prepared. A su- canid condition of mostly parallel bullar prameatal fossa is vaguely developed on the axes. Composition of the bullar elements is posterodorsal aspect of the meatal wall of the dif®cult to ascertain due to extensive fusions squamosal. Such a weak fossa is in contrast and ®ne cracks on the bullar surface. An ex- to that in the holotype, on which it is not tremely subtle groove seems to run from the only substantially deeper but also better de- base of the paroccipital process across the ®ned by a sharp rim along its lateral and ven- posterior aspect of the bulla, crossing slightly tral aspects. The fossa in MAE SG.95.8919 behind the ventral ¯oor of the bulla and is also less well developed than in Amphi- reaching toward the anterior carotid foramen cynodon teilhardi (see description below). (the anterior extent of this groove is less well The postglenoid process is broken off, ex- de®ned because surface marks are becoming posing the canal for the retroarticular vein, less clear). This narrow band of slightly which is of relatively small caliber. 20 AMERICAN MUSEUM NOVITATES NO. 3483

Mandible (®g. 7B±D): Discovery of the rating the parastyle from the paracone. The associated upper and lower jaws of MAE paracone is much higher than the metacone. BU.91.9187±90 allows us con®dently to re- The preprotocrista is low and lacks a proto- fer several ramal fragments to Amphicticeps. conule on the holotype but is swollen slightly Nonetheless, our knowledge of the angular at the base of the paracone to indicate an process and the ascending ramus is still in- indistinct protoconule in MAE BU.91.9187± complete. 90. The postprotocrista is clearly present and The mandible is short, thick, and deep, is oriented somewhat posteriorly. The post- with an average thickness of 6.0 mm and protocrista ends at the posterior border of the depth of 11.0 mm (both measured at the level M1 rather than at the base of the metacone. of the talonid basin of m1; N ϭ 5). On The lingual cingulum is moderately devel- AMNH 19127, the remaining ascending ra- oped. It is rather low as compared to the pro- mus suggests a rather erect anterior border, tocone, and is thickest along the posterolin- forming a 125Њ angle with the horizontal ra- gual border of the M1. The cingulum quickly mus. There are two mental foramina, one be- tapers off anterior and posterior to the pro- low the anterior edge of the p2 and another tocone, in contrast to a well-developed pos- between the two roots of the p3. terior ridge bordering a deep talon basin in Teeth (®gs. 2, 7): No upper incisor is pre- the oligobunines. M2 is double-rooted and is served. Only the root of right I3 is partially located lingually such that its lingual border intact on AMNH 19010. A robust upper ca- is at the same level as that of M1 whereas nine can be inferred from the large alveoli its labial border only reaches the middle of on both sides of the holotype. Immediately M1. The oval-shaped M2 has a prominent behind the canine is a small, single-rooted P1 paracone toward the labial margin, and a with a single main cusp. The double-rooted posterolingually located, but much smaller, P2 also has a single main cusp, which is sur- metacone at the posterior border of the tooth. rounded by a weak cingulum on the lingual A crestlike protocone is near the middle of side. This cingulum thickens on the posterior the tooth, and is surrounded lingually by a end and shows an incipient development of lingual cingulum. a cingular cusp. Both P3s are missing on the No lower incisors or canines are preserved type, but are well preserved in MAE on the holotype or referred specimens. The BU.91.9187±90. Like P2, P3 is single lower premolars are as robust as their upper cusped, although its cingulum is stronger counterparts. The p1 is single rooted and has than that on P2. The upper carnassial, P4, is a single main cusp. A cingulum is present transversely broad due to a lingually extend- around the anterior and posterior borders of ed protocone, which is near the anterolingual p2±p3, which are single cusped. The p4, corner of the tooth. The apex of the P4 pro- however, has a small posterior accessory tocone is relatively low and formed by a cusp behind the main cusp. The p4 cingulum raised lingual cingulum. This crestlike pro- nearly completely surrounds the tooth except tocone contrasts with that of the North Amer- the region between the roots on the labial ican oligobunines, which have a primitively side. The m1 is rather broad (transversely) tall, cusplike protocone (i.e., the apex is not and its trigonid is short. The trigonid cusps associated with the cingulum). A low crest is are low and blunt, and the metaconid is not present on the labial aspect of the protocone. greatly reduced. The lingual border between There is a narrow cingulum on the labial the paraconid and metaconid is slightly con- side, which continues in front of the tooth cave to give a somewhat sigmoid appearance and thickens slightly to become an indistinct in occlusal view. Most individuals have a la- parastyle. A carnassial notch is present. bial cingulum on the trigonid, whereas the M1 is transversely elongated, and its labial lingual cingulum is more reduced. But the border forms a steep angle, averaging 112Њ, lingual cingulum is usually present at the lev- with that of the P4. The M1 parastyle is el of the carnassial notch and may extend strong, rising to nearly the same height as the along the entire trigonid as in AMNH 21695, paracone. In MAE BU.91.9187±90, there is which occurs stratigraphically higher than a faint notch (absent in the holotype) sepa- the rest of the sample. The talonid of m1 is 2005 WANG ET AL.: CARNIVORANS FROM CENTRAL MONGOLIA 21 narrower than the trigonid, and consists of a Amphicynodon are still correct and their con- dominant hypoconid bordered lingually by a cept of the hypodigm still valid. However, low entoconid crest much like a cingulum. specimens from the 1922 collection lack an The anterior hypoconid crest, the cristid ob- m3, which naturally led Matthew and Grang- liqua, is oriented parasagittally. There is a er to conclude that absence of this last molar weak cingulum on the labial side of the hy- is one of the main distinctions between Am- poconid. The entoconid crest does not have phicticeps and Amphicynodon. a notch at the base of the trigonid as in Po- Our new discovery that MAE BU.91.9187± tamotherium and the oligobunines. The 90 has an unmistakable m3 adds a new wrin- double-rooted m2 is shortened and nearly kle to the interpretation of this character. As quadrate in outline. The protoconid and pointed out in the above descriptions about metaconid are large and distinct. There is no specimens that have preserved the posterior paraconid anterior to the protoconid and dental battery, an m3 is de®nitely absent in metaconid; in its place there is a low, trian- AMNH 19127 and is probably absent as well gular platform. The greatly reduced talonid in AMNH 19017. Given the tiny size of the consists of a small hypoconid along the pos- m3 in MAE BU.91.9187±90, it is quite pos- terolabial border of the tooth. The entoconid sible that individuals, such as AMNH 19217, takes the form of a narrow cingulum. A nar- could have had an m3 in an earlier part of row cingulum is also present along the labial their life, which was later broken and fully border of m2. The presence of a tiny m3 is healed without leaving traces of its root, a indicated by a small root in MAE situation common in carnivores with small BU.91.9187±90, but it is absent in other in- p1s (personal obs.). Whatever the actual sit- dividuals (de®nitely in AMNH 19127 and uation with AMNH 19217, taken at the face probably in AMNH 19017). value of existing materials, 30%±50% of in- COMPARISON: Although Matthew and dividuals have retained an m3, although our Granger's (1924) diagnosis of Amphicticeps sample size is obviously too small to allow shackelfordi indicated the presence of lower a true statistical sense of the ratios. A similar jaws, they did not elaborate the exact nature situation is better documented in the loss of of the specimens, nor did they illustrate a the M3 in the basal canid Hesperocyon gre- lower jaw of this species. Four lower jaws garius, in which about 7% of the Chadronian were probably available at the time of their individuals still retain a small, nonfunctional study (specimens that were collected during M3 and by Orellan time all have lost it the 1922 season), and what Matthew and (Wang, 1994: 30). However, the actual ratio Granger had in mind were probably AMNH may not be an important point in the present 19017 and 19128 because these two jaw analysis. The important phylogenetic impli- fragments possess an m1 but are missing the cation is that Amphicticeps represents a small m2, a combination that matches their de- clade that in its most basal species, A. shack- scriptions. Their descriptions of the lower elfordi, is on its way to losing its last molars, teeth, although very brief, must have been and this loss is yet another independent dis- important in their attempt to delineate vari- appearance of this molar among carnivorans. ous species. In particular, their contrasts be- It is also worth noting that AMNH 21695, tween lower carnassials of Amphicticeps and the only referred specimen of A. shackelfordi Amphicynodon (their Cynodon) must have from the Zavlia fauna well above the level been based on these referred lower jaws. of the persistent basalt, is also the most ro- With the naturally associated upper and bust individual known, both in terms of ra- lower jaws of MAE BU.91.9187±90, our mal construction and width of the m1. In ad- con®dence in the references of isolated lower dition, it is the only individual with a nearly dental materials to Amphicticeps shackelfordi complete cingulum on the lingual side of the is considerably increased. In light of the new trigonid, and its premolar alveoli indicate an materials, Matthew and Granger's (1924: 4) individual with a relatively shorter rostrum comparisons about the lower carnassials hav- compared to individuals from the Ulaan ing ``a narrower and shorter heel with more Khongil fauna below or immediately above distinct hypoconid crest'' relative to those of the lava. The reliability of these features as 22 AMERICAN MUSEUM NOVITATES NO. 3483 indications of a later stage of evolution of the mal fragment with m1±2, from ``Grand Can- species remains to be veri®ed by further sam- yon'', ®eld no. 531; AMNH 84211, right ra- ples from the Zavlia fauna. mal fragment with m1±2, ®eld no. 532; MAE SG.95.8919 offers the only bulla for AMNH 85217, left ramal fragment with m1 Hsanda Gol carnivorans, and despite its less and alveoli of p2±4, ®eld no. 538; AMNH than certain taxonomic status, is of consid- 85223, isolated left m1, ®eld no. 538; erable importance in our understanding of AMNH 85224, left maxillary fragment with the basal arctoids. That it belongs to the basal P3, ®eld no. 538; AMNH 85233, isolated left arctoids is certain. Of the main two arctoid m1, ®eld no. 538; MAE SG.91.9192, left ra- lineages in the Shand Gol that are likely can- mal fragment with c broken and p2±4, from didates, Amphicynodon teilhardi, the only locality MAE 91±82, Tatal Gol, below lava species so far known for the genus in Mon- in Tatal Member; MAE SG.95.8655, left ra- golia, is too small for MAE SG.95.8919. mus fragment with p3; MAE SG.97.3576, Species of Amphicticeps, on the other hand, isolated right m1; and MAE SG.9799, left encompass a size range that is consistent maxillar fragment with M1. with that of MAE SG.95.8919. More specif- ETYMOLOGY: Mongolian: dorog, badger. ically, the holotype of A. shackelfordi has the DIAGNOSIS: Amphicticeps dorog differs same bulla size (judging from the attachment from A. shackelfordi in its possession of the sites for the bulla) and general basicranial following derived characters: larger size and morphology as the MAE SG.95.8919. We more robust dentitions and jaws, lower and thus cautiously place MAE SG.95.8919 in A. more crestlike P4 protocone, more prominent shackelfordi. P4 anterior cingulum, more reduced M1 par- Overall, the holotype of Amphicticeps astyle, relatively larger and more labially lo- shackelfordi has a more robust construction cated M2, relatively shorter m2, and loss of than in MAE SG.95.8919, particularly in its m3. It is readily distinguishable from A. thicker nuchal crests and larger and more lat- makhchinus in its smaller size, less lingually erally extruded mastoid process. These pro- and posteriorly expanded P4 protocone crest, portional differences may seem conspicuous, more labially oriented M1 postprotocrista, but probably are all attributable to a stronger and less lingually expanded lingual cingulum development of the head±neck musculatures of M1. in the holotype. Even more extreme lateral DESCRIPTION: Our knowledge of this new expansions of the mastoid process can be species is still limited to isolated maxillar seen in Allocyon. In the absence of contra- and mandibular fragments and cheek teeth. dicting evidence, we tentatively treat such Upper teeth (®gs. 8A, C): Only a single differences as variations due to sexual di- isolated P3 (AMNH 85224) is available and morphisms in Amphicticeps. If our treatment it has a simple main cusp and a well-devel- is correct, the variations in the size of the oped cingulum. The P4 on the holotype is suprameatal fossa are also considerable. relatively wide due to a rather lingually expanded protocone. The protocone is low and Amphicticeps dorog Wang, McKenna, and its apex is located along the lingual margin Dashzeveg, new species and is continuous with the lingual cingulum Figure 8; Tables 3, 5 through crests on either sides of the cusp. There is also a low ridge on the labial side HOLOTYPE: MAE SG.9194, right maxillary of the protocone that ends at the base of the fragment with P4±M1 and M2 alveolus. paracone. A cingulum is strongly developed TYPE LOCALITY: Tsagan Nor Basin, eastern around the entire P4, and the anterior cin- Valley of Lakes, Obor-Khangay Province, gulum is especially strong to the point of al- Mongolian People's Republic. Top of Tatal most forming a parastyle. The labial cingu- Member, Hsanda Gol Formation, early Oli- lum is better developed than the lingual gocene. cingulum. The paracone is broad based and REFERRED SPECIMENS: AMNH 21656, left has a distinct anterior ridge leading down ramal fragment with m2 broken and m3 al- from the apex to the base. There is a well- veolus, ®eld no. 538; AMNH 21672, left ra- developed carnassial notch. 2005 WANG ET AL.: CARNIVORANS FROM CENTRAL MONGOLIA 23

The most distinguishing feature of the M1 conid is the largest and tallest cusp. The is its transverse elongation, mostly due to a metaconid and paraconid are of approxi- large paracone and parastyle. The paracone mately the same height. The metaconid is is the tallest cusp of the tooth, and is sub- lingual to, and slightly posterior to, the pro- stantially larger and taller than the metacone. toconid. The labial cingulum is narrow, and A large parastyle is formed by a prominent a short and indistinct lingual cingulum is pre- elevation of the labial cingulum surrounding sent between the paraconid and metaconid. the paracone. In contrast, the labial cingulum The tall trigonid is in contrast to a low tal- around the metacone is much narrower and onid, which is dominated by a large (at the lower. The metacone is on the posterolingual base), but relatively low hypoconid. The hy- aspect of the paracone. The protocone is poconid is crestlike. It is rather labially lo- about the same height as the metacone. A cated at its posterior end and stops anteriorly distinct pre- and postprotocrista converge at at the base of the protoconid, almost directly the apex of the protocone and form a sharp below the apex of the protoconid. The en- V-shaped crest. No protoconule or metaco- toconid is no more than a low crest, directed nule is present. The lingual cingulum sur- posteriorly at an angle with the long axis of rounds the protocone but is asymmetricalÐ the tooth. The entoconid crest is decorated its posterolingual corner behind the proto- with ®ne wrinkles along its top edge. An in- cone is more swollen than its anterolingual distinct labial cingulum surrounds the talonid corner. but no cingulum is present on the lingual No M2 is preserved. The double-rooted al- side. veoli on the holotype suggest an M2 that is The m2 is single rooted, very short, and probably transversely elongated, as is M1, almost equal in its length and width. The tri- but probably anteroposteriorly short because gonid is formed by two low cusps, the pro- of a short m2 and the absence of an m3 (see toconid and metaconid, which are set apart below). The location of the labial root indi- from each other. The two cusps are located cates an M2 that is not lingually shifted as almost on the lingual and labial borders of in Amphicticeps shackelfordi. the tooth. A hypoconid is barely distinguish- Lower teeth (®gs. 8B, D, E, F): Although able on the talonid. A vague cingulum is de- no associated upper and lower jaws are avail- veloped on the anterior half of the tooth. The able, our references of isolated lower jaws m3 is absent. are mostly based on their intermediate sizes, COMPARISON: Even on the basis of the corresponding to size differences of upper fragmentary materials at hand, the transition- teeth of different species of Amphicticeps, al nature of this species seems readily ap- and on their dental morphologies that are parentÐAmphicticeps dorog is in many consistent with those of the upper teeth. ways an intermediate form between the more Fragmentary lower jaws, such as in AMNH primitive A. shackelfordi and more derived 21672, indicate a robust mandible of deep A. makhchinus. Average length of the upper and thick horizontal ramus. carnassials is 15% longer than that of A. Lower premolars are best preserved in shackelfordi but 16% shorter than that of A. MAE SG.91.9192, which has p2±4. Both p2 makhchinus. In the lower carnassial length, and p3 are similar, with a simple main cusp A. dorog is 22% longer than that of A. shack- and an indistinct anterior cingular cusp, al- elfordi. Such size differences are comparable though the latter is larger and less asymmet- to those among modern sympatric species of rical in lateral view. A narrow cingulum sur- some desert canids (Dayan et al., 1989, rounds much of the crown of these premo- 1992), which offer a quantitative criterion for lars. The p4 has added a moderate posterior identi®cation of fragmentary materials. accessory cusp as well as a posterior cingular These overall size differences, in addition to cusp. Its anterior cingular cusp is also better the fact that the two species cluster by them- developed than those of anterior premolars. selves without intermediate individuals to The p4 cingulum also becomes more distinct. bridge the gap, strongly suggests a separate The m1 trigonid is relatively short and its species for A. dorog. shearing blade bends lingually. The proto- Qualitative morphological differences also 24 AMERICAN MUSEUM NOVITATES NO. 3483 2005 WANG ET AL.: CARNIVORANS FROM CENTRAL MONGOLIA 25

TABLE 5 Measurements of Lower Teeth of Amphicticeps dorog (in mm)

AMNH AMNH AMNH AMNH AMNH MAE MAE 21672 84211 85217 85223 85233 SG.91.9192 SG.95.8655 Length of p2 Ð Ð Ð Ð Ð 5.3 Ð Width of p2 Ð Ð Ð Ð Ð 2.6 Ð Length of p3 Ð Ð Ð Ð Ð 5.9 5.5 Width of p3 Ð Ð Ð Ð Ð 2.9 3.0 Length of p4 Ð Ð Ð Ð Ð 7.0 Ð Width of p4 Ð Ð Ð Ð Ð 3.7 Ð Length of m1 10.5 10.5 11.2 11.8 11.4 Ð Ð Trigonid length of m1 6.9 6.5 7.2 7.8 7.7 Ð Ð Trigonid width of m1 5.6 4.6 5.3 6.0 4.8 Ð Ð Talonid width of m1 4.9 4.5 Ð 5.7 5.0 Ð Ð Length of m2 4.2 3.8 Ð Ð Ð Ð Ð Width of m2 3.9 3.9 Ð Ð Ð Ð Ð indicate a transitional form for Amphicticeps the Tatal Member of the Hsanda Gol For- dorog. In the following characters A. dorog mation, early Oligocene. is almost exactly intermediate between A. REFERRED SPECIMENS: Holotype only. shackelfordi and A. makhchinus: the crestlike DIAGNOSIS: As the largest and possibly the P4 protocone, the development of the P4 an- most derived species of the genus, Amphic- terior cingulum, the size of the M1 parastyle, ticeps makhchinus is distinguished from the the development of the posterior lingual cin- other two species of the genus, A. shackel- gulum of M1, and the angle between the la- fordi and A. dorog, in its larger size, a broad- bial borders of the P4 and M1. ened P3 with an extra lingual root, a low and lingually expanded P4 protocone crest, a Amphicticeps makhchinus Wang, slightly more reduced M1 parastyle, an en- McKenna, and Dashzeveg, new species larged M1 metaconule, and a more expanded Figure 9; Table 3 M1 lingual cingulum. HOLOTYPE: MAE 93±213 (AMNH cast ETYMOLOGY: Mongolian: makhchinus, 129862), right maxillary fragment with P4± meat eater, carnivore. M1, partial P3, and alveolus of M2. Collect- DESCRIPTION: Amphicticeps makhchinus is ed by James M. Clark on 16 August 1993. the least known of the three Hsanda Gol spe- TYPE LOCALITY: MAE 93±213 was found cies of the genus. We are limited to two and in the Tatal Gol (Ulaan Khongil or ``Grand a half teeth on the fragmentary right maxil- Canyon'') locality, 45Њ17Ј50ЉN, 101Њ37Ј16ЉE, lary of the holotype. The maxillary clearly Tsagan Nor Basin, eastern Valley of Lakes, shows a shortened infraorbital canal, imply- Obor-Khangay Province, Mongolian Peo- ing a shortened rostrum. Attached to this ple's Republic. maxillary fragment is the anterior-most part GEOLOGY AND AGE: MAE 93±213 was col- of the jugal. The well-delineated jugal- lected from the main exposure of the Tatal maxillary suture indicates that the anterior Gol locality, below the level of the lava, in jugal process stops at the antorbital rim and

← Fig. 8. Amphicticeps dorog, n.sp. A, lateral view of upper teeth, MAE SG.9194, holotype; B, lateral view of anterior ramal fragment, MAE SG.91.9192; C, occlusal view of upper teeth, MAE SG.9194, holotype, stereophotos; D, occlusal (stereophoto), E, lingual, F, labial views of lower jaw fragment, AMNH 21672. Scales ϭ 10 mm; top scale is for B, middle scale for A, C±D, and lower scale for E and F. 26 AMERICAN MUSEUM NOVITATES NO. 3483

Fig. 9. Amphicticeps makhchinus, n.sp, MAE 93±213, holotype, photographs of a polyester cast (AMNH 129862). A, occlusal view of upper teeth, stereophotos; B, lateral view of upper teeth and maxillary; and C, lingual view of upper teeth and maxillary. Scale ϭ 5 mm. is probably not in contact with the lacrimal of Amphicticeps, the P4 protocone is com- or frontal as in other species referred to this posed of a raised lingual cingulum. However, genus. the protocone is more expanded toward the Only the posterior half of the P3 is pre- lingual side than in the other species of the served, which has a well-developed cingu- genus. As in the other two species of Am- lum. The P3 has a signi®cantly broadened phicticeps, there is a low crest on the labial lingual side and appears to have an extra lin- side of the protocone. The broad-based para- gual (third) root, in contrast to the double- cone has an anterior ridge leading up to the rooted condition in other species of Amphic- cingulum. ticeps. The P4 is typical of the genus, with a Overall proportions of the M1 have an an- complete cingulum surrounding the entire teroposteriorly broadened appearance for a tooth. The anterolabial corner of the cingu- basal ursoid. The parastyle is large and rises lum is the strongest, but it does not elevate above the paracone, but does not reach to the to form a parastyle. Like that of other species same degree of expansion as seen in A. 2005 WANG ET AL.: CARNIVORANS FROM CENTRAL MONGOLIA 27 shackelfordi and is more similar to that of A. seen in Cephalogale. However, structural de- dorog. Likewise, the cingulum adjacent to tails of these features tend to argue against a the metacone shows no sign of reduction as true homology in the hypocarnivorous den- in A. shackelfordi. Consequently, the angle titions shared between A. makhchinus and between the labial borders of the P4 and M1 Cephalogale. For example, the P4 protocone remains a relatively large 124Њ,15Њ greater in all ursids (including Cephalogale)is than in A. shackelfordi but almost identical formed by a swollen lingual cingulum, often to that in A. dorog. A distinct pre- and post- in the form of a crest instead of a conical protocrista are present, the latter being slight- cusp, that has receded far back from the an- ly more posteriorly directed than in A. shack- terior border of the tooth, in contrast to an elfordi and A. dorog. There is no protoconule essentially conical protocone located on the (paraconule) and the metaconule is only in- anterolingual corner of P4 in A. makhchinus. dicated by a vague platform (probably suf- Another derived character for the Ursidae is fered from some wear) slightly raised above a posteriorly oriented postprotocrista of M1. the surrounding areas. The M1 internal cin- This is a highly consistent feature among all gulum is broad and thick, much more ex- known ursids. Such a condition is lacking in panded than in A. shackelfordi. An anterior A. makhchinus (although wear in this region spur of this cingulum is present near the base in the holotype of A. makhchinus renders our of the preprotocrista. M2 is missing. Its par- observation less certain). Finally, all ursoids, tial roots, however, indicate a transversely including the commonly acknowledged basal broadened M2 whose lingual border is more ursoids such as Amphicynodon, have a highly internal than that in the M1. Its labial border reduced parastyle and lingual cingulum on is ¯ush with that of M1, similar to that in A. M1, in sharp contrast to a still relatively dorog but in contrast to a lingually shifted prominent parastyle in A. makhchinus. M2 in A. shackelfordi. Conversely, everything about Amphicti- COMPARISON: Amphicticeps makhchinus is ceps makhchinus is consistent with other spe- the largest species of the genus so far known. cies of Amphicticeps, despite its modest de- It is 16% larger than A. dorog and 32% larg- viations toward the direction of hypocarni- er than A. shackelfordi (based on measure- vory. Our inclination to assign it to Amphic- ments of P4 labial length). It is 62% larger ticeps is further helped by the transitional than Amphicynodon teilhardi. Besides its nature of A. dorog between A. shackelfordi large size, A. makhchinus is also the most and A. makhchinusÐin just about every as- hypocarnivorous species in the genus. Dental pect of its dental morphology A. dorog bridg- features that indicate such hypocarnivory in- es the gap between the extremes in A. shack- clude an enlarged but low-crowned P4 pro- elfordi and A. makhchinus. In the ®nal anal- tocone, a reduction of M1 parastyle, expan- ysis, given what we know, it is easily con- sion of M1 lingual cingulum, a reduced angle ceivable that a series of three endemic between lingual borders of P4 and M1, and species of Amphicticeps form a clade in the an enlarged M2. early Oligocene of central Asia. Dental morphology of Amphicticeps makhchinus is reminiscent of certain ursids, Amphicynodon Filhol, 1881 particularly a basal ursid such as Cephalo- gale, so far known mostly in the Oligo- COMMENT: Cirot and Bonis (1992) recently Miocene of Eurasia and North America. In revised the taxonomy of the genus Amphi- particular, the French early Oligocene Quer- cynodon and furnished a cladistic relation- cy ®ssure ®lls produced some of the most ship for included species. However, their di- primitive forms (e.g., Cephalogale minor). agnosis of the genus is almost entirely based Similarities between A. makhchinus and Ce- on primitive characters (Cirot and Bonis, phalogale include an enlarged grinding part 1992: 105): ``arctoide primitive; craÃne alongeÂ of the dentition (M1±2) at the expense of the et bas, bulle ossi®eÂe, fosse suprameÂatale su- shearing part (P4). More speci®cally, A. per®cielle, canal de l'alispheÂnoide present,'' makhchinus has a low, shel¯ike P4 protocone which essentially describe the morphotypical and a quadrate outline on M1, features often condition for a basal arctoid but shed no light 28 AMERICAN MUSEUM NOVITATES NO. 3483 of whether or not the genus forms a natural beds'') of Hsanda Gol Formation, early Oli- clade. As such the concept of Amphicynodon gocene. remains a grade of small, primitive arctoids REFERRED SPECIMENS: AMNH 19014, right that are not easily placed in other genera of ramal fragment with p1±3, ®eld no. 73; carnivorans of similar ages. AMNH 19129, left ramal fragment with m1, Ten species of Amphicynodon were rec- from 10 mi west of Loh, ``Grand Canyon''; ognized by Cirot and Bonis (1992; see Bas- AMNH 21628, left maxillary fragment with kin and Tedford, 1996, for a possible North P2±4, ®eld no. 532; AMNH 21673, left ra- American species). With the exception of A. mal fragments with p2, m1±3, and alveoli of teilhardi and A. mongoliensis (see below), all c±p1, ®eld no. 531, ``Grand Canyon''; are from Europe and most are produced from AMNH 84198, isolated left m1, ®eld no. the classic Quercy ®ssure ®lls. As de®ned by 531, ``Grand Canyon''; AMNH 84212, left Cirot and Bonis, their concept of Amphicy- ramal fragment with m2±3 and m1 alveolus, nodon offers a measure of morphological ®eld no. 532; MAE SG.8162, left ramus with consistency, despite primitive status of most p2±m1 (broken); MAE SG.9198±201, basi- of their features. The limited scope of their cranium and maxillary fragments with left phylogenetic analysis (within the genus), P2 and P4±M2 (AMNH cast 129861), and however, does not permit a sense of overall right M1±2, from loc. MAE M-174, 2 mi relationships among basal arctoids, nor does southwest of the Loh, on east side of a ridge, it address the question of to what extent the 45Њ16Ј14ЉN, 101Њ46Ј02ЉE, found not far genus might be paraphyletic, given that cer- above a brown sandstone layer, a local equiv- tain derived forms (such as Pachycynodon alent of the basalt lava (MAE M-174 is still and Cephalogale) may have arisen from within the Ulaan Khongil fauna that contains within the genus. Such questions are dif®cult most specimens of Amphicticeps shackelfor- to address because most of the species of di in Shand Member); MAE SG.9193, right Amphicynodon are still represented by frag- maxillary fragment with P4±M2, from local- mentary jaws and teeth only. A comprehen- ity MAE 95-M-50, the Main Camp Locality, sive analysis of basal arctoids at the species Tatal Gol, in Tatal Member; MAE SG.95.7488, level is not feasible. Our Mongolian materi- left ramal fragment with broken p4 and m1; als, though signi®cantly improved over those PST 17/34, anterior half of skull with com- available for previous studies, are still not as plete right upper incisors, broken left and complete as their European counterparts. right canines, and complete left and right P2± M2, from Tatal Gol (Ulaan Khongil); PIN Amphicynodon teilhardi (Matthew and 475±3016, holotype of Cynodictis mongo- Granger, 1924) liensis (Janovskaja, 1970: ®gs. 2±3), partial Figures 10±14; Tables 2, 6, 7 skull with complete left and right upper teeth Cynodon (Pachycynodon) teilhardi Mat- and left ramus with p2±m3, from Tatal Gol thew and Granger, 1924: 9, ®g. 6D. (®g. 14); PIN 475±1388, partial palate with Amphicynodon teilhardi (Matthew and P3±M2 (Janovskaja, 1970: ®g. 4); ZPAL Granger): Mellett, 1968: 11; Lange-BadreÂ MgM III/96, right ramal fragment with p3± and Dashzeveg, 1989: 139; Cirot and Bonis, m2 and m3 alveolus, Tatal Gol (Lange-BadreÂ 1992: 119, ®g. 13; Dashzeveg, 1996: 3. and Dashzeveg, 1989: 139); and ZPAL MgM Cynodictis mongoliensis Janovskaja, 1970: III/97, left ramal fragment with p3±m1, Tatal 73, ®gs. 2±6. Gol. Amphicynodon mongoliensis (Janovskaja): DISTRIBUTION: Early Oligocene of north- Cirot and Bonis, 1992: 119. central Mongolian People's Republic. Dash- HOLOTYPE: AMNH 19007, left ramal frag- zeveg (1996: ®g. 1) reported that Amphicy- ment with m1±2 and m3 alveolus. nodon teilhardi occurs in both the lower Ta- TYPE LOCALITY: Loh, in Tsagan Nor Basin, tal Member and upper Shand Member of the eastern Valley of Lakes, Obor-Khangay Hsanda Gol Formation. An undescribed re- Province, in north-central Mongolian Peo- cord was reported in the early Oligocene ple's Republic. In Tatal Member (``lower red Khatan-Khayrkhan locality of Altai Province 2005 WANG ET AL.: CARNIVORANS FROM CENTRAL MONGOLIA 29 of Mongolia by Russell and Zhai (1987: process is small and does not have the dis- 324). tinct protrusion seen in Amphicticeps. The EMENDED DIAGNOSIS: As the only known orbit is relatively large and is of approxi- species from Asia, Amphicynodon teilhardi mately the same size as that of Amphicticeps, differs from all European species of the ge- which has a much larger skull. The distance nus, except A. velaunus, in its shortened m2, between the postorbital process and postor- along with a correspondingly reduced M2, in bital constriction is 7 mm on PST 17/34 and contrast to a primitively long m2 and large 9 mm on MAE SG.9198±201, signi®cantly M2 in most European species. A. teilhardi shorter than the 12 mm of Amphicticeps, and primitively retains a distinct posterior acces- thus has a far shorter temporal region than sory cusp on p4, which is lost or extremely the latter. Furthermore, the postorbital con- reduced in the European A. gracilis, A. spe- striction is not so narrow as in Amphicticeps. ciosus, and A. velaunus. A. teilhardi further The temporal crests are very indistinct, and differs from European A. typicus, A. gracilis, they converge more slowly toward the sag- and A. crassirostris in its relatively low hy- ittal crest than in Amphicticeps. In PST 17/ poconid of m1 with wrinkled enamel, in con- 34, the temporal crests do not fully converge trast to a trenchant talonid in the latter three at the posterior edge of the broken skull, and species. the sagittal crest, if present, is not preserved. DESCRIPTION: PST 17/34 offers the best In lateral view, the orbital region is best cranial morphology among all materials. Al- preserved on the right side of PST 17/34 though it is missing the posterior one-third (®g. 11). It is complemented by the partially of the skull, the remaining skull of PST 17/ preserved left orbital region of MAE 34 is nearly perfectly preserved and offers SG.9198±201. The orbital mosaic is quite ®ne details of bony and dental structures. similar to that of Amphicticeps in several re- Another cranial fragment, MAE SG.9198± spects: a short infraorbital canal, presence of 201, is less complete on the anterior part a shallow fossa in front of the antorbital rim (consisting of a heavily crushed partial ros- (less developed in PST 17/34), a small, trum plus left orbital region) but preserved a nearly rounded lacrimal bone with a lacrimal partial left basicranium. In addition, we are foramen near its anterior aspect, anterior pro- in possession of a cast of a partial skull and cess of jugal in contact with the lacrimal, and mandible from the collection of the Russian other topographic relationships among indi- Paleontological Institute, PIN 475±3016 (ho- vidual bony elements. Perhaps of phyloge- lotype of Cynodictis mongoliensis). In com- netic signi®cance is the shorter postorbital bination, much of the skull, except the pos- area between the postorbital process and terodorsal portion, is known. postorbital constriction. Skull (®gs. 10, 11): The overall proportion In ventral view, the incisive foramen (pal- of the skull is less specialized than that of atine ®ssure) is short and located at the level Amphicticeps. The rostrum is not shortened of the anterior aspect of the upper canine. A and broadened and the temporal region is not tiny foramen is present along the midline su- elongated, as in the latter. The premaxillaries ture at the junction of the premaxillary and form a thin blade on either side of the nasal maxillary (foramen palatine medialis of Sto- opening. The entire premaxillary is pre- ry, 1951), as is commonly seen in arctoids. served. The posterior process of the premax- The maxillary±palatine suture is mostly illary does not touch the anterior process of fused and dif®cult to recognize. The palatine the frontal, and ends near the posterior tip of foramen is somewhat behind the level of the the canine root at the level of P1. Both nasals P4 protocone. The posterior border of the are broken anteriorly, and their posterior tips palatine bone is anterior to the posterior bor- end at roughly the same level as the maxillary- der of the M2 and is distinctly indented by frontal suture. The frontal process inserts be- a semicircular notch on either side of the tween the nasal and maxillary and ends an- midline suture. teriorly at the level of the P2 main cusp. The The width of the rostrum across P1s (mea- frontal is slightly domed, in contrast to a ¯at sured on the lingual edge of the alveolus) forehead in Amphicticeps. The postorbital measured 9.4 mm in PST 17/34. That for 30 AMERICAN MUSEUM NOVITATES NO. 3483

Fig. 10. Amphicynodon teilhardi, PST 17/34, referred specimen. A, dorsal, B, lateral, and C, ventral views of skull. 2005 WANG ET AL.: CARNIVORANS FROM CENTRAL MONGOLIA 31

Fig. 11. Dorsolateral aspect of skull in Amphicynodon teilhardi (PST 17/34), showing anatomy of the orbital region. Abbreviations are the same as in ®gure 4, except for Nas, nasal and Prem, premaxillary. 32 AMERICAN MUSEUM NOVITATES NO. 3483

Fig. 12. Ventrolateral view of the left side of the basicranium of Amphicynodon teilhardi, MAE SG.9198±201. Cross-hatched areas indicate unprepared matrix. ac, alisphenoid canal; bo, basioccipital; ce, caudal entotympanic; cica, canal for internal carotid artery; ec, ectotympanic; fo, foramen ovale; fv, fenestra vestibuli (fenestra ovalis); hm, head of malleus; in, incus; la, lamina; ma, manubrium; mup, muscular process; pgf, postglenoid foramen; pgp, postglenoid process; plf, posterior lacerate foramen; pr, promontorium; sf, suprameatal fossa. the laterally crushed rostrum on MAE materials of MAE SG.9198±201 offer the SG.9198±201 measured 11 mm (restored only information about the basicranium of from distorted left and right halves). These Amphicynodon teilhardi. Although heavily compare with 15.2 mm for the same mea- crushed dorsoventrally, the left side of the surement in Amphicticeps shackelfordiÐthe basicranium of MAE SG.9198±201 pre- rostrum of Amphicynodon teilhardi is on av- serves several key anatomical features absent erage 49% narrower than the former. Given in Amphicticeps. The overall basicranial a size difference of 22% for the average morphology of Amphicynodon teilhardi is length of P4 between these two species, the somewhat similar to that of Amphicticeps. width of the rostrum in A. teilhardi is also The most obvious similarities are the pres- relatively narrower than that of the type spe- ence of a shallow suprameatal fossa and a cies. See additional cranial measurements laterally expanded squamosal blade for the (table 2) for PST 17/34. dorsal roof of the external auditory meatus. Basicranium (®g. 12): The fragmentary Much of the posterior half of the mastoid

→ Fig. 13. Teeth and lower jaw of Amphicynodon teilhardi. A, lateral view of upper teeth, PST 17/ 34; B, occlusal view of upper teeth, PIN 475±1388; C, occlusal view of upper teeth, PST 17/34, 2005 WANG ET AL.: CARNIVORANS FROM CENTRAL MONGOLIA 33

stereophotos; D, occlusal view of lower teeth, AMNH 19007, holotype, stereophotos; E, lingual, and F, labial views of lower jaw, AMNH 19007, holotype. All, except E and F, are photographs of a polyester cast. Scales ϭ 10 mm. 34 AMERICAN MUSEUM NOVITATES NO. 3483

Fig. 14. Stereophotos of a cast of PIN 475±3016 (holotype of Cynodictis mongoliensis Janovskaja, 1970). A, occlusal view of upper teeth and B, occlusal view of lower teeth. Scales ϭ 10 mm. process is lost. However, the process is less vated into the squamosal as in Amphicticeps, laterally protruded than in Amphicticeps, and it does not have so clear-cut a lateral rim judged from a more vertically oriented lateral as in the latter. wall of the braincase that forms a 90Њ angle The basisphenoid area is fragmented, and with the horizontal squamosal shelf, in con- the ventral ¯oor of the alisphenoid canal is trast to Amphicticeps, in which far more in- broken off. However, enough is preserved in clined lateral braincase walls (in posterior the area surrounding the foramen rotundum view) almost continue into the mastoid pro- to indicate the presence of a canal, that is, cess. The suprameatal fossa is slightly less the presence of a deep groove on the ali- well developed than in Amphicticeps. In par- sphenoid that probably forms its posterior ticular, its lateral half is not so deeply exca- opening. A small foramen opens into the me-

TABLE 6 Measurements of Upper Teeth of Amphicynodon teilhardi (in mm)

MAE AMNH SG.91.9198- MAE PST PIN PIN 21628 201 SG.9193 17/34 475-1388 475-3016 Length of C1 Ð Ð Ð 4.9 Ð 3.7 Width of C1 Ð Ð Ð 2.7 Ð 2.9 Length of P1 Ð Ð Ð Ð Ð 2.9 Width of P1 Ð Ð Ð Ð Ð 1.7 Length of P2 4.4 4.0 Ð 3.9 Ð 3.7 Width of P2 2.3 1.9 Ð 1.7 Ð 2.0 Length of P3 5.0 Ð Ð 4.6 5.4 4.8 Width of P3 2.9 Ð Ð 2.7 2.7 2.6 Labial length of P4 8.0 8.0 7.8 7.4 7.6 7.7 Lingual length of P4 (protocone to metastyle) 9.2 9.1 8.9 8.2 9.4 8.9 Anterior breadth of P4 5.8 5.9 5.3 4.9 5.4 5.4 Labial length of M1 Ð 5.6 5.3 5.8 6.2 6.0 Max. transverse width M1 Ð 8.6 7.9 8.2 8.8 Ð Longitudinal length of M2 Ð 2.1 2.4 2.3 2.5 2.5 Max. transverse width M2 Ð 3.7 3.7 4.4 4.5 4.3 Alveolar distance P1±M2 Ð Ð Ð 25.7 Ð Ð 2005 WANG ET AL.: CARNIVORANS FROM CENTRAL MONGOLIA 35

TABLE 7 Measurements of Lower Teeth of Amphicynodon teilhardi (in mm)

MAE MAE ZPAL ZPAL ZPAL AMNH AMNH AMNH AMNH AMNH AMNH SG. SG.95. MgM MgM MgM 19007 19129 21673 84212 84198 19014 8162 7488 III/96 III/97 III/98 Length of p1 Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Width of p1 Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Length of p2 Ð Ð 4.4 Ð Ð 4.3 Ð Ð Ð Ð Ð Width of p2 Ð Ð 2.2 Ð Ð 2.3 Ð Ð Ð Ð Ð Length of p3 Ð Ð Ð Ð Ð 4.8 4.5 Ð Ð Ð Ð Width of p3 Ð Ð Ð Ð Ð 2.6 2.3 Ð Ð Ð Ð Length of p4 Ð Ð Ð Ð Ð Ð 5.2 Ð 5.5 5.5 5.5 Width of p4 Ð Ð Ð Ð Ð Ð 2.7 Ð 2.6 2.5 2.9 Length of m1 8.6 7.9 8.7 Ð 8.9 Ð Ð 8.0 7.7 Ð Ð Trigonid length of m1 5.4 4.9 5.3 Ð 5.4 Ð Ð 5.1 Ð Ð Ð Trigonid width of m1 4.1 3.6 4.2 Ð 4.0 Ð Ð 3.6 3.8 Ð Ð Talonid width of m1 3.8 3.3 3.9 Ð 4.1 Ð Ð 3.6 Ð Ð Ð Length of m2 3.0 Ð 3.3 3.9 Ð Ð Ð Ð 2.8 Ð Ð Width of m2 3.2 Ð 3.4 3.9 Ð Ð Ð Ð 2.6 Ð Ð Length of m3 Ð Ð 2.2 2.0 Ð Ð Ð Ð Ð Ð Ð

dial wall of the canal at the level of its pos- The promontorial part of the petrosal has terior opening. been pushed dorsally toward the brain cavity, The postglenoid process is large, forming and its ventral surface is unnaturally rotated a long ventral hook for articulation with the into a vertical orientation. The ventral pro- condyle of the mandible. At the posterior montorial surface has thus become laterally base of the postglenoid process, behind the facing and is partially hidden beneath the lat- postglenoid foramen, there is a triangular eral edge of the basioccipital. The postmor- piece of ectotympanic still ®rmly attached to tem crushing of the promontorium into the the basicranium. Part of the anterior ectotym- brain cavity, however, is apparently bene®- panic ring for attachment of the tympanic cial for the preservation of a nearly complete membrane is also preserved. The ectotym- malleus and incus, which occupy the space panic does not extend laterally far beyond the originally occupied by the promontorium. postglenoid foramen, indicating no or a very The malleus lies on its medial side with short bony external auditory meatus. On the the lateral surface (for attachment of the tym- medial side, along the suture between pro- panic membrane) of the manubrium facing montorium and basioccipital, several broken ventrally. The malleus has a rather slender pieces of the entotympanic are still preserved construction but is of large sizeÐthe head to and cover the internal carotid canal. The ven- manubrium tip (broken) measures 5.8 mm. tral ¯ooring of the canal, presumably formed The lateral process is inconspicuous. The by the caudal entotympanic, forms a gentle muscular process (for insertion of m. tensor curve in a typical primitive arctoid fashion, tympani) is broken near its base and its main leading toward the middle lacerate foramen. body sticks out between the anterior process The presence of this medially positioned ca- and the manubrium instead of its original po- rotid canal further con®rms our conclusion sition, pointing away from the plane of tym- that there is no promontorial artery in Am- panic membrane. This slightly dislocated phicynodon and Amphicticeps, despite of the muscular process is very large and has a sulci on ventral surface of the promontorium broad distal end, which contrasts with the in the holotype of Amphicticeps shackelfordi much reduced condition in living ursids (Se- (see description under that species). No sul- gall, 1943). The neck is slender and forms a cus is visible on the promontorium of A. teil- smooth curve between the head and the ma- hardi. nubrium. The head is not enlarged as in pin- 36 AMERICAN MUSEUM NOVITATES NO. 3483 nipeds (Wyss, 1987). The sharp-tipped an- protocone apex is clearly continuous with the terior process completes the anterior rim of lingual cingulum. The cuspidate protocone a circular lamina. Much of the incus is buried has an indistinct ridge on the labial side of beneath the head of the malleus. Only the the cusp. The paracone has a distinct anterior processus brevis is fully exposed. ridge reaching up to the parastyle, and a less Upper teeth (®gs. 13A±C, 14): Most of the distinct anterolingual ridge that reaches to upper dentition is now known. Upper inci- the base of the protocone. A deep carnassial sors in PST 17/34 are all broken and the notch separates the paracone from the me- crown morphology is no longer preserved. tastylar blade. The roots indicate progressively enlarged in- The overall construction of M1 is less hy- cisors from I1 to I3, with the I3 almost twice percarnivorous than that of Amphicticeps. as large as the I1. All incisor roots are me- The M1 parastyle is substantially reduced as diolaterally compressed. The left and right compared to those of Amphicticeps, even for I3s in PIN 475±3016 are preserved and are the least developed M1 parastyle in A. makh- slightly precumbent in lateral view, as is also chinus. A vague notch separates the parastyle the case in PST 17/34. The upper canines and paracone. The labial cingulum is slightly on both sides are also broken in PST 17/34, swollen around the paracone in PIN 475± preserving only the roots. The canine roots 3016 and around the right M1 of PST 17/ are oval in cross section. The canines in PIN 34, similar to the swellings in Amphicticeps 475±3016 are present. The canine crowns are shackelfordi. The labial cingulum near the smooth surfaced and their tips curve back- metacone is not so reduced as in Amphicti- ward slightly. ceps; this is especially so in MAE SG.9193. The cheek teeth are evenly spaced with Together, the smaller parastyle and less re- short alveoli between all premolars in PST duced labial cingulum along the metacone 17/34 but slightly more tightly spaced in PIN give M1 a more quadrate look. The labial 475±3016, in contrast with crowded premo- borders of P4 and M1 form an angle of 127Њ, lars of Amphicticeps due to its shortened ros- 18Њ larger than in Amphicticeps shackelfordi. trum. P1 is only seen in PIN 475±3016, and The differences in size and height between is single rooted. It has a simple main cusp paracone and metacone are also relatively and an indistinct cingulum anteriorly and smaller than in Amphicticeps. There is a posteriorly. The P2s are single cusped and well-developed preprotocrista and postpro- are more slender than that of Amphicticeps. tocrista. There is no protoconule, except a A rather tall and erect main cusp has a pos- slight swelling in PST 17/34, which is ab- terior ridge and an anterolingual ridge. The sent in the Russian specimens. The postpro- anterior and posterior cingula are slightly tocrista is essentially posteriorly oriented, more distinct than in the P1s, but there is no particularly so in PST 17/34, leaving a cingular cusp on either end. The cingulum is broad valley between the postprotocrista and continuous on the lingual side and discontin- metacone. A metaconule is only vaguely sug- uous on the labial side. The main distinction gested by a low and indistinct swelling at the between P3 and P2, besides a larger size and posterior end of the postprotocrista and by a more prominent cingulum in the P3, is a slight notch toward the posterior end of the slight swelling on the posterolingual cingu- postprotocrista. The internal cingulum (hy- lum of the tooth, but there is no extra root pocone) surrounds the entire protocone, al- beneath this swelling. The morphology of P4 though its anterior segment is narrower. This is similar in overall construction to that of anterior extension in front of the protocone, Amphicticeps, except for a much smaller pro- less well developed in MAE SG.9193, has a tocone. A distinct cingulum surrounds the narrow spur near the base of the preproto- entire tooth. The anterior cingulum is partic- crista. ularly well developed, as is the parastyle. M2 has the same distinct shape as in Am- However, the parastyle is not cusplike but phicticeps, reaching the same stage of reduc- more like a wide cingulum. As in Amphic- tion and acquiring the same peculiar cusp ticeps, the anterior border of the P4 proto- pattern as in the latter, although this tooth cone is slightly ahead of the parastyle. The tends to vary more than does M1. A large 2005 WANG ET AL.: CARNIVORANS FROM CENTRAL MONGOLIA 37 paracone is located at the labial border of the largely crestlike and is oriented at a slight tooth, whereas the metacone is reduced to a angle from the anteroposterior axis of the faint cusp at the posterior border (that in tooth. Anteriorly, the hypoconid crest ends at PST 17/34 is more distinct). The protocone the base of the protoconid just below its is relatively large and is in the middle of the apex. The entoconid consists of a low ridge, tooth, followed lingually by a broad cingular rather like a cingulum. Together with the hy- shelf. The main feature to be distinct from poconid, the entoconid encloses a broad ba- that of Amphicticeps is its less lingually shift- sin on the talonid. The talonid cusps are dec- ed position. Instead of being ¯ush with the orated with ®ne wrinkles. The m2 is small lingual border of the M1 as in Amphicticeps and distinctly short; its length and width are shackelfordi, the lingual border of M2 in Am- nearly identical. No paraconid is visible. A phicynodon is slightly lateral to the M1 lin- protoconid and metaconid are of equal height gual border. and positioned on the borders of the tooth, Lower teeth (®gs. 13D±F, 14): Lower jaws such that a central valley essentially runs ®gured by Janovskaja (1970: ®gs. 3, 5, 6) through the length of the tooth. A small cris- substantially improved the knowledge of tid connects between the protoconid and Amphicynodon teilhardi over the original metaconid in the holotype but is absent in topotype series. The following descriptions PIN 475±3016. On the talonid, the hypocon- of the ramus are based on ®gures published id is far better developed than the entoconid, by Janovskaja. Her dental illustrations, how- which is absent in the holotype. A small m3 ever, lack suf®cient details for useful com- is present in all specimens that preserve this parisons, and have exaggerated the length of part of the jaw. It is formed by a small, the m2 (see table 7 and comparison below), rounded, peglike structure with a surround- a critically important feature of this species. ing cingulum but without a distinct cusp pat- Our dental descriptions are mainly based on tern. a cast of the PIN 475±3016, as well as more COMPARISONS: The topotype series of Am- fragmentary materials from the AMNH. phicynodon teilhardi consists of a few jaw The horizontal ramus is relatively slender fragments and lower teeth without upper compared to that of Amphicticeps. The as- teeth. Based on such meager materials, Mat- cending ramus has a rather erect anterior bor- thew and Granger (1924: 8) were initially der. The angular process is slender and point- ambivalent in their assignment of this species ed. No lower incisor is preserved. The lower to Cynodon (Amphicynodon of current us- canine hooks backward slightly. The lower age): ``This species can be referred only pro- premolars are relatively more slender than visionally until better specimens are avail- those of Amphicticeps. The p1 is not pre- able. It appears to fall within Pachycynodon served. The p2 has a simple main cusp and rather than the typical Cynodon, by Teil- a very vague cingulum. The p3 begins to hard's key to the Phosphorite genera,'' in ref- have a tiny posterior accessory cusp and a erence to Teilhard de Chardin's (1915) slightly more distinct cingulum. The p4 pos- monographic revision on Quercy carnivorans terior accessory cusp is further enlarged, and that dealt with these genera. The ®rst major its cingulum surrounds the entire tooth. The breakthrough in the state of knowledge of anterior cingulum has a hint of developing this species came by a crushed but associated into an anterior cingular cusp, and the pos- skull and lower jaw (PIN 475±3016; ®g. 14) terior cingulum is also slightly elevated. along with a few more specimens collected The m1 trigonid is tall crowned. The meta- by the 1946±1949 Expeditions of the Soviet conid is approximately the same height as the Academy of Sciences. The Russian collec- paraconid. The metaconid is on the lingual tion was described as a new species, Cynod- side of the protoconid, not trailing behind the ictis mongoliensis, by Janovskaja (1970). protoconid as seen in most ursids. A cingu- While describing three additional jaw frag- lum is present on the entire labial margin of ments collected by the Polish±Mongolian Pa- m1 but is only vaguely present on the lingual leontological Expeditions in the 1960s, side of the paraconid. The m1 talonid is low, Lange-BadreÂ and Dashzeveg (1989: 141), especially the hypoconid. The hypoconid is however, argued that ``there are no signi®- 38 AMERICAN MUSEUM NOVITATES NO. 3483 cant morphological or biometrical differenc- nodictis mongoliensis is synonymous with es between C. mongoliensis and A. teilhardi Amphicynodon teilhardi. The combined ma- and those that exist represent no more than terials from AMNH, PIN, ZPAL, and MAE intraspeci®c variation.'' More recently, Cirot allow more con®dent assignments of frag- and Bonis (1992), in their phylogenetic anal- mentary specimens and, as a result, increased ysis of the species of Amphicynodon, chose morphological cohesion of this species. to leave Amphicynodon mongoliensis as a Lange-BadreÂ and Dashzeveg (1989: 141) valid species, and in their cladogram, placed chose to compare the Mongolian form with it next to A. teilhardi as the sister-species three Quercy species of Amphicynodon: A. forming an Asiatic clade. typicus, A. leptorhynchus, and A. gracilis. Although we are unable to examine the They concluded that A. teilhardi was closer three jaw fragments in the ZPAL collection to A. leptorhynchus or A. typicus than to A. (Lange-BadreÂ and Dashzeveg's measure- gracilis. Characters that were cited to indi- ments for the holotype of A. teilhardi are ap- cate such a relationship include the wrinkled parently an underestimate), we have access enamel and a reduced m2 paraconid. A re- to casts of two PIN specimens in the topo- vision of the systematics and phylogeny of type series of C. mongoliensis. With the ad- Amphicynodon by Cirot and Bonis (1992: dition of even better materials from the MAE 119 and ®g. 16), on the other hand, recog- collections, we are in a position to evaluate nized 10 valid species, 8 European and 2 morphological variations of at least 20 spec- Asian, and suggested a relationship almost imens (see table 7). Although the increased opposite to that suggested by Lange-BadreÂ sample size naturally leads to a slight in- and Dashzeveg. Cirot and Bonis placed A. crease in variations, the coef®cient of varia- teilhardi (along with A. mongoliensis; see tion for most dental measurements generally comments above) as the sister-taxon to the falls between 5% and 8%, a range not un- terminal clade formed by A. gracilis and A. common for small carnivorans (e.g., Wang et cephalogalinus, whereas A. leptorhynchus al., 1999). The only exceptions are for the and A. typicus are further down the tree in upper canines, which tend to be more di- more basal positions. A critical synapomor- morphic among arctoids, and the m2s, which phy cited in support of above relationship have the least occlusal constraint because of was a ``trigonide de M/1 disjoint'' shared by their ¯at grinding surfaces. Cirot and Bonis A. teilhardi, A. gracilis, and A. cephalogali- (1992: 119 and ®g. 16) noted the rather long nus (Cirot and Bonis, 1992: ®g. 16, node 10). m2 in A. mongoliensis, apparently on the ba- In their remarks on A. teilhardi (Cirot and sis of Janovskaja's (1970: ®g. 3) illustration Bonis, 1992: 119), this character was ex- of the holotype, and assigned this presumed plained as an open trigonid due to a reduc- long m2 as an autapomorphy for the species. tion and a posterior position of the metaconid Our own examination of a plaster cast of PIN of m1 (``un trigonide ouvert en raison de la 475±3016 shows no elongation of the m2 reÂduction et de la position reculeÂe du meÂta- (®g. 14). In fact, our own measurements on conide''). Lange-BadreÂ and Dashzeveg's the cast indicate a shorter length than the character, a reduced m2 paraconid, was width for the m2 (3.0 mm in length and 3.2 pushed further down the tree by Cirot and mm in width), in sharp contrast to an elon- Bonis and was shared by A. velaunus, A. lep- gated m2 shown in Janovskaja's ®gure. It torhynchus, A. teilhardi, A. gracilis, and A. seems clear that the m2 length in Janovska- cephalogalinus. While it is beyond our scope ja's illustration was exaggerated (the lack of to reevaluate species relationships of Amphi- morphological details in her illustration of cynodon, we note that Cirot and Bonis's the m2 further undermines the reliability of character of a disjoint m1 trigonid is not her published line arts). In our examinations readily apparent in their illustrations of three of the rest of the dentitions, we failed to de- of the four species of Amphicynodon that are tect any substantial difference, either in size supposed to share it. On the other hand, their or shape, between the Russian collection and illustration of A. leptorhynchus (Cirot and the rest of the samples. We thus fully agree Bonis, 1992: ®g. 3), a species that was sup- with Lange-BadreÂ and Dashzeveg that Cy- posed to possess a primitive condition for 2005 WANG ET AL.: CARNIVORANS FROM CENTRAL MONGOLIA 39 this character, shows a more posteriorly dis- extremely shortened m2 of the latter. Indi- placed m1 metaconid than in any other spe- cating its primitive status, MAE SG.97.5396 cies. Our own examination of some of Cirot has a paraconid platform in front of the pro- and Bonis' materials in the UniversiteÂde toconid, and the paraconid is separate from Poitiers, which form the partial basis of their the protoconid by a weak notch. A paraconid systematic revision, fails to substantiate the is absent in the Hsanda Gol Amphicticeps validity of this character. Such a character, if and Amphicynodon. Furthermore, the proto- it does exist, must be quite subtle at this stage conid and metaconid are not widely separate of its evolution. In our observations, the from each other, as in both of the above gen- presence and absence of an m2 paraconid era. The morphological condition in MAE does seem to be a valid character that unites SG.97.5396 evokes that of a more primitive some species of Amphicynodon, including A. carnivoran, and the fact that it has an m3 teilhardi. alveolus tends to suggest a basal arctoid As for the generic status of the Mongolian among known lineages in this period of time. species, Lange-BadreÂ and Dashzeveg (1989: In size and overall proportions, MAE 141) rejected the initial suspicion by Matthew SG.97.5396 is closest to a recently described and Granger (1924) that the Hsanda Gol form small arctoid, Pachycynodon tedfordi Wang was closer to the European Pachycynodon and Qiu, 2003, from the early Oligocene than to Amphicynodon:``C. teilhardi belongs Saint Jacques locality of Chinese Inner Mon- unquestionably to the genus Amphicynodon. golia (see further comments below about its It differs from Pachycynodon in the situation generic status). Unfortunately, the m2 on the of the metaconid on m1, in the open basin, in Chinese specimen is too worn to permit de- the wrinkled enamel, in the reduced paraconid tailed comparisons. and entoconid in m2 and in the ratios of the talonid and trigonid on m2.'' Cirot and Bonis PARVORDER MUSTELIDA TEDFORD, 1976 (1992) more explicitly placed A. teilhardi among the rest of the European species of the SUPERFAMILY MUSTELOIDEA FISCHER DE genus. In light of the entire dentition available WALDHEIM, 1817 in this study, A. teilhardi falls within the over- Pyctis inamatus Babbitt, 1999 all parameters of the genus. However, since Amphicynodon, as a basal ursoid, was long COMMENTS: Pyctis inamatus was named on suspected to have given rise to other clades the basis of a left jaw fragment with p3 (bro- (e.g., Teilhard de Chardin, 1915), it is likely ken)±m1 and partial m2 alveolus from the a paraphyletic genus in a strict cladistic Tatal Gol area in a ``red claystone unit, SW senseÐvarious species may ultimately be of Maikant, 16.9 m above the lava layer in shown to be more closely related to other the Shand Member'' (Babbitt, 1999: 791). clades. Until a comprehensive, species-level With an m1 length of approximately 11 mm, phylogenetic analysis is done, current con- Pyctis is somewhat smaller than Amphicti- cepts of Amphicynodon remain largely gra- ceps makhchinus (arctoids tend to have a dational. See table 1 for more contrasts of somewhat shorter P4 than m1; the reverse is morphological differences between Amphicti- true between Pyctis and A. makhchinus) but ceps and Amphicynodon. falls in the range of A. dorog (see measurements in tables 3, 5). Pyctis is highly hyper- ?Cephalogale sp. carnivorous with an extremely trenchant m1 Figure 15A±C talonid, which has a single, centrally located hypoconid, along with the complete loss of REFERRED SPECIMEN: MAE SG.97.5396, m1 metaconid. This dental pattern is almost left ramal fragment with m2. exactly opposite to that in A. makhchinus, COMMENTS: MAE SG.97.5396 (®g. 15A± which tends toward the hypocarnivorous di- C) is clearly not assignable to the Hsanda rection. Pyctis is also more hypercarnivorous Gol ursoids discussed above. The size of the than A. dorog, which still retains a substan- m2 indicates an animal of the size of Am- tial m1 metaconid and entoconid. Another phicticeps dorog, but it does not have the important difference between A. dorog and 40 AMERICAN MUSEUM NOVITATES NO. 3483

Fig. 15. A, stereophotos of occlusal view of left m2, B, labial view of jaw, and C, lingual view of jaw of MAE SG.97.5396, ?Cephalogale sp. D, stereophotos of occlusal view of right m1±2, E, labial view of jaw, and F, lingual view of jaw of IVPP V10527.1, ?Amphicticeps sp. G, stereophotos of right M1, ?Amphicticeps sp. Scales ϭ 10 mm.

Pyctis is the presence of a posterior acces- Europe, such as Plesiogale and Paragale, sory cusp on the p4 of the former, which is and she also made comparisons with later lacking in the latter. Eurasian forms, such as Eomellivora and Babbitt (1999) compared Pyctis with some Mellivora. Given the poor condition of the basal mustelids from the early Miocene of holotype, she did not draw de®nitive system- 2005 WANG ET AL.: CARNIVORANS FROM CENTRAL MONGOLIA 41 atic conclusions, except placing it in the fam- ®g. 11), and A. crassirostris (personal obser- ily Mustelidae. vation on an uncataloged specimen of MuseÂe Pyctis also shares some resemblance to de Gaillac). Another prominent feature of certain oligobunines (a group of basal mus- IVPP V10527.1 is its high-crowned p4 that teloids) in the late Oligocene through early lacks of a posterior accessory cusp, a char- Miocene of North America (Baskin, 1998a), acter also relatively rare among known spe- such as Oligobunis and Megalictis, which cies of Amphicynodon. Only two European also have a highly hypercarnivorous lower species of Amphicynodon approach this con- carnassial and reduced posterior accessory ditionÐA. gracilis and A. speciosus (Cirot cusp on p4. On the other hand, Ischyrictis and Bonis, 1992: ®gs. 7, 11) lost the poste- and Hoplictis from early to middle Miocene rior accessory cusp, and a third species, A. of Europe also exhibit tendencies toward hy- velaunus, has nearly lost this cusp (Cirot and percarnivory (Ginsburg and Morales, 1992). Bonis, 1992: ®g. 1). It thus appears that IVPP The systematic status of Pyctis will remain V10527.1 is best compared with A. gracilis uncertain without additional materials, par- in terms of the p4. Our own comparisons ticularly its upper teeth. with a referred specimen of A. gracilis (MM 11, see Teilhard de Chardin, 1915: pl. VIII, COMMENTS ON CHINESE BASAL URSOIDS ®g. 2) further con®rm the above observation, although this specimen is much smaller than Huang (1993) described a right ramal frag- the Inner Mongolian form. On the other ment with p4±m1 (IVPP V10527.1) and a hand, the overall morphology of the m1 of left ramal fragment with m1 (IVPP IVPP V10527.1 is closest to that of Amphic- V10527.2) from the early Oligocene Ulan- ticeps, particularly in its narrowed talonid tatal locality, Ulantatal Formation, Inner relative to the laterally bulging trigonid, a Mongolia (®g. 1). He tentatively referred it condition Amphicynodon does not have. Giv- to? Amphicynodon sp. Huang considered the en the fragmentary nature of the Ulantatal Chinese materials to have a broad m1 talonid materials, it is dif®cult to choose between a and therefore similar to that of A. teilhardi. small but aberrant (in terms of the absence However, he was rightly cautious in his as- of a posterior accessory cusp of p4) form of signment of these jaw fragments to Amphi- the local Amphicticeps or a taxon close to the cynodon because of the poor preservation of European Amphicynodon gracilis. What the materials. Indeed, our own examination seems to be certain is that the Inner Mon- of a cast of IVPP V10527.1 (®g. 15D±F) led golian form is not closely related to A. teil- us to the following observations that the hardi. Ulantatal materials do not compare well with Most recently, Wang and Qiu (2003a) de- known Asian species of Amphicynodon and scribed three species of ursoids, based also Amphicticeps. on rather poor materials, from the early Ol- Contrary to Huang's description, the m1 igocene Saint Jacques locality (®g. 1), Wul- talonid is not broadÐit is substantially nar- anbulage Formation, Inner Mongolia. A left rower than that of Amphicynodon teilhardi. lower jaw with c±m2 (IVPP V12426) was Correlated to this narrow talonid, IVPP named a new species, Pachycynodon tedfor- V10527.1 also has a relatively trenchant m1 di; a maxillary fragment with M1 (IVPP talonid, that is, high crowned m1 hypoconid V12428) was referred to Amphicynodon sp.; dominating over a low entoconid crest, and and an isolated left M2 (IVPP V12429) was is in contrast to a ¯at, basin-like talonid in referred to Cephalogale sp. Of these, P. ted- A. teilhardi, as well as in most known Eu- fordi has the best material available and is ropean species of Amphicynodon. Of the 10 also the least similar to known forms in Asia. species of Amphicynodon recognized by Cir- For one thing, it has a rather long m2, unlike ot and Bonis (1992), only three European any known Asian Amphicynodon or Amphic- species are known to have a high hypoconid: ticeps, but is consistent with an elongated m2 A. typicus (FSP PC 46 from Pech Crabit, Cir- talonid in Pachycynodon. However, P. ted- ot and Bonis, 1992: ®g. 14), A. gracilis fordi also has narrow and trenchant cheek (MNHN QU 9192 in Cirot and Bonis, 1992: teeth, in sharp contrast to European species 42 AMERICAN MUSEUM NOVITATES NO. 3483 of Pachycynodon that tend to have broad gence point where their respective clades premolars and molars. Such relatively hyper- originated, and possess many derived char- carnivorous features are also seen in certain acters that tend to obscure true relationships. European specimens of Cephalogale. In par- In Carnivora, a limited morphological rep- ticular, a posteriorly expanded m2 hypoconid ertoire in which a clade can vary ensures that is frequently associated with a posteriorly few morphological characters are truly oriented M2 postprotocrista, a diagnostic fea- unique. The result is an overwhelming num- ture for Cephalogale. This long hypoconid ber of homoplasies that are dif®cult to dis- occludes with the posteriorly opened trigon tinguish from true synapomorphies whenever basin of the M1 due to the reorientation of the subjects have diverged from the point of its postprotocrista. IVPP V12426 is particu- origin for a suf®ciently long time. larly close to Cephalogale minor. For ex- A solution to such a problem seems to lie ample, FSP PFRA 32, from the Pech du in the studies of basal members of major Fraysse locality of the Quercy district in the clades, where the addition of fossils can re- late Oligocene (Paleogene biochronological veal relationships previously obscure (Gau- level MP 28; BiochroM'97, 1997), has pro- thier et al., 1988). Close to the point of di- portions very similar to the lower cheek teeth vergence, basal taxa did not undergo long pe- of IVPP V12426. C. minor, as currently con- riods of differentiation as the terminal mem- ceived, shows substantial variation (Beau- bers did, and thus are far less affected by mont, 1965; Bonis, 1973). The possibility homoplasies and more likely to reveal the that IVPP V12426 and V12429 may be con- true morphotypical status of the clade. The speci®c cannot be ruled out at the moment. disadvantage in such an approach, however, As for the M1 in IVPP V12428 (®g. 15G), is that basal members are often small, gen- it is larger and more robust than those of all eralized forms that commonly have poor fos- species of Amphicynodon that have pre- sil records, a problem that usually eases served upper teeth (e.g., A. velaunus, A. lep- somewhat over time as fossil records im- torhynchus, and A. teilhardi). Furthermore, prove. Such is the case in the basal arctoids its parastyle area is also better developed as afforded by recent additions in Hsanda than those of all known species of Amphi- Gol materials. However, even where the fos- cynodon. In that sense, IVPP V12428 is clos- sil records are becoming adequate, it is im- er to Amphicticeps dorog than to Amphicy- perative to ascertain the memberships of the nodon. The locations of the M2 roots also basal forms in the major clades. It is there- indicate an M2 equivalent in size to that of fore a matter of considerable importance to A. dorog. Beside such tentative observations, make sure that our selections of basal taxa however, a more de®nite judgment will have are truly representative of their clades, and to await further discovery. the following discussion attempts to detail our concept of the basal taxa selected for PHYLOGENETIC ANALYSIS analysis. Our classi®cation scheme follows that by McKenna and Bell (1997). CONCEPTS OF MAJOR CLADES AND Cynoidea: The infraorder Cynoidea con- SELECTIONS OF BASAL TAXA tains living and fossil relatives of foxes, Our phylogenetic analysis of the Arctoidea wolves, and jackals in the family Canidae. stems from observations that past studies of Derived from the archaic, paraphyletic fam- higher level relationships are often based on ily Miacidae, Canidae is the ®rst to arise living taxa (crown groups) or those that be- among living families of Carnivora and tends long to terminal clades. While such an ap- to have the most conservative morphology. proach has the bene®t of completeness of in- The basal canids are represented by excel- formation (morphological as well as molec- lently preserved materials from the late Eo- ular) and taxonomic clarity (memberships of cene to early Oligocene of North America, living taxa are rarely in doubt), with the ex- and a common basal taxon, Hesperocyon, ception of a few ``living fossils'' (such as serves as an ideal outgroup for the present African palm civet Nandinia), living taxa too analysis both in terms of its primitive mor- often are far removed from the initial diver- phology and its closeness to the basal di- 2005 WANG ET AL.: CARNIVORANS FROM CENTRAL MONGOLIA 43 chotomy of the Cynoidea and Arctoidea burg and Morales, 1995; Hunt, 1998c) and is (Tedford, 1978; Wang, 1994; Wang and Ted- ideally suited as a representative basal taxon ford, 1994, 1996; Tedford et al., 1995). for the family. Even among its primitive spe- Arctoidea: The infraorder Arctoidea rep- cies, such as C. minor from the French Quer- resents the largest group of predators, with cy phosphate ®ssures, such ursid synapo- 211 extinct and 56 living genera (McKenna morphies as posteriorly oriented M2 postpro- and Bell, 1997). It includes living families of tocrista, elongated m2, and reduction of pre- bears (Ursidae), seals, sea lions, and walrus molars are already recognizable. (Phocoidea ϭ Pinnipedia), red pandas (Ail- Phocoidea (ϭ Pinnipedia): Extreme uridae), raccoons (Procyonidae), weasels, aquatic adaptations in the pinnipeds tend to badgers, and otters (Mustelidae), as well as obscure the primitive conditions of ancestral the extinct bear dogs (Amphicyonidae). Arc- forms. Nowhere is this more apparent than toids feature an unsurpassed diversity of ter- in the secondarily undifferentiated teeth of restrial and aquatic adaptations and the wid- living pinnipeds, which bear practically no est extremes of dental specializations. Per- resemblance to those of terrestrial arctoids. It haps because of these extremes in morpho- is not surprising that pinniped relationships logical and ecological variations, arctoid have been the most controversial among arc- relationships have been highly controversial toids. Controversies still exist regarding the and consistent resolutions have proven elu- monophyletic versus diphyletic origins of the sive so far. Attempts at phylogenetic synthe- pinnipeds (e.g., Mitchell and Tedford, 1973; sis are usually con®ned to forms that are well Tedford, 1976; Wiig, 1983; Wyss, 1987, de®ned enough within a clade, often leaving 1988; Berta and Wyss, 1994; Flynn and Ned- the critically important basal forms unattend- bal, 1998; Berta and Adam, 2001) or whether ed. We will include most early arctoid genera pinnipeds arose from ursoids or musteloids in our analysis (see below). (e.g., Wolsan, 1993; Hunt and Barnes, 1994). Ursida (ϭ Ursoidea): Within the infraor- For the present analysis, we assume the ter- der Arctoidea, parvorders Ursida and Mus- restrial Pachycynodon and its sister taxon Al- telida represent two ®rst-order dichotomous locyon to be stem groups in the pinniped clades. Ursida includes the living Ursidae clade (Tedford et al., 1994), although the and the aquatic Phocoidea (ϭ Pinnipedia), as aquatic Potamotherium has been argued to well as the extinct bear-dogs (Amphicyoni- play a role in the origin of the pinnipeds dae). The early Oligocene Amphicynodon is (e.g., Tedford, 1976; Wolsan, 1993) or not probably one of the most primitive members (Wyss, 1991). We also include Potamother- of Ursida known, and it plays a critical role ium in this analysis. in anchoring the Ursida clade. A North Amphicyonidae: Traditionally thought to American group of basal Ursida includes the be closely related to canids, mostly due to genera Parictis, Subparictis, and Nothocyon their conservative dental morphology (e.g., from the Chadronian to early Arikareean Matthew, 1924; Kuss, 1965), amphicyonids (late Eocene to late Oligocene; Clark and (bear-dogs) have a rather arctoid basicranium Guensburg, 1972; Wang and Tedford, 1992; (Hough, 1948; Hunt, 1974, 1977, 1996a, Baskin and Tedford, 1996). Placed in a small 1998a). More speci®cally, many amphicyon- clade of its own, Subparictidae (Baskin and ids share with ursids a greatly enlarged in- Tedford, 1996) is still too poorly known to ferior petrosal sinus that houses a double- be included in this analysis. looped internal carotid artery, presumably a Ursidae: Living members of the ursids are counter current exchange mechanism for the morphologically well de®ned by their hypo- cooling of the brain. It has become increas- carnivorous dentitions. Fossil ursids, how- ingly accepted that amphicyonids are arc- ever, include some rather hypercarnivorous toids and may be closely related to the ursids taxa but have never achieved the extreme hy- (Flynn et al., 1988; Viranta, 1996; Ginsburg, percarnivory seen in mustelids (Hunt, 1999). The most primitive amphicyonids ap- 1998c). The mesocarnivorous Cephalogale is pear to be the late Eocene Daphoenus of widely regarded as the most primitive ursid North America and Cynodictis of Europe; (e.g., Beaumont, 1965; Bonis, 1973; Gins- both gave rise to later clades in their respec- 44 AMERICAN MUSEUM NOVITATES NO. 3483 tive continents (Kotsakis, 1980; Hunt, 1996a, lationship (e.g., McLaren, 1960; Tedford, 1998a, 2001). Unfortunately, the basicranial 1976; Muizon, 1982; Wolsan, 1993) or lack systems of these basal forms have not been of it (e.g., Wyss, 1991) between Potamoth- studied adequately, and we exclude them erium and phocid pinnipeds is out of the from our analysis. scope of this paper, recent morphological and Mustelida (ϭ Musteloidea): Outside the molecular studies seem mostly in favor of a clades represented by Mustelidae and Pro- monophyletic Pinnipedia with an ursoid ori- cyonidae lie the basal musteloids, which, gin (see Berta and Wyss, 1994, for a recent lacking all or parts of the derived features, summary), and Potamotherium may thus be do not ®t neatly in any particular pattern and decoupled from discussions about pinniped presumably occupy a place in a basal bush phylogeny (Wyss, 1991), that is, aquatic ad- (Wolsan, 1993). This loosely de®ned muste- aptations in Potamotherium are hypothesized loid stem group includes the archaic Am- to be derived independently (but see Wolsan, phictis, Mustelictis, Bavarictis, Stromeriella, 1993, for an alternative argument). and Plesictis from the Oligo±Miocene of Eu- A very primitive musteloid, Mustelavus, rope, in addition to some less well known from the late Eocene (Chadronian) to early taxa. Recent attempts to resolve relationships Oligocene (Orellan) of North America, ap- of these forms have met with limited success parently occupies the most basal position in (Schmidt-Kittler, 1981; Cirot and Bonis, the Mustelida (see Baskin and Tedford, 1993; Wolsan, 1993), because of the increas- 1996). Although the original holotype skull ing realization that nearly all characters tra- was crushed, contributing to its relative ob- ditionally thought to be consistent and reli- scurity since its original description, a new, able are subject to homoplasies of greater or undescribed skull is available in the AMNH lesser degrees. In North America, a similar collection that supplies important informa- group of archaic musteloids occurs in the tion in the present analysis. As shown in our Arikareean through Barstovian time (late Ol- phylogeny, Mustelavus occupies the most igocene to middle Miocene). Referred to here basal position of all Mustelida, and helps to as the Oligobuninae (Baskin, 1998a), it in- establish its morphotypic status for the entire cludes Promartes, Oligobunis, Aelurocyon, Mustelida clade. Megalictis, Zodiolestes, and Brachypsalis. Ailuridae: The highly derived and enig- Riggs (1945) proposed a linear transforma- matic Asiatic red (lesser) panda, Ailurus (and tion series from the primitive Promartes, closely related Parailurus), and the related through the intermediate Oligobunis,tothe hypercarnivorous Simocyon form a small hypercarnivorous Aelurocyon and Megalic- clade of their own, and their controversial tis. Brachypsalis is derived in its own pecu- phylogenetic position (e.g., Ginsburg, 1982; liar way (presence of a P4 parastyle, enlarged Wang, 1997) is increasingly pushed toward M2, broadened lower molars, etc.), and its the base of the Arctoidea in recent studies of relationship with respect to Riggs' phyloge- molecular phylogenies (Vrana et al., 1994; netic framework is unclear. Zodiolestes,on Flynn et al., 2000). Unfortunately, fossil re- the other hand, has a deeply excavated su- cords for Ailuridae are extremely rare and prameatal fossa and lacks a posterior crest on recent, with the exception of some dental the mastoid process, and has been argued to fragments suggestive of red panda in the be a procyonid (Hough, 1948; Schmidt- middle Miocene of Europe (Ginsburg et al., Kittler, 1981). 2001), the earliest being the single tooth of Another part of the basal musteloid radi- Pristinailurus from the late Miocene or Pli- ation leads to an aquatic form, Potamother- ocene of North America (Wallace and Wang, ium, in Eurasia and North America (Seman- 2004) leaving a long gap of possibly ϳ30 torinae Orlov, 1933). Potamotherium (see My in the fossil record. We have no choice Savage, 1957, for a detailed description) is but to use the living genus Ailurus in the pre- sometimes cited as a primitive pinniped, and sent analysis. However, its highly hypocar- thus is a contentious taxon in the controversy nivorous dentition is obviously far removed on monophyletic/diphyletic origin(s) of the from the morphotypical condition of the ail- Phocoidea. While study of a sister-group re- urid clade. The Ailuridae is the only instance 2005 WANG ET AL.: CARNIVORANS FROM CENTRAL MONGOLIA 45 in our analysis in which a basal taxon is not (Schmidt-Kittler, 1981; Bryant et al., 1993; available, and our result cannot be complete- Wolsan, 1999), despite its apparently more ly free of the in¯uence of homoplasies. Its primitive upper molars, bullar compositions, sister group Simocyoninae has somewhat and brain morphology (Radinsky, 1973; better fossil records, which go back to Alo- Hunt, 1974; Wang and Qiu, 2004). Recent pecocyon in the middle Miocene of Eurasia molecular studies (e.g., Ledje and Arnason, (Viret, 1951; Ginsburg, 1961; Wang et al., 1996a, 1996b; Dragoo and Honeycutt, 1997; 1998). Alopecocyon is mostly known by Flynn and Nedbal, 1998), however, have in- fragmentary materials. We thus use the ter- creasingly indicated a more basal position for minal genus Simocyon, which is relatively the skunks in the Mustelida or even Arcto- better known (Wang, 1997). idea, raising Mephitinae to family rank. Mor- Procyonidae: Although presently con®ned phological studies, on the other hand, have to North and South America, procyonids not kept up with the apparent molecular ad- probably originated in the Old World (Bas- vancement. However, a middle Miocene ge- kin, 1982, 1989, 1998b; Wolsan, 1993). Liv- nus from Europe, Palaeomephitis, has been ing and most fossil procyonids can be diag- identi®ed as the earliest mephitid (Wolsan, nosed by an elongated m2 talonid and a mod- 1999). Palaeomephitis features a highly au- est suprameatal fossa (e.g., Schmidt-Kittler, tapomorphous dental morphology (e.g., 1981; Baskin, 1982, 1989, 1998b; Flynn et Fraas, 1870; Wegner, 1913; Wolsan, 1999), al., 1988; Wyss and Flynn, 1993). The most which almost certainly rules out its being in primitive forms of this clade appear to be the a basal position of the family. Until a more late Oligocene to early Miocene European appropriate stem group is identi®ed, we ten- Pseudobassaris, Angustictis, and Broiliana tatively leave the mephitids aside, pending (Wolsan, 1993). We use Broiliana as a mor- further investigations. photypical synecdoche for the Procyonidae, which is agreed upon by recent workers (e.g., CHARACTER ANALYSIS Baskin, 1982; Wolsan, 1993). Discrete morphological characters are cod- Mustelidae: Living members of the family ed to compile a 22 taxa ϫ 39 characters data Mustelidae constitute the largest and most di- matrix (table 8) for parsimony analysis. Most verse group of terrestrial arctoids. Most taxa characters used in this study have been men- can be diagnosed by a loss of a carnassial tioned by previous authors, although often in notch, ventrally enclosed suprameatal fossa, a different context and carrying different and loss of the M2 (e.g., Tedford, 1976; meanings. Our own contribution is mainly in Schmidt-Kittler, 1981; Flynn et al., 1988; the reexamination of the characters in a Bryant et al., 1993; Wyss and Flynn, 1993; broader context, thus showing their relevance Baskin, 1998a), although some homoplasies in the basal relationships of the arctoids. By for these characters may have existed (Qiu necessity only cranial and dental characters, and Schmidt-Kittler, 1982; Wolsan, 1993). commonly the best preserved in fossil re- The absence of the alisphenoid canal, shared cords, are used in this analysis. See Appen- by living members of mustelids and pro- dix 1 for a list of character descriptions and cyonids, is sometimes cited as a synapomor- their states that correspond to those in the phy uniting the two families, although it may data matrix. have been independently acquired (Wolsan, 1993; Wang, 1997). We follow Wolsan PARSIMONY ANALYSIS OF BASAL ARCTOIDS (1993) and accept that European early Mio- cene Paragale and Plesiogale represent the Using Miacis and Hesperocyon as out- most primitive mustelids (mustelines of Wol- groups, we performed a maximum parsimo- san, 1993). These genera, along with a re- ny analysis on the 22 taxa ϫ 39 characters cently discovered Kinometaxia from China data matrix compiled in the above Character (Wang et al., 2004), are included in the anal- Analysis using the computer program PAUP ysis. (Swofford, 1993). Searches using the Mephitidae: The skunks are traditionally ``Branch and Bound'' option yielded 36 treated as a subfamily in the Mustelidae shortest trees of 89 steps in length. A boot- 46 AMERICAN MUSEUM NOVITATES NO. 3483

TABLE 8 A 22 Taxa ؋ 39 Characters Data Matrix Used in the PAUP Analysisa

123 12345 67890 12345 67890 12345 67890 12345 6789 Miacis 00000 0?000 00000 00000 00000 00000 ?0000 0000 Hesperocyon 00000 0?000 01000 00000 00000 00001 10001 1110 Mustelavus 01000 ??000 01000 ?0000 01000 ??101 ?100? ??11 Amphictis 01001 0?000 02000 00031 01020 00001 ?1001 1?11 Simocyon 01001 1?001 02000 31031 21021 11001 01101 1011 Ailurus 01001 00100 22000 30231 21221 11001 00001 1011 Broiliana 00112 00000 12000 30231 21200 00001 ??001 1?1? Stromeriella 00111 01001 12000 00031 01000 ??001 ?100? ??1? Mustelictis 00002 00000 02010 00100 011?0 ??001 ?1001 1?11 Bavarictis 00001 00000 02000 00100 011?? ?0001 ??001 1?1? Pseudobassaris 00011 00000 02000 00100 0110? 00001 0?001 101? Plesictis 11112 00000 02000 10110 011?0 00001 ??001 1?11 Promartes 01110 00000 01000 10110 11100 00101 ?1001 1011 Potamotherium 01110 00001 11001 10110 11100 00101 01011 1011 Kinometaxia 11113 ??110 0110? 2???? 1?10? ??101 ?100? ??11 Paragale 01113 0?010 0110? 21120 11100 00101 ??01? ??11 Plesiogale 01113 01010 0110? 22120 111?? ??101 ??00? ??11 Amphicticeps 00001 0?000 01001 10010 10010 00101 ?1011 1?11 Amphicynodon 00000 0?000 02000 00000 00010 00011 ?1011 1?11 Cephalogale 00000 ??001 02000 30030 20010 0?011 01111 1?11 Pachycynodon 01000 ??001 12000 30000 2001? ?0011 01111 1011 Allocyon 01000 0?001 12000 30000 200?0 ?00?1 01111 1?11 a Character numbers correspond to those listed in appendix 1. strap analysis (®g. 16) shows relative support phylogeny thus captures a sense of realism for individual nodes and a strict consensus of to re¯ect the early radiations, something that the 36 trees is shown in ®gure 17. Despite a is not possible by studying terminal members relatively poor bootstrap support, many of of the clades. Such a sense of intimacy in the basal arctoids are well resolved in the observing early diversi®cations, as afforded consensus tree, except parts of the basal bush by the richness of fossil record, is a strength of the Mustelida, which have been dif®cult of paleontology even in light of the increas- to resolve due to an unusually high diversity ingly large data set afforded to neontology. in the initial radiation of the Mustelida in the The following paragraphs are brief narra- late Oligocene to early Miocene of Europe tives to highlight salient features of our phy- (see Wolsan, 1993). logeny. As shown in the character optimi- Most of the major clades in our phylogeny zations in ®gure 17, the crown portion of the (®g. 17) are supported by one or two derived suborder Caniformia (Cynoidea ϩ Arcto- characters. Such a small number of synapo- idea) is well differentiated from the archaic morphies is not necessarily an indication of ``miacids'' by several derived characters: weak support for the respective clades. In- presence of an ossi®ed bulla, a medially po- stead, this is a re¯ection of the morphologi- sitioned internal carotid artery, reduction of cal closeness of all basal taxa analyzed (with M1 parastyle, and loss of M3. Infraorder Cy- the exception of Ailurus). Such a morpho- noidea is also well supported by autapomor- logically closely spaced array of taxa natu- phies that are unique to the canids, such as rally yields few characters. We would argue a type B bulla with enlarged caudal entotym- that our use of basal taxa closely approxi- panic and a semi-septum within the bullar mates the early divergence of arctoids and chamber (Hunt, 1974; Wang and Tedford, the strength of our phylogeny lies in the short 1994). morphological distance between taxa. Our Infraorder Arctoidea is supported by two 2005 WANG ET AL.: CARNIVORANS FROM CENTRAL MONGOLIA 47

based on living taxa alone often list the loss of the alisphenoid canal and an elongated external auditory meatus as key synapomorphies for the arctoids (e.g., Flynn et al., 1988; Wyss and Flynn, 1993). These characters are almost certainly independently derived within various lineages of arctoids, when they are seen in the broader perspectives that include basal arctoids shown here. Within the arctoids, the commonly accepted Ursida±Mustelida dichotomy is also re- ¯ected in our own phylogeny. The Mustelida includes all taxa that have lost the last lower molar (m3), as a commonly used synapomorphy (e.g., Tedford, 1976; Schmidt-Kit- tler, 1981). This loss of the last lower molars, which happens numerous times in mammals, was justi®ably suspected to be a weak character (Schmidt-Kittler, 1981), and our dem- onstration in this study that Amphicticeps has independently lost its m3 adds further evidence of the homoplasticity of such a character. Nonetheless, it remains as the only derived character uniting the Mustelida. The North American Mustelavus is at the base of the entire Mustelida, and appears to be a stem taxon outside the mustelid±procyonid± ailurid clade. European basal Mustelida are united by their possession of a small suprameatal fossa and reduced M1 parastyle, as demonstrated by Schmidt-Kittler (1981) and Wolsan (1993). A dichotomy between mustelids plus stem groups and procyonids plus stem groups (including ailurids) is supported by an enlarged m2 with an elongated talonid on the procyonid side, and a lingual notch in front of m1 entoconid and loss of M1 postprotocrista on the mustelid side. Presence of a suprameatal fossa has long been considered as a key synapomorphy for procyonids (Riggs, 1942, 1945; Segall, 1943; Hough, 1944, Fig. 16. Bootstrap analysis based on the data 1948), but this character has been further matrix in table 8. Numbers at the nodes indicate elaborated in the mustelids plus their stem percent bootstrap support from heuristic search groups to be the basis of a new phylogenetic based on 100 replicates. Only groups with a fre- scheme of the mustelids (Schmidt-Kittler, quency of greater than 50% are retained. 1981; Wolsan, 1993). Elsewhere, Wang et al. (2004) explored the basal mustelid relationships in greater detail than is afforded here, synapomorphies: a reduction or loss of pos- including additional stem taxa not analyzed terior accessory cusps on premolars and a here. widening of the basioccipital between the Two derived characters help to push Am- bullae. Traditional phylogenetic assessments phicticeps toward the Ursida side: a dorso- 48 AMERICAN MUSEUM NOVITATES NO. 3483

Fig. 17. Cladistic relationships of basal arctoids based on a strict consensus tree from 36 shortest trees (tree length ϭ 89 steps) that were recovered from Branch and Bound method in the PAUP program (Swofford, 1993 version 3.1.1) on the 22 taxa ϫ 39 characters data matrix (table 8). Character distri- butions are optimized using ClaDos program (version 1.2 by Kevin Nixon): solid bars are synapomorphies and open bars are homoplastic characters (parallelisms and reversals); numbers above the branches are character numbers and those below the branches are character states. 2005 WANG ET AL.: CARNIVORANS FROM CENTRAL MONGOLIA 49

ventrally compressed, laterally expanded par- RELATIONSHIPS WITHIN AMPHICTICEPS: Giv- occipital process and a long distance between en the limitation of the fragmentary materials the postorbital process and the postorbital in two of the three recognized species of Am- constriction. Such characters are often ho- phicticeps, a comprehensive phylogenetic moplastic and, in the present case, are not analysis using the full spectrum of cranial enough to withstand the bootstrap analysis and basicranial characters listed above is not (®g. 16). The highly autapomorphic Amphic- feasible. Instead, we tacitly assume that cra- ticeps does not easily conform to any exist- nial and basicranial characters for A. dorog ing scheme of systematics, as shown by pre- and A. makhchinus are similar to those of A. vious controversies (see above), and the pre- shackelfordi and that they mainly differ by sent placement within the Ursida is also dental morphology and size. Based on dental somewhat tentative. In favor of this place- characters, relationships among species of ment, however, is a laterally expanded mas- Amphicticeps appear to fall in a linear pro- toid process in Amphicticeps, a character gression, starting from the small A. shackel- most prominently developed in the ursids fordi, transitioning through the intermediate and pinnipeds, but here treated as an auta- A. dorog, and ending in the large A. makh- pomorphy by the parsimony program be- chinus. cause of a relatively small mastoid process Using Miacis and Hesperocyon as out- in Amphicynodon. groups, character transformations within Am- As recovered in the bootstrap analysis, un- phicticeps can be deduced to show the fol- doubted Ursida starts at Amphicynodon with lowing trends. A basal status for A. shack- its incipient development of a posteriorly ori- elfordi is indicated by such primitive char- ented postprotocrista of M1 and a very re- acters as relatively less robust dentitions, duced parastyle, characters that are further small but high-crowned P4 protocone, large elaborated in ursids. The position of Cephal- M1 parastyle, and the presence of an m3 in ogale as a basal ursid has been widely ac- some individuals. A. shackelfordi may have cepted (e.g., Beaumont, 1965; Bonis, 1973; one autapomorphy of its own: a reduced and Ginsburg and Morales, 1995; Hunt, 1998c). lingually shifted M2, a condition that paral- Its sister relationship to the phocoids is sup- lels that in the basal musteloid Potamother- ported by a crestlike P4 protocone that is ium. posteriorly positioned, enlarged M2, and en- Amphicticeps dorog is obviously more de- larged mastoid process. These derived char- rived than A. shackelfordi in the following acters exclude Amphicynodon from being characters: larger size and more robust den- placed within the Ursidae±Phocoidea clade, titions and jaws, a more crestlike P4 proto- and renders the Amphicynodontidae (Simp- cone that is relatively lower crowned, a son, 1945; Cirot and Bonis, 1992; Baskin slightly more distinct P4 anterior cingulum, and Tedford, 1996; Hunt, 1996b) a paraphy- a reduced M1 parastyle (and consequently a letic group. larger angle between the labial borders of the Here, we assume that the Phocoidea is P4 and M1), and the loss of an m3. A. dorog monophyletic, based on recent morphologi- also has a larger and more labially positioned cal and molecular studies (Wyss, 1987; Wyss M2. and Flynn, 1993; Berta and Wyss, 1994; Amphicticeps makhchinus is apparently Vrana et al., 1994; Flynn and Nedbal, 1998). the most derived species of the genus. Com- The basal phocoid Enaliarctos has been pared to A. shackelfordi and A. dorog, A. demonstrated to be more ursoid than muste- makhchinus is the largest in size, has the loid in its basicranium and dentition (Hunt broadest P3, possesses the lowest and most and Barnes, 1994) and the terrestrial Pachy- enlarged P4 protocone crest, and features the cynodon and Allocyon are probably the clos- most reduced M1 parastyle, the largest M1 est stem taxa for the phocoids (Tedford et al., metaconule, and the greatest expansion of 1994). Under our phylogenetic scheme, Po- M1 lingual cingulum. A. makhchinus also tamotherium must have acquired its aquatic hints at a relatively large and labially posi- adaptations independently from the pho- tioned M2. Most of the above derived char- coids. acters in A. makhchinus suggest a modest 50 AMERICAN MUSEUM NOVITATES NO. 3483 tendency toward hypocarnivory, paralleling As for the more basal amphicyonids, North some dental characters of the ursids. American Daphoenus is similar in stage of evolution to the European Cynodictis, and ZOOGEOGRAPHIC COMMENTS both of them ®rst appear in the late Eocene (Hunt, 1996a, 1998a). Although some au- The theater of evolution for early arctoids thors are inclined to treat the above North centers around the northern continents (Hol- American basal arctoids as occasional im- arctic region), at a time when southern con- migrants from Eurasia (e.g., Hunt, 1996b), tinents are wholly or partially isolated from given that the North American forms are ei- the northern continents. On the caniform side ther more primitive and/or earlier in occur- of the Carnivora, the archaic and paraphylet- rence, a North American origin of the Arc- ic family Miacidae, from which arctoids are toidea seems an equally likely scenario. In presumably derived, was present in all three favor of such an argument is the sister rela- northern continents: Europe, Asia, and North tionship between cynoids and arctoids, and America (Hunt, 1996b; McKenna and Bell, the former is undoubtedly North American 1997). First to arise from the miacids is the in origin. Cynoidea (including family Canidae) from The three species of Amphicticeps appar- the Bridgerian and Uintan (middle Eocene) ently form an endemic clade con®ned to cen- in the form of ``Miacis'' sylvestris and ``Mia- tral Asia. Their zoogeographic origin is cur- cis'' gracilis of North America (Wang and rently unknown. Amphicynodon has a much Tedford, 1994) followed by the true canids higher diversity in Europe than in Asia, and Prohesperocyon and Hesperocyon in the Du- phylogenetically the Asian A. teilhardi seems chesnean and Chadronian (late Eocene; to be nested within the European congeneric Wang, 1994). The rest of canid history was species, indicating an eastward dispersal for mostly played in the North American conti- this group and linking the European ``Grande nent until late Miocene, when some members Coupure'' and the Asian ``Mongolian Recon- ®nally crossed the Bering Strait to enter into struction'' events as land connection began Eurasia (Crusafont-PairoÂ, 1950), and later in to form between Asia and Europe in the early the Pliocene, when canids crossed the Pana- Oligocene (Meng and McKenna, 1998). manian Isthmus to arrive in South America (Berta, 1987, 1988). ACKNOWLEDGMENTS The case for the origin of arctoids is less clear. Almost all major clades of the arctoids We are greatly indebted to Richard H. Ted- (except pinnipeds) have their basal members ford for his unfailing generosity of his time ®rst appearing in Europe, such as Cynodictis and his knowledge of carnivoran systematics. for basal amphicyonids, Cephalogale for He served as an excellent sounding board for basal ursids, Broiliana (and other genera) for many ideas developed in this paper, and thor- basal procyonids, and Plesiogale (and others) oughly critiqued several drafts of this paper. for basal mustelids. However, North America A preliminary draft of this paper was also also has several basal arctoids that may pre- reviewed by Louis de Bonis, Harold Bryant, date the European counterparts. On the Mus- and Mieczysøaw Wolsan. We are especially telida side, Mustelavus occupies the most grateful to the latter two reviewers for their basal position in the musteloids. Its ®rst ap- detailed criticisms and insightful comments. pearance in the Chadronian is probably ear- We have also bene®ted greatly from the crit- lier than most of the Quercy musteloids. On ical reviews by two anonymous reviewers for the Ursida side, the North American subpar- the latest draft of this paper. ictines are comparable to European amphi- We thank Edward Pederson and Jane cynodontines in their stage of evolution, and Shumsky of the AMNH for their skillful their occurrence in the Chadronian may also preparation and casting of the basicranium of predate the European counterparts (many of the holotype of Amphicticeps shackelfordi the classic Quercy ®ssure collections from and other relevant specimens. Howell Thom- France cannot be dated precisely and gener- as, Theodore Conner, and Norman Rosenfeld ally range from late Eocene to Oligocene). of the Natural History Museum of Los An- 2005 WANG ET AL.: CARNIVORANS FROM CENTRAL MONGOLIA 51 geles County made additional preparations Journal of Vertebrate Paleontology 9(1): 110± and casting of many comparative materials. 117. We thank the following individuals for their Baskin, J.A. 1998a Mustelidae. In C.M. Janis, permission to access collections under their K.M. Scott, and L.L. Jacobs (editors), Evolu- tion of Tertiary mammals of North America, care: Louis de Bonis (UniversiteÂ du Poitiers), Volume 1: Terrestrial carnivores, ungulates, and Burkart Engesser (Naturhistorisches Museum ungulatelike mammals: 152±173. Cambridge: Basel), Meng Jin and Richard Tedford Cambridge University Press. (American Museum of Natural History), Mi- Baskin, J.A. 1998b. Procyonidae. In C.M. Janis, chael Morlo and Doris Nagel (joint Austrian± K.M. Scott, and L.L. Jacobs (editors), Evolu- Mongolian collection in Museum of Natural tion of Tertiary mammals of North America, History of Vienna), Qiu Zhanxiang and Volume 1: Terrestrial carnivores, ungulates, and Wang Banyue (Institute of Vertebrate Pale- ungulatelike mammals: 144±151. Cambridge: ontology and Paleoanthropology), Alexander Cambridge University Press. Averianov (Institute of Paleontology, Rus- Baskin, J.A., and R.H. Tedford. 1996. Small arctoid and feliform carnivorans. In D.R. Prothero, sian Academy of Sciences), and Mieczysøaw and R.J. Emry (editors), The terrestrial Eo- Wolsan (Institute of Paleobiology, Polish ceneÐOligocene transition in North America, Academy of Sciences). Pt. II: Common vertebrates of the White River Field work for securing the new materials chronofauna: 486±497. Cambridge: Cambridge and for the new synthesis of the stratigraphic University Press. relationships, by the joint Mongolian Acad- Beaumont, G.d. 1964. Essai sur la position tax- emy of Sciences±American Museum of Nat- onomique des genres Alopecocyon Viret et Sim- ural History Paleontological Expeditions, ocyon Wagner (Carnivora). Eclogae Geologicae was funded by the Frick Laboratory Endow- Helvetiae 57: 829±836. Beaumont, G.d. 1965. Contribution aÁ l'EÂ tude du ment Fund of the AMNH, Philip McKenna genre Cephalogale Jourdan (Carnivora). Foundation, National Geographic Society, Schweizerische PalaÈontologische Abhandlun- and by the National Science Foundation gen 82: 1±34. (DEB-9300770 to M. Novacek and others). Beaumont, G.d. 1968a. Note sur la reÂgion auditive We gratefully acknowledge funding for trav- de quelques carnivores. Archives des Sciences els and research by X. Wang from the Na- (Geneva) 21: 213±223. tional Science Foundation (EAR-0446699, Beaumont, G.d. 1968b. Note sur l'osteÂologie DEB-9420004, and DEB-9707555), Chinese craÃnienne de Plesiogale Pomel (Mustelidae, National Natural Science Foundation (Nos. Carnivora). Archives des Sciences (Geneva) 21(1): 27±34. 40232023 and 40128004), and National Geo- Berggren, W.A., D.V. Kent, J.D. Obradovich, and graphic Society (Nos. 6004±97 and 6771± C.C. Swisher, III. 1992. Toward a revised Pa- 00). leogene geochronology. In D.R. Prothero, and W.A. Berggren (editors), Eocene±Oligocene REFERENCES climatic and biotic evolution: 29±45. Princeton, NJ: Princeton University Press. AntoÂn, M., M.J. Salesa, J.F. Pastor, I.M. SaÂnchez, Berkey, C.P., and W. Granger. 1923. Later sedi- S. Fraile, and J. Morales. 2004. Implications of ments of the desert basins of central Mongolia. the mastoid anatomy of larger extant felids for American Museum Novitates 77: 1±16. the evolution and predatory behaviour of sa- Berkey, C.P., and F.K. Morris. 1927. Geology of bretoothed cats (Mammalia, Carnivora, Feli- Mongolia. Reconnaissance report based on the dae). Zoological Journal of the Linnean Society investigations of the years 1922±1923. Natural 140(2): 207±222. History of Central Asia, vol. II. New York: Babbitt, C. 1999. Pyctis inamatus, gen. et sp. American Museum of Natural History, 475 pp. nov., a new mustelid from the Hsanda Gol For- Berta, A. 1987. Origin, diversi®cation, and zoo- mation, Oligocene, Mongolia. Journal of Ver- geography of the South American Canidae. In tebrate Paleontology 19(4): 791±792. B.D. Patterson and R.M. Timm (editors), Stud- Baskin, J.A. 1982. Tertiary Procyoninae (Mam- ies in Neotropical mammalogy: essays in honor malia: Carnivora) of North America. Journal of of Philip Hershkovitz, Fieldiana Zoology, n. Vertebrate Paleontology 2(1): 71±93. ser. 39: 455±471. Baskin, J.A. 1989. Comments on New World Ter- Berta, A. 1988. Quaternary evolution and bioge- tiary Procyonidae (Mammalia: Carnivora). ography of the large South American Canidae 52 AMERICAN MUSEUM NOVITATES NO. 3483

(Mammalia: Carnivora). University of Califor- of the Chadron Formation in the Big Badlands nia Publications in Geological Science 132: 1± of South Dakota. Annals of Carnegie Museum 149. 25: 261±350. Berta, A., and P.J. Adam. 2001. Evolutionary bi- Clark, J., and T.E. Guensburg. 1972. Arctoid geology of pinnipeds. In J.-M. Mazin, and V.d. netic characters as related to the genus Parictis. BuffreÂnil (editors), Secondary adaptation of tet- Fieldiana Geology 26(1): 1±71. rapods to life in water: 235±260. MuÈnchen: Crusafont-PairoÂ, M. 1950. El primer representante Verlag Dr. Friedrich Pfeil. del geÂnero Canis en el Pontiense eurasiatico Berta, A., and A.R. Wyss. 1994. Pinniped phy- (Canis cipio nova sp.). Boletin de la Real So- logeny. In A. Berta and T.A. DemeÂreÂ (editors), ciedad EspanÄola de Historia Natural Seccion Contributions in marine mammal paleontology Geologica 48: 43±51. honoring Frank C. Whitmore, Jr., Proceedings Dalrymple, G.B. 1979. Critical tables for conver- of the San Diego Society of Natural History: sion of K-Ar ages from old to new constants. 33±56. Geology 7: 558±560. BiochroM'97. 1997. Paleogene biochronology. In Dashzeveg, D. 1996. Some carnivorous mammals J.-P. Aguilar, S. Legendre, and J. Michaux (ed- from the Paleogene of the eastern Gobi Desert, itors), Actes du CongreÁs BiochroM'97, Me- Mongolia, and the application of Oligocene moires et Travaux de l'EPHE 21: 769±805. carnivores to stratigraphic correlation. Ameri- Montpellier: Institut de Montpellier. can Museum Novitates 3179: 1±14. Bonis, L.d. 1971. Deux nouveaux carnassiers des Daxner-HoÈck, G. 2000. Ulaancricetodon bada- Phosphorites du Quercy. Annales de PaleÂonto- mae n. gen., n. sp. (Mammalia, Rodentia, Cri- logie 57: 117±127. cetidae) from the Valley of Lakes in Central Bonis, L.d. 1973. Contribution aÁ l'eÂtude des Mongolia. PalaÈontologische Zeitschrift 74(1/2): MammifeÁres de l'Aquitanien de l'Agenais, ron- 215±225. geurs-carnivores-perissodactyles. MeÂmoires du Daxner-HoÈck, G. 2001. New zapodids (Rodentia) MuseÂum National d'Histoire Naturelle Nou- from Oligocene±Miocene deposits in Mongo- velle SeÂrie 28: 1±192. lia, Part 1. Senckenbergiana Lethaea 81(2): Bonis, L.d. 1997. PreÂcisions sur l'aÃge geÂologique 359±389. et les relations phyleÂtiques de Mustelictis oliv- Daxner-HoÈck, G., V. HoÈck, D. Badamgarav, G. eri nov. sp. (Carnivora, Mustelidae), carnassier FurtmuÈller, W. Frank, O. Montag, and H.P. de l'OligoceÁne infeÂrieur (MP 22) des Phospho- Schmid. 1997. Cenozoic stratigraphy based on rites du Quercy (France). Geobios 20: 55±60. a sediment-basalt association in central Mon- Bryant, H.N., A.P. Russell, and W.D. Fitch. 1993. golia as requirement for correlation across cen- Phylogenetic relationships within the extant tral Asia. In J.-P. Aguilar, S. Legendre, and J. Mustelidae (Carnivora): appraisal of the cladis- Michaux (editors), Actes du CongreÁs tic status of the Simpsonian subfamilies. Zoo- BiochroM'97, Memoires et Travaux de logical Journal of the Linnean Society 108: l'EPHE, 21: 163±176. Montpellier: Insitut de 301±334. Montpellier. Bryant, J.D., and M.C. McKenna. 1995. Cranial Daxner-HoÈck, G., and W.-y. Wu. 2003. Plesios- anatomy and phylogenetic position of Tsagan- minthus (Zapodidae, Mammalia) from China omys altaicus (Mammalia, Rodentia) from the and Mongolia: migrations to Europe. In J.W.F. Hsanda Gol Formation (Oligocene), Mongolia. Reumer, and W. Wessels (editors), Distribution American Museum Novitates 3156: 1±42. and migration of Tertiary mammals in Eurasia. Cirot, E. 1992. EÂ tude phylogeÂneÂtique de quelques A volume in honour of Hans de Bruijn, Deinsea genres d'Arctoidea de l'OligoceÁne Eurasia- 10: 127±151. tique, comparaison des donneÂes morpholo- Dayan, M., D. Simberloff, E. Tchernov, and Y. giques et moleÂculaires. Ph.D. thesis, UniversiteÂ Yom-Tov. 1992. Canine carnassials: character de Poitiers, Poitiers, 152 pp. displacement in the wolves, jackals, and foxes Cirot, E., and L.d. Bonis. 1992. ReÂvision du genre of Israel. Biological Journal of Linnean Society Amphicynodon, carnivore de l'OligoceÁne. Pa- 45: 315±331. laeontographica Abteilung A 220: 103±130. Dayan, M., E. Tchernov, Y. Yom-Tov, and D. Cirot, E., and L.d. Bonis. 1993. Le craÃne Simberloff. 1989. Ecological character dis- d'Amphictis ambiguus (Carnivora, Mammalia): placement in Saharo-Arabian Vuples: outfoxing son importance pour la compreÂhension de la Bergmann's rule. Oikos 55: 263±272. phylogeÂnie des musteÂloõÈdes. Comptes Rendus Dehm, R. 1950. Die Raubtiere aus dem Mittel- de l'AcadeÂmie des Sciences Paris 316: 1327± MiocaÈn (Burgidalium) von Wintershof-West 1333. bei EichstaÈtt in Bayern. Bayerische Akademie Clark, J. 1937. The stratigraphy and paleontology der Wissenschaften Mathematisch-Naturwis- 2005 WANG ET AL.: CARNIVORANS FROM CENTRAL MONGOLIA 53

senschaftliche Klasse Abhandlungen, N. F. 58: National d'Histoire Naturelle SeÂrie C Geologie 1±141. 9: 1±187. Dragoo, J.W., and R.L. Honeycutt. 1997. System- Ginsburg, L. 1982. Sur la position systeÂmatique atics of mustelid-like carnivores. Journal of du petit panda, Ailurus fulgens (Carnivora, Mammalogy 78(2): 426±443. Mammalia). Geobios MeÂmoire SpeÂcial 6: 247± Ducrocq, S. 1993. Mammals and stratigraphy in 258. Asia: is the Eocene±Oligocene boundary at the Ginsburg, L. 1999. Order Carnivora. In G.E. RoÈs- right place? Comptes Rendus de l'AcadeÂmie sner and K. Heissig (editors), The Miocene des Sciences Paris 316: 419±426. land mammals of Europe: 109±148. MuÈnchen: Erbajeva, M.A., and G. Daxner-HoÈck. 2001. Pa- Verlag Dr. Friedrich Pfeil. leogene and Neogene lagomorphs from the Val- Ginsburg, L., O. Maridet, and P. Mein. 2001. Un ley of Lakes, Central Mongolia. Lynx (Prague), Ailurinae (Mammalia, Carnivora, Ailuridae) n. s. 32: 55±65. dans le MioceÁne moyen de Four (IseÁre, France). Evans, H.E., and G.C. Christensen. 1979. Miller's Geodiversitas 23(1): 81±85. anatomy of the dog. Philadelphia: W. B. Saun- Ginsburg, L., and J. Morales. 1992. Contribution ders, 1181 pp. aÁ la connaissance des MusteÂlideÂs (Carnivora, Evernden, J.F., D.E. Savage, G.H. Curtis, and G.T. Mammalia) du MioceÁne d'Europe Trochictis et James. 1964. Potassium-argon dates and the Ischyrictis, genres af®nes et genres nouveaux. Cenozoic mammalian chronology of North Comptes Rendus de l'AcadeÂmie des Sciences America. American Journal of Science 262: 315(2): 111±116. 145±198. Ginsburg, L., and J. Morales. 1995. Zaragocyon Flynn, J.J., and M.A. Nedbal. 1998. Phylogeny of daamsi n. gen. sp. nov., Ursidae primitif du the Carnivora (Mammalia): congruence vs. in- MioceÁne infeÂrieur d'Espagne. Comptes Rendus de l'AcadeÂmie des Sciences 321: 811±815. compatibility among multiple data sets. Molec- Helbing, H. 1930. Zwei oligocaene Musteliden ular Phylogenetics and Evolution 9(3): 414± (Plesictis genettoõÈdes Pomel. Palaeogale an- 426. gustifrons Pomel). Abhandlungen der Schweiz- Flynn, J.J., M.A. Nedbal, J.W. Dragoo, and R.L. erischen Palaeontologischen Gesellschaft 50(3): Honeycutt. 2000. Whence the Red Panda. Mo- 1±36. lecular Phylogenetics and Evolution 17(2): HoÈck, V., G. Daxner-HoÈck, H.P. Schmid, D. Bad- 190±199. amgarav, W. Frank, G. FurtmuÈller, O. Montag, Flynn, J.J., N.A. Neff, and R.H. Tedford. 1988. R. Barsbold, Y. Khand, and J. Sodov. 1999. Ol- Phylogeny of the Carnivora. In M.J. Benton igocene±Miocene sediments, fossils and basalts (editor), The phylogeny and classi®cation of from the Valley of Lakes (Central Mongolia), the tetrapods, vol. 2: Mammals, Systematics An integrated study. Mitteilungen der OÈ sterrei- Association Special Volume 35B: 73±116. Ox- chischen Geologischen Gesellschaft 90(1997): ford: Clarendon Press. 83±125. Fraas, O. 1870. Die Fauna von Steinheim. Mit Hough, J.R. 1944. The auditory region in some RuÈchsicht auf die MiocaÈnen SaÈugethier- und Miocene carnivores. Journal of Paleontology Vogelreste des Steinheimer Beckens. Jahres- 22: 573±600. hefte des Vereins fuÈr VaterlaÈndische Naturkun- Hough, J.R. 1948. The auditory region in some de in WuÈrttemberg 26: 145±306. members of the Procyonidae, Canidae, and Ur- Gabuniya, L.K.E.V.D., and M.M. Rubinshtein. sidae. Bulletin of the American Museum of 1975. Data on the absolute age of Cenozoic Natural History 92: 73±118. continental deposits of Asia and their biostrati- Huang, X.-s. 1993. Note on some carnivorous re- graphic signi®cance. Doklady Akademii Nauk mains from the middle Oligocene of Ulantatal, SSSR 225(4): 895±898. Nei Mongol. Vertebrata PalAsiatica 31(4): Gauthier, J.A., A.G. Kluge, and T. Rowe. 1988. 294±303. Amniote phylogeny and the importance of fos- Hunt, R.M., Jr. 1974. The auditory bulla in Car- sils. Cladistics 4: 105±209. nivora: an anatomical basis for reappraisal of Geisler, J.H. 2004. Humeri of Oligoscalops (Pros- carnivore evolution. Journal of Morphology calopidae, Mammalia) from the Oligocene of 143: 21±76. Mongolia. In G.C. Gould and S.K. Bell (edi- Hunt, R.M., Jr. 1977. Basicranial anatomy of Cy- tors), Tributes to Malcolm C. McKenna: his nelos Jourdan (Mammalia: Carnivora), an students, his legacy. Bulletin of the American Aquitanian amphicyonid from the Allier Basin, Museum of Natural History 285: 166±176. France. Journal of Paleontology 51: 826±843. Ginsburg, L. 1961. La faune des carnivores Mio- Hunt, R.M., Jr. 1996a. Amphicyonidae. In D.R. ceÁnes de Sansan (Gers). MeÂmoires du MuseÂum Prothero and R.J. Emry (editors), The terrestrial 54 AMERICAN MUSEUM NOVITATES NO. 3483

Eocene±Oligocene transition in North America: Palaeontological ExpeditionsÐPart V. Middle 476±485. Cambridge: Cambridge University Oligocene rodents from Mongolia. Palaeonto- Press. logia Polonica 30: 147±178. Hunt, R.M., Jr. 1996b. Biogeography of the Order Kuss, S.E. 1965. Revision der europaÈischen Am- Carnivora. In J.L. Gittleman (editor), Carnivore phicyoninae (Canidae, Carnivora, Mamm.) behavior, ecology, & evolution 2: 485±541. Ith- ausschlieûlich der voroberstampischen Formen. aca, NY: Cornell University Press. Sitzungsberichte der Heidelberger Akademie Hunt, R.M., Jr. 1998a. Amphicyonidae. In C.M. der Wissenschaften 1: 1±168. Janis, K.M. Scott, and L.L. Jacobs (editors), Lange-BadreÂ, B. 1970. Mustelictis piveteaui, mus- Evolution of Tertiary mammals of North Amer- teÂlideÂ nouveau des Phosphorites du Quercy. ica, Volume 1: Terrestrial carnivores, ungulates, Annales de PaleÂontologie (VerteÂbreÂs) 56: 73± and ungulatelike mammals: 196±227. Cam- 91. bridge: Cambridge University Press. Lange-BadreÂ, B., and D. Dashzeveg. 1989. On Hunt, R.M., Jr. 1998b. Evolution of the aeluroid some Oligocene carnivorous mammals from Carnivora: diversity of the earlier aeluroids Central Asia. Acta Palaeontologica Polonica from Eurasia (Quercy, Hsanda-Gol) and the or- 34: 125±148. igin of felids. American Museum Novitates Ledje, C., and U. Arnason. 1996a. Phylogenetic 3252: 1±65. analyses of complete cytochrome b genes of the Hunt, R.M., Jr. 1998c Ursidae. In C.M. Janis, order Carnivora with particular emphasis on the K.M. Scott, and L.L. Jacobs (editors), Evolu- Caniformia. Journal of Molecular Evolution 42: tion of Tertiary mammals of North America, 135±144. Volume 1: Terrestrial carnivores, ungulates, and Ledje, C., and U. Arnason. 1996b. Phylogenetic ungulatelike mammals: 174±195. Cambridge: relationships within caniform carnivores based Cambridge University Press. on analyses of the mitochondrial 12S rRNA Hunt, R.M., Jr. 2001. Small Oligocene amphi- gene. Journal of Molecular Evolution 43: 641± cyonids from North America (Paradaphoenus, 649. Mammalia, Carnivora). American Museum LeÂveÃque, F. 1993. Correlating the Eocene± Novitates 3331: 1±20. Oligocene mammalian biochronological scale Hunt, R.M., Jr., and L.G. Barnes. 1994. Basicra- from SW Europe with the marine magnetic nial evidence for ursid af®nity of the oldest pin- anomaly sequence. Journal of the Geological nipeds. In A. Berta and T.A. DemeÂreÂ (editors), Contributions in marine mammal paleontology Society, London 150: 661±664. honoring Frank C. Whitmore, Jr., Proceedings Li, C.-k., and S.-y. Ting. 1983. The Paleogene of the San Diego Society of Natural History: mammals of China. Bulletin of Carnegie Mu- 57±67. San Diego. seum of Natural History 21: 1±98. Hunt, R.M., Jr., and R.H. Tedford. 1993. Phylo- Matthew, W.D. 1924. Third contribution to the genetic relationships within the aeluroid Car- Snake Creek fauna. Bulletin of the American nivora and implications of their temporal and Museum of Natural History 50: 59±210. geographic distribution. In F.S. Szalay, M.J. Matthew, W.D., and W. Granger. 1924. New Car- Novacek, and M.C. McKenna (editors), Mam- nivora from the Tertiary of Mongolia. Ameri- mal phylogeny: Placentals: 52±73. New York: can Museum Novitates 104: 1±9. Springer-Verlag. McKenna, M.C. 1995. Biostratigraphy of the type Janovskaja, N.M. 1970. New cynodictines from Hsanda Gol Formation, Oligocene of Mongo- the middle Oligocene of Mongolia and Ka- lia. Journal of Vertebrate Paleontology zakhstan and the evolution of Carnivora of the 15(suppl. 3): 42A. subfamily Canidae. Materialy po Evolyutsii McKenna, M.C., and S.K. Bell. 1997. Classi®ca- Nazemnykh Pozvonochnikh: 71±84. tion of mammals above the species level. New Kellner, A.W.A., and M.C. McKenna. 1996. A York: Columbia University Press. leptictid mammal from the Hsanda Gol For- McKenna, M.C., D. Dashzeveg, and J.D. Bryant. mation (Oligocene), central Mongolia, with MS. Hsanda Gol Formation and Loh Forma- comments on some Palaeoryctidae. American tion, Oligocene and Miocene, Valley of Lakes, Museum Novitates 3168: 1±13. Mongolia. Kotsakis, T. 1980. Revisione sistematica e distri- McLaren, I.A. 1960. Are the Pinnipedia biphylet- buzione stratigra®ca e geogra®ca del genere ic? Systematic Zoology 9: 18±28. Cynodictis Bravard & Pomel (Carnivora, Mam- Mellett, J.S. 1968. The Oligocene Hsanda Gol malia). Bolletino della SocietaÁ Paleontologica Formation, Mongolia: a revised faunal list. Italiana 19(2): 259±273. American Museum Novitates 2318: 1±16. Kowalski, K. 1974. Results of the Polish-Mongolian Meng, J., and M.C. McKenna. 1998. Faunal turn- 2005 WANG ET AL.: CARNIVORANS FROM CENTRAL MONGOLIA 55

overs of Palaeogene mammals from the Mon- lower Miocene carnivores. Field Museum of golian Plateau. Nature 394: 364±367. Natural History Geological Series 7: 59±62. Mitchell, E., and R.H. Tedford. 1973. The Ena- Riggs, E.S. 1945. Some early Miocene carnivores. liarctinae a new group of extinct aquatic Car- Field Museum of Natural History Geological nivora and a consideration of the origin of the Series 9(3): 69±114. Otariidae. Bulletin of the American Museum of Russell, D.E., and R.-j. Zhai. 1987. The Paleogene Natural History 151: 203±284. of Asia: mammals and stratigraphy. MeÂmoires MoÈdden, C. 1991. Bavarictis gaimersheimensis n. du MuseÂum National d'Histoire Naturelle, SeÂr- gen. n. sp., ein fruÈher Mustelide aus der ober- ie C, Sciences de la Terre 52: 1±488. oligozaÈnen SpaltenfuÈllung Gaimersheim bei In- Savage, R.J.G. 1957. The anatomy of Potamoth- golstadt. Mitteilungen der Bayerischen Staats- erium, an Oligocene lutrine. Proceedings of the sammlung fuÈr PalaÈontologie und Historische Zoological Society of London 129: 151±244. Geologie 31: 125±147. Schmidt-Kittler, N. 1981. Zur Stammesgeschichte Muizon, C.d. 1982. Les relations phylogeÂneÂtiques der marderverwandten Raubtiergruppen (Mus- des Lutrinae (Mustelidae, Mammalia). Geobios teloidea, Carnivora). Eclogae Geologicae Hel- MeÂmoire SpeÂcial 6: 259±277. vetiae 74: 753±801. Nagel, D., and M. Morlo. 2001. The endemic car- Schmidt-Kittler, N. 1989. A biochronologic sub- nivorous fauna from the Oligocene of Taatsiin division of the European Paleogene based on GolÐCreodonta and Carnivora. Conference on mammalsÐreport on results of the Paleogene Distribution and Migration of Tertiary Mam- Symposium held in Mainz in February 1987. In mals in Eurasia in Honor of Hans de Buijn: 38. E.H. Lindsay, V. Fahlbusch, and P. Mein (edi- Nagel, D., and M. Morlo. 2003. Guild structure tors), European Neogene mammal chronology, of the carnivorous mammals (Creodonta, Car- NATO (Advanced Science Institute) Series A: nivora) from the Taatsiin Gol area, Lower Ol- Life Sciences 180: 47±50. New York: Plenum igocene of Central Mongolia. In J.W.F. Reumer, Press. and W. Wessels (editors), Distribution and mi- Scott, W.B., and G.L. Jepsen. 1936. The mam- gration of Tertiary mammals in Eurasia. A vol- malian fauna of the White River Oligocene. ume in honour of Hans de Bruijn, Deinsea 10: Princeton University Bulletin 28: 1±980. 419±429. Segall, W. 1943. The auditory region of the arc- Orlov, Y.A. 1933. Semantor macrurus (Ordo Pin- toid carnivores. Field Museum of Natural His- nipedia, Fam. Semantoridae fam. nova) aus den tory Zoological Series 29(3): 33±59. Neogen-Ablagerungen Westsibiriens. Travaux Simpson, G.G. 1945. The principles of classi®- de l'Institut PaleÂozoologique, AcadeÂmie des cation and a classi®cation of mammals. Bulletin Sciences, URSS 2: 165±268. of the American Museum of Natural History 8: Petter, G. 1967. Paragale huÈrzeleri nov. gen., nov. 1±350. sp., MusteÂlideÂ nouveau de l'Aquitanien de Story, H.E. 1951. The carotid arteries in the Pro- l'Allier. Bulletin de la SocieÂteÂGeÂologique de cyonidae. Fieldiana: Zoology 32: 477±557. France, SeÂrie 7, 9: 19±23. Swisher, C.C., III, and D.R. Prothero. 1990. Sin- Piveteau, J. 1961. TraiteÂ de paleÂontologie, Tome gle-crystal 40Ar/39Ar dating of the Eocene±Ol- VI, Vol. I, L'origine des mammifeÁres et les as- igocene transition in North America. Science pects fondamentaux de leur eÂvolution. Paris: 249: 760±762. Brodard-Taupin Imprimeur-Relieur, 1138 pp. Swofford, D.L. 1993. Phylogenetic analysis using Prothero, D.R., and C.C. Swisher, III. 1992. Mag- parsimony (PAUP), version 3. 1.1. Urbana: Il- netostratigraphy and geochronology of the ter- linois Natural History Survey. restrial Eocene±Oligocene transition in North Tedford, R.H. 1976. Relationship of pinnipeds to America. In D.R. Prothero and W.A. Berggren other carnivores (Mammalia). Systematic Zo- (editors), Eocene±Oligocene climatic and biotic ology 25: 363±374. evolution: 46±73. Princeton, NJ: Princeton Uni- Tedford, R.H. 1978. History of dogs and cats: a versity Press. view from the fossil record. Nutrition and Man- Qiu, Z.-x., and N. Schmidt-Kittler. 1982. On the agement of Dogs and Cats: chap. M23. St. Lou- phylogeny and zoogeography of the leptarcti- is: Ralston Purina Co. nes (Carnivora, Mammalia). PalaÈontologische Tedford, R.H., L.G. Barnes, and C.E. Ray. 1994. Zeitschrift 56: 131±145. The early Miocene littoral ursoid carnivoran Radinsky, L. 1973. Are stink badgers skunks? Im- Kolponomos: systematics and mode of life. In plications of neuroanatomy for mustelid phy- A. Berta and T.A. DemeÂreÂ (editors), Contribu- logeny. Journal of Mammalogy 54(3): 585± tions in marine mammal paleontology honoring 593. Frank C. Whitmore, Jr.: 11±32. San Diego: San Riggs, E.S. 1942. Preliminary description of two Diego Society of Natural History. 56 AMERICAN MUSEUM NOVITATES NO. 3483

Tedford, R.H., B.E. Taylor, and X. Wang. 1995. Wang, X., and Z.-x. Qiu. 2003b. A primitive lep- Phylogeny of the Caninae (Carnivora: Cani- tarctine (Carnivora: Mustelidae) from the early dae): the living taxa. American Museum Nov- Miocene of western Gansu, China and zooge- itates 3146: 1±37. ography of early mustelids. Journal of Verte- Teilhard de Chardin, P. 1915. Les carnassiers des brate Paleontology 23(suppl. 3): 107A. Phosphorites du Quercy. Annales de PaleÂonto- Wang, X., and Z.-x. Qiu. 2004. Late Miocene logie 9: 101±191. Promephitis (Carnivora, Mephitidae) from Chi- Tsubamoto, T., M. Takai, and N. Egi. 2004. Quan- na. Journal of Vertebrate Paleontology 24(3): titative analyses of biogeography and faunal 721±731. evolution of middle to late Eocene mammals in Wang, X., Z.-x. Qiu, and B.-y. Wang. 2004. A East Asia. Journal of Vertebrate Paleontology new leptarctine (Carnivora: Mustelidae) from 24(3): 657±667. the early Miocene of the northern Tibetan pla- Viranta, S. 1996. European Miocene Amphicyon- teau and implications of the phylogeny and idaeÐtaxonomy, systematics and ecology. Acta zoogeography of basal mustelids. Zoological Zoologica Fennica 204: 1±61. Journal of the Linnean Society 142: 405±421. Viret, J. 1951. Catalogue critique de la faune des Wang, X., and R.H. Tedford. 1992. The status of mammifeÁres mioceÁnes de la Grive Saint-Alban. genus Nothocyon Matthew, 1899 (Carnivora): Nouvelles Archives du MuseÂum d'Histoire Na- an arctoid not a canid. Journal of Vertebrate turelle de Lyon 3: 1±104. Paleontology 12: 223±229. Vislobokova, I.A., and G. Daxner-HoÈck. 2002. Wang, X., and R.H. Tedford. 1994. Basicranial Oligocene±early Miocene ruminants from the anatomy and phylogeny of primitive canids and Valley of Lakes (Central Mongolia). Annalen closely related miacids (Carnivora: Mammalia). des Naturhistorischen Museums in Wien American Museum Novitates 3092: 1±34. 103(A): 213±235. Wang, X., and R.H. Tedford. 1996. Canidae. In Vrana, P.B., M.C. Milinkovitich, J.R. Powell, and D.R. Prothero and R.J. Emry (editors), The ter- W.C. Wheeler. 1994. Higher level relationships restrial Eocene±Oligocene transition in North of the arctoid Carnivora based on sequence data America, Pt. II: Common vertebrates of the and ``total evidence''. Molecular Phylogenetics White River Chronofauna: 433±452. Cam- and Evolution 3: 47±58. bridge: Cambridge University Press. Wallace, S.C., and X. Wang. 2004. Two new car- Wang, X., R.H. Tedford, and B.E. Taylor. 1999. nivores from an unusual late Tertiary forest bi- Phylogenetic systematics of the Borophaginae ota in eastern North America. Nature 431: 556± (Carnivora: Canidae). Bulletin of the American 559. Museum of Natural History 243: 1±391. Wang, B.-y. 1992. The Chinese Oligocene: a pre- Wang, X., J. Ye, J. Meng, W.-y. Wu, L.-p. Liu, liminary review of mammalian localities and and S.-d. Bi. 1998. Carnivora from middle local faunas. In D.R. Prothero and W.A. Berg- Miocene of northern Junggar Basin, Xinjiang gren (editors), Eocene±Oligocene climatic and Autonomous Region, China. Vertebrata Pal- biotic evolution: 529±547. Princeton, NJ: Asiatica 36(3): 218±243. Princeton University Press. Wegner, R.N. 1913. Tertiaer und umgelagerte Wang, B.-y. 2001. On Tsaganomyidae (Rodentia, Kreide bei Oppeln (Oberschlesien). Palaeonto- Mammalia) of Asia. American Museum Novi- graphica 60: 175±272. tates 3317: 1±32. Werdelin, L., and N. Solounias. 1991. The Hyaen- Wang, B.-y., and Z.-x. Qiu. 2003a. Notes on early idae: taxonomy, systematics and evolution. Oligocene ursids (Carnivora, Mammalia) from Fossils and Strata 30: 1±104. Saint Jacques, Nei Mongol, China. In L.J. Wiig, O. 1983. On the relationship of pinnipeds Flynn (editor), Vertebrate fossils and their con- to other carnivores. Zoologica Scripta 12(3): text: contributions in honor of Richard H. Ted- 225±227. ford. Bulletin of the American Museum of Nat- Wolsan, M. 1993. Phylogeny and classi®cation of ural History 279: 116±124. early European Mustelida (Mammalia: Carniv- Wang, X. 1994. Phylogenetic systematics of the ora). Acta Theriologica 38: 345±384. Hesperocyoninae (Carnivora: Canidae). Bulle- Wolsan, M. 1999. Oldest mephitine cranium and tin of the American Museum of Natural History its implications for the origin of skunks. Acta 221: 1±207. Palaeontologica Polonica 44(2): 223±230. Wang, X. 1997. New cranial material of Simocyon Wyss, A.R. 1987. The walrus auditory region and from China, and its implications for phyloge- the monophyly of pinnipeds. American Muse- netic relationship to the red panda (Ailurus). um Novitates 2871: 1±31. Journal of Vertebrate Paleontology 17: 184± Wyss, A.R. 1988. On ``retrogression'' in the evo- 198. lution of the Phocinae and phylogenetic af®ni- 2005 WANG ET AL.: CARNIVORANS FROM CENTRAL MONGOLIA 57

ties of the monk seals. American Museum Nov- Wyss, A.R., and J.J. Flynn. 1993. A phylogenetic itates 2924: 1±38. analysis and de®nition of the Carnivora. In F.S. Wyss, A.R. 1991. What, then, is Potamotherium Szalay, M.J. Novacek, and M.C. McKenna (ed- (Mammalia, Carnivora)? Journal of Vertebrate itors), Mammal phylogeny: placentals: 32±52. Paleontology 11(suppl.): 63A. New York: Springer.

APPENDIX 1

CHARACTERS SCORED IN THE CHARACTER MATRIX 18. Entoconid on m1: 0, poorly developed; 1, (TABLE 8) FOR PAUP ANALYSIS presence of a lingual notch; 2, cuspidate. 19. Size of m2: 0, not enlarged or reduced; 1, Most of these characters have been discussed reduced; 2, lost; 3, enlarged. and illustrated previously (e.g., Hunt, 1974; Wol- 20. Talonid of m2: 0, short; 1, elongated. san, 1993; Wang and Tedford, 1994), although we 21. Mastoid process: 0, small; 1, laterally ex- may code them differently. panded; 2, ventrally expanded. 1. Temporal crest: 0, converge to sagittal 22. Reduction of m3: 0, m3 not reduced; 1, crest; 1, parallel without a merged sagittal m3 reduced or absent. crest. 23. M1 postprotocrista: 0, present; 1, absent; 2. Posterior border of palatine: 0, short and 2, enlarged metaconule. anterior to posterior border of last upper 24. Paroccipital process: 0, rodlike; 1, dor- molars; 1, elongated and posterior to pos- soventrally compressed; 2, laterally com- terior border of last upper molars. pressed. 3. Alisphenoid canal: 0, present; 1, absent. 25. Ascending ramus: 0, reclined; 1, anteri- 4. Posterior carotid foramen: 0, within fossa orly inclined. for posterior lacerate foramen; 1, anterior 26. Zygomatic arch: 0, horizontally ¯at; 1, to the fossa. dorsally arched. 5. Suprameatal fossa: 0, absent or a mere 27. Posterior promontorium process: 0, ab- depression on squamosal; 1, small; 2, dor- sent, 1, present. sally deep; 3, partially covered anteriorly. 28. M1 anteroposterior length: 0, not short- 6. Epitympanic recess: 0, not reduced; 1, re- ened; 1, shortened. duced. 29. M1±2 postprotocrista: 0, laterally orient- 7. Postlateral sulcus of brain: 0, absent; 1, ed; 1, posteriorly oriented. present. 30. Bullar ossi®cation: 0, unossi®ed; 1, ossi- 8. P1: 0, present; 1, absent. ®ed bony bulla. 9. Carnassial notch on P4: 0, present; 1, ab- 31. Bullar composition of Hunt (1974): 0, sent. type A; 2, type B. 10. P4 protocone: 0, conical; 1, crescent. 32. Premolar accessary cusps: 0, well devel- 11. P4 lingual cingulum: 0, absent or small; oped; 1, absent or poorly developed. 1, enlarged; 2 enlarged to form a hypocone. 33. P4 protocone: 0, anterorly positioned; 1, 12. M1 parastyle: 0, large; 1, reduced. posteriorly positioned. 13. M1 anterior and posterior borders:0, 34. Posterior orbital constriction: 0, short; 1, not parallel anterior and posterior borders; long. 1, parallel anterior and posterior borders. 35. Position of internal carotid artery:0, 14. M1 anterior and posterior cingula:0, transpromontorial; 1, medial. continuous; 1, discontinuous. 36. Petrosal: 0, isolated from basioccipital and 15. M2 position relative to M1: 0, buccal; 1, basisphenoid; 1, medially expanded to fuse lingual. with basioccipital and basisphenoid. 16. M2 size: 0, three-rooted and small; 1, re- 37. Semi-septum of bulla: 0, absent; 1, pre- duced; 2, absent; 3, enlarged. sent. 17. Metaconid on m1: 0, equal or higher than 38. M3: 0, present; 1, absent. paraconid; 1, lower than paraconid; 2, ab- 39. Basioccipital: 0, not broadened; 1, broad- sent. ened between bullae. Complete lists of all issues of the Novitates and the Bulletin are available at World Wide Web site http://library.amnh.org/pubs. Inquire about ordering printed copies via e-mail from [email protected] or via standard mail from: American Museum of Natural History, LibraryÐ Scienti®c Publications, Central Park West at 79th St., New York, NY 10024. TEL: (212) 769- 5545. FAX: (212) 769-5009.

a This paper meets the requirements of ANSI/NISO Z39.48-1992 (Permanence of Paper).