THE RELATIONSHIPS OF THE ARCTOSTYLOPIDAE (MAMMALIA) NEW DATA AND INTERPRETATION
RICHARD L. CIFELLI,' CHARLES R. SCHAFF,- AND MALCOLM C. McKENNA
CONTENTS Abstract. The dental morpholog steini, hitherto known onl) from pari Abstract cheek- tooth series is desi rilx-d on the ba Introduction collected specimen including .i near!) Acknowledgments per and lower dentition \U arctoMylopid • Abbreviations from North America principally from th Paleontology 5 Systematic Basin, Wyoming, are apparentl) referable l Order new 5 Arctostylopida, gle species, which therefore ranges from the lati 5 Family Arctostylopidae luiiian through the ( larklorkian Other an I 6 Matthew, 1915 to the \si.m Arctostylops genera are restricted I'.ileogi A. steini Matthew, 1915 6 A previously described 5pe< ies "l Matthew and Granger, Palaeostijlops sty/ops is placed in a new genus Gashal 1 1 1925 Arctostylopidae and its constituent subordii P. iturus Matthew and Granger, are diagnosed, and a h\ pothesis "I relationshi| 1925 1 5 in the family is presented B) comparison witl 15 new l>\ / Gashatostylops genus ungulate morphohpe as represented et .. 15 G. macrodon (Matthew al., 1929) latum, Asiostylops spumes is hypothesized 20 and 1976 ... Sinostylops Tang Yan, most primitive member ol the family; Both 1976 ... 20 S. promissus Tang and Yan, notios and B. progressus retain man) primitivi Bothriostylops Zheng and Huang, tures but clearK hear some ol the spec lalizatkx 22 other 1986 in Palaeostijlops, Arctostylops, and and 1986 ... North \mt B. notios Zheng Huang, genera. Of these derived taxa, and 1976) sister taxou to the n B. progressus (Tang Yan, tostylops may be the 1978 22 are in distributioi Anatolostylops Zhai, genera, all of which Vsiatfc 23 A. dubius Zhai, 1978 ato/osfy/opi and an as \et unnamed sp< 1979 23 taxa that be most dote!) n Asiostylops Zheng, specialized sister ma) A. spanios Zheng, 1979 to Gashatostylops th. Kazachostylops Nesov, 1987 Comparison of morphotypes tor K. occidentalis Nesov, 1987 daeand for southern Notoungulata sug Irorn within th.- oUm Arctostylopidae?, incertae sedis ivation of one group is unlikeb This Allostylops Zheng, 1979 rentlv known, compai 1979 most similai A. periconatus Zheng, indicates that notoungulate indet. acquin Arctostylopidae, genus and species tostylopids were independent!) the Arcto- Comparative Dental Morphology of basis for an exclusive relationship taxa is a d. to Notoungulata as sister single stylopidae and The Notoungulata of South America acter. The ankles ol arctostylopids and shan Discussion are divergently specialized eutherian morpnotype Possible Relationships not present in a ^ is. herefon Remaining Resemblances family Arctostvlopulae for th, Distinctness of Arctostylopida Notoungulata Relatives unknowi Holarctic faunas remain members of the group r. i 1 and De- although Oklahoma Museum of Natural History .uammaUtrom.h. I Nor- er enigmatic of Zoology, University of Oklahoma, mo. partment Because it is a well-defined man, Oklahoma 73019. relationship I Uni- out obvious elose 2 of Zoology, Harvard Museum Comparative the M,t opidaei 02138. groups, versity, Cambridge, Massachusetts IV. iati Arctostylopida 3 of Natural History, Central American Museum removes the i York 10024. Notoungulata Park West at 79th Street, New York, New
Zool 152 Bull. Mus. Comp Bulletin Museum of Comparative Zoology, Vol. 152, No. 1
to Neo- leocene or early Eocene link of Holarctic the Arctostylopidae have been integral to mammal faunas and in accordance tropical suggests, the development of correlations of early with other evidence now available, that whatever strata (Dashzeveg, 1982; Ginger- inter American connections of mammal faunas oc- Tertiary curred must have been earlier in time. The geometry ich and Rose, 1977; Matthew and Granger, ol li\ pothesized relationships among the Arctostylop- 1925; Szalay and McKenna, 1971). was most abundant id. te and the fact that the group Arctostylops, represented by the type .Hid diverse in Asia suggest an Asian, rather than (and only) species, A. steini, was described \.>rth or South American, origin for the family. by Matthew (1915), based on a partial low- er jaw from the "lower Gray Bull beds, Clark Fork This local- INTRODUCTION Basin, Wyoming." ity is probably, but not certainly, Clark- Since the studies of Gaudry (1902, 1904, forkian in age (Rose, 1981). Matthew re- 1906, 1908) and Scott (1904), it has been ferred the genus without question to the widely accepted that South America's fau- Notoungulata, hitherto known only from na is largely autochthonous, a result of that South America, placing it in the "Entelo- continent having been isolated by sea nychia," a mixed assemblage that then barriers from the rest of the world for most contained the most primitive of known no- of the Tertiary. Endemism at high taxo- toungulates. Matthew believed Arctosty- nomic levels is particularly conspicuous lops to be early Eocene in age, which may among the land mammals, which under- well be the case, but is a matter of defi- \\ ent their great diversification and radia- nition. Further materials of the species tions largely within the span of the Ter- were not forthcoming for another 50 years, tiary. It thus came as a great surprise when, when a specimen was reported nearby from in the first part of this century, apparent the Silver Coulee beds of the Polecat Bench members of South America's largest and Formation near Princeton Quarry. This lo- most characteristic group of hoofed mam- cality is late Paleocene (late Tiffanian) in mals, the Notoungulata, were described age (Jepsen and Woodburne, 1969). Inten- from specimens recovered in Wyoming sive collecting by Gingerich, Rose, and as- Matthew, 1915) and Asia (Matthew and sociates in Clarkforkian beds of the Clarks Granger, 1925; Matthew, Granger, and Fork Basin has produced four additional Other Simpson, 1929). possible close rel- specimens, consisting of dentulous lower atives Holarctic among and Nearctic jaw fragments and isolated teeth (Ginger- mammal faunas had been and have con- ich and Rose, 1977; Rose, 1981). The single tinued to be suggested (Ameghino, 1906; report of Arctostylops steini from outside Gingerich, 1985; McKenna, 1981). None- the Clarks Fork Basin is that of McKenna theless, none of the proposed relationships (1980), who recorded the species from beds seemed so based on characteristic certain, of Clarkforkian age at Togwotee Pass, synapomorphies, as in the case of these northwestern Wyoming. for ungulates, the Holarctic Arctostylopi- However, related mammals had in the 'l.ir a possess strongly specialized dentition meantime been recovered from Asia. Pa- that resembles alone notoungulates among leontological work at Gashato in Mongolia mammals. For this reason, the Arctosty- by the American Museum of Natural His- have lopidae figured prominently in dis- tory's Central Asiatic Expeditions led to cussions of the origin and early dispersal the description of two species, Palaeosty- of South America's native land mammal lops iturus Matthew and Granger, 1925 fauna McKenna, 1981; Simpson, 1951, and "P." macrodon Matthew, Granger, L978 L980) and of in zoogeography gen- and Simpson, 1929. These species are (Colbert, 1973; 1957; Darlington, probably latest Paleocene in age (Szalay Simpson. 1965). In because of addition, and McKenna, 1971). More recent addi- their presence in North America and Asia, tions to the group have come from slightly •
at deposits Naran Bulak, Mon- • younger Hua i (Gradzinski et al., 1969), the Paleo- golia Sinostylops pi cene and Eocene (or possibly The most Oligocene) recenl addition to the famil of China and (Tang Yan, 1976; Zhai, 1978; Kazachostylo) Zheng, 1979; Zheng and 1986), Neso\ Huang, L981 From the late l and the Paleocene of the USSR tashkent (Nesov, Svita of Kazakhstan i S5 1987), where seven additional described Since the initial descriptii species, placed in six genera, bear witness and stylops Palaeostylops Matthew I to a modest radiation of Matthew Arctostylopidae and Granger 1 identsl in the of Asia. 4 and early Tertiary Tang realized that these Holan ti< form Yan described includ- (1976) Sinostylops, some respects, more primitive than two from the late Paleocene ing species, known South American Notoungul of Anhui Province, China. S. while in other promissus respects the) are uniquel) the Dou-mu the (from Formation), type specialized. Their primitiveness is refl< is based on a mandibular ramus ed ol species, by placement the famil) \n t< with eight teeth; S. progressus (collected lopidae in Simpson's archaic notoungulate in the Shuang-ta-si Group and later trans- suborder Notioprogonia Simpson 1 ferred to a new genus, Bothriostylops) from 1945). However, the relationships ol Hoi- six jaw fragments. Anatolostylops dubius arctic to South American forms have n< was described by Zhai (1978) from the pu- been considered in detail Foi this reason tative early Eocene (but see below) Shi- a variety of opinions exist as to the place san-jian-fang Formation of the Turpan Ba- of origin of notoungulates and theii sub- sin, Xin-jiang Province, China. The species sequent dispersal patterns, with authors is known from a maxillary fragment with variously favoring northern Patterson 2 3 \-i.ui well-preserved M . Two additional gen- 1958) or. speeitiealb Matthew era and species were published by Zheng 1928; Zheng. L979 Smith American Marshall (1979). Asiostylops spanios, from the late (Hoffstetter, 1970: de Muizon and ( Paleocene Lan-ni-kong Member of the Chi- and Sige, 1983; Simpson 1951 en- l and I l »7' jiang Formation, Jiang-xi Province, China, tral American (Gingerich is based on a skull and associated mandible centers of origin. the data base for preserving much of the dentition. Because Until recently, making an assessment has been rather limited of its primitiveness with respect to other such South American no- members of the family, Zheng (1979) The early Tertiar\ have received monographs placed Asiostylops in its own monotypic toungulates treatment 1948, 1967 rheHol- subfamily. Allostylops periconatus Zheng, (Simpson. the was 1979, from the late Paleocene Wang-wu arctic radiation, Vrctostylopidae the lowei Member of the Chi-jiang Formation, Jiang- long represented onl) b) single dentition described lor \ortl xi Province, is known from an incomplete originall) American Arctostylops rteini and b) den- rostral part of a skull with poorly preserved to Vsiai 2 3 also from titions of two species referred P to M . Bothriostylops notios, some tonus remain the Wang-wu Member of the Chi-jiang laeostylops. Although Vsian arctostylopid taxa de- Formation, was described by Zheng and poorly known, scribed in recenl years add substantial!) t knowledge of morphological divers! 4 has An additional, undescribed genus and species a dramaticall) within the group, offering been from the late Paleocene Da-tang Mem- reported basis for comparison with N< ber of the Nung-shan Formation, Guang-dong Prov- improved ince, China (Li and Ting, 1983). Dashzeveg (1982) toungulata dentition recorded an undescribed species of "Arctostylops Herein we describe the of the Naran Bulak For- has • from the Bumban Member steini, much of which local tostylops mation, in the section than the Mongolia, higher hitherto unknown, based on a new!) occurrence of Palaeostylops iturus. Vol. 152, No. 1 4 Bulletin Museum of Comparative Zoology,
BUCCAL
ANTERIOR - POSTERIOR
LINGUAL Mesostyle
Metastyle
Parastyle Ectoloph Fossette Postmetaconule crista
Preprotocrista Postprotocrista
Protocone Pseudohypocone
Precingulum Postcingulum Sulcus pseudohypocone
Lingual cingulum
Ectocingulid cristid obliqua (selenolophid)
Talonid basin Paracristid
Protoconid Hypoconulid
Entolophid Precingulid""^ \yr A \ Entoconid Protocristid
Trigonid Metaconid Talonid notch notch
Figure 1 . Dental terminology used in describing arctostylopid molars, based on Palaeostylops iturus (after Szalay, 1969).
lected and remarkably complete specimen diagnoses are presented for previously de- from the late Tiffanian of Wyoming. This scribed taxa; we refer "Palaeostylops" specimen ( <>rms the basis for a revised di- macrodon to a new genus. Formal descrip- agnosis of the genus and species and for a tion of a hitherto unknown species of arc- comparison w Lth Vsian Axctostylopidae and tostylopid from the Yan-ma-tou Forma- South American Notoungulata. Revised tion, Hunan Province, China, is currently Arctostylopids • lia) ( ifelit ei al
in for progress; comparative purposes, we leontology and Paleoanthropolog briefly review some of its morphological jing, People's Republic oi China M( / features. Another new and genus species, Museum ol Comparath the from Member of the vard I Da-Tang Nung- I \\ I niversit) niversit) -I Mi. I, shan Formation, is de- Guang-dong, . being igan; YPM-Pl Yale Peabod) Museum scribed others elsewhere. by Princeton I niversit) ( oil.-, tion
Dental is terminology illustrated in I ACKNOWLEDGMENTS ure 1. We thank Dr. Donald Baird for loan of SYSTEMATIC PALEONTOLOGY a Princeton specimen under his care; Dr. Anne Roe Simpson, Mr. Will Downs, and Order Arctostylopida. new Mr. for us with trans- Meng Jin providing Distribution. Extinct; presentl) known lations of some Chinese important papers; only from the Paleocene, 1 ocene .md|H,v Mr. Sauer for translation of Kenneth a work sibly the Oligocene <.l \sia; Kit. Palea ene in Ms. and Drs. Russian; Ting Su-yin Ken- and possibly earl> Kmnic <>1 North Amer- neth Rose and Louis L. Jacobs for various ica. comments and and information; especially Diagnosis. Small mammals with uppei Dr. D. for Philip Gingerich providing us and lower dentitions forming an evenl) with casts of Asian and for canines rl\ or not Arctostylopidae graded series; | dif- facilitating access to areas currently under ferentiated and without diastemal investigation by the University of Michi- rating them from adjacent teeth Posterioi gan Field Program. Dr. Zheng Jia-Jian gra- upper premolarssomewli.it molarized ex- ciously permitted us access to an unde- cept in Asiostyl(J))s. I", at least, with a scribed specimen under study by him. Part metacone. Upper molars with well-devel- of this research was undertaken while Ci- oped centrocrista. becoming a salient, in felli was a Research Associate with the Bi- straight ectoloph advanced g< nera ological Survey, New York State Museum, parastyle usually prominent Pn and and the support of that organization is postprotocristae of upper molars strong, gratefully acknowledged. Exploration of conules lacking; upper lars primitive!) 1-2 Silver Coulee sites was supported by Na- triangular but M becoming quadrate in tional Geographic Society grant no. 2057 advanced forms 1>\ the addition <>t .. |x>s to Schaff. We extend our thanks to the terolingual cusp (pseudohypocone In- Churchill family of Powell, Wyoming, for terior lower premolars seriall) tricuspid, lowei mo- their assistance and generous hospitality with strong shearing surfaces; biselenodonl with over the years, in connection with field lars primitively para- work by Schaff. Field assistance was pro- cristid lost and various accessor) trigonid structures in advanced taxa vided by Dr. Farish A. Jenkins, Jr., and by acquired Messrs. William Amaral, Mark Goodwin, Lower molar hypoconid indistinct; ent< conid transverseK expanded and. m ad- Charles R. Schaff, Jr., and Rick Schaff. The into an antero- drawings were prepared by Mr. Laszlo L. vanced forms, developed Meszoly and Ms. Coral McCallister, and bucally oriented entolophid. the photographs were taken by Mr. Al- Schlosser. 1923. phonso H. Coleman. We thank also Mrs. Family Arctostylopidae 614 Lillian W. Maloney for her assistance in p. Arctostylopmae Zheng. 1979. preparation of the manuscript. (=Subfamily p. 391) ABBREVIATIONS Type Genus. Arctostylops Matth< 429 AMNH, of Vertebrate Pa- 1915. p. Department Genera Anatolostyi American Museum of Natural Other Included leontology, Z\ Pa- Zhai 1978, p 1"'' isiostylopi History; IVPP, Institute of Vertebrate Vol. No. 1 6 Bulletin Museum of Comparative Zoology, 152,
a sulcus two 1979, 388; Matthew and upper molars with separating p. Palaeostylops 12 and on M . molars differ Granger, 1925, p. 2; Sinostylops Tang lingual cusps Upper and further from those of in ^ an, 1976, p. 91; Bothriostylops Zheng Asiostylops having Ne- a and in lack- Huang, 1986, p. 121; Kazachostylops strongly developed ectoloph an a fold. Lower molars differ sov, 1987, p. 212; Gashatostylops, new; ing paracone and unnamed genus; and, with some doubt, from Asiostylops, Kazachostylops, in re- Allostylops Zheng, 1979, p. 391. Bothriostylops having paracristid Distribution. Paleocene, Eocene, and duced, prominent ectocingulid with shear Paleo- surface from cris- possibly the Oligocene of Asia; late descending protoconid, a labial cene and possibly early Eocene of North tid obliqua achieving pronounced America. attachment to the trigonid, and entolophid Metacones on Diagnosis. \s for the order. stronger and more oblique. 23 Zheng (1979) divided the Arctostylopi- P lacking or not so well-developed as in dae into two subfamilies: the Arctostylop- Palaeostylops and Gashatostylops; a lin- 4 inae, which included "typical" genera; and gual cingulum is present on P and is more in those 1 2 the Asiostylopinae, containing only Asio- salient than genera. M more less in occlusal stylops itself. While we are in agreement transverse, quadrate view; 2 that this last-named genus is the most M sulcus between protocone and pseu- primitive of known forms, we choose not dohypocone not so well-developed as in to recognize a higher taxon (subfamily) on Palaeostylops or Gashatostylops. Meta- that basis alone. Moreover, the description conid of lower molars not forming a dis- of species "intermediate" between Asio- tinct column within the talonid basin as in stylops spanios and advanced forms (see those two genera. Pre- and postprotocris- Zheng and Huang, 1986) largely occludes tae of upper molars high and variably en- the morphological hiatus distinguishing the closing a very transient fossette, as occa- proposed subfamilies, so that they are not sionally seen in Palaeostylops and even clearly defined grades. Nesov (1987) Gashatostylops, but not so strongly devel- distinguished two further arctostylopid oped as in Anatolostylops. subfamilies, Sinostylopinae and Kazacho- Arctostylops steini Matthew, 1915 stylopinae. On the basis of evidence now Figures 2, 8, 9 in hand, we do not believe that such di- Arctostylops steini Matthew, 1915, vision of the group is warranted. p. 429; Jepsen and Woodbume, 1969, 546; 96 5 Arctostylops Matthew, 1915, p. 429 p. Rose, 1981, p. AMNH left mandibular Type Species. Arctostylops steini Mat- Holotijpe. 16830, ramus with P3 to M 3 . thew, 1915, p. 429. Included Species. The type only. Referred Material. MCZ 20004, asso- Distribution. Late Tiffanian to late ciated mandible and anterior part of skull Clarkforkian, and possibly Wasatchian, with nearly complete upper and lower North America. dentitions; YPM-PU 20397, poorly pre- Diagnosis. Large arctostylopid differing From Palaeostylops and all other members 5 The of this as stei- of the family in having a salient lingual listing species "Palaeostylops ni" by Thenius (1985, to rib on the lower a caption Fig. 1, p. 151) de- canine, molarized P., serves mention, although a text explanation is lacking with a low, recurved talonid that ex- loph and we are thus uncertain as to whether this is a tends at or lingually the posterior margin of lapsus implied synonymy. The figure itself is dia- the tooth, and a prominent anterolabial grammatic but suggestive of Palaeostylops iturus rather than A. steini (for which cingulum (ectocingulid). Distinct, where well-preserved upper molars have not been known, from previously reported otherwise). primitive genera (Asiosty- Vs indicated in the diagnoses, the species are clearly lops, in Bothriostylops) having quadrate distinct; regardless, Arctostylops is the prior name. • ARCTOSTYLOPIDS M wim \i.i \ ( //, |/i ,., a l
cm
B
stein,. MCZ 20004 dentitions of Arctostylops of upper (A) and lower (B) Figure 2. Stereophotographs Vol. No. 1 8 Bulletin Museum of Comparative Zoology, 152,
with a distal served incomplete skull and mandible; UM is mitten-shaped, prominent heel. A on the labial side 650^4 left dentarv fragment with worn cingulum, lacking with P of the tooth, is well-defined on the lingual \1 and dentarv 3 ; right fragment 2 with of the crown. I is represented only I \1 66707, right dentary fragment portion a of the crown. As M 2 ; fragmentary part M, and partial M2 ; UM 68863, right by are with the teeth, P is single-root- I \1 69280, right P3 (UM specimens preceding not ed. The crests from the single cited from Rose, 1981, p. 96, and have descending are a small heel is A been studied bv us); and AMNH 88141, cusp sharp; present. weak labial is a trigonid of left M,. cingulum present; lingual to have been well-de- Horizons and Localities. The type was cingulum appears obscures most of collected in the "Lower Gray Bull beds, veloped, but breakage tooth. The canine is Clark Fork Basin, Wyoming" (Matthew, this side of the upper similar to the incisors and, unlike those of 1915, p. 429), of probable late Clarkfork- and which ian (Rose, 1981) or, possibly, Wasatchian Palaeostylops Gashatostylops, in size to is (Jepsen and Woodburne, 1969) age. Re- are subequal adjacent teeth, 1 than P and P . The root is f erred specimens have been collected from larger single the Willwood Formation at University of round to oval in cross-section and is not the ad- Michigan localities SC-19, 116, 188, and well-differentiated from those of 203 in the Plesiadapis cooki and Phenac- jacent teeth. The crown bears sharp mesial odus-Ectocion zones, Clarkforkian, Clarks and distal crests, is labiolingually corn- is inclined Fork Basin, Wyoming (Rose, 1981, p. 96); pressed, and somewhat poste- is the in the "lower variegated sequence" (Love, riorly; the labial surface convex and 1947) of an unnamed formation, Clark- lingual surface is slightly concave. The dis- forkian, near Togwotee Pass, Wyoming tal coronal crest bears a small, compressed faint heel. cin- (McKenna, 1980, p. 330); Silver Coulee cusp followed by a The beds, Polecat Bench Formation, Plesia- gulum is well-defined both lingually and dapis simonsi zone, Tiff anian (Jepsen and labially; the posterolabial part bears poorly Woodburne, 1969, p. 546), Wyoming. The defined cuspules. There are no diastemata specimen described below, MCZ 20004, adjacent to the canine,
1 was collected by Charles Schaff and Mark P is single-rooted and bears a single cusp. Goodwin in 1977, approximately 5 m from The tooth is labiolingually compressed, the Princeton Quarry site (Jepsen, 1930). with a faint lingual bulge, and closely re- The specimen was excavated from a gray- sembles the larger canine. The lingual cin- green siltstone 2.5 m below the Princeton gulum is prominent. Salient crests descend Quarrv level. The localitv (MCZ number from the anterior and posterior ends of the 1/77WYO; SE Va sec. 21, T56N, R100W) tooth to the single cusp. These evidently is about 24 km northwest of Powell, Park were important shearing structures, as a ( .'()., Wyoming, on the west side of Polecat well-defined wear surface is developed on Bench. the lingual side of the tooth, obscuring any Diagnosis. As for the genus. detail that may originally have been pres- ent. P 2 also , anteroposteriorly elongate, has _ _ two roots and is triangular in coronal view; .i . DESCRIPTION ,, , r . i . j. i. the serial homologue or the protruding hn- and 1 The upper lower dentitions form gual cingulum on P is here developed into w evenl) graded series, ithout diastemata a protocone. Labially, the ectoloph is sup- or marked structural gaps between teeth, ported by a single prominent cusp, the is ti.»t preserved in place in MCZ 20004. paracone, from which the loph descends two isolated However, upper incisors, one anteriorly and posteriorly. The anterior o\ which has been lost, were found in as- surface is moderately worn, with the facet iation with the upper dentition and angled sharply with respect to the plane this probabl) represenl tooth. The crown of occlusion. This facet is continuous with AR( • lLIA) < ifelti ei al
1. Table Dental measurements of .
AMNH MCZ PU 16830 20004 20397 65024* • ' .-.»
\. Lower dentition Vol. No. 1 10 Bulletin Museum of Comparative Zoology, 152,
seems to have been unfused. Small mental This ectoloph outer wall also has a postero- foramina are located below the right P 4 inferiorlv developed bulge, probably cor- I or and below left I, and 2 , respectively, responding to the base of a metacone is not in MCZ 20004. I 2 is metastvle. The inner face of the ectoloph I, preserved and with a on each molar bears a very well-developed procumbent spatulate, long is round in cross-section, wear surface, oriented, as on the premo- straight root that to the An traverses the lingual sur- lars superolingually at a steep angle oblique ridge sulcus face of the crown. I3 and the lower canine plane of jaw occlusion. A lingual I in not much resemble 2 , being separates the protocone from another cusp differing The canine is thus incisiform, distal and somewhat appressed to it; this procumbent. and not latter cusp we believe not to be a true structurally undifferentiated, sep- from teeth diastemata. cingulum hvpocone, for reasons developed arated adjacent by of bears a below. The crests linking protocone to The crown C, well-developed to column; to this, two cus- parastyle (preprotocrista) and protocone lingual posterior a are "pseudohvpocone" (Gregory, 1920; Simp- pules, separated by notch, present, of son, 1929) to the posterobuccal angle of P! is missing as a result postmortem in 20004 and is the tooth (postprotocrista) maintained a damage MCZ represented small remnant of one heel. The primitive triangular arrangement with re- only by a P is a tooth spect to the ectoloph and were evidently tooth was single-rooted. 2 larger strongly developed, because a small rem- and is double-rooted. It is buccolingually nant of a fossette enclosed by them persists compressed and bears three principal cusps 2 line with each the on the left \T and right M . These heavily that are nearly in other, 6 worn crests descend buccally from the middle of which is the tallest. The ante- protocone to their junction with the de- riormost two cusps are separated by a dis- scending wear surface of the ectoloph de- tinct notch; the third cusp lies on the pos- veloped on the labial face of the trigon terior slope of the middle cusp and has basin, so that the molars appear to be been reduced in this specimen by wear. notched when viewed anteroposteriorly. Behind this the central crest slopes infe- M\ somewhat damaged on both sides of riorly before rising to a sharp heel at the the specimen, is smaller and more trian- distal margin of the tooth. A slight bulge 2 gular in outline than M . An accessory crest, is present on the inferolabial side of the 12 apparently lacking on M but perhaps not tooth, but this is not distinctly formed into seen because of heavy wear on those teeth, a cingulum. P3 to M 3 are similar to those sweeps posterolabially from the midpoint of the holotype, AMNH 16830, as figured of the to the base of the like is trenchant postprotocrista by Matthew (1915). P3 , P 2 , metacone (or metastvle). A small accessory and is similar to that tooth except for being the is is crest, postmetaconule crista, present larger. P 4 submolariform. The paraconid 3 on the left M" (the right M is damaged), is lower than and directly mesial to the As with the more anterior molars, a distinct protoconid; the metaconid is lingualfy lingual cingulum is present and appears to placed. The protoconid and the metaconid be confluent around the base of the pro- are subequal in size. The cristid obliqua tocone attaches to the trigonid somewhat nearer The mandible is shallow and somewhat to the metaconid than to the protoconid I at the shaped symphysis. The symphysis and extends superiorly to a level near the apices of these cusps. The talonid is formed by a simple, crescentic crest that termi-
, nates at the f „ posterolingual angle of the I "i teeth we lollow in . .i • • uppei convention using A n i i i i / too h - A sma11 anterolabial (ec- cend," .l.-cend," "superior," "inferi f cingulum
I so in a is forth, sense relative to the way the) tocingulid) present. ith reference to orientation in life. The lower molars are morphologically Aw . , , al n
similar to each other. This series may differ slightly from that of the holotype, AMNH Palaeostylops Matthew and Granger,
in that is 1925 - 2 16830, M 2 more distinctly the P- of the three. The largest paraconid and its s Type Specit Palai ostylop Hurt crest are linking altogether lacking, and thew and Granger L92i the is near the anterolabial protoconid Included .Sp, margin of the tooth. From this cusp a crest Distribution Late Paleo descends anterolabially, forming a distinct Eocene (fide Li and Tin at that corner of the ridge (ectocingulid) Diagnosis. Dentall) advi -d arcto the is also tooth; protoconid slightly ex- stylopid generall) similar to in into an panded anterolingually developed but differing in the lack ol a heel The cristid has an lack of ridge. obliqua extremely paracolic Folds on the « tolophs 5 labial attachment to the at P '. lack trigonid; i.e., of a lingual rib on ( and in the the From this at which lesser size protoconid. point, differentiation ol the uppei it is nearly as high as the trigonid, it ex- nine from adjacent teeth. Molars \o\ tends as a distally sharp, straight loph, be- crowned than in Anatolostylops; uppei fore to fossette curving somewhat lingually end at molar more rapidK li»s| b) dental the hypoconulid. A hypoconid as such is wear. Differs from Gashatostylops ma lacking. The entoconid is transversely de- don, the most closel) similar form, in de- into a a sulcus veloped loph (entolophid), which ing strong separating tin lingual extends anterolabially to join the principal cusps of M\ three upper incisors and an in talonid loph (cristid obliqua and postcris- unconstricted snout, and lacking ( us- tid) at about its mid-point. Measurements pules on upper molar lingual cingula and are given in Table 1. relative enlargement of the upper and low- Available materials of Arctostylops are er second molars, inadequate to properly assess specific vari- Both Palaeostylops Hunts and Gasha- ability. All specimens in the hypodigm in- tostylops macrodon (herein separated from elude teeth also represented in the type of Palaeostylops) were described from the A. steini (AMNH 16830) and are suffi- type Gashato Formation Matthew and ciently similar to them in known morpho- Granger 1925; Matthew. Crangei and
Both but i logical features to cause us to consider all Simpson, 1929). species, n sam- specimens to belong to the same species, cially P. iturus, are know from large four of rather dental materials P3 and Mj are represented by speci- pies complete three Further remains ol both ies have beei mens each; M 2 and M 3 are known by spe< in the Naran Bulak I ormati teeth each. Of these, M 2 shows a marked recovered to Soviet and ei variability in length (Table 1). P3 seems by Polish-Mongolian c l l >d ). an vary considerably in proportion of length tions (Gradzinski et al., Szala) to the to width, but the significance of this cannot McKenna, 1971) Nemeg now be determined. about 250 km W'SW ol Gashato; in the Formation, near Nom As thus recognized, the species A. steini Nomogen Chinese workers is known from sediments of late Tiffanian Mongol, by Chow and 1978); and i (Plesiadapis simonsi zone) through Clark- 1977; Qi, Formation Nei forkian (Phenacodus-Ectocion zone) or van Ulan Individuals oi these species possibly Wasatchian age. This is a wide 1979). sent far the most abundant stratigraphic range for mammalian species by of the Gashato and of that age; however, several other species, curious fact considering tl including the abundant phenacodontids Arctostylops, Phenacodus primaevus, P. vortmani and close relative, North Amer Ectocion are believed to have temporaneous osbornianus, »i.iui^ • i'\iM .i» t<> tin- -i /d ioqi 99 9 log(M 1 area) area vs. M for iturus (dots) and Gashatostylops macrodon (squares). Figure 3. Plot of log-transformed M, 2 length Palaeostylops from P. itu- the genus Palaeostylops, its contained distinguished "P." macrodon rus its size and its species, and the relationships of those by larger proportionately lower molars. species to Arctostylops steini. Simpson larger second upper and (1936a) indicated that the species P. iturus Comparison of type and referred materials and "P." macrodon might be considered reveal several other consistent morpholog- in the as closely allied but distinct genera; Dash- ical differences, summarized diag- zeveg (1982) referred both to the North noses and description given below. The American genus Arctostylops. The super- most obvious difference in specimens as- ab- ficially close similarity of the Asian species signed to the two species, other than (except for size) and the fact that they solute size, is the aforementioned dispro- always co-occur suggested to us, at the out- portionately large upper and lower second set of this study, the possibility that a sin- molars of "P." macrodon (Table 2). Length gle, sexually dimorphic, species was rep- measurements of these teeth do not even resented. Detailed qualitative and overlap in range, which would be expected quantitative comparisons, presented be- if the difference were due to sexual di- low, together with previously unknown morphism. In most mammals (Gingerich, morphology provided by a new specimen, 1974), M, is the least variable lower molar; uphold Simpson's view. To explore the dif- in Palaeostylops iturus, the species for Ferences between these superficially sim- which samples are most nearly adequate, ilar we examined available is between and speeies, variability comparable M l of VMNH) samples arctostylopid denti- M 3 (Table 2). Because M, is represented tions from the Gashato and Nomogen lo- by larger samples in both species, this tooth calities and performed univariate and was chosen as a basis for comparison of multivariate analyses on tooth dimension second molar proportionate size. A plot width' length, data, using the Systat mi- (Fig. 3) of log-transformed M^ area (length software x iputer package. width) against M 2 length indicates that ( Matthew . Granger, and Simpson (1929) the difference in relative length of the sec- • al A -t— 2 - 1 - *- o <0 • • 1 -. -2 - -+- -1 factor 2 B -i- 1.0 0.5 .V o o.o ^ w, w • 1 u V_ 10 M L 0.5 - 3 2 M W -1.0 - 3 0.5 1.0 -1.0 0.5 0.0 factor 1 Unoi variables of and Cash Figure 4. PCA loading plots for six lower molar Palaeostylops = = B. for the six vanables on first two axes (dots P. iturus: squares G. macrodon), loadings Vol. 152, No. 1 11 in Museum of Comparative Zoology, AND GASHATOSTYLOPS (MM; p = 7 KND SI MMAHV STATISTICS OF PALAEOSTYLOPS TABI 1 MEASUREMl NTS P. ITURUS, G = G. macrodon). Arctos ds • (Mammalia i IfeUietal variables (M,L, M 2 L, M 3 L), which have Tabli 3 Homogeneity , M ,.. ( LOWER in i very high loadings along the first axis; a k result consistent with the univariate anal- ysis. Thus, on a statistical basis, the differ- ences between P. iturus and "P." macro- don are significant and are not attributable to size alone. In addition, dental and cra- nial features indicate greater structural dif- ferences between the species than has hith- erto been appreciated. We consider these differences to be worthy of generic sepa- ration. Palaeostylops iturus Matthew and Gran- ger, 1925, p. 2 Arctostylops iturus Dashzeveg and Rus- sell, 1988, p. 131 Figures 7, 8 Holotype. AMNH 20414, right mandibular ramus I with broken I, 2 and with 3 to M 3 complete. Referred Specimens. The type, and the following AMNH specimens, consisting of dentulous upper and lower jaws or portions thereof: 20415, 20417, 22143, 101967, 101968 (uppers); 20429, 21723, 101983, 101985; and AMNH 109522 A-J, casts of 10 uncatalogued lower jaw specimens in the IVPP. The AMNH specimens are from Gashato; the IVPP specimens were col- lected at Nomogen. Additional materials referable to the species are housed at the Polish Academy of Sciences, Warsaw, and at the Paleontological Institute, Moscow. These specimens are not listed here be- cause we were not able to compare them directly with the fossils listed above. Horizon and Localities. Late Paleo- cene; Gashato, Bayan Ulan, Naran Bulak, and Nomogen formations, Nei Mongol. Diagnosis. As for the genus. Gashatostylops, new genus Type Species. Palaeostylops macrodon Matthew, Granger, and Simpson, 1929, p. 11. Etymology. Gashato-, for the original locality of the type species; -sty lops (Gr.), arc- pillarlike, a commonly-used suffix for Vol. No. 1 16 Bulletin Museum of Comparative Zoology, 152, 23 referred and M ) and other ma- leeted by McKenna; AMNH 21742, two complete terials from the leave no doubt isoalted calcanea; AMNH 21726, isolated type locality as to reference of this specimen to Gash- astragalus; and the following AMNH right macrodon. specimens consisting of dentulous upper atostylops As in AMNH 109521, the and lower jaw fragments: 22142, 101967, preserved 101979, 101977, 101963 (maxillary); snout is short and constricted, flaring 101980, 101987, 101984, 101982, 101981, broadly at the root of the zygomatic arch, the tooth row as- 20416, 21740, 21741, 21723, and 21716 so that in palatal aspect a double curvature. The form of the (mandibular). The AMNH specimens were sumes that seen collected at Gashato, the IVPP specimen dental arcade thus contrasts with in iturus, which curves is from Bayan Ulan. As with Palaeostylops Palaeostylops front to back. In iturus, additional specimens (not listed gently from palatal view, of the maxilla forms here) are in the collections of the Polish the posterior margin Academy of Sciences, Warsaw, and the a curved process that almost completely a small foramen to the Paleontological Institute, Moscow. encloses lingual 2 3 Horizon and Localities. Late Paleo- junction of M and M . This foramen in all cene; Gashato, Bayan Ulan, Naran Bulak, probability housed the minor palatine and Nomogen formations, Nei Mongol. branch of the maxillary artery, as it does in certain Diagnosis. As for the genus. in many living mammals and Although a diagnosis appeared in the notoungulates, such as Notopithecus (see infraorbital original publication (Matthew, Granger, Simpson, 1967, fig. 23). The 3 and Simpson, 1929), the morphology of foramen, located above the junction of P 4 this species has never been described. Im- and P about halfway between the base of portant details are provided by the spec- those teeth and the anteroinferior margin imen represented by AMNH 109521 from of the orbit, is small. The root of the zy- 2 Bayan Ulan, which preserves the left side gomatic arch arises at the base of M . It is of the rostrum, including the orbit and massive and dorsoventrally expanded, flar- left the roots to an inferior at the zygomatic root, C-M 3 , and ing prominence squa- of the incisors on both sides. In addition, mosal suture, suggesting relatively pow- further preparation of the original IVPP erful development of the masseteric specimen by one of us (Schaff) revealed musculature. The nasals are long and nar- the presence of part of the left mandible, row, flaring posteriorly, with the median including M 3 and the condyle, and a right processes of the f rontals deeply projecting astragalus lodged within the broken cra- between them. Small, isolated foramina are nial cavity. These are presumed to be as- present in each nasal. The premaxillary- sociated with the skull fragment itself. All maxillary suture is located in the most an- teeth are in full eruption but wear is light, terior quarter of the snout, just posterior 2 indicating that the animal was a young to I . The maxilla is extensive, incorporat- adult. The specimen is more or less split ing three-quarters of the snout region, and sagittally, except that both premaxillae are extends to the base of the orbit. The max- preserved. The palate, nasal, and frontal illary-jugal suture is oblique and runs above are so that 3 regions crushed, the corre- the base of M . sponding bones are somewhat fragmented. Although upper incisors are not pre- Comparison with the dentition preserved served in the specimen represented by in original paratypesof the species (AMNH AMNH 109521, roots preserved in the pre- 3 2 1 221 I t Ml P -M ; 22142, right broken M maxillae clearly indicate that only two were itostylops macrodon and associated partial left mandible (uncataloqued IVPP specimen; cast. A iorsal (A), ventral (B), and left lateral (C) views. 17 Vol. 152, No. 1 is Bulletin Museum of Comparative Zoology, contrast to the the lingual sulcus is strong, so present on each side, in protocone the tooth is bilobed. A ac- three known for Palaeostylops iturus and that prominent both in- lies in a median position at Arctostylops steini. The roots of cessory cusp the base of the to the cisors are subround and approximately ectoloph, posterior 3 canine M is similar to those equal in size. The base of the upper parastyle. generally of the of Arctostylops steini and Palaeostylops is larger than the roots incisors, ap- 1 Whether iturus, that the lingual cingulum is proximating the base of P in size. except bears an eminence or not this reflects a notable difference in complete and directly to the size between canine and lateral incisor lingual protocone. crowns cannot be determined; however, A nearly complete lower dentition is in Pa- in AMNH 21741 from Ga- root development is comparable represented a left with C, and laeostylops Hunts, whose anterior teeth shato, dentary I^, P^, bears a nonetheless form an evenly graded series M^; the last molar moderately talonid. The horizontal ramus is (cf. Matthew, Granger, and Simpson, 1929, damaged 1 Its shallow, with a horizontal p. 12). P is single-rooted. crown, gen- nearly sym- that to have been unfused. erally similar to those of corresponding physis appears are located below and teeth in Palaeostylops and Arctostylops, is Small foramina P[ a P . The three incisors are similar to those buccolingually compressed and bears 4 of iturus. \ is with sharp mesiodistal crest, which ascends me- Palaeostylops x spatulate a and a median vertical dially to the apex of the single cusp. A rounded point less than faint lingual cingulum, not developed into ridge, the crown being compressed 2 steini. I is a heel as in Palaeostylops, is present. P is in Arctostylops 2 larger and more 1 I with the an- double-rooted. It is larger than P and laterally compressed than ls of structurally similar to it, except that a pro- terior part the crown more expanded tocone, smaller than that of Arctostylops and the median vertical ridge better de- is in and equal to that of Palaeostylops, is de- veloped. I3 similar size and morphol- crests de- to I for the of an veloped lingually. Well-defined ogy 2 , except presence scend from this cusp to the anterior and incipient posterior lobe on the median posterior margins of the tooth. The lingual ridge. The canine is subequal in size to I3 surface of the coronal crest, or ectoloph, and somewhat larger than P,; no diaste- is steep and bears well-marked wear facets, mata separate it from those teeth. The 3-4 as seen in succeeding teeth. P are suc- crown of the canine is tricuspid and com- cessively larger and more molariform, with pressed; lingual crests are associated with more fully developed protocones. As in Pa- each cusp. In these respects it generally laeostylops iturus but in contrast to Arc- resembles Palaeostylops iturus rather than 4 tostylops steini, P is noticeably smaller Arctostylops steini. 1 than M . The molars bear The P is sharp, straight single-rooted { morphologically ectolophs with well-developed parastyles. similar to the canine, although the three On M', the lingual sulcus posterior to the coronal cusps are somewhat more distinct. is faint, unlike the condition is protocone seen P 2 double-rooted and significantly larger in Palaeostylops. Lingual cingula are well- than P,, with the tricuspid pattern clearly on all vari- defined. P is similar to but developed upper molars; cusps, 3 larger than P 2 , in 12 able development, are present on M . w ith the protoconid being the tallest cusp. M in tin is specimen represented by AMNH P 4 submolariform, with a serially tricus- L09521 bears two such cusps, one lingual pid trigonid and a small, crested heel. The to the and protocone another, larger, pos- paraconid and metaconid are equal in size, terolingual to thai cusp and in a hypoconal and the protoconid is the tallest cusp. The position \; is much larger than preceding cristid obliqua attaches somewhat labial to or teeth and bears succeeding three cusps the metaconid. There is no ectocingulid on the lingua] cingulum. Posterior to the present on any of the lower premolars. ' • mmalia ( tfeUietal cm Figure 6. Stereophotographs of left mandible of Gashatostylops macrodon. AMNH 21 741 The most notable feature of the lower reasonably inferred, resembles to the point molars is the extremely salient, blade-like, of identity the isolated specimens from labially-placed cristid obliqua. The molars Gashato believed on the basis <>l m/<- a\h\ are morphologically similar to each other, relative abundance to belong to Gasha- with to be M 2 appearing disproportionate- tostylops macrodon. In known respects than ly larger preceding and succeeding the ankle of Palaeostylops iturus is similai teeth. The entoconid is expanded into an to that of G. macrodon. and it is therefore obliquely-oriented entolophid that con- not described separately. Descriptive ter- tacts the cristid obliqua in about the mid- minology follows that of Cifelli h< v dle of the talonid. On all three molars, the Relative terms in the description arc based ectocingulid is developed as a distinct an- on comparison with ankle bones referred terolingual ridge at the junction of pro- to ProtUngulatUm and similar ta\a. u lin li toconid and cristid obliqua. The protoco- are assumed to approximate a eutherian nid is the tallest in and 1974 cusp except on M 2 , morphotype (Szala\ Decker, which the hypoconulid is larger. Szalay, 1977). No directly associated, articulated post- As represented b> VMNH 21726, a right cranial elements are yet known for any of astragalus from Gashato. the astragalai the Arctostylopidae, but proximal ankle body is mediolateralh compressed, with bones may now be referred to Palaeosty- nearly vertical sides: there is little or no lops iturus and Gashatostylops macrodon development of a fibular shell on the lat- with little doubt. These species are by a eral side. The body is more or less i ylin- considerable margin the most abundant drical, with the tibial trochlea marked b) raised taxa known from Gashato. Isolated astrag- a median groove and well-defined ali and calcanea, of appropriate size for P. borders. An astragalar foramen is appar- is on the tibial iturus and G. macrodon, occur there in ently lacking; a pit present the same relative abundances as dental re- trochlea of AMNH 21726 but appears to caused !>\ corrosion mains of these species. Furthermore, an have been diageneti< on the other side ol the astragalus was found lodged within the of the fossil (as such is common on cranial cavity of a specimen from Bayan same specimen; pitting Gashato The neck is ol mod- Ulan referred to G. macrodon (see below). fossils from so erate but is notabl) constricted, This astragalus, for which association is length No. 1 20 Bulletin Museum of Comparative Zoology, Vol. 152, Primitive dif- that the head is clearly demarcated. The Diagnosis. arctostylopids in head itself is subround with, however, the fering from Asiostylops having higher- a metaconid on P . navicular facet developed as a flattened crowned molars and ?3 band that does not extend onto its sides. Distinct from advanced genera such as a The navicular facet extends far superiorly, Arctostylops in retaining paracristid on onto the dorsal side of the bone, and is the lower molars. Similar to Bothriostylops to developed so that movement between as- in having the cristid obliqua attaching the of lower molars in a tragalus and navicular would have been trigonid lingual but differs from that in subparallel rather than highly oblique to position, genus molars and a more that between astragalus and tibia. There having higher-crowned P . is no observable separate facet for the cu- slender, elongate 3 boid. The tarsus might thus tentatively be regarded as "serial" (see discussion by Os- Sinostylops promissus Tang and Yan, de- born, 1889), although this cannot be 1976, p. 92 ascertained until a well-pre- finitively Holotype. IVPP V4263, right mandibular ramus with foot is discovered. The served, articulated eight partial or complete teeth. sustentacular facet is unremarkable, ex- cept that it is somewhat larger and better Hypodigm. The type only. developed distally than it is in Protun- Horizon and Locality. IVPP locality gulatum. The ectal facet is very steeply 71017, Dou-mu Formation, Anhui Prov- inclined, and the interarticular sulcus sep- ince, People's Republic of China; late Pa- arating the two facets is deep. leocene. The calcaneus, as represented by AMNH Diagnosis. As for the genus. 21742 (complete left calcaneus and right With the removal of referred species calcaneus lacking the tuber, almost cer- "Sinostylops" progressus to Bothriosty- tainly not from the same individual), is lops, the concept and affinities of Sino- notable in having a short neck (that part stylops become problematic. The identi- anterior to the astragalocalcaneal facets) ties of the eight teeth in the holotype and relative to the tuber. The ectal prominence only specimen of Sinostylops promissus is dominated by a very strongly developed are open to doubt. Because the third from fibular facet, which forms a broad, antero- the last tooth is remarkably low-crowned posteriorly oriented, semicylindrical sur- and long, unlike either preceding or suc- face. Medial to this lies the ectal facet, ceeding teeth, we believe it to be decid- which is strongly inclined with respect to uous. The penultimate tooth, although the fibular facet. The sustentaculum is un- much smaller than the ultimate, is mor- usual in lying at or near the distal end of phologically similar to it; both are badly the bone; a prominent "beak" is developed damaged but apparently were bicrescen- on the superior distomedial corner of the tic, which is not the case for the more bone. The cuboid facet is developed at a anterior teeth. We therefore believe the moderate with to the teeth in this to be I angle respect long specimen 3-P3 , dP4 , and asix of the calcaneus. to Comparisons other Mi.2) although other interpretations are are taxa deferred until the discussion. possible. Available materials of Sinosty- lops promissus and Bothriostylops pro- gressus suggest further differences be- Sinostylops Tang and Yan, 1976, p. 91 tween the species beyond those listed in Type Species. Sinostylops promissus the diagnoses, but because of the uncertain ( Tang and Yan, l )76, p. 92. identities of the teeth in IVPP 4263 and Included The Species. type only. because of postmortem damage to that Distribution. Kate Paleocene (fide Li specimen, we are unable to evaluate the and Ting, 1983), Asia. significance of these differences. Arctostylopids (1 ,,, .< ifeUtetal fibular facet B cuboid facet fibular facet sustentaculum sustentaculum D ectal facet tibial trochlea sustentacular facet navicular facet mm Figure 7. Right astragalus (unnumbered IVPP specimen associated with rostrum of skull, a cast of which b numbered AMNH 109521) and left calcaneus (AMNH 21742) of Gashatostylops macrodon. A. B. C: calcaneus in distal, dorsal, and planter views. respectively; D, E, F: astragalus in plantar, dorsal, and distal views, respectively. Vol. 152, No. 1 .->.-> Bulletin Museum of Comparative Zoology, 1986, with the paracristid less truncated, Bothriostylops Zheng and Huang, gually, than in B. progressus. p. 121 notios and Type Species. Bothriostylops Bothriostylops progressus (Tang Yan, Zheng and Huang, 1986, p. 122. 1976) and Both- Referred Species. The type, Figures 8, 9 and 1976, riostylops progressus (Tang Yan, Sinostylops progressus Tang and Yan, p. 92). 1976, p. 92 Asia. Distribution. Late Paleocene, Bothriostylops progressus Zheng and with Diagnosis. Primitive arctostylopids Huang, 1986, p. 127 teeth, differing from Asio- brachydont mandib- in Holotype. IVPP V4264.1, fragment of right stylops, which they generally resemble, ular ramus with M 2 . P and from having a crescentic 4 talonid; with an The and all known genera in having M 3 Beferred Specimens. type, mandible elongate talonid, the hypoconulid forming IVPP 4264.2, right fragment mandible a distinct lobe. with worn M 2_3 ; 4264.3, right were list- mandible A number of other characters fragment with P3_ 4 ; 4264.4, right of the mandible ed in the diagnosis and description fragment with P34 ; 4264.5, right Of these, with and with broken M 3 ; genus (Zheng and Huang, 1986). fragment M 2 bro- the presence of a deep median labial groove 4264.6, right maxillary fragment with 2-3 and convex labial wall on lower molars ken M 1 and with M well-worn. were cited as important similarities to Horizon and Locality. IVPP locality Asiostylops. We see no distinction of arc- 71071, Shuang-ta-si Group, Anhui Prov- tostylopid genera on this basis but, lacking ince, People's Republic of China; late Pa- access to the original specimens (especially leocene (Li and Ting, 1983) or early Eocene the type of Bothriostylops notios), we de- (Zheng and Huang, 1986). fer to Zheng and Huang (1986). Nonethe- Diagnosis. Cheek-teeth higher-crowned less, we observe that a deep median ex- than in B. notios. Entolophid of lower mo- ternal groove is present on lower molars lars complete and more fully developed of Sinostylops promissus. Even on the ba- than in that species; trigonid of lower mo- sis of the single, enigmatic specimen avail- lars more compressed, with paracristid able, it is clear that this latter species is more truncated, than in B. notios. rather divergent and not obviously con- Zhai, 1978, 109 generic with other known taxa. Anatolostylops p. Anatostylops, Schaff, 1985, p. 593 dubius Bothriostylops notios Zheng and Huang, Type Species. Anatolostylops 1978, 109. 1986, p. 122 Zhai, p. Included Species. The type only. IVPP V7642, of left mandible with Holotype. portion Distribution. Late early Eocene or early P4-H, middle Eocene (fide Li and Ting, 1983) Referred Specimens. The type only. or, perhaps, Oligocene (fide Zhai, personal Horizon and Locality. Wang-wu Mem- communication); Asia. ber, Chi-jiang Formation; late Paleocene. Diagnosis. Differs from all other genera, North of Zhulin Hill, Dayu County, Jiang- excepting an unnamed form, in having \i Province, People's Republic of China higher-crowned cheek-teeth; ectoloph of (cited from Zheng and Huang, 1986). upper molars elongate, smooth and fea- Diagnosis. Cheek-teeth lower-crowned tureless, with a large parastyle but no para- than in R. progressus. Entolophid of lower stylar fold. Pre- and postprotocristae sa- so molars not well-developed as in that lient, enclosing a fossette that persists • at least sp< nd, on \1 , incomplete; through more advanced wear than in other 2 trigonid ol lower molars more open lin- forms. Sulcus on lingual side of M crown ARCTosTYLOPins • (Mammalia) ( :„//, ei at. not so broad as in Arctostylops or Palaeo- Kazachostylops Nesov, 1987. p. 212 stylops. Differs from a closely similar un- Type Species. named genus and species in a lin- Kazachostylops occid* r». having talis division of 2 Nesov, 1987, p 212 gual M , and in lacking the Included Spr.-i, . The great anteroposterior expansion of the ec- type only Distribution. Late toloph and the strong of the Paleocene western development Asia. postcingulum seen on upper molars of that Diagnosis (from Nesov genus. 1987, p. 21 Small with arctostylopids long, tall pai cristid on lower molars; Anatolostylops dubius Zhai, 1978, p. 109 premetacristid and postmetacristid reduced Figure 8 to absent I lophidof VI, ,long,nearl) transverse and Holotype. IVPP V4357, fragment of left maxilla with with the joined talonid loph. ( Irestsol Hi- lar teeth form practically uninterrupted Hypodigm. The type only. cutting edges. Horizon and access to the Locality. Shi-san-jian-fang Lacking two, relatively good of the Formation, Turpan Basin, Xin-jiang Prov- specimens type and only species ol ince, People's Republic of China; Eocene Kazachostylops, we defer to Neso> s briel of the or Oligocene (see above). diagnosis genus, and mint it from the detailed Diagnosis. As for the genus. comparisons and discussion below. presented From the figures, Ka- Asiostylops Zheng, 1979, p. 388 zachostylops appears to be rather similai to and, Type Species. Asiostylops spanios Bothriostylops perhaps, Sinosty- lops, particularly in the well-de- Zheng, 1979, p. 388 elongate, the veloped M 3 , strong, crescentic Included Species. The type only. para- cristid, and in the lingual attachment ol Distribution. Late Paleocene (fide Li cristid to at the meta- and Ting, 1983), Asia. obliqua trigonid (i.e., conid). Diagnosis. Distinct from all other arc- tostylopid genera in the more transverse P2 3 with a lesser occidentalis Nesov. 1987. , development of the pro- Kazachostylops 212 toconal region; upper molars lacking a pos- p. Holot Hi terolingual cusp or other secondary coro- ype. Specimen number 12455 indicated b) Nesov as in the Ts.N.1 G.F nal complications; metaconid lacking on (1987) being deposited Museum, Kazakhstan, Dzhilu.i I ssli , onsisting ol P3 . Lower molars in the primitive retaining \1 a right dentary with C, F, ,. and paracristid, as in Bothriostylops and Si- but not other cristid ob- nostylops genera; Hypodigm. The type, and at least one to in a median liqua attaching trigonid more dentulous jaw fragment figured l>\ position. Entolophid feebly developed and Nesov (1987), number 12 L2455, consist- oriented. 1 transversely ing of a right maxilla with P to M Horizon and Locality. Marginal marine 1979, 388 Asiostylops spanios Zheng, p. deposits of the Pretashkenl Svita, late Pa- 9 Figures 8, leocene; site TDA-2. Kazakhstan, Dzhilga Nesox. 212 Holotype. IVPP V5042, cranium and associated left USSR (fide L987, p. mandible. Diagnosis. As for the genus Hypodigm. The type only. Arctostylopidae?, incertae sedis Horizon and Locality. IVPP locality Allostylops Zheng, 1979, p. 391 73039, Lan-ni-kong Member, Chi-jiang Formation, Jiang-xi Province, People's Re- Type Species Allostylops periconatus public of China. Zheng, 1979, p. 391 Diagnosis. As for the genus. Included Species.The type onl) Zoology, Vol. 152, No. 1 2 I Bulletin Museum of Comparative "^TmJllfflmjmrr'Tnr. standardized to size Figure 8. Comparative series of arctostylopid (A-H) and primitive notoungulate (I) upper dentitions. Teeth and reversed where necessary. A, Arctostylops steini (MCZ 20004); B, Palaeostylops iturus (AMNH 22143); C, Gashatostylops 1 IVPP macrodon (cast, AMNH 09521 ); D, Anatolostylops dibius (IVPP V4357); E, undescribed genus and species (unnumbered specimen); F, Bothhostylops progressus (IVPP V4264.6); G, Allostylops periconatus (IVPP V5043); H, Asiostylops spanios (IVPP V5042); I. Peripantostylops minutus (AMNH 28494). 2 Distribution. Late Paleocene {fide Li ing a hypocone on M . Differs from ad- and Ting, 1983), Asia. vanced arctostylopids (Palaeostylops, Diagnosis. Generally primitive ?arcto- Arctostylops, Gashatostylops, Anatolo- stylopids similar to Asiostylops spanios in stylops) in having lower-crowned cheek- the low -crowned the small size a smaller P3 smaller cheek-teeth, teeth, , upper molar <>f P\ and the presence of a paracone fold parastyles, and a broadly expanded pos- on the of at 1 ectoloph least some upper mo- terior cingulum on M . Allostylops is dis- lars, hut differing from that species in hav- tinct from all forms in the family save series of ;omparative arctostylopid (A-E) and primitive notoungulate (F) lower dentitions. Teeth standardized to size j reversed where necessary. A. Arctostylops steini (MCZ 20004); B, Palaeostylops iturus (AMNH 20414); C, Gashatostylops rodon(AMNH 21741); D, IVPP Bothhostylops progressus (P^, V4264.4; M 2 , IVPP V4264.1; M 3 outline, IVPP V4264.2)- E. Asiostylops spanios (IVPP V5042); F, Peripantostylops minutus (AMNH 28494). 25 Vol. 152, No. 1 Bulletin Museum of Comparative Zoology, IVPP of P3-M 3 of undescribed and species of Arctostylopidae (unnumbered specimen). Figure 1 0. Stereophotographs genus in a prominent genera in the great anteroposterior expan- Gashatostylops having 3 3 on sion of the crest on P -M and in pericone anterolingual to the protocone ectoloph molars. the strong development of the postcingu- upper ' 1 2 of the lum on M . Indeed, the ectolophs 391 cheek-teeth are so devel- Allostylops periconatus Zheng, 1979, p. upper strongly oped that the rest of each tooth appears IVPP V5043, rostral por- Holotype. badly preserved comparison to have been constructed tion of cranium. by as an afterthought. A lingual division of The 1-2 in all other Asio- Hypodigm. type only. M , seen genera except Horizon and Locality. IVPP locality stylops, is lacking. The second upper mo- For- 73041, Wang-wu Member, Chi-Jiang lar is notably larger than the first. Re- mation, Jiang-xi Province, People's The single specimen representing this late Paleocene. public of China; species was plotted into a measured section Diagnosis. As for the genus. (Unit 21 of Zhu-chen, 1986) of beds re- ported to be of Cretaceous age. The basis and Arctostylopidae, genus species for this surprising age determination is not indet. A entirely clear, but it seems to involve fossil In addition to the foregoing previously remains believed to be dinosaur eggs (list- described species, an unnamed arctosty- ed as Elongatoolithus and other taxa) lopid is represented by an upper dentition which, apparently, bracket the arctosty- collected from the Yan-ma-tou Formation, lopid specimen. Other fossils from this sec- Hunan Province, People's Republic of tion are listed merely as "animal bones" ( !hina. While full description of this species or "animal teeth," and are therefore of is in progress, we briefly note some of its little help in age determination. However, morphological features here in order to a mammalian axis vertebra is larger than facilitate comparison among other mem- that of any Chinese Paleocene mammal bers of the family and to aid in assessing and would be totally out of place in the their relationships. Cretaceous, as would a large anterior tooth The taxou in question is a small, dentally of some ungulate-sized mammal. We be- dvanced arctostylopid similar to Anato- lieve on this basis that the locality is much lostylops in having high-crowned cheek- younger than Cretaceous, perhaps even tt i tli and a smooth ectoloph on the upper Eocene in age, whether dinosaurs were ii ilars, but it differs from this and all other present or not. Indeed, the advanced mor- ARCTOSH 4AMMAL1 ,,/ :: of the phology small, distinctive arctosty- (strong hiah ,,-„.. lopid from this site is suggestive of the cone nectingpara- Eocene or 1^ possibly Oligofene Anatolo- ^^X^S^S^T^ stylops dubius. duC€Hut'/* stylar prominences; blade-like I' Arctostylopidae, genus and species serially tricuspidate . with notches a indet. B aratmg the l\ with a cusps; trigonid i > (n Another arctostylopid, which we have 11 ' t ''IT* ' Kfffi^ X^^ ' not seen, occurs in the late Paleocene Da- l lower m( r s i ! tang Member of the Nung-shan Forma- i .. . • ZZonipciidtuiiiu in a, , dian )os 7^ media i , tvt i /-. i 'T""i S turn and ' turn, Nan-xiong basin Peo- Guang-dong, onid of cristid consisting obli „ land , ,' pies Republic of China. The animal is cristid united into a ,„„„ s' a e , *"*" ^ and L?aLiandnd TinWl983 fl3) ^ *»** '"' ^ ling (1983, p. 13). tinct. The^^ entoconid of lower molars is iso- lated from the postcristid and developed int ° a t tr COMPARATIVE n en ' 0p " ^ extends DENTAL MORPHOLOGY f^? f. f S! ^ !' OF THE ARCTOSTYLOPIDAE anterolabially to the tal dcrescenl Man) °r these features are shared b\ pKMiinal.K Review of dental variations the among unrelated groups of mammals, but the se- and assessment of the rel- Arctostylopidae rially tricuspid anterior lower premolars ative of various primitiveness character and the transversely developed entoconid states is based on comparison with an un- (entolophid) of the lower molars are ratlin gulate morphotype as represented by Prot- distinctive characters. and various ungulatum comparable oxy- Arctostylops, Palaeostylops, and Ga- claenine are Arctocyonidae (Cifelli, 1983a). shatostylops distinctively more spe Some of the features that are represented cialized. Advanced characters of th< in available materials of the three known species genera with respect to Asiostylops are summarized in Table 4. The most include higher-crowned posterior premo- primitive arctostylopid for which good lars and molars; an expanded protocone on 3 materials is P are available unquestionably ; upper molars with a high, flat ectoloph Asiostylops spanios Zheng, 1979. Zheng wall including parastylar and metastylar (1979) referred Asiostylops to the Notoun- folds only (Asiostylops has a distinct para- 2 based on the cone at is bifid gulata biselenodont lower fold); M , least, linguall) molars, with shortened trigonid and loph- with high pre- and postprotoeristae that odont entoconid, and on the upper molar enclose a very transient trigon fossette l>nt ectoloph, with parastyle developed. He which are rapidly reduced l>\ hcaw weai considered Asiostylops to be primitive The lower molars of these three genera are within the order because the cheek-teeth distinctive in a number of respects, such are low-crowned, the premolars are not as: 1) the presence of a salient, pillar-like with a wear surface dea end- molarized (in particular, P4 lacks an ento- ectocingulid conid), the lower molars have a pro- ing along its face; 2) the reduction ol nounced paraconid, and the upper molars trigonid by loss of the paracristid 3) the lack the secondary coronal complications presence of a high, shearing talonid seen in Henricosbornia and more ad- cent (cristid obliqua), which joins the tri- ol the vanced South American notoungulates. gonid labial to position prota onid; en- Compared to an ungulate morphotype and 4) the strongK developed, oblique have an ante- represented by Protungalatum, Asiosty- tolophid. All three genera unlike \sio fops spanios has a greater development of riorly placed P, paraconid, the protocone on P3 \ with a metacone on stylops. The polarities ol some feati Arctos those teeth; upper molars with an ectoloph P< are uncertain. No. 1 Bulletin Museum of Comparative Zoology, Vol. 152, an in Palaeostylops, Gashatostylops, and most tostylopids in having elongate M 3 forms a other Arctostylopidae in the presence of a which the hypoconulid separate differs from lobe. also to be true of lower crescentic P., talonid loph, and (This appears Gashatostylops macrodon (but not Pa- molars belonging to Kazachostylops occi- have not examined laeostylops iturus and several other species) dentalis, which we firsthand. are unable to consider the in the lingual placement of its P4 meta- We conid. Arctostylops steini is probably au- species further here, but note that the tapomorphous in having a stronger ecto- above-mentioned feature and several oth- P a rib on er lower molar characters a close cingulid on 34 , stronger lingual suggest 1 heel on I and a to We as- C,, a prominent , slightly relationship Bothriostylops spp.) 2 . for the of that larger protocone on P Palaeostylops and sume, purpose comparison, Gashatostylops appear to be derived with these two species form an exclusive unit respect to Arctostylops in the lesser dif- within the family. Thus conceived, Both- 1 lack of in several interme- ferentiation of C , the paracone riostylops is, respects, of P34 and the diate beteen on the one hand folds on the ectolophs , pres- Asiostylops ence of a shearing notch on P 4 . Palaeo- and advanced arctostylopids (Arctosty- stylops and Gashatostylops differ from lops, Gashatostylops, Palaeostylops) on the Arctostylops also in the more quadrate, other. As in Asiostylops, Palaeostylops, and 12 in less transverse nature of M and the Arctostylops, the metaconid on P4 is lin- fact that the sulcus between the two in- gually placed (we are uncertain of the con- ternal cusps is better developed, at least on dition in B. notios). The talonid crest of 1 . in M The P 4 cusps Gashatostylops ma- that tooth is curved in Bothriostylops spp, crodon are more or less anteroposteriorly although not so strongly as in Arctostylops. aligned, as with the more anterior pre- The lower molar trigonids of B. progressus molars of all genera; the talonid crest is a are anteroposteriorly compressed, as in the straight, bladelike structure. G. macrodon derived genera, but unlike those forms, is also distinctive in that the upper and part of the paracristid remains, as in Asio- lower second molars are greatly enlarged, stylops. In B. notios, the trigonid retains in the variable development of one or more a more open arrangement, with the para- cuspules on the lingual cingulum, and in cristid little reduced. The cristid obliqua the reduction or absence of a lingual suclus attaches to the trigonid at a lingual posi- on 1 M (further distinctions are given in the tion, near the apex of the metaconid, un- diagnoses provided above). Thus, Arcto- like either Asiostylops on the one hand or stylops, Gashatostylops, and Palaeosty- Palaeostylops/Arctostylops on the other. share lops presumed synapomorphies with The ectocingulid is feebly developed and respect to Asiostylops. Within this clade not expanded into an occlusal structure. of advanced genera, there is some evi- The entolophid varies from well-devel- dence to that and as in suggest Gashatostylops oped (B. progressus) , the advanced shared a Palaeostylops more recent com- forms, to weak and incomplete (B. notios). mon ancestor than either did with Arcto- A partial, very worn, upper molar series Because of stylops. uncertainty in mor- is available for Bothriostylops progressus, of several the phocline features, possibility but it adds little to knowledge of the species. of lineal between of relationships any the The ectoloph appears to have been high; included cannot be species evaluated. as far as can be determined, paracone and The remaining species of Arctostylopi- metastylar folds are lacking although a dae are known from less 1 complete mate- parastylar fold is well-developed. M has ils and there is. accordingly, some un- a sulcus separating two lingual cusps; this certainty as to various character states. not to 2 appears have been true of M , which rather the two is Although primitive, species triangular in outline (as with Asiosty- ol Bothriostylops are unique among arc- lops), but excessive wear has obscured de- • Arctostylopids (Mammalia) Cifelli ei al. fi C bC o c u cd O C wi O C "TV " — bC " — c - a cb O b "_ yo "*-"3 —~ _ H CD CO CO CO ft B**-j 5 . £. _2 _2 ._£_:;- 0-- C»- fL. Cl. fL. cl. o-. O-. CT-. o_. O, > o3 jj > o3 ~3 Sjb > 2i = bC CD O 3 " bO — bl 3 _b C « -S O C bC < __ — bC C 3 u 7 £_ S3 <> cS .5 c cd _c a, H ''I D . co cd SP cd s .5 3= o ID .5 -C ai « 'S aj i fc co 03 03 Cb_3 03 o3 Rii i t/> — M-. tii- J H oC/3 C — H C 03 -S U bC -J U li 3 i- *; ^ « Ml rt ° 2 c « .2 o C ai -C « « ™ '- d c M bC o. ™e J3 _D _D JD < nj -^-Q — VI « ai ax « «J a<« ox ac H O z o 5 _ bJD _ o. M 2 ,_ c c £ c 0) 3 ^c» ^ W5 ¥ ^ .2 S c« -2 4; )) liX « 41 bo I C bD O • o-. f>-. o-' ; o-- o. < a- S o . a-. a^= ? cx a- o3 H Z w Q c O .2 C ~ - C j CQ CLifL. ^l. < C 3^0 u Sj 6C'o3 x Si 03 i—i (/i -a -o "S '3 o o -is o3 a3 C o 4J "S-2 -T T —CO — — No. 1 30 Bulletin Museum of Comparative Zoology, Vol. 152, 2 have and lacks a fold; unlike those tails of crown morphology. M may paracone forms, the is otherwise feature- been slightly larger than the adjacent teeth, ectoloph as in Ga- a fold or basal but it is not greatly enlarged less, lacking parastylar bulges in the of and shatostylops macrodon. regions parastyle metastyle. coronal crests and As for Sinostylops promissus Tang and The lingual (pre- post- are and enclose a fos- Yan, 1976, poor preservation of the type protocristae) strong sette that into a and only known specimen leaves various probably persists fairly of wear. The sulcus be- character states open to question. It cannot advanced stage 2 on is not so be determined if an entolophid was pres- tween the lingual cusps M cres- as in or ent on M l2 . It appears that a trigonid deep Palaeostylops Gashatosty- cent was retained, as in Asiostylops, and lops but, as in those genera, it probably the cristid obliqua attaches to the trigonid persists to advanced wear. A lingual cin- 3 is or on as in Both- at the metaconid, as in Bothriostylops spp. gulum weak lacking M , The antemolariform teeth form a graded riostylops progressus; as in Gashatosty- 2 series and are long, narrow, and bladelike, lops macrodon, M is considerably larger 3 . is most . are than M especially dP 4 The premolars serially Anatolostylops closely and tricuspid, with a straight, crested heel. Si- similar to the unnamed genus species, nostylops promissus lacks advanced fea- with which it shares several derived char- tures of the lower molars seen in Palaeo- acters not found in other Arctostylopidae. stylops and Arctostylops. The morphology The ectoloph is anteroposteriorly elongate, of the premolars would seem to indicate with labial plications reduced or lost. The pertinence to the Arctostylopidae; within lingual division of upper molars is poorly the family, Sinostylops promissus is sim- marked in Anatolostylops and absent in ilar only to Bothriostylops spp in the lin- the unnamed form; because these genera gual attachment of cristid obliqua to tri- otherwise appear to be closely related to gonid. forms in which it is well-developed (e.g., Allostylops periconatus Zheng, 1979, Palaeostylops), we believe this to repre- about which little can be said, is repre- sent reduction or loss rather than retention sented by the rostral part of a skull with of a primitive condition (as in Asiostylops) . the dentition very poorly preserved. The The cheek-teeth of the undescribed genus upper molars resemble those of Asiosty- and Anatolostylops are higher-crowned lops, and are therefore presumably prim- than in other genera, and the pre- and itive, in lacking an enlarged parastyle and postprotocristae better developed, enclos- in retaining paracone and metacone folds ing a more persistent fossette than in other on the ectoloph. There was, apparently, no members of the family. Although Gasha- 1-2 on a is posterointernal cusp M ; prominent tostylops autapomorphous in several re- anterolingual cusp (pericone) is present on spects, notably in the development of ac- the lingual cingulum, as is variably present cessory cuspules on the lingual cingulum on upper molars of Gashatostylops mac- and base of the ectoloph of upper molars, rodon. 1-2 The posterior cingulum of M is it is similar to Anatolostylops and the un- so that the broadly expanded, molars are described form in several other respects. in occlusal subquadrate aspect. The den- These include a reduction of the lingual tition as preserved gives little indication of sulcus on at least the first tooth of the upper to this and the affinity group, position of molar series and the great size of the sec- is therefore Allostylops indeterminate. ond molar relative to that of adjacent teeth. dubius Anatolostylops Zhai, 1978, Among advanced Arctostylopidae, Ana- 2 known from M is a -\ clearly rather spe- tolostylops is divergent in having double i ialized Form and may be significantly opposition of upper to lower teeth, as in- than younger the other genera. As in Pa- dicated by the presence of a distinct wear laeostylops, Arctostylops, Gashatostylops, facet in the mesostylar area of the upper and the is Bothriostylops, ectoloph high molar (this would correspond to a facet Arctostylopids • (Mammalia) Cifelli ei al. 3] 1 1 . the at 1 Figure Hypothesized relationships among Arctostylopidae. Characters nodes (see Table 4): ) metaconid added to P , at 3 pseudohypocone on least one upper molar, upper molar paracone fold lost, ectocingulid developed on lower molars, on P , P talonid ectocingulid developed 4 4 curved, upper molar parastyle enlarged; 2) lower molar cristid obliqua attaches Imgually to rear of trigonid; 3) M 3 elongate? (condition unknown in S. promissus), 4) lower molar entolophid well developed, lower molar cristid attaches to rear of obliqua labially trigonid, lower molar ectocingulid strong and pillarlike, P 4 paraconid shifted labiaily. lower molar 3^ paracristid lost; 5) shearing notch developed on P4 talonid, P paracone fold lost, canines lesser differentiated. 6) second molars 1 2 9 enlarged, pseudohypocone lost on M ?; 7) upper molar protocristae salient, M pseudohypocone reduced , parastyle fold on ectoloph of posterior upper cheek teeth lost, ectoloph of upper molars anteropostenorly elongate, cheek teeth very high crowned. anterior to the protoconid on the lower periconatus, poorly known and of doubt- molars, which are not known for Anato- ful affinities, have been omitted from this lostylops). This facet is lacking in Arcto- phylogeny). Asiostijlojis spanios is the most stylops, Palaeostylops, and Gashatosty- primitive taxon known and i^ considered lops, which apparently had singly opposing to represent the sister group of all remain- upper and lower cheek teeth. ing taxa. Bothriostylops spp, unique in .it An hypothesis of interrelationships of the least one character (the presence o\ .i hy- 7 Arctostylopidae is given in Figure ll (Ka- poconulid lobe on \1>. generall) resemble zachostylops occidentalis, which we have Asiostylops in their retention "t primitive t" share not examined first-hand, and Allostylops features, but nonetheless appear several derived features with the remain- taxa. these are the ol 7 ing Among presence The absence of a chronologic dimension is due to a P metaconid, a curved P< talonid. a lin- uncertainties of relative age, not our lack of appre- 3 of 1 and the loss of the ciation for this consideration. gual division M , Vol. No. 1 Bulletin Museum of Comparative Zoology, 152, with an morphotype, of which paracone fold and the presence of a large ungulate is a approximation, narastvle on the upper molars. Sinostylops Protungulatum good all these notoungulates of the earliest fau- nromissus (poorly known and lacking nas share a number of dental speciahza- much of the most' diagnostic morphology is similar tions 81, 9F). The posterior upper i„ the type and only specimen) (Figs. 34 are somewhat molanzed, onlv to' Bothriostylops spp in its lingual premolars (P ) it is with supporting antenor cristid obliqua-trigonid attachment; large protocones sister tax- and lingual cingula and trigonal very tentatively regarded as the posterior the teeth are dominated labially by on of Bothriostylops spp. The remaining crests; de- a which is separated Vrctostylopidae clearly are united by prominent paracone, from the also and rived morphology not found in Asiostylops well-developed parastyle A metacone, as far as is known or Bothriostylops. These features include metastyle. cheek- does not on upper premolars of mainly specializations of the lower develop Illustrated of teeth, such as the labial attachment of the notoungulates. specimens Henricosbornia cristid obliqua, the presence of a pillar- lophodonta (Simpson, and de- like ectocingulid, and the loss of the para- 1948, figure 53) Oldfieldthomasia have cristid. Among advanced genera, Arcto- bilitata (Simpson, 1967, plate 5) metacones on the teeth indicated to be P\ stylops appears to be the most primitive, but with other materials be- lacking specializations such as a shearing comparison in and to these indicate that the notch on ? 4 , found Palaeostylops longing species are deciduous. Gashatostylops. Within the group formed teeth in question probably molars bear a b\ the remaining genera, the undescribed The upper strong ectoloph sulci form and Anatolostylops possess several whose labial wall is marked by sep- and meta- svnapomorphies (mainly features related arating parastyle, paracone, 1 "2 to the hvpertrophied ectoloph of upper cone. M are quadrate in occlusal view, molars) and both share with Gashatosty- with a posterolingual cusp (hypocone) sep- the a sulcus. M 3 lops an enlarged second molar. arated from protocone by does not a hypocone, but variants OF SOUTH develop THE NOTOUNGULATA among even prim itive taxa may show AMERICA strong development of the cingulum in this The early Tertiary Notoungulata of region. The crest linking protocone to South America have been fully reviewed paracone (preprotocrista) is strong and is by Simpson (1948, 1967). Additions to developed into a protoloph; on the first two knowledge since publication of these molars, at least, and variably on M\ a monographs have been principally the Ita- metaloph joins hypocone and metacone boraian to Casamayoran notoungulates of (Fig. 12). The metaconule of upper molars Itaborai, Brazil (Paula Couto, 1952, 1954, is expanded anterolabially into the trigon 1978) and of northwestern Argentina basin as a crochet; various other cuspules (Bond, 1981; Pascual, Vucetich, and Fer- and crests characterize this part of most nandez, 1978; Vucetich, 1980). As recog- notoungulate upper molars (see Patterson, nized by Simpson, the major advanced no- 1934; Simpson, 1948). Cingula are present toungulate suborders Toxodonta and anteriorly and posteriorly but not lingual- T\ potheria (including Hegetotheria; see ly. The posterior lower premolars (P3_ 4 ) are ( lifelli, 1985a) were differentiated by the molarized (P3 somewhat less than P 4 ): the late Paleocene, \\ ith 5 families collectively trigonid is crescentic, with crests directed represented. Simpson grouped two other anteriorly and posterolingually from the families of the earliest faunas (Riochican protoconid; the talonid is much shorter than ', the Henricosborniidae the trigonid and also bears a crescentic into his paraphyletic crest. The construction of the lower molar . When compared trigonids is extraordinary, and the homol- • Arctostylopids (Mammalia) Cifelli et al. ogies of some parts are open to question. DISCUSSION A crest (paracristid?), variable in length, In the original description ol thespei ies, extends anteriorly or anterolingually from Matthew (1915) referred Arctostylops the a low anterior crest or cin- protoconid; steini to the order "Entelonychia" and. on the anterior face of the tooth, gulum, within that group, plated the species w ith with this in wear so may connect heavy some doubt in the Isotemnidae. At thai that the (?) appears to run to paracristid time, "notoungulate" t<> many students the of the tooth. The lingual margin great- (see, e.g., the influential classifications oi variation occurs in the of the est region Gregory, 1910; Osborn, L910; and Scotl metaconid. That be cusp may anteropos- 1904) was equivalent to "indigenous South terior^ expanded (Henricosborniidae, American ungulate," and did not explic- Oldfieldthomasiidae), an an- some bearing itly refer to that group in the sense it is directed crest. 8 terolabial-posterolingually defined today . "Entelonychia" was a sub- variant involves the of Another presence 1 i in- order proposed by Ameghino 1 89 to an anterior accessory cusp, which some- clude the aberrant, clawed Homalodo- of a times bears the appearance paraconid therium (a Santacrucian, mid-Miocene that has lost the paracristid connecting it form shown by Patterson, L936, to be tox- in to the protoconid (most notably Isotem- odont-like in the construction of its eat in some Oldfieldthomasi- nidae but also region and since universally placed in the idae). Notostylopids are characterized by Toxodonta, a suborder of the Notoungu- the crest an accessory cusp on linking pro- lata) within the "Aneylopoda," thus unit- so that toconid to metaconid (protocristid), ing it with the similarly clawed chalico- de this crest is serially tricuspid. Marshall, theres of Holarctic faunas. Ameghino had Muizon, and Sige (1983) propose homol- abandoned the use of the term "Entelon- which ogies for these trigonid structures, ychia" by the time of his final (1906) clas- a basic no- oth- they argue are variations about sification, but by this time had placed and toungulate pattern that included a pre- er notoungulate families (Isotemnidae and in its postmetastylid. The talonid consists, Leontiniidae, both currently recognized as of a simplest form (Henricosborniidae), belonging to the Toxodonta) with the crescent (cristid obliqua and postcristid) Homalodotheriidae in the "Aneylopoda." which all these fam- uniting hypoconid and hypoconulid, The dentition of members of the nonetheless are retained as distinct, cusp- ilies are relatively primitive within No- is workers like entities. The entoconid developed toungulata. Thus, later ignored and other transversely (entolophid) and, in advanced Ameghino's reference of these to the and in- forms, joins the postcristid anterior notoungulates to Holarctic groups, condi- hypoconulid. The most primitive stead resurrected his term "Entelonychia tion of this feature is seen in henricosbor- to include generally primitive notoungu- the niids such as Henricosbornia itself. The lates. (Scott, 1913, for instance, placed K\ the entolophid is incompletely developed, pos- Notostylopidae under this heading. more "Entel- terobucally oriented, and is somewhat time of Matthew's (1915) writing, on the to notoun separated from the hypoconulid onychia" referred primitive then, as now, the Isotem- posteriormost molar of Henricosborniidae. gulate mammals; to be basal meml On M,, however, this crest runs labially nidae were considered to ra the hypoconulid or to a point just anterior of the South American notoungulate that henricosborniids are to that cusp, and it therefore appears diations (although the the crest to be somewhat the entolophid is homologous to generally acknowledged connecting entoconid and hypoconulid (a more primith of the postcristid), and becomes portion current of the Notoungulata now distinct as a separate loph by migrating -The concept clear Roth, I had, however, been made b) anteriorly. Vol. 152, No. 1 3 t Bulletin Museum of Comparative Zoology, then Matthew and Granger (1925) recog- 1967), was that the Henricosborniidae, from the Riochican and nized that Palaeostylops iturus was strong- known only early in (Cifelli, 1985b), or putative ly specialized having high-crowned Casamayoran cheek-teeth with well-developed shearing late Paleocene and early Eocene (Marshall, Hoffstetter, and Pascual, surfaces, and in having reduced lower mo- 1985; Marshall, indicated the most primitive of lartrigonids. In this respect, they 1983), represent as an- known this interpreta- that (pp. 4-5), "it may be regarded Notoungulata. By arose in South America from cestral to Arctostylops and through that tion, the order the same stock which rise genus to some of the South American "ungulate" gave also to the other of South Eocene Notoungulata (e.g., Leontinia, groups indigenous of a Notostylops, etc.) but to the latter only in American ungulates. Migration prim- to North America and a broad way, as no one of the genera of itive notoungulate the Deseado fauna can be cited as clearly thence to Asia would thus provide the for the following the line indicated by Palaeo- source Arctostylopidae (Simpson, and stylops-Arctostylops." Nonetheless, as im- 1951, 1965, 1978, 1980). Szalay followed in this plied in the foregoing statement, they re- McKenna (1971) Simpson that molars of then known garded Palaeostylops as more primitive in respect, noting a number of features (for instance, the sim- arctostylopids were more advanced than in the earliest South American no- pie premolars) than the earliest of the South any American notoungulates or Arctostylops. toungulates. Apparent support for a south- They thus believed the Asian genus to be ern origin of the Notoungulata, on both ancestral, at least in a general sense, to all morphological and temporal grounds, is New World forms, and that "the South lent by the proposed referral of Peru- American Tertiary hoofed mammals were therium, from the Late Cretaceous of Peru, originally derived from the north, al- to the order (Marshall, de Muizon, and though undergoing a great secondary evo- Sige, 1983). Placement of this genus, which lution in the Neotropical region" (p. 2). is based largely on two broken molars of Simpson (1934) clearly defined the No- the type and only species, has been a mat- toungulata and its contents. He removed ter of considerable dispute since its initial the Arctostylopidae and Notostylopidae (a description (Grambast et al., 1967), with group of primitive South American no- workers variously suggesting arctocyonid toungulates) from the "Entelonychia" and (Grambast et al., 1967), didolodontid placed them with the Henricosborniidae (Tedford, 1974), periptychid (Van Valen, in a then new paraphyletic suborder, No- 1978), and even marsupial (Hoffstetter, tioprogonia, defined on the basis of prim- 1981) affinities. Marshall, de Muizon, and itiveness of its constituent taxa. This left Sige (1983) suggested that Perutherium the "Entelonychia" as Ameghino had orig- possesses, in common with notoungulates, inally conceived it except that Simpson re- a pre- and postmetastylid in the trigonid moved the Leontiniidae to the Toxodonta. of the lower molars, and that the genus is Thus recognized, the Notoungulata com- a morphologically appropriate antecedent prised four suborders: Notioprogonia, to both the South American notoungulates "Entelonychia," Toxodonta, and Typothe- and the Arctostylopidae. ria. On the basis of further studies (Pat- Patterson (1958; Patterson and Pascual, terson, 1936; Simpson, 1936b), Simpson 1972), on the other hand, followed Mat- later (1945) removed the remaining con- thew (1928; Matthew and Granger, 1925) tents of the "Entelonychia" (Isotemnidae in believing that notoungulates arose in the and I lomalodotheriidae) to the Toxodon- north and, along with several mammalian ta. where they have since remained. companions, colonized South America in aborated in his two the earliest Tertiary, later to radiate and irliest South flourish on that continent. The basis for tunas (Simpson, 1948, this opinion is unclear, but it is likely that Arctostylopids Mammali fellietal Patterson, like Matthew before him, was fications and some marked changes in the of impressed by early records Arcto- later, more specialized forms. These in North America stylopidae (then thought apparently diagnostic characters an to be early Eocene) and Asia (latest Paleo- not known in am other defined ordei several of cene), and by the strikingly of mammals Nevertheless, these arc dental features found in mem- unlike primitive South American notoungulates bers of that An Asian for the in detail and family. origin one cannot absolutely was also Notoungulata suggested by Nesov exclude the possibility ol conver- and Rose (1987). Gingerich (1977) pro- gence." (Simpson, L978, p. 325 posed yet another possibility, that the No- toungulata arose in Central America Possible (where evidence bearing on this issue is Relationships and from there both north- Evaluation of lacking) spread these contrasting v iews on ward and southward. the origin and subsequent dispersal ol the Because of the inferred primitiveness of Notoungulata, of great interest in both Asiostylops within the Notoungulata (sim- zoogeographical and paleobiologies terms, ple premolars, triangular upper molars is dependent on determination ol mor- lacking a hypocone, simple molar lophs, phocline polarity sequences and the ro- unreduced anterior wing of lower molar bustness of the phylogenetic framework trigonids), Zheng (1979) suggested that the derived therefrom. The issue of funda- order originated in Asia and, more specif- mental interest, one which remains to be ically, in southern China. Earliest records examined in detail, is the phylogenetic po- need not infallibly indicate centers of or- sition of the Arctostylopidae with respect igin, however. Van Valen (1988) consid- to South American Notoungulata Assum- ered Asiostylops to be sufficiently primi- ing notoungulate monophyly, inclusive of tive to be structurally antecedent to the Arctostylopidae, three possibilities trigonostylopids (an archaic group of As- present themselves: 1) arctostylopids took trapotheria, which are endemic to South origin from a southern notoungulate as that America). group is known (southern origin): 2) the Several recent studies have emphasized southern notoungulates derived from a the profound differences in dental spe- form that falls within the Arctostylopidae cializations between the Notoungulata and as that group is here conceived (northern Arcto- the Arctostylopidae, and on this basis have origin for the order); and 3) the South American tentatively disassociated Holarctic from stylopidae and known or South American forms (Cifelli, 1983a, Notoungulata are sister taxa (northern .1985a; Schaff, 1985; Thenius, 1985). It is southern origin). of the cranial well worth pointing out that it was Simp- Even without knowledge of suite ol son who first flirted with this possibility, morphology arctostylopids (a characterizes tins before returning to a more traditional view synapomorphies region there is in the same paper: in notoungulates; Simpson. 1948), rather imposing evidence, in the dentition No- and proximal ankle, that the southern constitute a as- "A possibility that seems not to have toungulata monophyletic been considered but perhaps should semblage. Derivation of the Arctostylopi- within the order as it iscmreiitK be is that Arctostylops, Palaeostylops, dae from would simplifi- and Sinostylops, although quite sure- recognized require man) cations in the dentition, because ly related among themselves, might (reversals) in cases and all arcto- not after all be true notoungulates. Asiostylops many instances are more Their dentitions do have derived stylopids in some prim- southern not,. un- characters that occur in almost all ear- itive than any known modi- The most significant ol th< ly notoungulates with various gulate. Vol. No. 1 Bulletin Museum of Comparative Zoology, 152, and that characters are in the upper molars. All de Muizon, Sige (1983) suggest the various structures ot southern notoungulates have secondary accessory tngonid a cro- be homologized with complications, consisting of at least notoungulates may in a and a and that these chet (Patterson, 1934; Simpson, 1948) pre- postmetastylid molars and a are for the order. Unlike typical the trigon basin on all upper primitive 12 and the all lack South American notoungulates liNpocone on M ; arctostylopids Perutherium lacks an en- the first character and at least Asiostylops Arctostylopidae, on its lower molars. The absence among that family lacks either a hypocone tolophid of a and in Asiostylops or hypocone-like structure. By analogy pre- postmetastylid 3 would therefore re- with a series of variants in M of Henri- and Bothriostylops of loss of these cosbornia lophodonta (Fig. 12), which are quire postulation secondary in forms which otherwise seem not quadritubercular but which illustrate structures for the to be rather in the construction a plausible character state series primitive molars. Thenius ac- addition of the posterolingual cusp on of their lower (1985) the lower molar and primitive notoungulate anterior upper cepted pre- postmeta- as a of no- molars, the posterointernal cusp of south- stylid pattern synapomorphy excluded em notoungulates appears to be a deriv- toungulates, and arctostylopids it from ative of the cingulum and therefore a from the order because was lacking "true" hypocone (Simpson, 1929). By con- "Palaeostylops steini." for the trast, in arctostylopids which have quad- The morphotype notoungulate 12 ankle bones is not ritubercular M , the posterointernal cusp proximal strongly spe- for to un- is encircled basally by the cingulum and cialized (as compared, instance, Pe- appears to have originated as a transverse, gulate groups such as the Litopterna, and lingual extension of the metacrista from rissodactyla, Artiodactyla, Hyracoidea, 9 modified at first the region of the metaconule . Thus, the all of which are highly Nonethe- posterolingual upper molar cusp of south- appearance in the fossil record). of ern notoungulates and arctostylopids ap- less, it is characterized by a number pears to have been acquired indepen- synapomorphies which render it readily dently and in a nonhomologous fashion. recognized (Cifelli, 1983b). These features Even the most primitive of southern No- include a long, constricted astragalar neck, toungulata (Henricosborniidae) have sub- with an oblique dorsal crest; astragalar molariform posterior lower premolars; P 4 body with a median (tibial) protuberance; has a complete, curved talonid crescent, astragalar foramen with posterolateral sul- Although the serially multicuspate, blade- cus interrupting continuity of tibial troch- like lower premolars of such forms as Pa- lea and flexor tendon groove; and well- laeostylops may reflect specialization for developed sustentacular-navicular facet shearing (secondary simplification), Asio- contact on the astragalus. stylops lacks the degree of molarization Except for a constricted astragalar neck, seen even in henricosborniids. none of these features is shared with known The proposed addition of Perutherium arctostylopid ankle regions (Gashatosty- to the (iltiplanense Notoungulata (Mar- lops macrodon and Palaeostylops iturus) , shall, de Muizon, and Sige 1983) presents which bear specializations contrasting with further problems for an origin of the Arc- those of notoungulates. The arctostylopid tostylopidae within that group. Marshall, ankle is advanced in having an astragalus with a cylindrical, vertically-walled body, the tibial trochlea extensivelv developed I considered the posteroin- lack of a fibular shelf; s anteroposterior^; il i " "I l arctostylopid upper molars to be a • i r . i i i .1 .1 navicular facet so that the axis elj expanded metaconule; it developed lowever, that the upper molar conules were of movement along the midtarsal joint would have been roughly parallel (rather Arctostylopids • (Mammalia) Cifelli ei al. mm 3 12. M variants in Henhcosbornia Figure lophodonta, AMNH 28964, from the early Casamayoran Canadon Vaca local fauna addition of illustrating hypothesized hypocone through linking of postcingular cusp and metaloph. than to that at the ankle oblique) proximal Granger, 1925; Patterson, 1958; Zheng, cuboid facet lost joint; astragalar (?); ectal 1979) is also contradicted by the available facet inclined with steeply respect to in- morphological evidence. Neotropical \<>- ferior surface of calcaneal fib- astragalus; toungulata have a different style of upper ular facet strongly developed into a semi- premolar molarization from that of arc- cylindrical surface; and sustentaculum of tostylopids and lack a metacone on P calcaneus distally located, at or near distal The lower molars of henricosborniids show (cuboid) end of the bone. Most of these a very primitive state in the development ankle modifications are usually associated of the typical notoungulate talonid: the with restriction of lateral and inversion/ major cusps (entoconid, hypoconid. hy- eversion movement, with concomitant poconulid) remain distinct; the entolophid greater capability for flexion/extension, at is weak. The placement and orientation ol the proximal and mid-tarsal joints. Such the entolophid suggest that it was derived specializations are commonly found among from the entoconid to hypoconulid pari "I terrestrial mammals (Cifelli, 1983b). The the postcristid. Even in primitive arcto- extreme distal position of the astragalo- stylopids (e.g.,Asiostylops), the hv poconid calcaneal facets on the calcaneus (a prim- is indistinct, having been merged into tin itive condition?), implying poor mechan- talonid crescent. The entolophid <>1 arc- ical advantage for rapid flexion of the pes tostylopids is advanced in being more an- by the gastrocnemius and soleus muscles, teriorly placed and is oriented anterola- ^ is enigmatic in this regard, and contrasts bially (Schaff, 1985). If Perutherium a with the condition seen in terrestrial sal- notoungulate, as argued l>\ Marshall, de tators or cursors. In any event, regardless Muizon, and Sige (1983), then derivation of the paleobiological implications of this of South American taxa from arctostylo- unusual ankle morphology, it is clear that pids would require independent acquisi- notoungulates are uniquely derived with tion of the entolophid in the Neotropical structure is in respect to arctostylopids, and vice versa. forms, because that lacking Derivation of southern notoungulates Perutherium. is that known \rc- from the Arctostylopidae (Matthew and The final possibility No. 1 Bulletin Museum of Comparative Zoology, Vol. 152, unnamed genus and tostvlopidae and southern Notoungulata (Anatolostylops; which lacks folds other than those are sister taxa: that they shared an ancestor species), and is remi- that was exclusive to them and no other for the parastyle metastyle, would niscent of that of such as groupof mammals. This hypothesis notoungulates with which was be compatible with all existing scenarios Notostylops (comparison and dis- the basis for the and regarding the geographic origin genus family-group the addition names of the northern forms) and various persal of notoungulates. With but members of to the Arctostylopidae of primitive forms Leontiniidae, primitive and southern No- such as Asiostylops and Bothriostylops, both the Arctostylopidae have lower, more nearly all of the similarities shared by toungulata complexly and ?Pa- southern notoungulates and advanced arc- folded ectolophs. Arctostylops in- also resemble some southern tostylopids would have been acquired laeostylops in dependency and therefore represent par- notoungulates, especially Notostylops, allelisms. These include the reduction of the high talonid crescent, which achieves attachment with the the lower molar trigonids, the addition of an anterior trigonid labial is not accessory trigonid structures to those teeth at a very position; this, again, a condition shared more (the homology of these structures, termed by primitive of either pre- and postmetastylid by Marshall, de members group. Muizon, and Sige, 1983, is open to some Remaining Resemblances question, even among the taxa restricted to South America), the development of a With the dismissal of many arctostylo- the molar crown similarities as conver- talonid on ? 4 , and upper pid-notoungulate pattern, which is superficially similar but gent acquisitions within each group, it is appears on other grounds to include non- relevant to evaluate the uniqueness of re- homologous features, as discussed above, semblances that remain. The most striking What is known of the ankle region in arc- of these is the transversely developed, lo- tostylopids indicates that they are diver- phate entoconid (entolophid) of the lower gently specialized from notoungulates. One molars. This is an unusual but not excep- specialization of the arctostylopid ankle, tional feature among mammals: it surely the development of the calcaneal fibular developed independently in the Astra- facet into a large, semicylindrical surface, potheria and twice among the Litopterna is found among a group of advanced tox- (Cifelli, 1983a; Cifelli and Soria, 1983). odont Notoungulata (the monophyletic Among Holarctic mammals, an entolophid group including Notohippidae, Leontini- or similar structure developed indepen- idae, and Toxodontidae), but this was dently in numerous rodent lineages (L. L. clearly developed independently by them. Jacobs, personal communication). Without Certain other notoungulate resemblances knowledge of more primitive forms, it is of arctostylopids, which undoubtedly in- not possible to determine if the arctosty- fluenced early workers in their compari- lopid entolophid arose, as in the southern sons and in their speculation regarding re- notoungulates, from part of the postcristid lationships, evidently represent derived or if it is demonstrably non-homologous ( liaracter states within both groups and are (the entolophid of astrapotheres, for in- almost certainly convergent. These in- stances, appears to be a de novo structure), elude the presence of a labial ectocingulid, If, as argued by Marshall, de Muizon, and which is characteristic of most toxodont Sige (1983), Perutherium is a notoungu- lower molars and premolars and of ad- late, then independent acquisition of the vanced forms (e.g., Palaeostylops, Ga- entolophid in the Arctostylopidae is sug- and shatostylops, Arctostylops) among the gested by the fact that they primitively The Vrctostylopidae. smooth ectoloph of lack the accessory trigonid structures pos- y lopid upper molars sibly shared by that genus with Neotrop- • Arctostylopids (Mammalia) Cifelli et al. 39 ical Other shared dental Notoungulata. ambiguously homologous, and in part be- of and features arctostylopids notoungu- cause the evidence of relationship has often lates, derived with respect to an ungulate been based on shared primitive features include a crescentic morphotype, might rather than uniquely derived specializa- lower molar trigonid (this condition is tions. The Arctostylopidae have been im- uncertain in southern notoun- somewhat mune to such controversy because, despite gulates, as the trigonid is already reduced some unique aberrancies and retention ol at first appearance), reduced upper molar a few primitive features, the advanced stylar shelf and lobes, and slightly raised genera Arctostylops and Palaeostylops centrocrista between paracone and meta- strikingly resemble notoungulates and no cone on the upper molars ("incipient" ec- other mammals in certain aspects of their latter features are not in toloph). These dental anatomy. Evaluation of the realit) themselves or collectively diagnostic, as of this relationship and its precise nature they represent generalized, almost gradal was long hampered by insufficient knowl- trends in many different groups of Paleo- edge of arctostylopid morphologic diver- cene and Eocene ungulate-like mammals, sity and of the structure and relationships Of the three alternatives of arctostylo- of the most primitive notoungulates of pid-notoungulate relationships discussed South America. With these circumstances above, the most permissive, that they rep- now dramatically improved, considerable resent sister taxa, is the most likely. (This doubt is cast on the close relationship of is true by definition, as the other two pos- the two groups, accepted without question sibilities are more specific and therefore for most of this century. A common no- more susceptible to falsification.) Yet, be- toungulate/arctostylopid ancestor (i.e., a cause most similarities of arctostylopids to morphotype for the two groups, consid- notoungulates must have arisen indepen- ered as sister taxa) might have been suf- dently, whether by parallelism or conver- ficiently primitive to have given rise to In gence, the evidence that they collectively many other orders of mammals. rec- comprise a monophyletic unit with respect ognition of this, and considering the ample to other mammals is slim: it amounts, in evidence for monophyly of the Arctosty- the to its fact, to one possible character (entolophid) lopidae, we have referred family the that is known to have developed indepen- own order. Thus recognized, group Asian radiation that dently several times among other, unre- would represent an to North America, lated groups. This is hardly secure docu- managed to disperse The mentation of monophyly. Other evidence, possibly in the late Paleocene. geo- such as that provided by the ankle region, graphic distribution of arctostylopid taxa, the to \ (>rtn suggests that a common ancestor of the two and hypothesized immigration 13. It is in- groups would have been exceedingly America, are given in Figure to note that most of the primitive and, probably, not exclusive. teresting primitive forms are more southerly in distribution. Of while Distinctness ArctOStylopida being found in south China, spe- in Since the time of Ameghino, many close cialized taxa are generally northerl) dis- relationships of South American with Hoi- tribution. broader oi Vrctosty- arctic forms have been proposed (see, e.g., The relationships Mammalia are em summaries by Simpson, 1978; McKenna, lopida among the dental morpho- 1981; and Gingerich, 1985). With the ex- matic. The arctostylopid bears some to several \muii ception of marsupials, the controversy sur- type similarity debatable affinities, such as Lan- rounding all ordinal and lower level re- taxa of and Petrolemur ferrals of South American to Holarctic taxa tianius (Cifelli, 1983a) has been considerable, in part because de- although contrasting specializations the latter are rived similarities are incomplete or not un- as loss of premolars in genusj Zoology, Vol. 152, No. 1 [Q Bulletin Museum of Comparative 100° 120° T CHI-JIANG BASIN NAN-XIONG BASIN Asiostylops span/os nov Arctostylopidae gen.+ sp. Allostylops periconah Bothnostylops notios ^\P£\ CHINA BASIN gx Vhun^n qiaNSHAN """ - Sinostylops promissus_ _ , - TURPAN BASIN ^r' Anatolostylops dubius -" XUAN- CHENG ~=? -anhui u BASIN ^-N /& XINJIANG NEMEGT BASIN ,'°*J . Bothnostylops propressusi ^ Palaeostylops iturus ^-Beijing' Gashatostylops macrodon , 140 • Arctostylopids (Mammalia) Cifelli et al. 41 Both forms were re- evident. originally rather than early Eocene (Gingerich and ferred to the Primates; the ankle of known Rose, 1977) age for Asian faunas, such as arctostylopids is completely dissimilar to Gashato, which include Palaeostylops and any belonging to that order. Arctostylopid Gashatostylops. ankle specializations are shared, as best we are able to determine from published fig- LITERATURE CITED ures (Sulimski, 1968; Szalay, 1977, fig. 16), AMEGHINO, F. 1894. Enumeration des with the Asian late Paleocene Pseudictops. synoptique especes de mammiferes fossiles des formations in been considered to This taxon has, turn, eocenes de Patagonie. Boletin d<- L Vcademia be part of "an endemic Cretaceous and Nacional de Ciencias de Con 1( il>a 13:259 l~>_: . 1906. Les formations sedimentaires du < early Tertiary Asian radiation, whose clos- tace superieur et du tertiare de Patogonie av« est living relatives are the Lagomorpha" un parallcMc entre leurs faunes mammalogiques (Szalay and McKenna, 1971, p. 301). et celles de l'ancien continent Anales del Museo Whatever the constituents of this radiation Nacional, Buenos Aires, 15(ser. 3. \<>l 8 I 568 M. 1981. (see also McKenna, 1975; Novacek, 1986; Bond, Un nuevo Oldfieldthomasiidai (Mammalia, Notoungulata) del Eoceno inferior Szalay, 1977), we note that lagomorphs and (Fm. Lumbrera, GrupoSalta idelNW ^rgentino some of their allies are special- suspected Anales II Congreso Latino-Americano de Pa- locomotion ized for saltatory (Szalay, 1977; leontologia, Porto Alegre: 521-536. see Bleefeld and McKenna, 1985, for de- Bleefeld, A. R., and M.C. McKenna 1985. Skel- etal of rodens scription of some lagomorph ankle spe- integrity Mimolagus (Lagomor- —which pha, Mammalia). American Museum Novitates, cializations); arctostylopids may 2806: 1-5. in this — just be primitive regard appar- Chow Min-chen, Qui Tao. vnd Li Vi \<. 1977 ently were not, as indicated by the lever Paleocene stratigraphy and faunal characters of mechanics of the calcaneus. mammalian fossils on Nomogen Commune, Si- zi-wang-qi, Nei Mongol. Vertebrata PalAsiatica. The diversity and abundance of arcto- 14: 228-233 [in Chinese; English abstract]. in Asian faunas, stylopids early Tertiary Chow Min-chen, and Qui Tao. 1978. Paleocene coupled with the proposed close relation- mammalian fossils from Nomogen Formation of 77- ship of North American Arctostylops to Inner Mongolia. Vertebrata Pal Asiatica 16: 85 abstract). Asian Palaeostylops as rather derived taxa (in Chinese; English ClFELLI, B. L. 1983a. The origin and affinities of within the that dispersal family, suggests the South American Condylarthra and early Ter- from west to rather than the reverse, Mu- east, tiary Litopterna (Mammalia). American is the most probable explanation for geo- seum Novitates, 2772: 1-49. the late Pa- . 1983b. Eutherian tarsals from graphic distribution of the group. Owing leocene of Brazil. American Museum Novitates, to endemism of Asian faunas older high 2761: 1-31. than those of the North American Wa- evolution . 1985a. South American ungulate and satchian, correlation of earliest Tertiary and extinction, pp. 249-266. In S. D. Webb The Great American Iiiotic mammalian assemblages between the two F. G. Stehli (eds.), New York: Plenum Pub! ( o itvii continents has been (Szalay Interchange. problematic + 532 of Arc- pp. and McKenna ,1971). The presence of the . 1985b. Biostratigraphy Casamayor- American Mu- tostylops in the Tiffanian (late Paleocene) an, early Eocene, of Patagonia of North America, the geometry of pro- seum Novitates, 2820: 1-26. 1983. ol the . and M. F. Soria. Systematic posed relationships among the Arctosty- Adianthidae (Litopterna. Mammalia American and the fact that more primitive i lopidae, Museum Novitates, 2771: 1 2 but not North An taxa are known from Asia Colbert, E. H. 1973 Wandering Lands and 323 America, suggest a late Paleocene (Dash- imals. New York: E. P. Dutton. pp. < P. 1957. Zoogeograph> The zeveg, 1982; Szalay and McKenna, 1971), Darlington, J. of Northern Hemisphere, showing arctostylopid ( Figure 13. Lamberts azimuthal equal area projection map hypothesized dispersal route to North America. No. 1 Bulletin Museum of Comparative Zoology, Vol. 152, Animals. New York: JEPSEN, G. L. 1930. New vertebrate fossils from the graphical Distributions of lower Eocene of the Bighorn Basin, Wyoming. Wile) and Sons. 675 pp. of the American So- DASHZEVEG, D. 1982. La faune de Mammiferes du Proceedings Philosophical cen- 69: 117-131. Paleogene inferieur de Naran Bulak (Asie ciety, et l'Ame- G. L., and M. O. Woodrurne. 1969. Pa- trale) et ses correlations avec l'Europe Jepsen, leocene from Polecat Bench For- rique du Nord. Bulletin de la Societe Geologique hyracothere Science, 164: 543-547. de France, 24(2): 275-281. mation, Wyoming. Li Chuan-kuei, and Ting Su-yin. 1983. The Pa- DASHZEVEG, D., AND D. RUSSELL. 1988. Paleocene Bulletin of the Car- and Eocene Mixodontia (Mammalia, Glires) of leogene mammals of China. 129- Museum of Natural 21: 1-98. Mongolia and China. Palaeontology, 31: negie History, D. 1947. of the 164. Love, J. Tertiary stratigraphy Jack- northwestern U.S. Geo- GAUDRY, A. 1902. Recherches paleontologiques de son Hole area, Wyoming. Oil and Gas Pre- \l Andre Tournouer en Patagonie. Bulletin de logical Survey, Investigations 27. la Societe d'Histoire Natural de Autun, 15: 117- liminary Chart 123. McKenna, M. C. 1975. Toward a phylogenetic Dentition de classification of the Mammalia, 21-46. In W. . 1904. Fossiles de Patagonie. pp. P. Luckett and F. S. of quelques mammiferes. Memoirs de la Societe Szalay (eds.), Phylogeny York: Plenum Press. Geologique de France (Paleontologie), 31: 1-27. the Primates. New sur un . 1980. Late Cretaceous and . 1906. Fossiles de Patagonie. Etude early Tertiary Pa- vertebrate reconnaissance, portion du monde antartique. Annales de paleontological Tog- wotee Pass northwestern 321- leontologie (Paris), 1: 1-42. area, Wyoming, pp. of Vertebrate 1908. Fossiles de Patagonie. De l'economie 343. In L. L. Jacobs (ed.), Aspects dans la nature. Annales de Paleontologie (Paris), History: Essays in Honor of Edwin Harris Col- 3: 1-28. bert. Flagstaff: Museum of Northern Arizona Press, + 407 GlNGERH II. P. D. 1974. Size variability of the teeth xv pp. the of in living mammals and diagnosis closely 1981. Early history and biogeography of of Pa- related sympatric fossil species. Journal South America's extinct land mammals, pp. 43- leontology, 48: 895-903. 77. In R. L. Ciochon and A. B. Chiarelli (eds.), . in the 1985. South American mammals Evolutionary Biology of New World Monkeys Paleocene of North America, pp. 123-137. In F. and Continental Drift. New York: Plenum Publ. G. Stehli and S. D. Webb (eds.), The Great Amer- Co. ican Biotic New York: Plenum Press, Interchange. Marshall, L. G. 1985. Geochronology and land xvii + 532 pp. mammal biochronology of the transAmerican \\i) K. D. Rose. 1977. report Preliminary faunal interchange, pp. 49-85. In S. D. Webb on the American Clark Fork mammalian fauna, and F. G. Stehli (eds), The Great American Biot- and its correlation with similar faunas in Europe ic Interchange. New York: Plenum Publ. Co. xvii and Asia. Geobios, Memoire Special, 1: 39-45. + 532 pp. Gradzinski, R., Kazmiercak, and Lefeld. J. J. Marshall, L. G., C. deMuizon, and B. Sige. 1983. 1969. Geographical and geological data from Perutherium altiplanense, un notoungule du the Polish-Mongolian paleontological expedi- Cretace superieur du Perou. Palaeovertebrata, tions. Polonica, 19: 33-82. Palaeontologica 13: 145-155. Grambast, L., M. Martinez, M. Mattauer, and Marshall, L. G., R. Hoffstetter, and R. Pascual. L. Thaler. 1967. Perutherium altiplanense 1983. Mammals and stratigraphy: geochronol- nov. gen., nov. sp., premier mammifere meso- ogy of the continental mammal-bearing Tertiary zoique d Amerique du Sud. Comptes Rendus de of South America. Palaeovertebrata, Memoire la Academie des Sciences, Paris, ser. D, 264: 707- 710 Extraordinaire, 1983: 1-93. GREGORY, W. K 1910. The orders of mammals. Matthew, W. D. 1915. A revision of the lower Bulletin ol the American Museum of Natural Eocene Wasatch and Wind River faunas. Part IV. Insectivora History, 27: 1-254. Entelonychia, Primates, (part). Bulletin of the 1920. On the structure and relations of American Museum of Natural 34: 429-483. Notharctus, an American Eocene primate. Mem- History, oirs of . 1928. The evolution of in the American Museum of Natural History, the mammals n.s. 3: 51-243. the Eocene. Proceedings of the Zoological Society i i of Hoi mi ii it, R. 1970. Radiation initiale de Mam- London, pt. 4, 1927: 947-985. mileres placentaires et biogeographie. Comptes Matthew, W. D., and W. Granger. 1925. Fauna Rendus Formation of Mongolia. American Museum Beartooth Region oi Wyoming and Montana Novitates, 376: 1-12. National Geographic Research Reports 1979: NESOV, L. A. 1987. Results of searches and inves- 589-595. tigations in the mammal-bearing Cretaceous and SCHLOSSER, M. 1923. Saugetiere, pp. 402-689 In the early Paleogene in the territory of USSR. F. Broili and M. Schlosser, eds., K V von Zittel, Annual of the All-Union Paleontological Society Grundziigeder Palaontologie, neubeari Munich 30: 199-218 [in Russian], M. R. Oldenbourg. 1986. skull of insec- 1904. Novacek, M. J. The leptictid Scott, W. B. The Miocene ungulates of 1 tivores and the higher-level classification of eu- agonia. Report to the British Association for tin- therian mammals. Rulletin of the American Mu- Advancement of Science: 589-590. seum of Natural History, 183: 1-112. . 1913. A History of Land Mammals in the Osborn, H. F. 1889. The Mammalia of the Uinta Western Hemisphere. New York: Macmillan I Formation. Part IV. The evolution of the un- 693 pp. Phil- gulate foot. Transactions of the American Simpson, G. G. 1929. Paleocene and lower Eocene osophical Society, new series, 16: 530-569. mammals of Europe. American Museum Novi- Osborn, H. F. 1910. The Age of Mammals in Eu- tates, 354: 1-17. rope, Asia, and North America. New York: Mac- . 1934. Provisional classification of extinct millan Co. xvii + 635 pp. South American hoofed mammals American Pascual, R., M. G. Vucetich, and J. Fernandez. Museum Novitates, 750: 1-21. 1978. Los primeros mamiferos (Notoungulata, . 1936a. A census of Paleocene mammals. Henricosborniidae) de la Formacion Mealla American Museum Novitates, 848: 1—15 (Grupo Salta, Subgrupo Santa Barbara). Sus im- . 1936b. Notas sobre los mamiferos mas an- Roth. Instituto del Museo plicancias filogeneticas, taxonomicas y cronolo- tiguos de la coleccion la Universidad Nacional de La Plata, Obra gicas. Ameghiniana, 15: 366-390. de Patterson, B. 1934. L^pper premolar-molar struc- del Cincuentenario, 2: 63-94. and ture in the Notoungulata, with notes on taxon- . 1945. The principles of classification of of mammals. Bulletin of the omy. Geological Series of the Field Museum a classification of Natural 85: l- Natural History, 6(6): 91-111. American Museum History, the ear in 350. . 1936. The internal structure of of . of the Mam- some notoungulates. Geological Series of the Field 1948. The beginning Age of the Museum of Natural History, 6(15): 199-227. mals in South America. Part 1. Bulletin 1958. Affinities of the Patagonian fossil American Museum of Natural History, 91: l- mammal Necrolestes. Breviora, 94: 1-14. 232. of Latin Amer- . of the fauna Patterson, B., and R. Pascual. 1972. South 1951. History Science American fossil mammals, pp. 247-309. In A. ica, pp. 369-408. In G. A. Baitsell (ed.), Haven: Yale Univ. Keast, F. C. Erk, and B. Glass (eds.), Evolution, in Progress (7th ser.). New Mammals, and Southern Continents. Albany: Press. of Evolution New State Univ. New York Press. 543 pp. . 1965. The Geography Paula Couto, C. de. 1952. Fossil mammals from York: Capricorn Books. 249 pp. of the of Mam- . The the beginning of the Cenozoic in Brazil. No- 1967. beginning Age mals in South America. Part 2. Bulletin of the toungulata. American Museum Novitates, 1568: 137: 1- 1-16. American Museum of Natural History. the Paleo- 259. . 1954. On a notostylopid from in South Ameri< a cene of Itaborai, Brazil. American Museum Novi- . 1978. Early mammals in and Proceedings tates, 1693: 1-5. fact, controversy, mystery. 122: 1978. Ungulados fosseis do Riochiquense of the American Philosophical Society, Anais de Itaborai, RJ, Brasil. III. Notoungulata. 328. the Curious 1 1 do Academia Brasileiro de Ciencias, 50: 219- 1980. Splendid Isolation: Mammals. New Haven 226. tory of South American Ter- Yale Univ. Press. 266 Qi Tao. 1979. A general account of the early pp. mammalian faunas of Shara Murun area, SULIMSKI, A. 1968. Paleocene genus Pseudictops tiary and 1929 [Mam Inner 2nd Congress of Stratigraphy, Matthew, Granger, Simpson, Mongolia. Polonica from Li and and its revision. Palaeontologica China, Beijing [in Chinese; cited malia) 19: 102-129. Ting, 1983, and not seen by us]. 1977. relationships and Rose, K. D. 1981. The Clarkforkian Land-Mammal SZALAY, F. S. Phylogenetk of the eutherian Mammalia, pp Age and mammalian faunal composition across a classification In M. K. Hecht, P. C Goody, and B the Paleocene-Eocene boundary. University of 315-374. m Vertebrate M. Hecht (eds.), Major Patterns Michigan Papers in Paleontology, 26: 1-197. Ana- Evolution. New York: Plenum Press. Roth, S. 1903. Los ungulados sudamericanos. R L. Decker 1974 Origins 1-36. Szalay, F. S., and les del Museo de La Plata, 5: Eocene evolution and function ol the pes m the Schaff, C. R. 1985. Paleocene mammals from the Bulletin Museum of Comparative Zoology, Vol. 152, No. 1 \dapidae (Lemuriformes, Primates), pp 239-259. ungulates in South America. Evolutionary The- 79 In F \ Jenkins, Jr. (eel), Primate Locomotion. ory, 10, pp. \r\\ 'lork: Academic Press. Vucetich, M. G. 1980. Un nuevo Notostylopidae SZA1 w F.S., \ND M. C. McKenna. 1971. Begin- (Mammalia, Notoungulata) proveniente de la ning of the Age of Mammals in Asia: the late Formacion Lumbrera (Grupo Salta) del noroeste PaJeocene Gashato fauna, Mongolia. Bulletin of Argentine Ameghiniana, 17: 363-372. the American Museum of Natural History, 144: Zhai Ben-jie. 1978. More fossil evidences favour- 269-318. ing an early Eocene connection between Asia and Tang Yim.-jun, and Yan De-fa. 1976. Notes on Neoarctic. Memoirs of the Institute of Vertebrate some mammalian fossils from the Paleocene of Paleontology and Paleoanthropology, Academ- Qianshan and Xuancheng, Anhui. Vertebrata ica Sinica, 13: 107-115 [in Chinese]. PalAsiatica, 14: 91-99 [in Chinese]. Zheng Jia-jian. 1979. The Paleocene notoungu- I i i >i OKI), B. H. 1974. Marsupials and the new pa- lates of Jiang-xi, pp. 387-394. In A Symposium leogeography, pp. 109-126. In C. A. Boss (ed.), on Cretaceous and Early Tertiary Bed Beds of Paleogeographic Provinces and Provinciality. So- South China. Beijing: Science Press, Academica ciety of Economic Paleontologists and Mineral- Sinica. [in Chinese]. 21. ogists, special publication , and Huang Xueshi. 1986. New arcto- Tin NIUS, E. 1985. Zur systematischen stellung der stylopids (Notoungulata, Mammalia) from the late Arctostylopiden (Mammalia, Notoungulata). Eine Paleocene of Jiangxi. Vertabrata PalAsiatica, 24: odontologische analyse. Bazprave IV. Bazreda 121-128 [in Chinese; English summary]. Sazu (Zbornik Ivana Bakovca), 26: 147-156. Zhu-chen. 1986. Late Cretaceous red beds in Hun- \\ V Valen, L. 1978. The beginning of the age of an. Journal of Stratigraphy, 10: 54-59 [in mammals. Evolutionary Theory, 4: 45-80. Chinese]. . 1988. Paleocene dinosaurs or Cretaceous