Chapter 2 Author(s): QIU ZHANXIANG Source: Bulletin of the American Museum of Natural History, :18-31. Published By: American Museum of Natural History DOI: http://dx.doi.org/10.1206/0003-0090(2003)279<0018:C>2.0.CO;2 URL: http://www.bioone.org/doi/full/10.1206/0003-0090%282003%29279%3C0018%3AC%3E2.0.CO %3B2

BioOne (www.bioone.org) is a nonprofit, online aggregation of core research in the biological, ecological, and environmental sciences. BioOne provides a sustainable online platform for over 170 journals and books published by nonprofit societies, associations, museums, institutions, and presses. Your use of this PDF, the BioOne Web site, and all posted and associated content indicates your acceptance of BioOne’s Terms of Use, available at www.bioone.org/page/terms_of_use. Usage of BioOne content is strictly limited to personal, educational, and non-commercial use. Commercial inquiries or rights and permissions requests should be directed to the individual publisher as copyright holder.

BioOne sees sustainable scholarly publishing as an inherently collaborative enterprise connecting authors, nonprofit publishers, academic institutions, research libraries, and research funders in the common goal of maximizing access to critical research. Chapter 2

Dispersals of Neogene Carnivorans between Asia and North America

QIU ZHANXIANG1

ABSTRACT Three major carnivoran dispersal waves of ®lter-bridge type between Eurasia and North America are recognized in the late Neogene. The ®rst is around 20 Ma, probably from 21 Ma to 19±18 Ma, during which intermittent dispersals might have occurred. The carnivorans mi- grating from Eurasia to North America included Cynelos, Ysengrinia, Amphicyon, Cephalo- gale, , , some small-sized mustelids, Potamotherium, an ancestral form of Edaphocyon, and Proailurus. The second wave occurred at about 7±8 Ma. The carnivorans migrating from Eurasia to North America included , , Simocyon, Eomel- livora, Plesiogulo, and Machairodus. The last wave took place in the early ϳ4 Ma. The Eurasian emigrants recorded in North America are , Parailurus, Lynx (?), Felis (?), Homotherium, and Chasmaporthetes. At about the same time Megantereon and Pannonictis migrated from North America to Eurasia.

INTRODUCTION hydriodon, Ursus, and Chasmaporthetes)as immigrants from Eurasia, although with Carnivorans frequently appear in the zoo- some degree of uncertainty. Similar endeav- geographic literature as a group of highly ors were made by Korotkevitch and Topach- mobile and widely distributed terrestrial evskii (1976) and by B. KurteÂn (1986). Ted- . Nevertheless, the exchange histo- ry of carnivorans between Asia and North ford et al. (1987) made the major contribu- America during the Neogene has not been tion in our understanding of Neogene mam- systematically untangled at the generic level. mal faunal exchange history, listing 38 North The major dif®culty herein is the recognition American Neogene carnivorans as exotic to of the same or very closely related genera on that continent. The majority of these exotic distant continents based on incomplete fossil forms was thought to have immigrated from evidence. Generation-long study may be Eurasia. Flynn and Swisher (1995) and Daw- needed to make sure of the identity of such son (1999), based mainly on Tedford et al. low-level taxa on different continents. An (1987) data, made further interpretations on early attempt by Repenning (1967) listed ®ve the Neogene dispersal events of land mam- genera (Simocyon, Indarctos, Agriotherium, mals between North America and Europe. Plionarctos, and Lutravus) as de®nite and Flynn and Swisher (1995) listed six major four (Eomellivora, Plesiogulo, Lutra, and interchange events (their 6±11, see below), Machairodus) as questionable migrants from with a number of unnamed events of smaller Eurasia to America during the Hemphillian scale for the late Neogene. Regrettably, many Age. Among the Blancan carnivor- of the above inferences were based on evo- ans, Repenning listed seven American genera lutionary considerations and phylogenetic in- (Lynx, Trigonictis, Canimartes, Enhydra, En- terpretation without solid fossil evidence.

1 Curator, Institute of Vertebrate Paleontology and Paleoanthropology, Chinese Academy of Sciences, 100044, Beijing, China.

18 2003 QIU: DISPERSAL OF NEOGENE CARNIVORANS 19

A reliable reconstruction of faunal ex- fossil record, the most reliable and decisive change history needs solid support from both among these criteria is the exclusive occur- the fossil records and a stable classi®cation. rence of immediately ancestral form(s) in im- Fortunately, considerable progress has been mediately earlier deposits in one region. made in both directions since Tedford et al.'s fundamental contribution. The classi®cation FOSSIL RECORDS RELATING TO within the Order has been exten- DISPERSALS sively studied in recent years. A number of taxonomic revisions of European and North ARCTOIDS American Neogene carnivorans have ap- AMPHICYONIDS peared (e.g., Hunt, 1996, 1998a, 1998b, 1998c; Werdelin, 1996; Baskin, 1998a, Following Hunt (1998b), the amphicyon- 1998b; Martin, 1998a, b; Ginsburg, 1999). ids are here viewed as a separate family of As a result, the classi®cation of the Order the arctoid carnivorans, comprising four sub- Carnivora becomes more re®ned in recent families: Daphoeninae, Temnocyoninae, years, and sometimes more complex, culmi- Haplocyoninae, and Amphicyoninae. The nating in inserting extra ranks (Infraorder, former two are restricted to North America, Parvorder, etc.) by McKenna and Bell the third one is endemic to Eurasia, and the (1997). To avoid entangling details of clas- last one is Holarctic in distribution. The am- si®cation of the Order Carnivora, the basic phicyonids originated probably in the Eocene subdivision of the order proposed by the of Eurasia, ¯ourished during the Oligocene great English anatomist Flower in 1869 (Arc- and early in both Eurasia and North toids, Canoids, and Aeluroids) is used here. America, and became ®nally extinct at ϳ8 On the other hand, the systematic situation Ma in Asia (vide infra). in Asia remains little improved. Although a Excluding Paleogene dispersals, three am- number of carnivoran fossils has been found phicyonine genera may have migrated from in early and middle Miocene deposits (Qiu Eurasia to North America in the early Mio- and Qiu, 1995), few of them have been stud- cene: Cynelos, Ysengrinia, and Amphicyon. ied so far. Since Beringia was the only route The earliest record of Cynelos is from the of dispersal between Eurasia and North Oligocene in Europe: Pech Desse, Quercy, America during the Neogene, some conclu- MP28, ϳ25 Ma (Remy et al., 1987). The ear- sions can safely be drawn, even with a lim- liest Ysengrinia is so far very poorly repre- ited record from Asia, if the evidence from sented in Europe: a mandible with p4±m2 of Europe and North America is suf®ciently Y. tolosanus from Le Cammas, France (ter- conclusive. minal Oligocene), a mandible with p3±m2 of The term ``carnivorans'' is here used to the same species from FloÈsheim, Germany denote the members of the Order Carnivora (MP30, ϳ24 Ma), and one m2 from TheÂzels, in the sense of Van Valen (1969), excluding Aquitan (MP30, de Bonis and Guinot, 1987). creodonts and the nonplacental predatory The ®rst records of both Cynelos and Ysen- mammals. The term ``dispersal'' is used here grinia in North America are from the upper as de®ned by Brown and Gibson (1983) as Harrison Formation, Nebraska, 19.2 Ma the ``movement of organisms away from (Hunt, 1998b). So far, no fossils of Ysengri- their point of origin''. To determine a dis- nia have been reported from Asia. However, persal event for a certain fossil form, pale- my personal examination of the undescribed ontologists should ®rst recognize the disjunct carnivoran specimens of the Sihong localities occurrence of the same or immediately an- revealed that some of them may belong to cestral/descendant genera, and then deter- Cynelos and/or Ysengrinia. The Sihong fauna mine the point of origin. The latter is a topic is roughly correlated with the European MN of much discussion. Savage (1958) listed ®ve 4, thus is ϳ18 Ma. The dispersal event of criteria for determination of origin of fossil both genera may have occurred from Eurasia mammals. Cain (1944) discussed altogether to North America at ϳ20 Ma. 13 biological criteria that have been used, as Amphicyon had long been a wastebasket cited by Brown and Gibson (1983). For the genus until the mid-1960s, when many gen- 20 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 279

Fig. 2.1. Migration events of Neogene carnivorans between Eurasia and North America. 2003 QIU: DISPERSAL OF NEOGENE CARNIVORANS 21 era were removed from it by Kuss (1965) from the late Eocene Nadu For- and Ginsburg (1966). In its restricted sense, mation, Guangxi: Cephalogale new species A. astrei is the earliest species. This species, and cf. Cephalogale sp. Unfortunately, the ®rst reported by Kuss (1962), was based on specimens cannot be located in the IVPP col- a mandible with poorly preserved teeth from lection. It is uncertain whether they really be- Garduch (terminal Oligocene, ϳ24 Ma). Ac- long to this genus or not. A de®nite record cording to Hunt (1998b), the ®rst North of early Oligocene Cephalogale is now America Amphicyon is represented by un- known from Saint-Jacques, Nei Mongol, described specimens from the Runningwater China (Wang and Qiu, this volume). Certain Formation (early Hemingfordian, 19±18 Cephalogale is known from the early Mio- Ma), Nebraska. They are close to the Euro- cene deposits at Bugti, Pakistan (Forster- pean A. giganteus, the earliest fossils of Cooper, 1923). Thus the dispersal of Ce- which were found from Wintershof-West phalogale from Europe to North America (MN 3, ϳ20±18 Ma), Germany. In Asia, the should have occurred at ϳ21 Ma. earliest Amphicyon, A. confucianus, is from Classi®cation of the hemicyonids has be- the Shanwang fauna (comparable to MN 5, come rather complicated. McKenna and Bell ϳ16 Ma). From Tongxin (comparable to MN (1997) and Ginsburg and Morales (1998) 6, ϳ15 Ma), a form very close to A. gigan- recognized them as a separate family, where- teus is present, but not yet described. The last as Hunt (1998a) considered them as a sub- appearance of the genus may be that from family of the Ursidae. Ginsburg and Morales Lufeng (comparable to MN 11, ϳ8 Ma) and (1998) further subdivided the Hemicyonidae Yuanmo sites that yield Lufengpithecus.At into two subfamilies: Phoberocyoninae (in- any rate, its ®rst dispersal from Eurasia to cluding Phoberocyon and ) and North America should have occurred at ϳ20 (including Zaragocyon, Hem- Ma. icyon, and ). The earliest occur- rence of the hemicyonids in Europe is from HEMICYONIDS MN 2 (22.8±20 Ma) of Spain: Phoberocyon hispanicus from Loranca and Zaragocyon Three genera are shared in common by daamsi from Cetina de Aragon. Deposits Eurasia and North America: Cephalogale, from then through MN 9 show that hemi- , and Agriotherium. The earliest re- cyonids lived continuously in Europe. The cord of Cephalogale in Europe is from the earliest hemicyonine so far found from North early Oligocene MP22 (Mas de Got, Quercy, America is Phoberocyon johnhenryi from France, ϳ33 Ma), after the ``Grande Cou- Thomas Farm local fauna, Florida, 19±18 pure'' according to Remy et al. (1987). The Ma. Referral of the Thomas Farm species to genus is well represented in MP28 (Pech the genus Phoberocyon is con®rmed by di- Desse, Quercy, ϳ25 Ma) and MN 1±2 (Paul- agnostic features of the lower dentition: pos- hiac, Saint Gerand-le-Puy, 23.8±20 Ma) by terior accessory cusps in lower premolars numerous species (eight, according to de and paraconid in m2. However, it is certainly Bonis, 1973). In North America, there are more derived than Phoberocyon hispanicus only a few records of Cephalogale (Hunt, in being larger and having diastemata be- 1998a): the late Arikareean (Zia Sand For- tween the premolars. All the other American mation, New Mexico, ϳ20 Ma; Upper Har- hemicyonids are more advanced in character, rison Formation, Wyoming, 21±20 Ma; Ag- and are assigned to Plithocyon, known from ate Spring Local Fauna, Nebraska, 21±20 the Barstovian of California, New Mexico, Ma) and early Hemingfordian (Marsland Nebraska, and Wyoming. All could be as- Formation, 20±19 Ma; Runningwater For- signed to Plithocyon. In Asia, the earliest mation, Nebraska, 19±18 Ma). No compari- hemicyonid is Phoberocyon youngi from son at the speci®c level is possible, since Shanwang, Shandong (ϳ16 Ma). It is very none of these North American samples have close to the North American Phoberocyon been described. In Asia, the fossil record of johnhenryi, as pointed out by Qiu et al. Cephalogale is sparse. Tang and Qiu (1979), (1985). The latest occurrence in China is in a preliminary report, listed two forms of from the Tung-Gur Formation, Nei-Mongol 22 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 279

(ϳ13 Ma). This species was ®rst described Similar work on a less comprehensive scale as Hemicyon teilhardi, but according to the was carried out by Ginsburg and Morales morphology of the upper cheek teeth, it (1998) for the European forms. Although should be assigned to Plithocyon. The dis- slightly different in infrafamilial arrange- persal should thus have occurred from Eu- ment, these revisions have proven valuable rope to North America prior to 19 Ma. for de®ning dispersal events between Eurasia Whether the North American Plithocyon rep- and North America during Neogene time. resents another immigration event is hard to Ursavus, the earliest member of the Sub- say. It is more probable that the American family , was subdivided by Ginsburg Plithocyon might be the descendant of the and Morales (1998) into Ursavus and Bal- native Phoberocyon and belong to a new ge- lusia. The latter is slightly different from the nus. former in having stronger cingula on the up- Although very close to Indarctos in skull per molars, higher m1 trigonid, and oval in- and tooth morphology, Agriotherium is here stead of rectangular M2. The ®rst appearance considered a member of the family Hemi- of both genera is MN 3 of Europe (Elm and cyonidae. As pointed out previously (Qiu and Schmidt-Kittler, 1982; Qiu et al., 1991), Wintershof-West in Germany, Savigne-sur- the presence of a series of hemicyonine char- Lathan in France), ϳ20 Ma. Although not acters in Agriotherium (unelongated molars, very abundant, fossils of Ursavus are fre- premasseteric fossa, etc.) strongly substanti- quently found in the middle Miocene of Eu- ate its closer af®nity with hemicyonines rath- rope. The latest record in Europe is from the er than with ursines. Agriotherium has its Turolian (Euboa, Greece, 9±8 Ma). There- longest history in Asia. The most primitive fore, the distribution of Ursavus (and Ballu- and earliest form is from a Hipparion fauna sia) in Europe is from MN 3 to MN 11, 20 in Hezheng, Gansu, ϳ9 Ma (Qiu et al., to 8 Ma. 1991). All other forms of Agriotherium oc- In North America, a single record of this curred in the Pliocene (5.3 to ϳ2 Ma) in Eur- genus was questionably reported from the asia and Africa, leaving a hiatus in the record early Arikareean (John Day Formation, between 9 and 5.3 Ma. In North America, Oregon), 29 Ma. However, Tedford et al. Agriotherium appeared suddenly at the mid- (1987: 172) unequivocally stated that the dle Hemphillian and became extinct before Flint Hill local fauna (19 Ma) ``has yielded the Blancan. Unequivocal Agriotherium fos- the ®rst North American occurrence of the sils were found at Coffee Ranch, Texas, and little , Ursavus''. Hunt also listed the ear- its equivalent Guymon (Optima) Local Fau- ly Hemingfordian Batesland Formation, na, Oklahoma, and Mount Eden Local Fau- South Dakota, as producing the earliest Ur- na, California. The age of the Coffee Ranch savus. Ursavus became more common dur- Local Fauna is estimated as about 6.6 Ma ing the Barstovian Mammal Age. Hunt listed (Dalquest, 1986). A good sample of Agrioth- 10 localities yielding its fossils, mainly in the erium was recovered from Old Cabin Quar- area west of the Rocky Mountains (Oregon, ry, Quiburis Formation (late Hemphillian), Nevada, Saskatchewan) with a few from the Arizona. This material remains unstudied. area east of the Rocky Mountains (Nebraska, Hunt (1998a) listed 18 localities yielding Colorado). Taken as a whole, North Ameri- Agriotherium fossils, ranging from California can Ursavus was poorly represented. Little to Florida. The occurrence of Agriotherium in North and Central America is thus limited has been studied and published of the ®ve to the later half of the Hemphillian (ϳ7±5 forms (U. pawniensis, U. cf. brevirhinus, U. Ma). The most plausible hypothesis is that cf. primaevus, and two unnamed species) the American Agriotherium emigrated from from 13 localities listed by Hunt (1998a), Eurasia at the beginning of the second half which limits comparison with the Ursavus of the Hemphillian Mammal Age (ϳ7 Ma). material of Europe at the speci®c level. In East Asia, the only record of this genus URSIDS is that of Qiu et al. (1986) from Shanwang. Recently the fossil ursids of North Amer- It is apparently a quite aberrant form deserv- ica were revised by Hunt (1996, 1998a). ing to be classi®ed as a separate genus, as 2003 QIU: DISPERSAL OF NEOGENE CARNIVORANS 23 suggested by Ginsburg and Morales (1998). early Turolian deposits. From the middle Its age is early middle Miocene. Turolian, the genus is represented by the The conclusions that can be drawn from large I. atticus. So far as the Asian fossil re- the above data seem to be as follows. (1) If cord is concerned, the genus is restricted to the ®rst occurrence of Ursavus in the early the Baodean, which is roughly correlative Arikareean (ϳ29 Ma) in North America is with the European Turolian. There are three accurate, North America might be the con- species in China. The ®rst is the primitive I. tinent where the ®rst Ursavus emerged. lagrelii, intermediate between I. arctoides Then, during the interval from 29 to 20 Ma, and I. atticus in morphology. The second is it migrated to Europe. The later Ursavus of morphologically inseparable from the Euro- Hemingfordian and Barstovian age may ei- pean I. atticus. The third is new, larger, and ther be direct descendants of this native form more advanced in character. Two other spe- or immigrants from Europe. (2) Otherwise, cies, I. salmontanus and I. punjabiensis, are the ®rst appearance of Ursavus (and Ballu- recorded in the Indo-Pakistan subcontinent. sia) is about one m.y. earlier in Europe (20 Both are from the Dhok Pathan Formation. Ma) than in North America (19 Ma), which, According to Tedford et al. (1987), the lo- together with its comparatively rich represen- calities containing the ®rst American Indarc- tation in Europe, tends to show that the dis- tos are Mulholland Formation, California (7± persal might have occurred at the time span 6 Ma), Cambridge Local Fauna, Ash Hollow between 20 to 19 Ma, and in the direction Formation, Nebraska (7±6 Ma), Rattlesnake from Europe to North America. Formation, Oregon (6.6±6.8 Ma for the over- Except for Plionarctos and its possible de- lying ash), Smiths Valley Fauna, Nevada scendant American endemic tremarctines, the (corrected as ϳ7 Ma, Tedford et al., 1987), af®nities of which are still obscure, there is Box T Local Fauna, Texas (estimated at 6.5± only one early form of true ursine so far re- 6 Ma), and Withlacoochee 4A, Florida (7±6 ported from the Pliocene of North America, Ma). The earliest appearance of this genus in Ursus abstrusus, from the Hagerman and North America might be the specimen re- White Bluffs Local Faunas. The Hagerman ferred to ``Indarctos near oregonensis'', pos- local fauna has been dated as about 3.75±3.2 sibly from the very bottom of the Rattlesnake Ma (Lundelius et al., 1987) and the White Formation (Merriam et al., 1925). Judging Bluffs slightly older than it, probably not from its size and morphology, it is very close older than 4 Ma. In Europe, the Pliocene true to the European I. arctoides. The largest spe- bear, Ursus, is rather richly represented cies in North America, and in the World in (Weze, Baron-Copez, Rouscillion, etc.). The general, is I. oregonensis keithi, described by earliest record of Ursus is from the Mont- Schultz and Martin (1975), from Cambridge, pellier marine sands (MN 14, 5.3±4.2 Ma). Nebraska, and the skull from Florida de- The earliest Ursus in Asia is from the top scribed by Wolff (1978). There is little doubt part of the Gaozhuang Formation of the Yu- that Indarctos migrated from Europe to she area (4.5±4.1 Ma). It is likely that Ursus North America at about 7 Ma, as concluded migrated from Asia to North America at by Tedford et al. (1987). about 4 Ma. Indarctos is a separate branch of the sub- MUSTELIDS family Ursinae. It evidently originated from a form of Ursavus. The record of Indarctos The mustelids diversi®ed greatly during in Europe is the most complete in compari- Neogene time. Recognition of the same tax- son with those of Asia and North America. on from the fossils of different continents is The most primitive species, I. vireti, was re- often a dif®cult task. This renders the recon- corded from Can Llobateres and Can Pon- struction of the exchange history of the mus- sich, the type section of the Vallesian telids between the two continents often high- ``stage'' in the ValleÂs-PenedeÁs basin, near Sa- ly conjectural. Baskin (1998b) postulated 21 badell, Spain. Its age is estimated as ϳ11 immigration events for the North American Ma. The next most primitive form in Europe mustelids. Most of them were inferred from is I. arctoides, found in late Vallesian and possible ancestry-descendant relationships, 24 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 279 without detailing the occurrences of the same fa). The genus became rather prosperous in genera on the continents in question. the later half of the Miocene and the Pliocene The ®rst reliable record of migration of the in Eurasia. In North America, the earliest re- mustelids at the generic level is that of Po- cords of these two genera are from later tamotherium. The taxonomic position of this Hemphillian Mammal Age deposits, but no genus is highly controversial. It was tradi- earlier than 7 Ma. tionally placed in the subfamily Lutrinae. Re- cently it has been allocated variously: to pho- LEPARCTINES cids (Tedford, 1976; de Muizon, 1982), to One of the few carnivorans that migrated Oligobuninae (Baskin, 1998b), to a separate from America to Asia during the Neogene is subfamily (Ginsburg, 1999), or as a primitive Leptarctus. In Eurasia, this genus has so far mustelid (McKenna and Bell, 1997). Its ®rst been found only at Tung-Gur, China (late appearance is de®nitely in Europe, in the late middle Miocene, ϳ13 Ma). On the other Oligocene, with proliferation since the Aqui- hand, the leptarctine mustelids are well rep- tanian (ϳ23 Ma). In North America, it is ®rst resented by many forms ranging from early recorded in the Hemingfordian to late Hemphillian (20±6 (19±18 Ma). Thus, the dispersal event must Ma) in North America (but see Addendum). have occurred prior to 19 Ma. Of the Mellivorinae and Gulolinae (Is- LUTRINES chyrictini according to Baskin, 1998b) three genera are commonly shared by Eurasia and There are few identical genera of lutrines North America. Hoplictis has long been con- found in both continents. According to Bas- sidered a subgenus of Ischyrictis with hyper- kin (1998b), the mandibular fragment with carnivorous adaptations of the cheek teeth. It m1 from the Truckee Formation (late Clar- is comparatively well represented in Europe, endonian) described as Lutra sp. may belong from MN 3 (20 Ma) to MN 9 (ϳ11 Ma), but to Limnonyx, a genus created by Crusafont- only sparsely found in North America. The Pairo (1950) based on a mandible from the best specimen is a lower jaw from Ellensburg Vallesian type locality, Spain. If this proves Formation, Washington (late Clarendonian), true, the migration should have taken place originally described as Beckia grangerensis. at about 8.8±9.5 Ma (Baskin, 1998b), from Hoplictis sp. was also reported from a few Eurasia to North America. Mionictis had other Clarendonian Mammal Age localities long been considered a primitive lutrine of (Love Bone Bed and Juntura Formation). It Holarctic distribution until Ginsburg and is certain that Ischyrictis originated in Eu- Morales (1996) argued that the European rope, and its migration to North America can specimens once referred to that genus should only be postulated as prior to 11 Ma. On the be separated from Mionictis, and they estab- other hand, Plesiogulo and Eomellivora have lished two genera for them: Lartetictis and good records in Eurasia. They are now gen- Adroverictis. erally considered to have originated from the Trigonictis and Pannonictis have been Ischyrictis group, which is well established considered congeneric by many carnivore in the early and middle Miocene of Europe. students. If this proves true, the generic name Wolsan and Semenov (1996) recently revised Pannonictis should be adopted because of its Eomellivora based on a large sample from a year priority in occurrence. In Europe, it ap- limestone quarry of Turolian age near Grit- peared from the Villany (MN 16, ϳ3.4 Ma), sev, Ukraine. All the materials published by while in North America, the earliest record then, except those from the Siwaliks, were is from ϳ4 Ma. Therefore, Pannonictis may referred to two subspecies of E. wimani. The have had a North American origin and mi- ®rst appearance of this species is from the grated to Eurasia in the time span of 4 to 3.4 early Vallesian in Europe (Kalfa and ValleÂs Ma. The ®rst Martes almost simultaneously de Fuentidueha, ϳ11 Ma). It is well repre- occurred in Europe (MN 3) and North Amer- sented in the later half of the Miocene in Eur- ica (early Hemingfordian) at about 20 Ma. asia. The ®rst Plesiogulo appeared also in Since it is a form with mainly plesiomorphic Europe, from early Vallesian deposits (Kal- features, its real af®nity and the relationship 2003 QIU: DISPERSAL OF NEOGENE CARNIVORANS 25 between the forms of the two continents are long history there, starting at least from the still uncertain. Mustela similarly appeared middle Miocene. late in both continents, at the end of the Pli- Simocyon has long been referred to the ocene. The origin, time and direction of dis- family Canidae. Recently Wang (1997) and persal of both genera remain unsolved. Baskin (1998b) transferred it to the family Procyonidae, proposing a sister group rela- PROCYONIDS tion with the living lesser panda. Simocyon has a good record in Europe, from Vallesian The systematic position of the procyonids to Turolian (11.2±5.3 Ma). In China, only a within the arctoids has been highly contro- more advanced species, S. primigenius, has versial. Most systematists currently include been recorded in Baodean age deposits. in them with musteloids in their broader sense. North America, a few specimens were re- Since the extant procyonids (other than the ported. A mandible ®rst described by Thorpe lesser panda, Ailurus fulgens) are restricted (1921) as Araeocyon marshi from the Rattle- to and most of the procyonid fossils are snake Formation (Ͼ6.6±7.5 Ma), Oregon, found in the Americas, it has long been has been generally considered to belong to thought that the Procyonidae are Nearctic in Simocyon. Its migration from Eurasia to origin and no essential migration has ever North America should have occurred at ϳ8 taken place between Eurasia and North Ma. America. Baskin (1982) revised North Amer- ican Neogene procyonids, citing the earliest CANOIDS procyonid record in North America as Eda- phocyon lautus from the Runningwater For- There is no disagreement that North mation, Nebraska (18±19 Ma). In Europe, America was the center of origin and radia- Broiliana from the Wintershof-West (MN 3, tion of the canids. The earliest record of a 18±20 Ma) is widely accepted as an ancestral true European canid is that of Canis cipio form of the procyonids. Similarly, Amphictis from the Turolian deposits (MN 12, 8.2±7.1 from Quercy (Oligocene) and Wintershof- Ma) of Spain. It is rather perplexing that no West is also considered referable to Procyon- other canid material has since then been idae. Since the procyonids are distinct from found from the Turolian in Europe or equiv- all other carnivorans in dental morphology, alent deposits in Asia. On the other hand, it seems safe to postulate that the procyonids various canids, including Eucyon, Nyctereu- originated in Europe during Oligocene time tes, and Vulpes, became richly represented and migrated to North America at about 20 from the Ruscinian (5.3±ϳ3.4 Ma) in Europe Ma. and its equivalent in Asia (the early Yushean There is a general agreement in recent in China, Calta in Turkey, etc.). According years that the extant lesser panda, Ailurus to Tedford et al. (1987), ``Canis'' and Vulpes fulgens, and rare fossils like Parailurus con- ®rst appeared at the beginning of the Hem- stitute a separate group of procyonids. For phillian mammal age (ϳ9 Ma) in North quite a long time this group of procyonids America. There is no doubt that these two was thought to be restricted to Eurasia. Its genera migrated from North America to Eur- origin and distributional history remain long asia. If the ®nding of the Spanish Turolian enigmatic. Therefore, Tedford and Gustaf- Canis cipio can be veri®ed in the future, the son's report in 1977 on the discovery of a time of the migration should be ®xed at tooth of Parailurus from a Blancan assem- about 8 Ma. Otherwise it should be at about blage in Washington was highly unexpected. 5 Ma. The distribution of Nyctereutes, the Just recently, Ginsburg et al. (1997) de- raccoon dogs, is restricted to the Palaearctic scribed an ailurine m2 (Magerictis imperi- region. There is little doubt that its ancestor alensis) from the locality Estacion Imperial, is among the North American canids. This MN 4±5 (ϳ16±18 Ma), Spain. Based on the ``ancestor'' should remain very ``Canis''- highly derived characters of this Spanish like, lacking major diagnostic Nyctereutes form, Ginsburg et al. (1997) postulated that characters (the mandible subangular lobe, the ailurines originated in Europe and had a etc.). Migrating from North America to Eur- 26 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 279 asia at about the same time as ``Canis'' and Asia and Africa. Therefore, the migration of Vulpes, it rapidly evolved into true Nycter- the ®rst felids from Eurasia to North America eutes. may have occurred at 18±17 Ma. From then, the two lineages may have evolved separate- AELUROIDS ly. The of the extant felids is high- NIMRAVIDS ly controversial, as are the referral of some This group of carnivorans, the ®rst radia- Plio-Pleistocene felid fossils and the hypoth- tion of cat-like mammals, ¯ourished mainly eses on their relationships. This is particular- during the Paleogene in both Eurasia and ly true with the North American fossils re- North America. The only Neogene survivors ferred to Lynx and Felis. Schultz and Martin are members of the dirk-toothed Barbouro- (1972) referred a felid of Clarendonian age felines: Prosansanosmilus-Sansanosmilus from the Ogallala beds of Colorado directly lineage in Eurasia and Barbourofelis in to the genus Lynx: L. stouti. It was consid- North America. The latter is a group of high- ered the earliest record of the genus and its ly specialized dirk-toothed cat-like progenitor. Martin (1998b) transferred the suddenly appearing in the basal Clarendonian form to Pseudaelurus, following MacFadden (11 Ma). No possible ancestor of this group and Galiano's (1981) suggestion. Neverthe- occurs in North America. So far as is known, less, Martin maintained the North American the Eurasian middle Miocene Sansanosmilus origin of Lynx by including Felis rexroad- is the only possible candidate for its ancestral ensis in Lynx. The earliest record of L. re- form. Its dispersal from Eurasia to North xroadensis is the late Hemphillian (Upper America can thus be placed in the interval of Bone Valley Formation, Florida, 5 Ma). This 15 to 11 Ma. led Martin to think that the dispersal event of Lynx must have been from North America to Eurasia. Hendey (1974) and Werdelin FELIDS (1981) thought that Africa could be the con- Felids were considerably diversi®ed dur- tinent where Lynx originated, based mainly ing the late Cenozoic. Their dispersals be- on a mandible referred to Lynx aff. issiodor- tween Eurasia and North America must have ensis from Langebaanweg. The materials had occurred many times. The earliest true from both Africa and North America referred felids were found in the Oligocene in Europe to Lynx are inadequate. The European Lynx (Proailurus), but are only late early Miocene issiodorensis is more likely to be referred to in North America. According to the latest Lynx. However, it lacks the diagnostic Lynx study by Hunt (1998c), the earliest North features in its postcranial skeleton (short American true felids are close to the Euro- body but longer hind limb). It is hard to say pean Proailurus, with more plesiomorphic whether Lynx issiodorensis originated from characters in the auditory area. However, Felis rexroadensis or the Langebaanweg they appeared later than in Europe, in the Lynx aff. issiodorensis, based on presently Hemingfordian (at about 17 Ma). Pseudae- available material. True Lynx with the char- lurus is known from both Eurasia and North acteristic short body but long hind limb ap- America. In Europe, the earliest occurrence peared late, and the dispersal event (probably of the genus is in MN 3 (Wintershof-West, from Eurasia to North America) likely oc- etc., 20±18 Ma), while in Asia it is in Sihong curred in the late Pleistocene. (comparable to MN 4, ϳ18 Ma). In North Machairodus has a rather continuous re- America, Pseudaelurus evolved rather con- cord in Europe, from the basal Vallesian servatively, resulting mainly in size increase, through the Turolian. In Asia, Machairodus and became extinct prior to the Blancan was found only from the Baodean (no earlier Mammal Age (4.2 Ma). However, in Eurasia, than 10 Ma). Morphologically, the genus can especially in Asia, its counterpart evolved be traced back to the genus Miomachairodus into the more specialized Metailurus, Para- from the late middle Miocene (Yeni-Eskih- machairodus, and Dinofelis, the latter of isar, Turkey, ϳ12 Ma). In North America, the which lingered into the Pleistocene in both earliest record of this genus is from the 2003 QIU: DISPERSAL OF NEOGENE CARNIVORANS 27

Smiths Valley, Nevada (7 Ma; Tedford et al., summary of the then available data and rec- 1987, ®g. 6.2). Therefore, the dispersal event ognized two major phases of seaway opening of Machairodus from Eurasia to North in Beringia. The ®rst opening occurred dur- American should have occurred at the end of ing the late Miocene (10±12 Ma), when a the early Hemphillian Mammal Age. large number of Paci®c invertebrates and Homotherium (ϭIschyrosmilus and Dino- marine mammals reached the North Atlantic. bastis) did not appear until the Blancan The reopening of Bering Strait took place in Mammal Age. Its earliest appearance in the late Pliocene (3.5±4 Ma), as evidenced North America cannot be precisely dated. by extensive occurrence of the same mol- Well-dated material came from the deposits lusks on either side of the strait. This would of late Blancan Mammal Age (Cita Canyon, mean that, during the majority of Neogene Lisco) and the Pleistocene. Since Homoth- time other than the two episodes of reopen- erium species from Yushe (and also from the ing of the Bering Seaway, Beringia could Siwaliks) are morphologically more primi- serve as a land bridge connecting the Eur- tive than those of North America, it may be asian and North American continents. safe to say that Homotherium appeared ®rst A considerable step forward was made by in Eurasia, and then migrated into North Tedford et al. (1987). In using genera exotic America at probably the ®rst half of the to North America to de®ne the beginning of Blancan Mammal Age. ``age'' and ``subage'', they listed 13 Neogene The dirk-toothed cat, Megantereon, was appearance events for North America. It is reported recently by Berta and Galiano not fully clear what ``exotic'' entails, and do (1983) from the latest Hemphillian Mammal all these immigrants and their ®rst appear- Age (Upper Bone Valley Formation, Flori- ances de®ne distinct events? In a compre- da). It is dated as about 4.5 Ma. In Eurasia, hensive review of Cenozoic Era dispersal the earliest occurrence of this genus has been history, Woodburne and Swisher (1995) list- dated not older than 3.5 Ma (Les Etouaires, ed six major Neogene interchange events: (6) France). In Asia, the earliest record of Me- beginning of Hemingfordian, 19 Ma, (7) be- gantereon is probably the undescribed ma- ginning of late Hemingfordian, 17.5 Ma, (8) terial from Fanchang, Anhui, a Paleolithic beginning of Barstovian, 15.9 Ma, (9) begin- site dated as 2.4±2 Ma. If all the above data ning of Clarendonian, 11 Ma, (10) beginning are correct, the only possible interpretation is of Hemphillian, 9 Ma, and (11) beginning of the reverse direction of dispersal of the genus late Blancan, 2.5 Ma. Megantereon: from North America to Eura- From the point of view of carnivorans, it sia. seems that only three major dispersal waves of the ®lter-bridge type can be recognized. HYAENIDS The ®rst is around 20 Ma. The whole dura- tion of the dispersal wave could have lasted Of the hyaenids only the genus Chasma- 2 or 3 m.y., probably from 21 Ma to 19 or porthetes migrated from Asia to North even 18 Ma, during which time, intermittent America early in the Blancan Mammal Age dispersals might have occurred. The carni- (later than 4.2 Ma). In Asia, the fossil record vorans that migrated from Eurasia to North of this genus is quite long, dating back to ϳ6 America included Cynelos, Ysengrinia, Am- Ma (Yushe) and continuing to about 1.8 Ma phicyon, Cephalogale, Phoberocyon, Ursa- (Nihewan). vus, some small-sized mustelids, Potamoth- erium, an ancestral form of Edaphocyon, and CONCLUDING REMARKS Proailurus. According to Ginsburg's compi- Early in 1947, Simpson, one of the great- lation (1999), in Europe, the early Miocene est pioneer paleozoogeographers among ver- contains 37 carnivoran genera, of which at tebrate paleontologists, pointed out a pro- least 8 are recorded from North America. It nounced endemism of the Clarendonian fau- is noteworthy that these immigrants cover na of North America. This accords nicely the major carnivoran groups, leaving only with the evidence provided by marine pale- the viverrids unrepresented. There is little ontology. Hopkins (1967) made an excellent reason to consider the fossil viverrids radi- 28 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 279 cally different from the living ones ecologi- Hipparion at about 11.5 Ma, and Equus at cally. Their absence in the list of the migrants 2.5 Ma. Gomphotherium spread from Eurasia may indicate that the migration concerned to North America at 16 Ma. only nontropical elements of the fauna. It is The above migration scheme (®g. 2.1) also remarkable that no carnivorans were grossly accords with that deduced from ma- found to have migrated in the opposite di- rine paleontology, if slight adjustment is rection, that is, from North America to Eur- made. As stated above, the two episodes of asia. The only noncarnivoran emigrant from opening of the Bering Strait during Neogene North America to Eurasia during this period time, according to marine paleontology, were of time is Anchitherium. Woodburne and in the late Miocene (ϳ12±10 Ma) and the Swisher (1995) assigned this exchange event late Pliocene (4±3.5 Ma). Since the migra- to Simpson's corridor type. Judging from the tion of Hipparion took place at 11.5 Ma, limited number and the ecological constraints which is extensively veri®ed and in good ac- of migrants involved, it is more reasonable cordance with sea-level ¯uctuation data (at to view the above dispersal wave as by ®lter- the base of TB3), the late Miocene opening bridge. of the Bering Strait may have started later The next major dispersal wave occurred at than 12 Ma, probably later than the Hippa- about 7±8 Ma, the early part of Hemphillian. rion migration event. The carnivorans migrating from Eurasia to The effects of the migration differ widely North America included Indarctos, Agrioth- from taxon to taxon. Most of the carnivorans erium, Simocyon, Eomellivora, Plesiogulo, did not play important roles after their settle- and Machairodus. As is well known, the car- ment in the new continent, except two: ``Ca- nivorans of the Eurasian Hipparion fauna are nis'' and probably the descendants of Me- particularly well represented among fossils gantereon. The ``Canis'' species ¯ourished and rich in number of taxa, totaling 45 (Gins- rapidly and diversi®ed after entering Eurasia. burg, 1999). Nevertheless, the number of the It is commonly believed that this group of emigrants to North America constitutes only carnivorans was similar to fossil hyaenas a small fraction of the total, only about one- ecologically and probably played an impor- sixth. It is interesting to note that the emi- tant role in the of the hyaenas in grants are mostly large-sized carnivorans. the north part of the Palaearctic region. As The last major dispersal wave is early stated above, Smilodon may be a descendant Blancan, ϳ4 Ma. The Eurasian emigrants re- of Megantereon in North America. It became corded in North America are Ursus, Parail- the second dominant group of carnivorans at urus, Lynx (?), Felis (?), Homotherium, and Rancho La Brea, next to Canis dirus.Onthe Chasmaporthetes. At about the same time, other hand, herbivorous emigrants like Megantereon and Pannonictis migrated from equids and proboscideans played very im- North America to Eurasia. portant roles in changing faunal composition There are two small-scale migration events after having intermingled with native ele- of carnivorans in the opposite direction, that ments. is, from North America to Eurasia. The dis- persal concerns only single taxa. At about 13 ACKNOWLEDGMENTS Ma, Leptarctus migrated from North Amer- ica to Asia. Probably at the same time, an- This paper is especially dedicated to Dr. cestral Barbourofelis (a form like Sansanos- R. H. Tedford, to whom the author owes his milus) migrated from Eurasia to North Amer- progress, if any, in the studies of the Chinese ica. At ϳ5 Ma (or 7 Ma, vide supra), Canis Neogene mammals. Dr. Tedford's erudition and some of its closely related forms might and willingness to help others were so amply have crossed Beringia. Canids came to Eur- manifested during the execution of the Yushe asia apparently together with camels. Project, such that everyone who worked with For noncarnivoran mammals, at least four him bene®ted greatly. For reviews and im- other dispersal events are well established: provement of the manuscript, the author is the three equid genera emigrated from North particularly appreciative of the efforts of Drs. America to Eurasia: Anchitherium at 20 Ma, J. Baskin, L. J. Flynn, and J. Barry. 2003 QIU: DISPERSAL OF NEOGENE CARNIVORANS 29

REFERENCES bassin d'Aquitaine. MuÈnchner Geowissenschaf- liche Abhandlungen 10: 49±58. Baskin, J.A. 1982. Tertiary Procyoninae (Mam- de Muizon, C. 1982. Les relations phylogeÂneÂ- malia, Carnivora) of North America. Journal of tiques des Lutrinae (Mustelidae, Mammalia). Vertebrate Paleontology 2(1): 71±93. GeÂobios MeÂmoire SpeÂciale 6: 259±277. Baskin, J.A. 1998a. Procyonidae. In C.M. Janis, Flower, W.H. 1869. On the value of the characters K.M. Scott, and L.L. Jacobs (editors), Evolu- of the base of the cranium in the classi®cation tion of Tertiary mammals of North America. 1: of the order of Carnivora, an on the systematic Terrestrial carnivores, ungulates, and ungulate- position of Bassaris and other disputed forms. like mammals: 144±151. New York: Cam- Proceedings of the Zoological Society of Lon- bridge University Press. don 1869: 4±37. Baskin, J.A. 1998b. Mustelidae. In C.M. Janis, Flynn, J.J., and C.C. Swisher III. 1995. Cenozoic K.M. Scott, and L.L. Jacobs (editors), Evolu- South American land mammal ages: correlation tion of Tertiary mammals of North America. 1: to global geochronologies. In W.A. Berggren, Terrestrial carnivores, ungulates, and ungulate- D.V. Kent, M.-P. Aubry, and J. Hardenbol (ed- like mammals: 152±173. New York: Cam- itors), Geochronology, time scales and global bridge University Press. stratigraphic correlation. Society of Economic Baskin, J., and R.H. Tedford. 1996. Small arctoid Paleontologists and Mineralogists Special Pub- and feliform carnivorans. In D.R. Prothero and lication 54: 317±333. R.J. Emry (editors), The terrestrial Eocene-Ol- Forster-Cooper, C. 1923. Carnivora from the Dera igocene transition in North America: 486±497. Bugti deposits of Baluchistan. Annals and New York: Cambridge University Press. Magazine of Natural History, London (9) 12: 259±263. Berta, A. 1981. The Plio-Pleistocene hyaena Ginsburg, L. 1966. Les amphicyons des phospho- Chasmaporthetes ossifragus from Florida. rites du Quercy. Annales de PaleÂontologie Journal of Vertebrate Paleontology 1(3±4): 52(1): 1±44. 341±356. Ginsburg, L. 1999. Order Carnivora. In G.E. RoÈs- Berta, A., and H. Galiano. 1983. Megantereon sner and K. Heissig (editors), The Miocene hesperus from the late Hemphillian of Florida land mammals of Europe: 109±148. MuÈnchen: with remarks on the phylogenetic relationships Verlag Dr. Friedrich Pfeil. of machairodonts (Mammalia, Felidae, Ma- Ginsburg, L., and J. Morales. 1996. Lartetictis et chairodontinae). Journal of Paleontology 57(5): Adroverictis, nouveaux genres de Melinae 892±899. (Mustelidae, Carnivora, Mammalia) du Mio- Brown, J.H., and A.C. Gibson. 1983. Biogeogra- ceÁne de l'Ancien Monde. Bulletin du Museum phy. St. Louis-Toronto-London: C.V. Mosby. National d'Histoire Naturelle section C, Sci- Cain, S.A. 1944. Foundations of plant geography. ences de la Terre Paleontologie Geologie Mi- New York: Harper. neralogie (4) 18: 663±671. Crusafont-Pairo, M. 1950. Limnonyx, un nuevo Ginsburg, L., and J. Morales. 1998. Les Hemi- Lutrido del Mioceno espanol. Notas y Comun- cyoninae (Ursidae, Carnivora, Mammalia) et icaciones del Instituto GeoloÂgico y Minero de les formes apparenteÂes du MioceÁne in¯eÂrieur et EspanÄa 20: 1±14. moyen d'Europe occidentale. Annales de Pa- Dalquest, W.W. 1986. Lower jaw and dentition of leÂontologie 84(1): 71±123. the Hemphillian bear, Agriotherium (Ursidae), Ginsburg, L., J. Morales, D. Soria, and E. Her- with the description of a new species. Journal raez. 1996. DeÂcouverte d'une forme ancestrale of Mammalogy 67(4): 623±631. du petit panda dans le MioceÁne moyen de Ma- Dawson, M.R. 1999. Bering down: Miocene dis- drid (Espagne). Comptes Rendus de persals of land mammals between North Amer- l'Academie de Science, Paris, SeÂrie IIA, Sci- ica and Europe. In G.E. RoÈssner and K. Heissig ences de la Terre et des PlaneÁtes 325: 447±451. (editors), The Miocene land mammals of Eu- Hendey, Q.B. 1974. The late Cenozoic Carnivora rope: 473±483. MuÈnchen: Verlag Dr. Friedrich of the South-Western Cape Province. Annals of Pfeil. the South African Museum 63: 1±369. de Bonis, L. 1973. Contribution aÁ l'eÂtude des Hopkins, D.M. 1966. The Cenozoic history of mammifeÁres de l'Aquitanien de l'Agenais. MeÂ- BeringiaÐA synthesis. In D.M. Hopkins (edi- moires du Museum National d'Histoire Natu- tor), The Bering land bridge: 451±484. Stan- relle, n. ser. (C) 28: 1±192. ford: Stanford University Press. de Bonis, L., and Y. Guinot. 1986. Le gisement Hunt, R.M., Jr. 1996. Biogeography of the Order de verteÂbreÂs de TheÂzels (Lot) et la limite Oligo- Carnivora. In J.L. Gitttleman (editor), Carni- MioceÁne dans les formations continentales du vore behavior, ecology, and evolution 2: 485± 30 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 279

541. Ithaca and London: Comstock Publishing Scott, and L.L. Jacobs (editors), Evolution of Associates. Tertiary mammals of North America. 1: Terres- Hunt, R.M., Jr. 1998a. Ursidae. In C.M. Janis, trial carnivores, ungulates, and ungulatelike K.M. Scott, and L.L. Jacobs (editors), Evolu- mammals: 236±242. New York: Cambridge tion of Tertiary mammals of North America. 1: University Press. Terrestrial carnivores, ungulates, and ungulate- Martin, L.D. 1998c. Hyaenidae. In C.M. Janis, like mammals: 174±195. New York: Cam- K.M. Scott, and L.L. Jacobs (editors), Evolu- bridge University Press. tion of Tertiary mammals of North America. 1: Hunt, R.M., Jr. 1998b. Amphicyonidae. In C.M. Terrestrial carnivores, ungulates, and ungulate- Janis, K.M. Scott, and L.L. Jacobs (editors), like mammals: 243±246. New York: Cam- Evolution of Tertiary mammals of North Amer- bridge University Press. ica. 1: Terrestrial carnivores, ungulates, and un- McKenna, M.C., and S.K. Bell. 1997. Classi®ca- gulatelike mammals: 196±227. New York: tion of mammals above the species level. New Cambridge University Press. York: Columbia University Press. Hunt, R.M., Jr. 1998c. Evolution of the Aeluroid Merriam, J.C., C. Stock, and C.L. Moody. 1925. Carnivora: Diversity of the earliest aeluroids The Pliocene Rattlesnake Formation and fauna from Eurasia (Quercy, Hsanda-Gol) and the or- of eastern Oregon, with notes on the geology igin of felids. American Museum Novitates of the Rattlesnake and Mascall deposits. Car- 3252: 1±65. negie Institution of Washington Publication Korotkevich, E.L., and V.A. Topachevskii. 1986. 347: 43±92. Related elements among the Neogene mam- Qiu, Z-x., and G.Q. Qi. 1986. Ailuropod found malian fauna of eastern Europe, Asia and North from the late Miocene deposits in Lufeng, Yun- America. In V.L. Kontrimavichus (editor), Ber- nan. Vertebrata PalAsiatica 27(3): 153±169. ingia in the Cenozoic Era: 289±295. New Deh- Qiu, Z-x., and Z.D. Qiu. 1995. Chronological se- li: Rekha Printers. quence and subdivision of Chinese Neogene KurteÂn, B. 1986. Trans-beringian carnivore pop- mammalian faunas. Paleogeography, Paleocli- ulations in the Pleistocene. In V.L. Kontrimav- matology, Paleoecology, 116: 41±70. ichus (editor), Beringia in the Cenozoic Era: Qiu, Z-x., and N. Schmidt-Kitler. 1981. Agrioth- 339±345. New Dehli: Rekha Printers. erium intermedium (Stach, 1957) from a Plio- Kuss, S. 1962. Problematische Caniden des eu- cene ®ssure ®lling of Xiaoxian County (Anhui ropaÈischen TertiaÈrs. Berichte der Naturfor- Province, China) and the phylogenetic position schende Gesellschaft zu Freiburg in Breisgau of the genus. Paleovertebrata 13(3): 65±81. 52: 132±172. Qiu, Z-x., J.Y. Xie, and D. Yan. 1991. Discovery Kuss, S. 1965. Revision der europaÈischen Am- of late Miocene Agriotherium from Jiegou, phicyoninae (Canidae, Carnivora, Mammalia) Gansu, and its taxonomic implications Verte- ausschliesslich der voroberstampischen For- brata PalAsiatica 29(4): 286±295. men. Sitzungberichte der Heidelberg Akademie Qiu, Z-x., D. Yan, H. Jia, and B. Sun. 1986. The der Wissenschaften, Mathematisch-naturwis- large-sized ursid fossils from Shanwang, Shan- senschaften Klasse 1: 1±168. dong. Vertebrata PalAsiatica 24(3): 182±194. Lundelius, E.L., Jr., T. Downs, E.H. Lindsay, H.A. Qiu, Z-x., D. Yan, H. Jia, and B.Z. Wang. 1985. Semken, R.J. Zakrzewski, C.S. Churcher, C.R. Dentition of the Ursavus skeleton from Shan- Harrington, G.E. Schultz, and S.D. Webb. wang, Shandong province. Vertebrata Pal- 1986. The North American Quaternary se- Asiatica 23(4): 264±275. quence. In M.O. Woodburne (editor), Cenozoic Remy, J.A., J.Y. Crochet, B. SigeÂ, J. Sudre, L. de mammals of North America: 211±235. Berke- Bonis, M. Vianey-Liaud, M. Godinot, J.L. Har- ley-Los Angeles-London: University of Cali- tenberger, B. Lange-BadreÂ, and B. Comte. fornia Press. 1987. Biochronologie des phosphorites du MacFadden, B.J., and H. Galiano. 1981. Late Quercy: mise aÁ jour des listes faunistiques et Hemphillian cat (Mammalia, Felidae) from the nouveaux gisements de mammifeÁres fossiles. Bone Valley Formation of central Florida. Jour- MuÈchner Geowissenschaftliche Abhandlungen nal of Paleontology 55: 218±226. 10: 169±188. Martin, L.D. 1998a. Nimravidae. In C.M. Janis, Repenning, C.A. 1967. Palearctic-Nearctic mam- K.M. Scott, and L.L. Jacobs (editors), Evolu- malian dispersals in the late Cenozoic. In D.M. tion of Tertiary mammals of North America. 1: Hopkins (editor), The Bering land bridge: 288± Terrestrial carnivores, ungulates, and ungulate- 311. Stanford: Stanford University Press. like mammals: 228±235. New York: Cam- Savage, D.E. 1958. Evidence from fossil land bridge University Press. mammals on the origin and af®nities of the Martin, L.D. 1998b. Felidae. In C.M. Janis, K.M. western Nearctic fauna. In C.L. Hubbs (editor), 2003 QIU: DISPERSAL OF NEOGENE CARNIVORANS 31

Zoogeography. American Association for the Faunal succession and biochronology of the Advancement of Science, Publication 51: 97± Arikareean through Hemphillian interval (late 129. Oligocene through earliest Pliocene Epochs) in Schultz, C.B., and L.D. Martin. 1972. Two Lynx- North America. In M.O. Woodburne (editor), like cats from the Pliocene and Pleistocene. Cenozoic mammals of North America: 153± Bulletin of the University of Nebraska State 210. Berkeley-Los Angeles-London: University Museum 9(7): 197±203. of California Press. Schultz, C.B., and L.D. Martin. 1975. (Ur- Thorpe, M.R. 1922. Araeocyon, A probable Old sidae) from the Late Cenozoic of Nebraska. World migrant. American Journal of Science 3: Bulletin of the University of Nebraska State 371±377. Museum 10(1): 47±54. Van Valen, L. 1969. The multiple origins of the Simpson, G.G. 1947. Holarctic mammalian faunas placental carnivores. Evolution 23(1): 118±130. and continental relationships during the Ceno- Wang, X.M. 1996. New cranial material of Sim- zoic. Geological Society of America Bulletin ocyon from China, and its implications for phy- 58: 613±688. logenetic relationship to the red panda (Ailu- Steininger, F.F., W.A. Berggren, D.V. Kent, R.L. rus). Journal of Vertebrate Paleontology 17(1): Bernor, S. Sen, and J. Agusti. 1996. Circum- 184±198. Mediterranean Neogene (Miocene and Plio- Werdelin, L. 1981. The evolution of lynxes. Acta cene) marine-continental chronologic correla- Zoologica Fennica 18: 37±71. tion of European mammal units. In R.L. Bernor, Werdelin, L. 1996. Carnivores, exclusive of V. Fahlbusch, and H.-W. Mittmann (editors), Hyaenidae, from the late Miocene of Europe The evolution of western Eurasian Neogene and western Asia. In R.L. Bernor, V. Fahlbusch, mammal faunas: 7±46. New York: Columbia and H.-W. Mittmann (editors), The evolution of University Press. western Eurasian Neogene mammal faunas: Storer, J.E. 1975. Tertiary mammals of Saskatch- 271±289. New York: Columbia University ewan. III: The Miocene fauna. Royal Ontario Press. Museum Life Sciences Contribution 103: 1± Wolff, R.G. 1978. Function and phylogenetic sig- 134. ni®cance of cranial anatomy of an early bear Tang Y.J., and Z.D. Qiu. 1979. Analysis and dis- cussion of the vertebrate fossils from Bose, Gu- (Indarctos) from Pliocene sediments of Florida. angxi (in Chinese). In Institute of Vertebrate Carnivora 1(3): 1±12. Paleontology and Paleoanthropology and Nan- Wolsan, M., and Y. Semenov. 1996. A revision of jing Institute of Geology and Paleontology (ed- the late Miocene mustelid carnivoran Eomelli- itors), Mesozoic-Cenozoic red beds of south vora. Acta Zoologica Cracovensia 39(1): 593± China: 407±415. Beijing: Science Press. 604. Tedford, R.H. 1976. Relationships of pinnipeds to Woodburne, M.O., and C.C. Swisher, III. 1995. other carnivores (Mammalia). Systematic Zo- Land mammal high-resolution geochronology, ology 25(4): 363±374. intercontinental overland dispersals, sea level, Tedford, R.H., and E. Gustafson. 1977. First climate, and vicariance. In W.A. Berggren, D.V. North American record of the extinct panda Kent, M.-P. Aubry, and J. Hardenbol (editors), Parailurus. Nature 265(5595): 621±623. Geochronology, time scales and global strati- Tedford, R.H., M.F. Skinner, R.W. Fields, J.M. graphic correlation. Society of Economic Pa- Rensberger, D.P. Whistler, T. Galusha, B.E. leontologists and Mineralogists Special Publi- Taylor, J.R. Macdonald, and S.D. Webb. 1987. cation 54: 335±364.

Addendum. Just recently a leptarctine skull was found in early Miocene deposits (ϳ19.5 Ma) in the Danghe area, Gansu province, China (Wang et al., in press). This is about the same time as the earliest leptarctine material found in North America. Thus, it is not impossible that the North American leptarctines were immigrants from Eurasia during the ®rst migration wave at about 20 Ma.