Novitates PUBLISHED by the AMERICAN MUSEUM of NATURAL HISTORY CENTRAL PARK WEST at 79TH STREET, NEW YORK, N.Y

Home , Didolodontidae, Divisaderan, Riochican

AMERICAN MUSEUM Novitates PUBLISHED BY THE AMERICAN MUSEUM OF NATURAL HISTORY CENTRAL PARK WEST AT 79TH STREET, NEW YORK, N.Y. 10024 Number 2772, pp. 1-49, figs. 1-7, tables 1-5 November 9, 1983

The Origin and Affinities of the South American Condylarthra and Early Tertiary Litopterna (Mammalia)

RICHARD L. CIFELLII

ABSTRACT A previously unknown genus and species of Proectocion (including Oxybunotherium), previ- a primitive litoptem from the late Paleocene ously referred to the Didolodontidae, is placed in (Riochican) of Itaborai, Brazil, is described, and the advanced litoptern family Adianthidae. three previously described didolodontid and li- An ancestral morphotype for the Litoptema and toptern species from Itaborai are referred to new Didolodontidae is constructed on the basis oftheir genera. Comparative analysis of the dentition of known morphology and by comparison with didolodontids and primitive Litoptema suggests a primitive oxyclaenine arctocyonids such as Pro- number of revisions in their taxonomy. The two tungulatum. Relevant North American ungulate Itaborai didolodonts appear to be more closely families (Arctocyonidae, Hyopsodontidae, Phen- related to each other than to Patagonian forms; acodontidae, Periptychidae) are briefly reviewed Protolipterna, new genus, Miguelsoria, new genus, and their members compared with this hypothet- ?and Asmithwoodwardia are placed in the Litop- ical ancestor. The primitive litoptern/didolodont terna in a new family, Protolipternidae. Aniso- morphotype is closely comparable to and shares lambda (including Ricardolydekkeria and Joseph- derived conditions with Torrejonian mioclaenine oleidya) and allies seem to be related to true hyopsodonts such as Litaletes. Such an exclusive proterotheriids, and are placed in the Proterotheri- relationship implies that the indigenous South idae in a new subfamily, Anisolambdinae, whereas American ungulate orders may not represent a the Polymorphinae (including only the type genus) strictly monophyletic unit with respect to other appears to be related to the Macraucheniidae. Eutheria.

INTRODUCTION The pre-Oligocene Litopterna and their and Simpson, consists ofjaw fragments and presumed allies, the didolodontid condy- isolated teeth from Patagonia. Because of its larths, are poorly known and little under- age, geographic location, and faunal repre- stood. Most of their early record, brought to sentation, the faunule from Sao Jose de Ita- light largely through the efforts of Ameghino borai, Brazil, is extremely important to in-

1 Student, Department of Vertebrate Paleontology, American Museum of Natural History.

Copyright © American Museum of Natural History 1983 ISSN 0003-0082 / Price $3.80 2 AMERICAN MUSEUM NOVITATES NO. 2772 terpretation ofthe origin and early radiations mammals, I am grateful to Drs. M. Mc- of these and other indigenous South Ameri- Kenna, M. Soria, G. Simpson, and L. Mar- can mammals. shall, although they do not necessarily es- Most of the manmmals from Itaborai were pouse my conclusions. Comments and described in a series ofcontributions by Paula criticisms on an earlier draft of this paper Couto (1952a, 1952b, 1952c, 1952d, 1954, were provided by Drs. M. McKenna and F. 1961, 1962, 1963, 1970, 1978a, 1978b, Szalay, and their advice is warmly appreci- 1978c, 1979). Despite considerable ecologic ated. Partial funding for this research was variance with the Patagonian faunas (Paula provided by the Theodore Roosevelt Me- Couto, 1970), a Riochican (late Paleocene) morial Fund, Thomas J. Dee Fund, Walker age is indicated by the presence of Carodnia Johnson Fund, Department of Geological (Paula Couto, 1952a), which occurs in Pat- Sciences, Columbia University, and the So- agonia only in the earliest Riochican faunal ciety of Sigma Xi. zone (Simpson, 1935a). The fossils derive from fissure- and channel-fill argillaceous ABBREVIATIONS marl deposits cut into the Itaborai Forma- a limestone of Late Cretaceous ACM, Pratt Museum, Amherst College, Amherst, tion, probable MA. or early Paleocene age (Francisco and Souza AMNH, American Museum of Natural History, Cunha, 1978), and have been recovered dur- New York, NY. ing the process of commercial limestone DNPM (sometimes cited as DGM), Divisao Na- quarrying. The Itaborai Formation is local, cional de Producao Mineral, Rio de Janeiro, being restricted to a small tectonic basin about Brazil. 25 km. northeast of Niteroi, State of Rio de FMNH, Field Museum of Natural History, Chi- Janeiro, Brazil. Additional material collected cago, IL. since the studies of Paula Couto reveal the MACN, Museo Argentino de Ciencas Naturales presence of a new litoptern species, and de- "Bernardino Rivadavia," Buenos Aires, Argen- tailed with the type and referred tina. comparison MLP, Museo de La Plata, La Plata, Argentina. specimens from Patagonia indicates three MHN, Museum National d'Histoire Naturelle, previously described species to represent new Paris, France. genera. In the present offering I describe these MNRJ, Museu Nacional de Rio de Janeiro, Rio materials and reevaluate the interrelation- de Janeiro, Brazil. ships and affinities ofthe Didolodontidae and Litopterna. SYSTEMATIC PALEONTOLOGY ACKNOWLEDGMENTS ORDER CONDYLARTHRA COPE, 1881a2 FAMILY DIDOLODONTIDAE SCOTT, 1913 For permission to study specimens in their care and help in so doing, I thank Drs. M. Didolodidae Scott, 1913, p. 489. Bunolipternidae Schlosser, 1923, p. 525. Coombs (Amherst College), C. Schaff (Mu- Didolodontidae Simpson, 1934, p. 6. seum of Comparative Zoology), D. Baird (Princeton University), W. Tumbull (Field DIAGNOSIS: Dentition complete and mio- Museum of Natural History), E. Lindsay claenine-like in primitive forms, but a hy- (University of Arizona), R. Emry (United pocone is present on M3. Tarsus, where States National Museum), F. de Souza Cunha known, lacking astragalocuboid contact (se- (Museu Nacional de Rio de Janeiro), D. rial); astragalus with relatively deep body but Campos (Divisao Nacional de Producao Mineral), J. Bonaparte and M. Soria (Museo de Ciencas Naturales "Bernardino 2 This aggregate of primitive ungulates has long been Argentino recognized to be a horizontal or paraphyletic group in Rivadavia") and R. Pascual and M. Bond the formal taxonomic sense, an arrangement not wholly (Museo de La Plata). The fine halftone illus- satisfactory but maintained because available evidence trations of figure 4 are the work of Mr. H. is inadequate to permit distribution ofthe included groups Galiano. For informative exchanges of fact into more meaningful suprafamilial categories. See dis- and opinion regarding South American fossil cussion below. 1983 CIFELLI: CONDYLARTHRA 3 margins of the tibial trochlea not raised into Simpson (1948) has already shown. Proec- sharp crests; superior astragalar foramen ves- tocion, based on upper cheek teeth, I place in tigial or absent. Astragalar head medially and the Litopterna and refer to it the lower molars transversely expanded. Astragalar facet for described by Pascual (1965) as Oxybuno- the medial malleolus ofthe tibia vertical and therium. Stirton (1947)3 described a worn M3 turning sharply mediad anteriorly. Peroneal from the Oligocene of Colombia as Lophio- tubercle of calcaneum extending posterodor- dolodus chaparralensis, but pertinence to this sally and terminating at the base of the ectal group is not clear, and it may be sirenian, as prominence. speculated by McKenna (1956). Paula Couto Simpson (1948, p. 95) noted that, "it is at (1952a) described four species of didolodon- present impossible to frame a purely mor- tids from Itaborai: Ernestokokenia parayi- phological definition (i.e., of the Didolodon- runhor, E. protocenica, Asmithwoodwardia tidae) which would ... exclude the North scotti, and Lamegoia conodonta. Based on American condylarth families." The above pedal evidence presented elsewhere (Cifelli, pedal characters, all advanced with respect 1983), I here refer Ernestokokenia parayi- to the primitive ungulate condition seen in runhor to the Litopterna under a new generic Protungulatum (see Szalay and Decker, 1974; name. Ernestokokenia protocenica is pedally Cifelli, 1983), do distinguish the didolodonts, and dentally a didolodontid, but differs sig- but they are based on unassociated tarsals of nificantly from the Patagonian species ofthat only a few of the included genera (Cifelli, genus and therefore also warrants generic sta- 1983), and the diagnosis must therefore be tus; Asmithwoodwardia is very tentatively re- regarded as provisional. An M3 hypocone is moved to the Litopterna. The addition of characteristic of all didolodontids (unless previously described and new material to La- Asmithwoodwardia is included), and does megoia conodonta improves knowledge of distinguish them from litopterns and the that distinctive species. McKenna (1956) de- primitive members, at least, of the Holarctic scribed Megadolodus from the Friasian La condylarthran families. Venta locality in Colombia, based on a jaw Simpson (1948) has reviewed the taxo- fragment with Ml and broken P4. Pertinence nomic history ofspecies placed in this family, to the family is clear, but more precise affin- and repetition here is unnecessary. Simpson ities are problematic. Perutherium altipla- listed the following genera of Didolodont- nense, based on two fragmentary specimens idae as valid: Didolodus, Argyrolambda, from a Late Cretaceous faunule near Laguna Paulogervaisia, Proectocion, Enneoconus, Umayo, Peru, has been referred to the Arc- Asmithwoodwardia, Ernestokokenia, and Ar- tocyonidae (Grambast et al., 1967; Sige, chaeohyracotherium. Of these, Archaeohyr- 1972), Didolodontidae (Tedford, 1974), and acotherium is probably but not surely syn- Periptychidae (Van Valen, 1978), and has onymous with Ernestokokenia, and been the subject of much debate (e.g., Pat- Argyrolambda has been shown to be a syn- terson and Pascual, 1972; Bonaparte, 1978; onym of Didolodus (Simpson, 1967). Van Simpson, 1978; Kielan-Jaworowska, Brown, Valen (1978) considered Paulogervaisia and Lillegraven, 1979; McKenna, 1981). This Ameghino, 1901 and Enneoconus Ameghi- mammal is so poorly known that familial and no, 1901 as junior synonyms of Didolodus ordinal reference is unclear; Perutherium is Ameghino, 1897, and Ernestokokenia not even demonstrably eutherian (Hoffstet- Ameghino, 1901 and Archaeohyracotherium ter, 1981). Ameghino, 1906 as junior synonyms of Asmithwoodwardia Ameghino, 1901. None PAULACOUTOIA, NEW GENUS of these synonymies are based on restudy of TYPE: Paulacoutoia protocenica (=Ernes- the original materials; several are certainly tokokenia protocenica Paula Couto, 1952a). incorrect (Asmithwoodwardia and Ernesto- DISTRIBUTION: Riochican, Brazil. kokenia; Paulogervaisia, Enneoconus, and Didolodus) and the others cannot be evalu- 3Separates bear the date 1946, but 1947 is the correct ated with materials currently at hand, as year of publication. 4 AMERICAN MUSEUM NOVITATES NO. 2772

DIAGNOSIS: Moderate-sized didolodont; scribed by Paula Couto (1952a, 1978b), and generally primitive in having a simple P3, its major characters are given in the above lacking upper molar mesostyles, relatively diagnosis. At this time the only tooth which small M3 hypocone, and lower molar para- may be added to the series previously figured conids connate with but distinct from meta- is P4, although more of the dentition will conids; but P4 bearing a bicuspid talonid and probably become known when the DNPM upper and lower M 1-3 greatly enlarged from collections are thoroughly studied. Unlike first to third. Upper molar cingula generally Didolodus, P3 is a simple tooth, triangular in not encircling protocone lingually nor para- outline, with a large conical paracone and cone labially, lower molar entoconid lacking very small, posterolingually placed proto- anterior accessory cuspule, P4 with weak pro- cone; there is no trace of a metacone, styles, toconule but no metaconule. or conules. A low, crenulated cingulum en- ETYMOLOGY: For Carlos de Paula Couto, circles the base of the crown except in the in recognition of his contributions to knowl- midlabial region and around the protocone. edge ofSouth American mammals, with par- P4 is enlarged, with bulbous and subequal ticular reference to the Itaborai fauna. protocone and paracone; a small metacone is Paula Couto originally (1952a) placed this present on the posterior slope ofthe paracone species in the late Paleocene and early Eocene and tightly appressed to that cusp, as in Di- Patagonian genus Ernestokokenia, but the dolodus. The cingulum is complete except at characters cited above show it to be more the lingual margin of the protocone. The up- distinct than previously thought. In these re- per molars are notably enlarged from first to gards, P. protocenica more closely resembles third; the first molar is by far the smallest Lamegoia conodonta (also of the Itaborai and is only slightly larger than P4. Styles are fauna), so that generic identity with Ernes- lacking and the molar cusps are low and bu- tokokenia, which shares derived conditions noid; the para-, meta-, and protocones are with Didolodus and other Patagonian genera, arranged in a triangular pattern, with a well- cannot now be maintained. More detailed developed hvpocone present on Ml-2 almost comparisons and discussion are given below. directly posterior to the protocone. The hypocone on M3 is small and variable in de- Paulacoutoia protocenica velopment, but invariably present. Figures 1G, 2F P4 bears a low but lingually placed paraconid and distinct, high proto- and metaco- Ernestokokenia protocenica Paula Couto, 1952a, nids, and the trigonid is therefore complete p. 361; 1978b, p. 211; Cifelli, 1983, p. 15. and molarized. The talonid is defined labially TYPE: MNRJ 1430V, left maxillary frag- by a crest, analogous to the cristid obliqua of ment with alveoli for P2 and with P3-M3, now the molars, and posteriorly by two well-de- missing or mislaid and not seen by me. fined cusps; a small talonid basin is formed. HYPODIGM: As given by Paula Couto The first to third molar enlargement seen in (1952a, pp. 361-362), except that AMNH the upper dentition is remarkable in the lower 49821 is a left P4 (not P3) and AMNH 49825 molar series. The paraconid is connate with includes three right M2s. Additional speci- and tightly appressed to the metaconid but mens now referred to this species include is always distinct, most clearly so on M3. All AMNH 55386 (right astragalus), DNPM cusps are low and bunoid, and the trigonid LE 443 (left astragalus of an immature in- is little higher than the talonid. The cristid dividual), and AMNH 55390 (left calca- obliqua is low and its anterior attachment neum); see Cifelli (1983). with the trigonid is buccal to the metaconid, HORIZON AND LOCALITY: Riochican fissure unlike the more advanced but contempora- fills in the Itaborai Formation, near Sao Jose neous litoptems of the Anisolambda group, de Itaboral, Rio de Janeiro, Brazil. in which attachment is to the base of the DIAGNOSIS: Sole known species of the ge- metaconid. There is no centroconid on the nus. cristid obliqua. The hypoconid, entoconid, DESCRIPTION: This species was briefly de- and hypoconulid are distinct and unjoined, 1983 CIFELLI: CONDYLARTHRA 5 the last cusp projecting slightly on M1 2 and lated right P4 and left P4 (MNRJ 1841V and more strongly on M3. Unlike Didolodus, the 1842V, respectively), and several DNPM entoconid is not multicuspate. Faint basal specimens, including a left P3, left M2, left cingula are generally present anteriorly and M3, and left M3 (casts, AMNH 109552). The on the buccal and lingual flanks of the hy- resultant composite, given in figures 1H and poconulid. 2G, is a distinctive form with P4 greatly en- Measurements are given by Paula Couto larged and a large molar size increase from (1952a, p. 362). first to third. These trends are evident also A confounding factor in the distinction of in Paulacoutoia protocenica, which on other phyletic units among didolodontids and grounds appears to be a close relative of La- primitive litoptems is the fact that these an- megoia (Cifelli, 1983). imals are relatively rare, and most ofthe type With the referral ofthe two Brazilian species and referred material correspondingly con- to new genera, the genus Ernestokokenia in- sists only of single (often fragmentary) cheek cludes five species represented by isolated teeth which are often noncomparable as to cheek teeth from the Riochican and Casa- jaw or locus within the jaw. Such isolated mayoran of Patagonia (Simpson, 1948) and teeth, which vary one from the next, present a number of specimens not yet given specific few characteristics upon which a hypothesis designations (Simpson, 1964, 1967). These of relationships may be based. No certain fragmentary remains represent a closely knit associations can be made until more com- and very primitive group ofungulates central plete specimens are brought to light; how- to interpreting didolodont and litoptern re- ever, it does seem likely that some of these lationships, but even the collocation of ma- enigmatic and fragmentary forms-many terials currently assigned to various species from the same localities-will prove to be is of little help because so little of the den- synonymous, and in the meantime certain tition is known. I propose to unite the Casa- reasonable associations must be made as a mayoran E. patagonica and E. trigonalis, a hypothetical basis upon which to proceed un- synonymy considered probable by Simpson til more adequate materials are available. The (1948, p. 111), and believe that the type and associations proposed below, some of which only specimen of the former species (MACN are inferred synonymies and others merely 10687) is a right M2 and that of the latter collocations of species very similar in known (MACN 10688) is a left M'. The teeth are morphology, are predicated on extrapolation structurally similar, and the difference in size from similar species in which the dentition may be explained by tooth locus, based on is more complete, relative size and predicted comparison with Didolodus multicuspis. Two morphology, and geographic and temporal other M2s from another Casamayoran local- distribution. ity (MHN Toumouer Coll. no. 6; casts, Among the species here considered to per- AMNH 39593), described by Gaudry (1904) tain to the Didolodontidae, Didolodus mul- as Protogonia Euprotogonia patagonica and ticuspis (see Simpson, 1948, 1970) and Pau- by Simpson (1964, 1967) as Ernestokokenia lacoutoia protocenica are known from sp. are virtually identical with MACN 10687. relatively complete upper and lower denti- Various described lower teeth, almost cer- tions. The type of the large Itaborai species tainly pertaining to a closely related animal, Lamegoia conodonta, MNRJ 1463V, is a left are associated here with some doubt. Ernes- lower molar referred to with some question tokokenia nitida, the genotype, is known only by Paula Couto (1952a) as an M2 but con- by two lower molars in addition to the type sidered here to be M1. As mentioned else- (MACN 10735), a left M2 and broken M3. As where (Cifelli, 1983), in addition to the left noted by Simpson (1948), this species evi- MI and right M2 (MNRJ 1464V and 1465V, dently derives from a lower horizon than E. respectively) included in the original hypo- patagonica and E. trigonalis, and cannot be digm, I would refer to this species also the compared directly with them; nonetheless, specimens described by Paula Couto (1952a, they are the only known lower molars ap- p. 364) under Didolodontidae, indet., iso- propriate for this form, and it seems doubtful 6 AMERICAN MUSEUM NOVITATES NO. 2772

in any case that any grouping of the specimens under consideration can be referred to as species in a biologically or paleontologi- cally meaningful sense. MHN Toumouer CuCuC . . Coll. no. 7 (casts, AMNH 39591), unassociated teeth considered here to be left M2 and , c- C- C C-- M3, were also described by Gaudry and by _CldCuCuuCZ Y .-AI0 Simpson, and may belong here although they differ slightly from the type of E. nitida in cn CA ( VI Cl ,-c (--c -4 uCu e0. proportions and size. Other specimens from the Cerro Negro Casamayoran collection, (A.0 0 0 MHN Tournouer Coll. no. 8 (casts, AMNH 2 t-.C- -.-b- e! 39592) include a right P4 and M1. As with u the upper molars noted above, M1 is smaller - Cu ncu C- 0.-- than the presumed M2s. P4 iS similar to that co ofDidolodus, but the single posterolabial cusp

0 4- is much higher and the tooth is unusual among .04 . C-. -. c-* c'.2 "C13~~~~0 all didolodonts in its large size, being sub- .0 equal to M1 (assuming that both teeth from Cn this locality belong to the same species). The C Y.0. 0. two described Riochican species of Ernes- tokokenia (Simpson, 1935b) are distinct. E. .0 40 CZC Ccu CZ CZ CIS a0.a.0. L yirunhor, a small form with simple molars, "0I= is represented by a referred M3 (cast, AMNH cn 0 0 0 0 ct0A 27896; original in Feruglio coll.) in addition ¢" 0 Cu0.0.cucuc , cZ c- c cuuIS co I-vPCC to the type M2-3 (AMNH 27896), and E. H chaishoer, a larger species, by a referred M2? 0 *_O (cast, AMNH 27892a; original in Feruglio *;Pr" Coll.; see Simpson, 1935b, 1948). For the ba Y cu n C-- C c-uD of comparison and not in a formal ._ purposes 0 taxonomic sense, three Patagonian Ernesto- cu kokenia types are considered: E. spp. (Cas- Cu Cu vu Y.CZ CZ amayoran, including the genotype E. nitida, 0 Cl the probably synonymous E. patagonica and CZCu 0 .0= C.-l CZ E. trigonalis, and the E. sp. Cerro Negro ma- of .0 C.- C) t) C)U terial) and E. yirunhor and E. chaishoer C.- the Riochican. The artificial assemblage of Casamayoran materials made here may later be found to be unjustified and in fact various Cu to this form E specimens now assigned genus CIS may belong to distinctly different animals, but this arrangement is in accord with evidence now available. A composite ofthe Cas- .-0 amayoran Ernestokokenia spp. is given in figures 1 and 2B. Other forms currently referred to the Di- (ZE Q) e4o dolodontidae and available for comparison include Paulogervaisia mamma, represented by M2-3 and P3-M2 (MACN 10719), Enneo- conus parvidens, known only by M3 (MACN 10726), and Asmithwoodwardia scotti, of which nearly the entire dentition is repre- 1983 CIFELLI: CONDYLARTHRA 7

A Asmithwoodwardia scotti E Paulogervaisia mamma

B Ernestokokenia spp. F Enneoconus parvidens

C Ernestokokenia yirunhor G Paulacoutoia protocenica -N

D Didolodus multicuspis H Lamegoia conodonta

FIG. 1. Schematic comparative left upper dentitions of Asmithwoodwardia scotti and representative Didolodontidae, standardized to length of M2, with associations as discussed in text. A, DGM 358M (type). B, MI, MACN A10688 (type of Ernestokokenia trigonalis); M2, MACN A10687 (type of E. patagonica; reversed). C, AMNH 27896 (cast, original in Feruglio collection), D, MACN A10690 (type), with details added from AMNH 109617 and 28474. E, MACN A10719 (type; M2 reversed). F, MACN A10726 (type; reversed). G, P3, AMNH 49820; P4-M3, AMNH 109553 (cast, original in DNPM). H, P4, MNRJ 1842V; M', MNRJ 1464V; M2, MNRJ 1465V (reversed); M3, AMNH 109552 (cast, original in DNPM). sented by DGM 358M. This last species does didolodontid species and Asmithwoodwardia appear to be referable to the Patagonian ge- scotti is compared in table 1, and compara- nus, although the Casamayoran genotype A. tive series are illustrated in figures 1 and 2. subtrigona is known only from the type upper Many ofthe differences among the forms are molar and referred isolated upper and lower slight, gradational, or variable within a single molars (additional undescribed cheek teeth species, and even with the pooling of various probably pertaining to this genus are known materials into units for comparison there re- from the Riochican ofPatagonia). As alluded main fundamental gaps in knowledge. P3, a to above, the familial and ordinal reference simple tooth in Paulacoutoia, bears a distinct of Asmithwoodwardia is highly uncertain; it metacone in both Didolodus and Asmith- is here removed with some doubt to the Li- woodwardia; it is thus difficult to determine topterna. whether the cusp was present in a common Dental morphology of the better known ancestor of all three genera (a metacone is 8 AMERICAN MUSEUM NOVITATES NO. 2772 A Asmithwoodwardia(:~~~~~~~~~~~~~~scotti E Paulogervaisia mamma

B Ernestokokenia spp.

C Ernestokokenia yirunhor F Paulacoutoia protocenica ,i

D Didolodus multicuspis G Lamegoia conodonta

FIG. 2. Schematic comparative right lower dentitions ofAsmithwoodwardia scotti and representative Didolodontidae, standardized to length ofM2, with associations as discussed in the text. A, DGM 358M (type). B, P4-M1, AMNH 39593 (casts, originals in MHN); M2-3, MACN A10735 (type ofErnestokokenia nitida; reversed). C, AMNH 28539 (type). D, MACN A10690 (type), with details added from AMNH 28475, 109618. E, MACN A10719 (type). F, P4, AMNH 49822; Ml, AMNH 49823; M2-3, AMNH 49827 (all reversed). G, P3, AMNH 109552 (reversed; cast, original in DNPM); P4, MNRJ 1841V; M1, MNRJ 1463V (reversed); M2-3, AMNH 109552 (reversed; cast, original in DNPM).

also present on P2 of Didolodus; Simpson, bearing two cusps, in Lamegoia and Paula- 1970, but this tooth is unknown in other di- coutoia, as in Asmithwoodwardia and the dolodonts and early litoptems). As Simpson primitive litoptems. In Didolodus and Pau- (1948) observed, P3 of Didolodus is enlarged, logervaisia the cristid obliqua extends almost inflated, and complicated (a protoconule is directly posteriad from the trigonid, termi- variably present), so that the difference be- nating in a median cusp or heel; in Ernes- tween it and P4 iS slight. All forms have meta- tokokenia spp. the cusp is at the postero- cone on P4; a metaconule is present only in buccal angle of the tooth and a rudimentary Didolodus, and is variable in that genus. In talonid basin is formed. the lower premolar series, Paulogervaisia, Lamegoia and Paulacoutoia are distinctive Didolodus, and Lamegoia are characterized in the enlargement of the molar series from by an abbreviated P4 trigonid (also true, where first to third. Upper molar protosylar cusps known, of P3), in which that portion anterior vary individually and according to tooth lo- to the transversely aligned protoconid and cus and wear; they appear to be absent in metaconid is reduced, with loss of the para- Asmithwoodwardia, Paulacoutoia, and sev- conid. The talonid of P4 is more molarized, eral Ernestokokenia species. Upper molars of 1983 CIFELLI: CONDYLARTHRA 9

Didolodus, Paulogervaisia, and Enneoconus bear mesostyles, presumably an acquisition convergent with litopterns. All didolodonts have a hypocone on M3, a feature lacking in Asmithwoodwardia; in Ernestokokenia yirunhor (M3 is not known in any Casamayoran representative ofthe genus), Didolodus, Pau- logervaisia, and Enneoconus, the M3 hypocone is strongly developed and lingual in position, giving the tooth a trapezoidal profile. In the lower molars, Lamegoia and Paula- coutoia seem primitive in that the paraconids are more distinct and less fused to the meta- conids than in other forms (the large paraconid of Lamegoia lower molars may be a secondary development; see below). An ac- FIG. 3. Proposed relationships among the bet- cessory ridgelike cuspule anterior to the ento- ter known Didolodontidae. Numbers in parenthe- conid is present on lower molars of Ernes- ses refer to advanced morphological characters tokokenia and Didolodus; available specimens listed in table 1 and discussed in the text. of Paulogervaisia are too worn to determine the condition in this genus. lar paraconid and in the well molarized P4. A hypothesis of relationships of these di- The position of Megadolodus molariformis dolodontid genera is given in figure 3 and a (McKenna, 1956) is not clear. The expanded revised classification is listed in table 2. Two P4 talonid and relatively large size differential major groups are suggested: Lamegoia/Pau- between M1 and the socket for M2 suggest lacoutoia, characterized by a molarized P4 possible affinity with the Lamegoia group, talonid and sequential size increase in the but the molar paraconid is completely fused molar series, and all others, characterized by with the metaconid, a similarity to Asmith- a trapezoidal M3 (with lingually developed woodwardia, as McKenna observed. hypocone) and lower molar anterior accessory cuspule. Many of the features described TABLE 2 above are ignored here because their distri- Classification of the South American butions are so inadequately known. Thus it Condylarthra may later be found that premolar pattern in the Didolodus group (including enlargement Superfamily Didolodontoidea Scott, 1913, NEW RANK of P3, trigonid reduction of P31) is more Family Didolodontidae Scott, 1913 (=Didolodidae widely distributed. Possible or probable con- Scott, 1913, p. 489; Didolodontidae in Simpson, vergences include the presence of a P4 meta- 1934, p. 6) conule and reduction of the P4 trigonid in Didolodus Ameghino, 1897, p. 437 Lamegoia, characters found in Didolodus Paulogervaisia Ameghino, 1901, p. 389 (and, for the latter, in Paulogervaisia) but not Enneoconus Ameghino, 1901, p. 378 Ernestokokenia Ameghino, 1901, p. 380 Paulacoutoia. Both groups, that is the family Archaeohyracotherium Ameghino, 1906, p. 307 as a whole, have an M3 hypocone and some- Lamegoia Paula Couto, 1952a, p. 363 P4 what inflated para- and protocones; the Megadolodus McKenna, 1956, p. 737 former is lacking in Asmithwoodwardia and Paulacoutoia, NEW the latter only partially developed in that ge- Family Sparnotheriodontidae Soria, 1980, p. 194 nus. In these and some other respects, As- Victorlemoinea Ameghino, 1901, p. 383 (possibly mithwoodwardia is perhaps the most primi- but not demonstrably a senior synonym of tive ungulate known from the South American Sparnotheriodon Soria, 1980) continent, but it is obviously derived and Phoradiadus Simpson, Minoprio, and Patterson, probably convergent on some didolodonts in 1962, p. 249 p. 194 the connate and usually indistinct lower mo- Sparnotheriodon Soria, 1980, 10 AMERICAN MUSEUM NOVITATES NO. 2772

ORDER LITOPTERNA AMEGHINO, 1889 and metacones and with para- and meso- DIAGNOSIS: Manus and pes mesodactyl, styles, bilophodont lower molars with the with three or fewer functional digits in known cristid obliqua attaining anterior attachment forms. Astragalar body spool-like, lacking the at the base of the metaconid, and other fea- superior astragalar foramen, with salient, tures noted by Simpson (1948, pp. 1 14-115). subequal tibial and fibular crests, and with With the inclusion of the Itaborai species in trochlea extending far posteroinferiorly (dig- the order based on astragalocalcaneal mor- ital flexor groove not distinct); astragalar head phology, this dental distinction can no longer more or less semicylindrical (but somewhat be maintained, because the Brazilian forms flattened in later macraucheniids), with ma- are almost wholly primitive in the dentition. jor axis subparallel to that ofthe tibial troch- Perhaps the only features in which known lea; navicular facet extending onto superior didolodonts differ from primitive litopterns and inferior surfaces of the head, cuboid and are the slightly more molarized P4 talonid medial collateral ligament facets lacking; as- (?converged on by the Lamegoia group) and tragalar sustentacular facet anteroposteriorly the lack of an M3 hypocone of members in elongate, ectal facet deeply concave and with the latter group, but this merely emphasizes pronounced outward orientation. Necks of the dental primitiveness of these early litop- astragalus and calcaneum primitively elon- terns (an M3 hypocone does develop in later gate, the latter bearing a strongly oblique cu- forms, the Macraucheniidae and Adianthi- boid facet; sustentacular facet of calcaneum dae). Pertinence ofthe Itaborai species to the anteroposteriorly concave and following same Litopterna may be questioned by those ad- curvature as anterior part of ectal facet. Car- hering to a concept of the order based on pus of known forms with enlarged magnum more derived forms, but there are no strik- and medial displacement of distal elements, ingly didolodont characters suggesting ref- so that lunar-unciform contact is lost but cu- erence to that family, and their primitiveness neiform-magnum contact is achieved. Den- in the dentition is thus permissive ofthe view tition primitive and complete in early forms; advocated here and is not particularly sugges- upper molars later becoming quadrate with tive in any case. Comparisons are more fully crescentic paracone and metacone and with detailed in the discussion below. para- and mesostyles; lower molars becoming bicrescentic. FAMILY PROTOLIPTERNIDAE, NEW As with the didolodonts, the taxonomic TYPE: Protolipterna, new genus. history of the various early Tertiary genera DISTRIBUTION: Riochican to Casamayoran, now placed here has been very fully reviewed South America. by Simpson (1948). Other litoptern features DIAGNOSIS: Small, almost wholly primitive are listed in Simpson's definition ofthe order, litopterns. Lower canine enlarged; a slight di- but a revised diagnosis is given here in order astema between it and P1 is present in one to permit the inclusion of two small and oth- specialized genus. Upper and lower cheek erwise primitive species from Itaborai which teeth low crowned, with bunoid cusps. P3-4 bear the diagnostic litoptern tarsal special- (inflated in one genus) with small protostylar izations noted above (Cifelli, 1983). The dif- cingulum and lacking mesostyle and hypo- ficulty of dentally distinguishing early litop- cone; metacone present but barely distin- terns from their contemporaneous allies, the guishable from paracone. Upper molars didolodonts, has long been noted; Simpson simple and didolodont-like except that a hy- (1948, p. 119) remarked that, "the litopterns pocone is lacking on M3. Posterior lower pre- appear to represent the direct further evolu- molars becoming inflated in one genus; P3 tion ofthe Condylarthra, continuing in South simple; P4 distinguished from most Didolo- America after they became extinct in the rest dontidae in having a molarized, multicuspid ofthe world." Nonetheless, the early Tertiary talonid. Lower molars didolodont-like, lack- species included by Simpson in the Litopter- ing the bicrescentic pattern seen in more ad- na are distinct from the Didolodontidae in vanced litopterns; paraconid primitively having upper molars with crescentic para- connate with but distinct from metaconid, 1 983 CIFELLI: CONDYLARTHRA I1I cristid obliqua attaching to trigonid some- is otherwise molarized (the talonid is bicus- what labial to the latter cusp. pid), and the presence of these cusps among most other species considered here and in PROTOLIPTERNA, NEW GENUS most primitive North American ungulates TYPE: Protolipterna ellipsodontoides, new strongly suggests that their loss is secondary species. and is probably related to specialization (in- DISTRIBUTION: Riochican, Brazil. flation) of the premolar series in general. DIAGNOSIS: Very small, generally primitive litoptem, but with transversely narrowed Protolipterna ellipsodontoides, symphysis and enlarged, well-differentiated new species lower canines with a slight diastema between Figures 4, 5C, 6C C and P,. Posterior lower premolars inflated; TYPE: DNPM LE444A (cast, AMNH P3 simple, P4 lacking paraconid and meta- 109542), mandible with right C, alveoli of conid but with bicuspid talonid. Lower molar right and left P1I2, and with right and left P3- paraconid tightly appressed to metaconid and M3; associated right maxillary fragment with sometimes indistinct. P4 lacking conules and M1-3. with enlarged, inflated protocone subequal to HYPODIGM: The type, and DNPM LE444B and transversely aligned with the paracone; (cast, AMNH 109543), right mandible with metacone faintly indicated on the posterior alveolus of P2 and with P3-M3; DNPM slope of the latter cusp and close to its apex. LE444C (cast, AMNH 109544), right man- M3 with stronger postcingulum than in dible with alveoli of P1-2 and with P3-M3; Asmithwoodwardia scotti (a slightly smaller DNPM LE444D (cast, AMNH 109545), left species), but lacking a hypocone. mandible with P4-M3; DNPM LE444E (cast, ETYMOLOGY: In reference to the primitive- AMNH 109546), right mandible with M1l-3; ness of this form within the Litopterna. DNPM LE444F (cast, AMNH 109547), frag- Specimens of Protolipterna ellipsodon- ment of left maxilla with P4-M3; DNPM toides are extremely abundant in the DNPM LE444G (cast, AMNH 109548), mandibular Campos/Price collection of 1968, mentioned symphysis with broken canines and alveoli below, and are not surely known from other of right and left PI-2 and part of alveolus of lots collected at Itaborai. Material from this P3; DNPM LE444H (cast, AMNH 109549), sample forms a large and important collec- fragment of left maxilla with Ml-3; DNPM tion since it is known to derive from a single LE4441 (cast, AMNH 109550), fragment of fissure. Reference of this species to the Li- left maxilla with P4-M3; DNPM LE445, un- toptema is based on tarsal structure, as pre- associated calcaneum and astragalus. In ad- sented elsewhere (Cifelli, 1983). Protolipter- dition, there are approximately 430 other na, characterized by the generally primitive specimens of this species, consisting of iso- molar structure common to didolodonts and lated upper and lower cheek teeth, dentulous litopterns, is distinct from all known genera jaw fragments, and astragali and calcanea, in in the specializations of the premolars and the DNPM collections. anterior dentition, as listed in the generic di- HORIZON AND LOCALITY: Riochican fissure agnosis. The large, subequal and transversely fills in the Itaborai Formation, near Sao Jose aligned P4 paracone and protocone are rem- de Itaborai, Rio de Janeiro, Brazil. All spec- iniscent ofadvanced didolodonts such as Di- imens were collected by Diogenes Campos dolodus, but the development of P4 in that and Llewelyn Price in 1968. genus is otherwise differently specialized. Di- DIAGNOSIS: As for the genus. dolodus lacks the inflation ofthe main cusps, ETYMOLOGY: In allusion to the specialized and possesses conules and the cristae con- premolars, which strikingly resemble those necting anterior and posterior faces of pro- of small mioclaenine condylarths such as El- tocone to paracone. Certain features of Pro- lipsodon. tolipterna ellipsodontoides, such as the lack The lower incisors and their alveoli are un- of a paraconid and metaconid on P4, might known. A small ?lower incisor found with the be considered as truly primitive, but the tooth type specimen may actually belong to it, but 12 AMERICAN MUSEUM NOVITATES NO. 2772

C A

.i' 'A D[) _i \-,'.'

Pr."

B ' ...... - .' -k ..I .. joAi .i: -,.4 4,7 .1 IV j., \ I \ " ,. ,k. :\ ' '. j

cm E .- ) - .X'*Ar. s 111-U,1 (I F

4t.00-- .-$LI'""'.WI' I

FIG. 4. Protolipterna ellipsodontoides. A-D, DNPM LE444A (type). A, B, right lower (C, P3-M3) dentition in occlusal (A) and labial (B) views. C, D, right upper dentition (M'-3) in occlusal (C) and lingual (D) views. E, mandibular symphysis, DNPM LE444G. F, left mandibular ramus (DNPM LE444D) in lateral view (P4-M3 are in place on the specimen but have been omitted from the drawing). Measurements are given in table 3. the superior border ofthe mandible is broken also inflated and lacks both paraconid and in this region and the socket is not visible. metaconid. A trigonid as such is thus lacking, The canine ofthe type is very long and curved, the anterior part ofthe tooth consisting solely even when allowance is made for breakage of a large, bulbous protoconid. A ridge de- and the missing portions ofthe alveolar bor- scends the posterolabial margin of the pro- der of the mandible. As can be seen from toconid, terminating in a cusp at the postero- DNPM LE444G, the canines extended well labial angle of the tooth. Directly lingual to into the symphysis and converged toward the this and connected to it by a ridge is a second midline posteroinferiorly. A diastema sep- talonid cusp. P4 was evidently an important arated the canine from P1 (not known), which crushing tooth, since specimens indicate that had a large, single root; a small gap, variable it was rapidly reduced by apical wear on the in length and occasionally absent, as in protoconid. Ml 3 show a slight increase in DNPM LE444C, separated P1 from the two size from first to third. The cusps are bunoid rooted P2 (also unknown). The remainder of and the cristae connecting them are low and the cheek teeth are in continuous series. P3 rounded. A faint, low paracristid is present; is a simple tooth with a single bulbous and the paraconid lies on the anterior face of the inflated cusp. A faint crest along the postero- metaconid and in some specimens is com- median edge of this cusp descends to a small pletely fused to it and indistinguishable. The basal swelling at the heel of the tooth. P4 iS cristid obliqua extends posterobuccally from 1983 CIFELLI: CONDYLARTHRA 13

TABLE 3 Dental Measurements of Protolipterna ellipsodontoides Type and Hypodigma LE444A LE444B LE444C LE444E LE444F LE444H LE444I P3 L 4.10 3.63 3.71 - - - - W 2.49 2.34 2.35 - - - - P4 L 3.96 3.59 4.00 - - - - W 2.90 2.57 2.72 - - - - Ml L 3.88 3.78 3.81 3.80 - - - W 3.10 3.08 3.06 3.09 - - - M2 L 4.29 4.04 4.13 3.94 - - - W 3.74 3.58 3.49 3.49 - - - M3 L 5.04 4.50 4.68 4.81 - - - W 3.61 3.26 3.43 3.39 - - - P4 L - - - - 3.34 - 3.18 W - - - - 4.71 - 4.16 M' L - - - - 3.61 3.88 3.34 W - - - - 4.95 4.85 4.37 M2 L 4.67 - - - 4.33 4.38 3.94 W 5.91 - - - 5.72 5.87 5.04 M3 L 3.92 - - - 3.79 3.73 3.57 W 5.48 - - - 5.04 5.26 4.62 Size Variation of M2 Low High Range Mean S M2 L 3.7 4.5 .8 4.08 .2211 (n = 18) W 3.2 3.7 .5 3.57 M2 L 4.0 4.8 .8 4.40 (n= 9) W 5.3 6.3 1.0 5.80 a Measurements taken with a shop scope from epoxy casts. b Sample picked at random from additional DNPM specimens with M2 or M2 still in place in thejaw; measurements taken with dial calipers. the median part of the trigonid; the talonid which it is subequal in size. The presence of cusps are all visible although the hypoconulid a metacone is faintly indicated on DNPM and entoconid are close and connected by a LE444I, where it is represented by a small ridge, so that distinction of the latter cusp is swelling on the posterior slope of the para- usually faint. cone, close to the apex of that cusp. Conules The mandibular symphysis is elongate and and ridges connecting protocone to paracone narrow, being somewhat compressed trans- are entirely lacking. The antero- and poster- versely. The corpus ofthe mandible is slender obuccal corners of P4 are marked by small and shallow as in mioclaenines and unlike parastylar and metastylar spurs, respectively, primitive arctocyonids such as Oxyprimus or and a weak cingulum is present anteriorly, Ragnarok. The coronoid process is rather long posteriorly, and buccally. The upper molars but not hooklike, and the mandibular con- are of the common primitive litoptern and dyle is noticeably elevated above the level of didolodont type. The external cingulum is the tooth row. The angle is well developed incomplete around the base of the paracone, and projects somewhat posteriorly and in- and both anterior and posterior cingula are feriorly. present, with a small protostylar cusp (vari- The anterior upper dentition is not known. ably absent) placed just anterolingual to the The protocone ofP4 is enlarged, bulbous, and protoconule on the former. The lingual cin- transversely aligned with the protocone, with gulum is incomplete, unlike the condition 14 AMERICAN MUSEUM NOVITATES NO. 2772 generally pertaining to Ernestokokenia. There with alveoli of 11-2, root of 13, alveoli of is no M3 hypocone, but the cingulum is slight- P1 2, and with the C and P3-M3 present. ly enlarged in the posterolingual region. HYPODIGM: As given by Paula Couto Measurements are given in table 3. (1952a, pp. 362-363), and DNPM LE447 (casts, AMNH 109555), unassociated left as- MIGUELSORIA, NEW GENUS tragalus and calcaneum; AMNH 55396, right calcaneum; and AMNH 55394, right astrag- TYPE: Ernestokokenia parayirunhor Paula alus. In addition, there are approximately 135 Couto, 1952a. other known specimens pertaining to this DISTRIBUTION: Riochican, Brazil. species, consisting ofjaw fragments and iso- DIAGNOSIS: Small, almost wholly primitive lated teeth, in the DNPM collections. litoptern: upper molars lacking para-, meso- HORIZON AND LOCALITY: Riochican fissure and metastyles; protostylar cusp small on fills in the Itaborai Formation, near Sao Jose Ml-2 and lacking on M3. Anterior and pos- de Itaborai, Rio de Janeiro, Brazil. terior cingula narrow and slightly crenulated, DIAGNOSIS: As for the genus. external cingula usually continuous. M3 hy- The anterior upper dentition is not yet pocone usually lacking. Lower canine well known. The upper molars are most compa- differentiated, but diastemata lacking. P3 rable to those ofAsmithwoodwardia scotti and simple, uninflated, and with small heel; P4 a new species, described below, differing from uninflated, with complete and molariform them only in size, development of cingula, trigonid and multicuspid talonid. Lower mo- and the possible variable presence of a hy- lars with connate but distinct paraconid. pocone on M3. M'-2 do also closely resemble ETYMOLOGY: For my colleague, Miguel F. those of Patagonian Ernestokokenia species, Soria, in recognition of his contributions to as Paula Couto observed, which is to say that knowledge of the Litopterna. they are primitive for the entire group of un- This species is dentally similar to Ernes- gulates under consideration. The bunoid tokokenia, to which Paula Couto originally paracone, metacone, and protocone are ar- referred it. The tarsus, however, is fully li- ranged in a triangular pattern; the protoco- toptern in character (Cifelli, 1983), and Er- nule and metaconule are unreduced and re- nestokokenia is dentally closer to true dido- tain their primitive connections and positions. lodonts for which the ankle is differently The hypocone is well developed and sepa- specialized, so that generic and even ordinal rated from the protocone by a sulcus on the identity seems very improbable. Miguelsoria lingual surface of the tooth. The posterior, differs from Ernestokokenia in the triangular anterior, and labial cingula are continuous M3, lacking a hypocone, much more molar- (except occasionally at the base of the para- ized P4, more distinct lower molar paraconid, cone) and slightly crenulated; a small pro- greater joining of lower molar talonid cusps, tostylar cusp is present on the cingulum just and absence of an anterior entoconid acces- anterolingual to the protoconule. M2 is sim- sory cusp. Similarity is greatest to Asmith- ilar to M', but larger. M3, smaller, more woodwardia and Protolipterna ellipsodon- transverse, and more triangular in shape than toides, described above, but Miguelsoria the preceding molars, is unlike Ernestoko- differs from the former in the more molarized kenia in that it is completely encircled by a posterior lower premolars and more distinct weakly crenulated cingulum except, appar- lower molar paraconid, and lacks the special- ently, in DGM 305M, not seen by me (Paula ized anterior dentition of the latter. Couto, 1952a, pl. 32, fig. 5). This specimen differs also in having what appears to be a Miguelsoria parayirunhor small hypocone on M3, a feature not even Figures SB, 6B foreshadowed on the numerous other specimens available. Possibly, it represents a dis- Ernestokokenia parayirunhor Paula Couto, 1952a, tinct species of the same size, perhaps allied p. 362, 1978b, p. 212; Cifelli, 1983, p. 14. to Patagonian Ernestokokenia (although TYPE: MNRJ 1468V, right hemimandible DGM 305M does not have the characteristic 1 983 CIFELLI: CONDYLARTHRA 15 trapezoidal profile of M3 in that genus) as developed. P4 molarized, with lophate tal- maintained by Paula Couto, but there is no onid; P4-M3 with cristid obliqua achieving a further evidence to warrant separation ofma- lingual trigonid attachment (to the base of terials allocated to M. parayirunhor into two the metaconid); lower molars bicrescentic, or species, and in fact all available specimens nearly so, with lophate trigonids and talonids. are otherwise so closely similar in size and morphology that a single species is indicated. FAMILY PROTEROTHERIIDAE Despite this apparent anomaly, I consider the AMEGHINO, 1887 condition of M3 in M. parayirunhor to be as given in the above diagnosis and descripfion. DIAGNOSIS: P3 molarized, with well-devel- The lower dentition is most completely oped metacone in a triangular arrangement known in the genoholotype, MNRJ 1468V. with the paracone and protocone, and with The canine is strong and curved, and prob- mesostyle, protoconule, and metaconule ably would have been well differentiated, al- present. P3 molarized, with crescentic trigo- though the incisors and P1 are not known. P3 nid and talonid lophs. Protocone and hypo- is a simple, unicuspid tooth with a small tal- cone of Ml-2 joined at or nearly to their api- onid heel. P4 has a high, lingually placed para- ces, forming a lingual continuity with the conid (unlike Ernestokokenia or other di- protoloph and postcingulum; protocone to dolodonts) arranged in a triangular pattern metaconule connection lost on all upper mo- with the well-differentiated protoconid and lars. metaconid, and the trigonid is thus complete and molarized. A low cristid obliqua extends SUBFAMILY ANISOLAMBDINAE, NEW posterobuccally from the median part of the TYPE: Anisolambda Ameghino, 1901. metaconid, forming the buccal boundary of DISTRIBUTION: Riochican to Deseadan, a shallow talonid basin. There is a well- South America. marked cusp at the posterobuccal angle ofthe DIAGNOSIS: Generally primitive protero- talonid; the posterior margin of the talonid theres with cheek teeth of moderate crown basin is defined by a ridge which includes height. Paralophid of lower molars well de- another (variably two) small cusp. The lower veloped and strongly marked (but disrupted molars increase slightly in size from first to in one genus), terminating posterolingually in third, and all bear a weak anterior cingulum. a large, distinct paraconid. The paraconid of the molars is appressed to but incompletely fused with the metaconid. PARANISOLAMBDA, NEW GENUS The cristid obliqua extends posterobuccally from the median part of the metacristid to TYPE: Anisolambda prodromus Paula Cou- the hypoconid. Low ridges connect the hy- to, 1952a. poconid to hypoconulid and hypoconulid to DISTRIBUTION: Riochican, Brazil. entoconid, in contrast to the more distinctly DIAGNOSIS: Similar to Anisolambda (in- separate cusps seen in Ernestokokenia, Di- cluding Josepholeidya), but M3 with very dolodus, and Paulacoutoia. elongate talonid and M3 correspondingly Measurements are given by Paula Couto larger than M2. Lower molars lower crowned (1952a, p. 363). than Anisolambda, paralophid more open anteriorly and not connected to paraconid; entoconid disconnected from hypoconulid SUBORDER LOPHOLIPTERNA, NEW and especially isolated on M3. Upper molar DIAGNOSIS: Dentally advanced litopterns hypocone less united to protocone and with with lophate cheek teeth. P4 molar-ized, with low ridge extending toward metaconule. enlarged metacone in a triangular arrange- ETYMOLOGY: Allusion to its similarity to ment with the paracone and protocone, and Anisolambda. with protoconule and metaconule present. P4- Most ofthe features which distinguish Par- M3 with crescentic para- and metacones; anisolambda prodromus from the Patagonian parastyle and mesostyle columnar and well species ofAnisolambda, such as a lengthened 16 AMERICAN MUSEUM NOVITATES NO. 2772

M3, paraconid and paralophid disconnected, "Ml" is more similar to and continuous with and isolated entoconid, appear to be derived the preceding deciduous teeth, and I thus be- characters ofthe Brazilian species, but A. fis- lieve it to be dM4 and the anterior teeth of sidens (genotype) and A. amel are more pro- DGM 273M correspondingly to be right and gressive in their higher crowned lower mo- left dM1-3, not p2 and dM34. The same ar- lars. A.fissidens (with which I include, at least gument pertains to the lower dentition, and in part, Josepholeidya and Ricardolydekker- the genoholotype, DGM 262M (possibly but ia), the better known Patagonian species, is not certainly belonging to the same individ- closely similar to A. amel, but presents other ual as DGM 304M) therefore probably in- progressive characters, such as a labial cin- cludes dM1 4 in full eruption, not P2, dM34, gulum and shallower buccal slope on the low- and M1. The adult dentition of Paraniso- er molars, centroconid on the cristid obliqua, lambda prodromus is therefore known only and extension of the metaconid posteriorly by M2-3 (MNRJ 1479V, DGM 310M) and into the talonid basin. These features are M1 3 (M1, 1496V; M2, AMNH 49818; M3, suggestive ofthe Deseadan Protheosodon, al- MNRJ 1859V). though some contradictions in the upper molars are apparent if reference to the Casa- DISCUSSION mayoran species of specimens currently allocated to Ricardolykkeria is correct. Simpson (1948) divided the probably valid Riochican, Casamayoran, and Mustersan litoptem genera into two groups and tenta- Paranisolambda prodromus tively assigned them to the families Mac- (Paula Couto, 1952a) raucheniidae and Proterotheriidae: Figures 5D, 6D Macraucheniidae Proterotheriidae Anisolambda prodromus Paula Couto, 1952a, p. 368; 1978b, p. 216. Victorlemoinea Josepholeidya TYPE: DGM 262M, right mandible with Ernestohaeckelia Guilielmofloweria dM1- and with M1 in the crypt. Rutimeyeria Anisolambda HYPODIGM: As given by Paula Couto Amilnedwardsia Polyacrodon (1952a, p. 368) plus AMNH 55387, right as- Xesmodon tragalus; and DNPM LE448 (casts, AMNH Heteroglyphis 109569), unassociated right astragalus and Polymorphis calcaneum, right astragalar head, and left astragalar body. Wainka HORIZON AND LOCALITY: Riochican fissure Ricardolydekkeria fills in the Itaborai Formation, near Sao Jose The tarsus probably but not certainly be- de Itaborai, Rio de Janeiro, Brazil. longing to Victorlemoinea is not that of a DIAGNOSIS: As for the genus. litoptem but is closer to that of didolodonts, ,The principal characters are listed in the such as Paulacoutoia and Lamegoia (Cifelli, generic diagnosis. The species was well de- 1983). Although Victorlemoinea (known from scribed by Paula Couto in the publications isolated teeth and jaw fragments) has some cited, but the tooth homologies of the type specializations reminiscent of the Macrau- and certain referred specimens are reinter- cheniidae (such as a connecting crest between preted here. Paula Couto considered the hypocone and metaconule on the upper mo- paratypes DGM 304M and 273M to include lars), the genus is, in any case, aberrantly spe- M' as well as several of the preceding decid- cialized. True macraucheniids, represented uous teeth, and that M2 was represented by by the Cramaucheniinae, first become well MNRJ 1479V (isolated tooth) and DGM known in the Colhuehuapian (Soria, 1981), 31 OM (maxillary fragment with M2-3). In ad- although some specimens are known from dition to numerous important details of the Deseadan. At this time they are far more structure, the supposed M' is smaller by al- primitive than Victorlemoinea (an as yet un- most 50 percent than M2. In these respects, described species from the Deseadan, early 1983 CIFELLI: CONDYLARTHRA 17

Oligocene, ofLa Salla-Luribay, Bolivia, even viser, selected Anisolambda, and this prior lacks the upper molar hypocone-metaconule usage is adopted here. crest), and as detailed further below, appear Without large samples from a single hori- to have derived from a form similar to the zon and locality and, more importantly, com- Mustersan Polymorphis. Victorlemoinea plete specimens, it is impossible to surely seems to be allied to Phoradiadus (Soria, 1981, group the teeth under consideration into rec- and personal observ.) from the peculiar ?early ognizable species, although the types are dis- Deseadan Divisadero Largo fauna (Simpson, tinct enough, because of the uncertainties of Minoprio, and Patterson, 1962), and is pos- determining tooth locus. Specimens of De- sibly referable to the Sparnotheriodontidae, seadan and later derived proterotheres and which is known by a single but complete the Deseadan Protheosodon coniferus, a very mandible from the Casamayoran ofCafnadon primitive form strikingly similar in upper and Vaca (Soria, 1980). lower dental morphology to this group, are Ernestohaeckelia, Rutimeyeria, and available for comparison. Hypothetical re- Amilnedwardsia are inadequately known. The construction of the early forms is therefore type and only specimen of Ernestohaeckelia necessarily based on species which a priori acutidens, MACN 10731 (the genoholotype, would seem to be more progressive and thus of E. aculeata, has not been located in the somewhat inappropriate models. The type of Ameghino Collection), includes three broken Protheosodon coniferus, MACN 52-211, in- upper molars and two other tooth fragments. cludes a left P4 and right p2 and M2. A referred None ofthese appear to be litoptern, and the specimen, FMNH P13418, is a skull with presence of a metaconule-derived crest ex- right partial P2 and P3-M3, and left M2. tending into the trigon basin, with some elab- Loomis (1914) referred to this species the oration and connection to the ectoloph wall, mandible ACM 3001, with complete denti- is suggestive instead of a primitive henricos- tion and from the same locality (Cabeza Blan- borniid notoungulate. The same holds true ca) as the other specimens. Although not of the closely similar isolated molars de- based on direct association ofupper and low- scribed and figured by Ameghino (1901) as er dentitions, this assignment has been ac- Rutimeyeria conulifera Amilnedwardsia cepted by subsequent workers (e.g., Simpson, brevicula, MACN 12013 and 12011, respec- 1948) and is in accord with information now tively. It is ofinterest to note that a completed at hand; it is therefore assumed here. Other metaloph is lacking in these specimens, as relatively well known Oligocene litopterns Simpson (1948) observed, and if notoungu- available for comparison include Prolica- late they would be very primitive indeed phrium specillatum and Cramauchenia nor- (?perhaps they are deciduous teeth). malis. Ofthe Casamayoran genera maintained as Simpson's Ca-nadon Vaca sample of "Jo- valid and placed by Simpson (1948) in the sepholeidya" seems to represent one species, Proterotheriidae, Anisolambda is known from and his reconstructed molar series (Simpson, lower molars and Josepholeidya, Ricardoly- 1948, fig. 39) appears to be correct, although dekkeria, and Guilielmofloweria from iso- AMNH 28516 may be an M2 of a small in- lated upper cheek teeth. Simpson suggested dividual rather than an M' as figured. The that Anisolambda would probably be found largest tooth of the sample, AMNH 28518, to be synonymous with Josepholeidya or Ri- is 10.8 -percent longer than the smallest, and cardolydekkeria or both. The evidence pro- this is well within the range for M2 vs. M' of vided by remains of Paranisolambda pro- the Deseadan genera, and less than that for dromus is sufficient to consider the genera the M2 percentage size difference between Anisolambda and Josepholeidya synony- Anisolambda fissidens and A. amel from the mous, as Paula Couto (1952a) established. same locality. The Caniadon Vaca "Joseph- The names were published simultaneously by oleidya" teeth correspond well to MACN Ameghino (1901). Odreman Rivas (1969, p. 10682, the type of "J." deculca, which falls 58) employed Josepholeidya as the senior into the lower part of the Caniadon Vaca size name, but Paula Couto (1952a), the first re- range and is therefore probably M'. The ge- 18 AMERICAN MUSEUM NOVITATES NO. 2772 noholotype, of "J." adunca (MACN 10702), I follow the former course and believe the seems to me to be M3. It differs from M3s of two genera to be synonymous. It does seem the Cafiadon Vaca sample, AMNH 28512 curious that the premolars, especially P3, and 28513, in being somewhat higher would be so fully molarized at such an early crowned, more anteroposteriorly com- date, but this is an assumption not predicated pressed, and in having the parastyle and me- upon factual knowledge. These teeth entirely sostyle very fully developed. No other known lack a hypocone, presenting only a postcin- teeth are of this type. Three isolated molars gulum in the hypoconal region. This, despite from the Casamayoran ofCerro Negro, MHN their otherwise advanced aspect, would make Tournouer Coll. no. 9 (casts, AMNH 39589), them more primitive than any known un- including an M2 and two M3s, were described gulate from South America, including dido- by Gaudry (1904) and referred by Simpson lodontids and the primitive litopterns de- (1964, 1967) to "Josepholeidya" sp. M2 is scribed herein, ifthey represent molars which virtually identical with those from Cafiadon have not undergone secondary reduction. Vaca, and the M3s are very similar to but Correspondingly, the lingual roots of "Ri- distinctly larger than specimens from that cardolydekkeria" teeth are narrow, simple, sample. If the Tournouer molars belong to and premolariform, unlike the broadened, one species, that species is therefore distinct sulcus-bearing roots of other litoptern and from both "J." adunca and "J." deculca, al- didolodont molars. In other respects, these though probably closely related to the latter. teeth are ofappropriate size and morphology The Riochican specimen described by Simp- to represent posterior upper premolars, as son (1948), from Bajo de la Palangana (cast, Simpson observed, and in fact correspond AMNH 27890; original in Feruglio Collec- well to P4 of the much smaller contempora- tion), is indistinguishable from M2s I refer to neous Proectocion, although the mesostyle in "Josepholeidya" deculca, differing from "J." that genus is smaller. The AMNH Caiiadon adunca in the characters mentioned above Vaca sample of six "Ricardolydekkeria" teeth (this accords with Cabrera's, 1935, identifi- falls into two size groups, as described and cation of a specimen from the same locality figured by Simpson, and I hypothesize that as "Eulambda" deculca). It seems inadvisa- they represent P3 and P4 of Anisolambdafis- ble to refer it at this time to that species, sidens (including Josepholeidya deculca). however, because of the possible presence of Conceivably, all of these teeth are P4, and as a similar species at Cerro Negro, identical in with the Caniadon Vaca Anisolambda, rep- M2 but differing in M3. "Josepholeidya" de- resent two species differing in size. The larger cu/ca, represented by M'-3, I refer to Aniso- of these is comparable in size and mor- lambda fissidens, known from the same lo- phology to MACN 10701, the genoholotype, cality. There are no upper teeth from Cafnadon of "Ricardolydekkeria" praerupta, which Vaca or elsewhere of appropriate size to be- would hypothetically make that also a P4 of long to the smaller A. amel. "Josepholeidya" Anisolambda fissidens. "Ricardolydekkeria" adunca is known only by the type M3, and cinctula, MACN 10705, is somewhat antero- a possible third species is represented by posteriorly shorter than "R." praerupta, but M2-3. the tooth is rather worn and this appearance Simpson (1948, p. 127) presented several seems to be due to attrition with preceding lines of evidence, arguing that teeth ascribed and succeeding teeth in the series. I believe to Ricardolydekkeria represent posterior up- it also to be a synonym. This is also possibly per premolars of "Josepholeidya" (=Aniso- true of a specimen included in MHN Tour- lambda) but provisionally maintained them nouer Coll. no. 9, from Cerro Negro, de- as distinct genera because the smaller "Ri- scribed by Simpson (1964, 1967), although cardolydekkeria" teeth from Cafiadon Vaca the Cerro Negro "Josepholeidya" molars not- would probably be P3s, and it is unlikely that ed above may be distinct from "J. deculca," P3 would be so fully molariform in the Casa- and this may also apply to the "Ricardoly- mayoran; and because of a higher than pre- dekkeria" specimen from that locality if it dicted proportion of "Ricardolydekkeria" to pertains to the same species as the molars. "Josepholeidya" teeth in existing collections. The type of "R." profunda, MACN 10707, 1983 CIFELLI: CONDYLARTHRA 19 may represent another species, but it is bro- Another assemblage ofgenerally primitive ken and I did not see it. "Ricardolydekkeria" but dentally litoptern ungulates is the Mus- lunulata, MACN 55-4, is more clearly dis- tersan group including Polymorphis and its tinct from the foregoing in being less trans- supposed allies. Polymorphis itself, based on verse, lower crowned, and in having a well- the lower dentition, is known from two good developed premetaconule crista and the hemimandibles of the type species, P. lechei; metaconule more separated from the proto- the other species placed here by Simpson, P. cone. Thus, two or possibly three (or, less planus and P. alius, are based on poor ma- likely, four) species seem to be represented terials-and are inadequately characterized at by teeth referred to "Ricardolydekkeria." present, although Odreman Rivas (1969) has "Ricardolydekkeria" praerupta (including suggested that they may be synonymous and "R." cinctula), the genotype, is the best rep- possibly belong to Xesmodon. Polyacrodon is resented, and I believe the AMNH Cafiadon based on upper molars of the same age and Vaca sample to include P3 and P4 of this from the same localities (most specimens ap- species, that it is synonymous with "Jo- pear to be from Roth's Cretaceo superior de sepholeidya deculca," and that both are syn- Lago Musters, which is probably the Mus- onyms of Anisolambda fissidens. AMNH tersan level at Simpson's Cerro del Humo 28509, from the same locality, includes an locality; Simpson, 1948, p. 133), and it seems upper cheek tooth interpreted here as being likely that the two genera are synonymous in p2 of the same species. A hypothetical com- part or altogether, as Simpson suggested.4 The posite dentition is illustrated in figure SE. two species of Polyacrodon, P. ligatus (ge- Comparison with Protheosodon coniferus, the notype) and P. lanciformis, are based on up- most completely known litoptern compara- per cheek teeth, distinguished primarily by ble in molar structure, reveals strong simi- the complete lingual cingulum and larger cus- larities but also several contradictions. P2 of pule on the posthypocone cingulum of the Protheosodon is broken but in known parts latter species. Odreman Rivas (1969) main- is similar, although the protocone of the De- tained that the latter is generically distinct seadan form is more posteriorly placed. The from Polyacrodon and placed it in Decaco- most significant contrast is the lack of well- nus, to whose type (D. intricatus) it is closely developed mesostyles on P34 of Protheoso- similar, an action taken earlier by Simpson don. This may be a genuinely primitive con- (1936a). With some doubt, Simpson (1948) dition, but the salient P4 metastylar cusp, referred to P. lanciformis also a poorly pre- extending anteriorly almost to the middle of served maxillary fragment (AMNH 28492) the tooth, and a small ridge on the ectoloph with partial right P4-M2. In known respects, wall ofP3 suggest the possibility that this may the molars are quite similar to those of this be secondary. Strong P4 mesostyles are pres- species and of P. ligatus, except that the type ent in all other Deseadan and later litoptems; specimens have the hypocone more distinct a P3 mesostyle is developed in Deuterothe- from the protocone, projecting slightly at the rium, faint in Prolicaphrium, and absent in posterolingual angle of the tooth. This dif- Cramauchenia. ference appears to be due to tooth locus, and Ricardolydekkeria and Josepholeidya are I thus believe the types, AMNH 29481, and here regarded as synonyms of Anisolambda MACN 10897 to be M3, as Simpson sug- because, in sum, the three genera in question gested (1948, p. 137). Although the species are known from the same localities in ap- ofPolyacrodon and Polymorphis are not sure- proximately the same relative abundances; ly all conspecific (and in fact probably are they are of appropriate morphology and size not), they clearly are closely related, and this to represent, in at least one case, the same with more animal, based on comparison 4 A suggestion adopted, in fact, by Odreman Rivas, completely known structurally similar mam- whose statement that, "sus caracteres (i.e., of Polyacro- mals (with differences as noted); and because don ligatus, the genotype) son identicos a aquellos de there are no other known forms to which any Polymorphis lechei" (1969, p. 62) is nonetheless puzzling of the materials under consideration might because the types are from different jaws (upper and pertain. lower) and no mention of associated material is made. 20 AMERICAN MUSEUM NOVITATES NO. 2772 hypothetical generic synonymy presents a pertains to Victorlemoinea (also true of working hypothesis upon which to proceed. AMNH 28515, a smaller tooth from Calia- Polymorphis Roth, 1899 may be used as the don Vaca which is probably P3 of Victorle- senior synonym, being based on better ma- moinea). terial than Polyacrodon Roth, 1899. In the Wainka tshotshe is based on an upper cheek face of Odreman Rivas's statement to the tooth (AMNH 28505) and a referred M3 contrary, "Decaconus" intricatus and "D." (AMNH 29101) from the Riochican ofCerro lanciformis do not seem to be generically dis- Redondo (Simpson, 1935b, 1948). The lower tinct from "Polyacrodon" ligatus; I follow molar is very similar to that ofAnisolambda Simpson (1948) in considering them to be except that the anterior attachment of the synonymous. Xesmodon is too inadequately cristid obliqua is not so medial and the hy- known to be considered; the type and only poconulid is more projecting and less conical; specimen of Heteroglyphis dewoletzky (MLP the upper tooth is more distinct. The close 12-1462), which I interpret as M3, is distinct appression of paracone and metacone, with and enigmatic as to affinities. Simpson (1948, intervening mesostyle lacking, and centrally p. 140) noted that the posterior margin ofthe placed protocone with hypoconal cingulum tooth is broken and that a hypocone was feebly developed and equal to the protostylar probably present, but judging from what re- cingulum, suggest that the type specimen is mains ofthe posterior cingulum this does not an upper premolar, but it is rather unlike seem to have been the case. The protocone others pertaining to the Anisolambda group. to conule connections are unusual; otherwise, Perhaps Wainka represents a large and prim- resemblance is closest to Anisolambda and itive offshoot of this general assemblage. there is nothing to suggest exclusive or even To the roster of early Tertiary Litopterna close affinity to Polymorphis, as proposed by I append the Casamayoran genus Proecto- Odreman Rivas (1969). cion, placed by Simpson in the Didolodon- Several other forms placed by Simpson in tidae. The type species, P. argentinus, is based the Proterotheriidae, very poorly known, de- on a right M3 (MACN 10673), but the closely serve mention. Guilielmofloweria, from the similar ifnot conspecific P. precisus is known Casamayoran, is known by the single speci- from right P4-M3 (MACN 10679). The tri- men (MACN 10703) of the type species, G. angular cusp pattern on P4, with widely spaced plicata, which I believe to be M3. It is similar and subequal paracone and metacone, is un- to M3 of Anisolambda from Cainadon Vaca, known among didolodonts, and allies the ge- differing chiefly in its much larger size and nus with advanced litopterns. Oxybunothe- in being somewhat more transverse. Simpson rium praecursor was described by Pascual (1948, p. 129) also placed the isolated cheek (1965), who correctly recognized its similar- tooth AMNH 28514, from Ca-nadon Vaca, ity to Polymorphis but placed both in the in this genus, considering it a positional vari- Didolodontidae (McKenna, 1981, referred ate. In general form, this tooth is similar to Oxybunotherium to the Proterotheriidae). those ascribed to "Ricardolydekkeria" (pos- The genoholotype of Oxybunotherium, MLP tulated here to be posterior upper premolars 59-11-28-107, is a partial right M2 and com- of Anisolambda), which probably strength- plete M3 from the late Casamayoran of Paso ened Simpson's conviction that Guilielmo- de los Indios, but an indistinguishable spec- floweria is closely related to that genus. This imen, AMNH 28769, a left mandibular frag- form, with such features as the subround out- ment with M2.3, derived from the Casama- line with centrally placed protocone and very yoran beds at Colhue-Huapi. These upper small hypoconal cingulum subequal to the (Proectocion) and lower (Oxybunotherium) anterior cingulum, I consider to be due to molars are of appropriate size, morphology, tooth locus and, as for "Ricardolydekkeria," geographic and probably temporal distribu- believe AMNH 28514 to be a premolar, tion to belong to the same genus and probably probably P4. The extremely salient meso- species (if the two species of Proectocion are style, absence of a columnar parastyle, and synonymous), and there are no other known strong plications on the inner wall of the ec- mammals from these localities to which either toloph suggest, however, that the specimen is likely to pertain. I consider them synony- 1983 CIFELLI: CONDYLARTHRA 21

A Asmithwoodwardia scotti G Prolicaphrium specillatum

B Miguelsoria parayirunhor H Proectocion* Xspp. ~~~~~~

C Protolipterna ellipsodontoides I Tricoelodus bicuspidatus

D Paranisolambda prodromus J Polymorphis spp.

EAnisolambda ?fissidens K Cramauchenia normalis

F Protheosodon coniferus

FIG. 5. Schematic comparative left upper dentitions of various litopterns, standardized to length of M2, with associations as discussed in the text. A, DGM 358M (type). B, AMNH 109559 (cast, original in DNPM; M3 reversed). C, DNPM LE4441 (cast, AMNH 109550). D, MNRJ 1479V. E, P3, AMNH 28522 (reversed); P, AMNH 28519; MI, AMNH 28516; M2, AMNH 28518 (reversed); M3, AMNH 28512 (reversed). F, FMNH P13418 (reversed). G, AMNH 29730. H, P4-M2, MACN A10679 (type of Proectocion precisus); M3, MACN A10673 (type of P. argentinus), all reversed. I, MLP 61-IV-11-65 (MI, M3 reversed). J, P4-M2, AMNH 29482; M3, AMNH 29481, all reversed. K, FMNH P13293 (P4, M' reversed). 22 AMERICAN MUSEUM NOVITATES NO. 2772

TABLE 4 Distribution of Selected Dental Features among Some Litopternaa 1 2 3 4 5 6 7 8 9 10 Asmithwoodwardia scotti abs abs a a a abs abs weak a b Protolipterna ellipsodontoides abs abs a a a abs abs weak a a Miguelsoria parayirunhor abs abs a a a abs abs weak a ? Proectocion spp. pres pres b ? ? pres pres strong a c Tricoelodus/Proadiantus pres pres b b b pres pres strong a c Polymorphis spp. pres strong b b b pres pres weak a c Cramauchenia insolita pres strong b b b pres pres weak a c Paranisolambda prodromus pres pres c ? ? abs abs weak b c Anisolambda amel ? ? c ? ? abs ? ? ? ? Anisolambdafissidens pres pres c ? ? abs abs weak b c Protheosodon coniferus pres pres c c c abs abs weak b c Prolicaphrium spectabile pres pres d b c abs abs weak b c a See text for discussion of characters and associations of dentitions. 1, M- mesostyle; 2, M- parastyle columnar; 3, condition of M paraconid and paralophid (a, paraconid connate with metaconid and faint to absent; b, paraconid absent with paralophid truncated and terminating in a ventromedian position; c, paraconid large with strongly recurved paralophid); 4, P4 talonid (a, bicuspid; b, lophate and crescentic; c, lophate but 3 cusps visible); 5, P3 talonid (a, absent; b, simple crest; c, crescentic loph); 6, M metaconid columnar; 7, M3 hypocone; 8, M- conule cristae; 9, M- mous, and refer Oxybunotherium Pascual, developed and more or less columnar styles 1965, to Proectocion Ameghino, 1904 (I am (parastyle and mesostyle). P4 talonid and mo- indebted to M. F. Soria for this suggestion). lar trigonids and talonids are lophate, and the cristid obliqua of the lower molars achieves PHYLOGENY AND a lingual attachment to the trigonid, i.e., at CLASSIFICATION the base of the metaconid. Dental morphology of the better known Some ofthe early forms, as Simpson (1948) early Tertiary litoptems and some Deseadan observed, do present molar specializations and later forms is compared in table 4 and a suggestive of later proterotheres. True pro- hypothesis oftheir interrelationships is illus- terotheriids molarize the third premolars (P3 trated in figure 7. Some aspects ofan ancestral with a crescentic talonid loph; P3 with co- morphotype are difficult to reconstruct be- nules and with subequal and widely spaced cause of continuing gaps in knowledge. Pro- paracone and metacone), whereas macrau- tolipterna has, for instance, what appear to chenioids, including the Adianthidae and be aberrantly specialized posterior premo- Macraucheniidae, do not. In the upper molars; P3-4 of Miguelsoria are not yet known, lars of proterotheres, the protocone and hy- and those of advanced litopterns are rather pocone are joined, forming a lingual conti- specialized. Miguelsoria has a fully molarized nuity with the anterior crest of the trigon P4 and that tooth in Protolipterna has a bi- (protoloph) and with the postcingulum; the cuspid talonid, a condition postulated here connection between protocone and metaco- as having developed independently in the nule is disrupted and in one lineage (Deutero- Lamegoia group. All more advanced litop- therium, Thoatherium, and allies; see Simp- terns, as in the concept of the order given by son, 1932a) the conule itself is lost. Among Simpson (1948), are characterized by a num- the early forms placed in the Litopterna, An- ber ofdental modifications. P4, where known, isolambda and Paranisolambda bear these has an enlarged metacone arranged in a tri- specializations, although as mentioned above, angular pattern with the paracone and pro- certain contradictions are introduced by the tocone; protoconule, metaconule, and me- referral of Ricardolydekkeria to the former sostyle are present. The upper molars have genus. A P3 mesostyle is lacking in Protheo- crescentic para- and metacones, with well- sodon and is only weakly developed in Prol- 1 983 CIFELLI: CONDYLARTHRA 23

TABLE 4-(Continued) 11 12 13 14 15 16 17 18 19 20 21 22 23 24 abs abs abs abs abs abs abs abs abs abs abs a abs abs ? abs abs ? abs ? abs abs abs abs abs a abs abs ? abs ? ? ? ? abs abs abs abs abs a abs abs ? pres ? pres pres ? abs ?abs abs pres abs a abs abs abs pres pres abs pres abs abs pres abs pres abs a abs abs ? abs pres ? pres ? abs pres--abs pres abs b abs abs abs abs pres abs pres abs abs pres abs pres abs c pres abs ? abs ? ? ? ? abs ? ?abs pres abs ? abs abs ? abs ? ? ? ? pres ? ? pres abs ? abs abs both abs pres pres pres pres pres abs pres pres pres a pres pres both abs pres pres ?abs ?abs pres abs abs pres pres a pres pres both abs pres pres pres pres abs abs pres pres abs a abs abs conule connections (a, primitive; b, metaconule isolated); 10, P4 conules (a, absent; b, protoconule only; c, protoconule and metaconule); 1 1, P3 conules; 12, M3 trilobed; 13, P4 paracone and metacone well separated; 15, P4 mesostyle; 16, P3 mesostyle; 17, M metaconid posteriorly developed; 18, M- hypocone crescentic; 19, M- protocone and hypocone joined; 20, M cristid obliqua attaching to metaconid; 21, M centroconid; 22, M- trigon shifted anteriorly (a, absent; b, present on M3; c, present on M'-3); 23, M labial cingulum; 24, shallow labial slope on M-.

icaphrium, but is strong in teeth here referred crista, unlike that of proterotheres, which is to Anisolambda. Anisolambda and Parani- a crescentic loph as in P4 and the lower molar solambda are otherwise very similar to series. Polymorphis, Proectocion, and later Protheosodon. The peculiar and strongly de- macrauchenioids are more advanced than veloped lower molar paraconid, probably de- proterotheres in having a well-developed hy- rived from a Miguelsoria-like condition, pocone on M3 and in the form of the lower unites these forms as distinct from true pro- molar cusps. The cusps in general, but par- terotheres (to which they nonetheless seem ticularly the metaconid and entoconid, are most closely related), which develop a cres- tall and columnar; the "paraconid," no lon- centic trigonid loph, and primitive macrau- ger a true cone (primitively as in Miguelso- chenioids, which are differently specialized. ria), is truncated, ventrally directed, and does Anisolambda fissidens possesses several fur- not extend to the lingual margin of the tooth ther progressive features seen in Protheo- but instead terminates in a median position. sodon, such as a posterior prolongation ofthe The upper molar pattern of macrauche- lower molar metaconid (partially closing off niids, as exemplified by Cramauchenia, is the talonid basin lingually), increased crown distinctive. The hypocone is enlarged and height, and a shallower labial slope and pres- supports a strong postcingulum and anterior ence of a buccal cingulum and centroconid crest, the latter extending anterobuccally to on the lower molars. the position ofthe metaconule, which is lost. Polymorphis and Proectocion seem to be Unlike proterotheres, the trigonal connec- related to later macrauchenioids, which lack tions are maintained and even enhanced. The the above progressive characters and are dif- trigon is shifted slightly forward (seen most ferently specialized. P3 in Proadiantus/Tri- in anterior of the coelodus (see below) and Cramauchenia are notably the position pro- advanced, but P3 apparently never developed tocone and attachment of that cusp to the a trigonal protocone-paracone-metacone pat- metacone), so that an entirely bicrescentic tern, and the metacone-weakly ifat all pres- pattern is formed on the upper molars. The ent in a common ancestor with protero- most characteristic feature of macraucheniid theres- is lacking. P3 in these genera bears a lower molars is the presence of a transverse simple anteroposteriorly oriented talonid entoconid loph (entolophid), present in Cra- 24 AMERICAN MUSEUM NOVITATES NO. 2772

mauchenia only on Ml 2.5 M3 ofPolymorphis wear. As seen in Tricoelodus, the pre- and has a large, expanded hypocone supporting a postprotoconule and pre- and postmetacon- broad postcingulum and prehypocone crest, ule cristae achieve contact with the ectoloph, with the trigon shifted anteriorly, suggestive so that fossettes are isolated at the lingual of the bicrescentic pattern of later macrau- bases of the paracone and metacone and at cheniids, but the metaconule is still present, the labial base ofthe protocone. The anterior and the anterior upper molars tentatively re- cingulum isolates an anterolingual fossette, ferred to it (AMNH 29482; Simpson, 1948, and a posterolingual fossette is defined by the fig. 47) are more primitive. The lower den- postcingulum and prehypocone crista, which tition of Polymorphis is of the primitive attaches to the trigon midway between the macrauchenioid pattern, as outlined above, metaconule and protocone as in Polymorphis and definite macraucheniid specializations and not to the protocone as in proterothe- such as the entolophid are lacking. riids. Fossettes are not developed in the up- The poorly known pygmy litopterns, per cheek teeth of Proectocion (or, if so, they Adianthus and allies, have been variously are very transient), but the genoholotype considered as a subfamily ofthe Macrauche- (MACN 10673) shows strong development niidae (Bordas, 1939; Patterson, 1940; Simp- of the interconnecting cusp crests, particu- son and Minoprio, 1949) or as a separate larly the conule cristae. The lower molars of family, Adianthidae, as Ameghino (1891) Proectocion (represented by Oxybunother- originally proposed (Simpson, Minophio, and ium, here referred to that genus) are also more Patterson, 1962; Soria, 1981). Work in prog- primitive than those of Deseadan adianthids ress indicates that the upper dentition de- such as Proadiantus (for instance, in the more scribed by Patterson (1940) as Proadiantus feeble development of trigonid and talonid actually pertains to the closely similar but crescentic lophs) but, significantly, the pe- larger genus Tricoelodus from the same De- culiar trilobed M3 of Proadiantus is fore- seadan locality (Cabeza Blanca). Upper den- shadowed in Proectocion by an elongate and titions are known also for Proheptaconus projecting hypoconulid which, unlike M2, is (Bordas, 1936; see Soria, 1981) and the enig- disconnected from the entoconid. matic and not certainly allied Divisaderan In a synthetic classification which predated Adiantoides (Simpson and Minoprio, 1949; publication of his own extensive contribu- Simpson, Minoprio, and Patterson, 1962); tion to didolodont and litoptern systematics, lowers are known for the latter and the type Simpson (1945) grouped Wainka, Joseph- of the family, the Santacrucian Adianthus. oleidya, Ricardolydekkeria, Guilielmoflow- Tricoelodus and Proadiantus are known from eria, Anisolambda, Polymorphis, Megacro- good upper and lower dentitions, respective- don, Xesmodon, Polyacrodon, Decaconus, ly, and appear to be both closely related and and Protheosodon in a then new subfamily of primitive within the family; in the compar- the Proterotheriidae, Polymorphinae, but did isons a composite is used (Tricoelodus for the not mention or elaborate on this grouping in upper series, Proadiantus for the lower). As his 1948 work. Pascual (1965) noted that this described by Patterson (1940), the adianthid group is somewhat artificial, and Odreman upper cheek teeth are relatively high crowned Rivas (1969) rediagnosed and restricted the and are characterized by the raising and unit- Polymorphinae, including in it only the Mus- ing ofcrests connecting the various cusps and tersan Polymorphis and Heteroglyphis, there- cuspules, so that fossettes are formed with by leaving the remaining genera (except Protheosodon, which he considered condylarthran) in an uncertain position within the -'This feature was considered by Scott (1910) to be family. Soria (1981) accepted this arrange- primitive for the Litoptema, and his beliefthat litoptems ment, noting that the "Anisolambda group" and notoungulates are closely related seems to have been appears to be more closely allied to the Pro- predicated largely on this view. As Simpson (1948) has already shown, however, the transverse entoconid is terotheriinae and considering Xesmodon and lacking in most litoptems, notably proterotheres and the Protheosodon to be condylarths. primitive forms described herein. The entolophid ap- The dental comparisons discussed above pears to have been lost in the latest macraucheniids. suggest a radical departure from existing ar- 1983 CIFELLI: CONDYLARTHRA 25 A Asmithwoodwardia,~~~~~~~~~~~~~scotti G Prolicaphrium specillatum

B Miguelsoria parayirunhor H Proectocion spp.

C Protolipterna ellipsodontoides I Proadiantus excavatus

D Paranisolambda prodromus J Polymorphis lechei

E Anisolambda fissidens K Cramauchenia normalis

F Protheosodon coniferus

FIG. 6. Schematic comparative right lower dentitions of various litopterns, standardized to length of M2, with associations as discussed in the text. A, DGM 358M (type). B, MNRJ 1468V (type). C, DNPM LE444C (cast, AMNH 109554). D, M,, MNRJ 1496V; M2, AMNH 49818; M3, MNRJ 1859V, all reversed. E, MACN A10668 (type; M1 reversed). F, ACM 3001. G, FMNH P13306 (M2-3 reversed). H, AMNH 28769 (reversed). I, P3_M2, MACN A52-214; M3, MACN A52-215. J, MLP 12-2168 (reversed). K, FMNH P13293.

rangements in litoptern classification. With concept of the order is considerably broad- the recognition that some dentally primitive ened. With the greatest respect for Simpson's didolodont-like ungulates possessed the di- incomparable knowledge ofthese animals and agnostic litoptern ankle specializations, the for his synthetic work on litoptern phylogeny, 26 AMERICAN MUSEUM NOVITATES NO. 2772

.( E o Co)0, E u c i C) -20 Q- 0

-0 *0

LU LU 0 LU r- () 00 Q = Z: z LUir LU o I I

Q Q/) a) Q/) - 40 -o U)

c c o E a) 0 L 0 a) *0 E E *o o U) ._ Q 0 o- o o Co c olQi ._ Cn 0~ Q a : E oU0 E ct o cu cn a,) Q-.03 -50 <) 0 E. ai a)U -a o a) O 0 .2

-60 ,2,13,15,20)

-70 FIG. 7. Relationships among various Litoptema, showing distribution in time and morphological distance on vertical and horizontal axes, respectively. Absolute time scale (left margin) after Marshall et al. (in press). Numbers in parentheses refer to derived characters listed in table 4 and discussed in text. Nodes refer to subdivisions of advanced litoptems: A, Lopholipterna; B, Macrauchenioidea; C, Macraucheniidae; D, Adianthidae; E, Proterotherioidea, Proterotheriidae; F, Anisolambdinae. Abbre- viations: CAS, Casamayoran; COL, Colhuehuapian; DES, Deseadan; MUS, Mustersan; RIO, Riochican; ?c, possible character convergence; ?r, possible character reversal. See text for discussion and additional characters. Prolicaphrium specillatum and Cramauchenia normalis are advanced Proterotheriidae and Macraucheniidae, respectively, the contents of which are not here subdivided.

I further believe that placement of Victorle- specializations not found in the didolodont- moinea and other known pre-Mustersan gen- like genera, and it seems taxonomically jus- era in the Macraucheniidae is incorrect and tifiable to recognize them as a distinct sub- has been a conceptual stumbling block to the ordinal unit, both because they appear to be interpretation of the affinities ofthese poorly a monophyletic unit and because they rep- known mammals. True macrauchenioids and resent a significant, more advanced structural proterotherioids share a number of dental grade. For this group I propose the suborder 1983 CIFELLI: CONDYLARTHRA 27

TABLE 5 TABLE 5-(Continued) Classification of the Litopterna Subfamily Proterotheriinae Ameghino, 1887 Order LITOPTERNA Ameghino, 1889, p. 492 (first used as a subfamily by Simpson, 1945, plesion Family Protolipternidae, NEW p. 125; contents not here subdivided) Protolipterna, NEW Miguelsoria, NEW aUsually attributed to Gill (1872), to which this is Asmithwoodwardia Ameghino, 1901, p. 379 clearly antecedent. b Listed incorrectly by Soria as Suborder Lopholiptema, NEW (1981) Ameghino (1885) Superfamily Macrauchenioidea Gervais, 1855, NEW due to an emendation in the compilation of the Obras RANK Completas. This latter (Ameghino, 1913-1936, vol. 5, p. 104) lists Proterotherium cervoides in the "Protero- Family Macraucheniidae Gervais, 1855, p. 36a theridae," but in the original publication (Ameghino, Subfamily Polymorphinae Simpson, 1945, p. 1885, p. 146) this species was placed in the "Protocer- 124. vina," a subdivision (of uncertain rank) of the Artio- Polymorphis Roth, 1899, p. 385 (including dactyla. Polyacrodon Roth, 1899, p. 382; Deca- conus Ameghino, 1901, p. 378) Subfamily Cramaucheniinae Ameghino, 1902, Lopholipterna, defined by the characters giv- p. 92 (including, in part, Mesorhinidae en in diagnosis above. This introduces the Ameghino, 1891, p. 136, Theosodontinae, problem of placement of such forms as Ameghino, 1902, p. 92; contents not here Asmithwoodwardia, Protolipterna, and Mig- subdivided) uelsoria. Judged by the structural variations Subfamily Macraucheniinae (Gervais, 1855; first of these genera, it seems likely that they rep- used as a subfamily in the present sense by resent part of a significant early radiation of Ameghino, 1902; contents not subdivided small, primitive litopterns, but there is little here) positive evidence available at present to sug- Family Adianthidae Ameghino, 1891, p. 134 gest that they form an exclusive monophy- (=Adiantidae Ameghino, 1894, p. 283; letic unit. Possibly the only notable feature Adiantinae Bordas, 1939, p. 417; Adianthi- nae Patterson, 1940, p. 13; Simpson and these genera possess in common is the well- Minoprio, 1949, p. 6) differentiated lower canine, unknown in di- Proectocion Ameghino, 1904, p. 83 (includ- dolodonts, and possibly but probably not a ing Oxybunotherium Pascual, 1965, p. very archaic condition. I tentatively propose 59) to unite them into the horizontal or para- Adianthus Ameghino, 1891, p. 134 (=Adian- phyletic new family Protolipternidae. The tus Ameghino, 1894, p. 283) type genus, Protolipterna, is distinctive Proheptaconus Bordas, 1936, p. 110 enough alone to warrant erection of a new Proadiantus Ameghino, 1897, p. 455 family to contain it, and later additions to (=Proadianthus Loomis, 1914, p. 51) knowledge of primitive litopterns, either by Tricoelodus Ameghino, 1897, p. 454 Adiantoides Simpson and Minoprio, 1949, the discovery of new allies of Protolipterna p. 6 or by the removal ofAsmithwoodwardia and Superfamily Proterotherioidea Ameghino, 1887, Miguelsoria to other suprageneric groups, NEW RANK should not result in the abandonment of the Family Proterotheriidae Ameghino, 1887, p. 1 9b family based on this type genus. The inclu- Subfamily Anisolambdinae, NEW sion of Miguelsoria and Asmithwoodwardia Anisolambda Ameghino, 1901, p. 383 (in- is not a wholly satisfactory taxonomic pro- cluding Josepholeidya Ameghino, 1901, cedure, but it does accord with and is cor- p. 384; Ricardolydekkeria Ameghino, roborated by the close patristic affinity and 1901, p. 397) structural grade of all three genera. Alterna- Guilielmofloweria Ameghino, 1901, p. 397 tively, these forms might be allocated to the Xesmodon Berg, 1899, p. 79 Litopterna, incertae sedis, but this is essen- Heteroglyphis Roth, 1899, p. 387 tially a non-view and is in that it Wainka Simpson, 1935, p. 9 imprecise Paranisolambda, NEW does not express any hypothesis of relation- Protheosodon Ameghino, 1897, p. 453 ships (i.e., that these forms are far more primitive than lopholipternans and cannot be 28 AMERICAN MUSEUM NOVITATES NO. 2772 grouped with them) or contain any infor- should be erected so that, for instance, mation on general morphological grade. The Thoatherium and allies are segregated out). Protolipternidae might in turn be placed in a group of equal rank to the Lopholipterna AFFINITIES OF THE (i.e., suborder Bunolipterna), but this is not DIDOLODONTIDAE AND necessitated by the materials at hand. In a LITOPTERNA strictly cladistic classification, the three genera in question could be designated "ple- Of all the indigenous South American un- sions" and their sequence of listing in a clas- gulates, didolodonts and primitive Litopter- sification of the Litopterna would indicate na most closely approximate in known struc- hypothesized position in the phylogeny ofthe ture Holarctic mammals generally considered group (Patterson and Rosen, 1977). How- basal ungulates, and their comparative mor- ever, there is no evidence at present to in- phological analysis is thus of paramount dicate any particular arrangement of these interest to understanding the origin and genera, and designation of the ?paraphyletic biogeographic deployment of native Neo- family Protolipternidae as a unit plesion is tropical mammals. In his treatise on Patago- proposed here. nian fossil faunas, Ameghino summarized his Within the Lopholipterna, the families view of South American mammal relation- Adianthidae and Macraucheniidae do appear ships as follows: "les diffierents groupes de to be closely allied, and this exclusive rela- [sic] anciens Mammiferes de Patagonia aient tionship is expressed in the superfamily Mac- [sic] une origine commune avec les groupes rauchenioidea. As noted above, Polymorphis semblables des autres regions de la terre. C'est- appears to be closer to macraucheniids than a-dire que les ressemblances seraient le re- to proterotheres (as Ameghino, 1904, rec- sultat de la parente ou de l'unite d'origine" ognized), and the Polymorphinae (containing (1906, p. 19). The various forms under con- only the type genus) is here included in the sideration were accordingly placed in various Macraucheniidae. Proectocion is here re- Holarctic families; most genera currently al- moved from the Didolodontidae and placed located to the Didolodontidae (e.g., Asmith- in the Adianthidae. If this allocation is cor- woodwardia, Enneoconus, Ernestokokenia; rect, a suprageneric category might be erected Didolodus itself was for reasons unknown for the genus in recognition of its distinc- omitted from this classification) were re- tiveness from all other, far more advanced ferred to the Phenacodontidae and those later adianthids, but the family is at present so removed by Simpson to the Litopterna were poorly understood that it is more conserva- placed in the "Catathleidae" (=Periptychi- tive not to subdivide it at this juncture, dae; e.g., Josepholeidya), Pantolambdidae avoiding the unnecessary proliferation of (e.g., Heteroglyphis, Ricardolydekkeria, Gui- names that may be discarded in the near fu- lielmofloweria), and Palaeotheriidae (e.g., ture. Most of the remainder of Simpson's Victorlemoinea, Anisolambda). Pauloger- (1945) Polymorphinae, the "Anisolambda vaisia and Proectocion, included by Simpson group," appear to represent a radiation of (1948) in the Didolodontidae, were allocated primitive proterotheres (as Simpson recog- to the Carolozitteliidae (which Ameghino nized) that persisted until the Deseadan considered to be Proboscidea) and "Hyra- (Protheosodon) before becoming extinct cotheriidae" (primitive equid perissodac- without issue. With the transfer of Polymor- tyls), respectively. The Litopterna have been phis to the macraucheniids, a new suprage- considered as a distinct order since Scott's neric name is needed for the "Anisolambda (1910) monograph on Santacrucian forms, group," for which the Anisolambdinae is pro- and most subsequent workers have consid- posed. The remainder of the proterotheres ered litopterns and didolodonts to be related, are here placed in the Proterotheriinae as Ameghino (e.g., 1904) clearly recognized; Ameghino, 1887 (sensu Simpson, 1945). some (Scott, 1913; Loomis, 1914) including These contents are not further subdivided al- didolodonts within the Litopterna and others though generic clusters are evident (tribes (Osborn, 1910; Gregory, 1910; Scott, 1937; 1983 CIFELLI: CONDYLARTHRA 29

Simpson, 1945, 1948) following Ameghino which comparison with other ungulate groups in placing them in the Condylarthra. may be made. Using this assumption of In general, the primitive South American monophyly, a hypothetical ancestral com- forms have been found to compare most fa- posite didolodont/litoptern morphotype may vorably with the Phenacodontidae. Ameghi- be constructed by comparison with a hypo- no (1901) at first referred the species Ernes- thetical ungulate morphotype, and the for- tokokenia patagonica and E. trigonalis to mer, in turn, compared with the various "Euprotogonia" (= Tetraclaenodon), a course groups of Holarctic "protungulates" to de- followed by Gaudry (1904). As the name im- termine if any special relationships are sug- plies, Proectocion was postulated by Amegh- gested. As noted above, comparison is most ino (1904, p. 82) as the ancestor of Ectocion. appropriate with members of the order Con- Simpson compared the better known dido- dylarthra which, as generally conceived, is a lodont genera with North American condy- horizontal and somewhat artificial assem- larths, noting that Asmithwoodwardia (not blage of primitive ungulates, and probably then well known) resembles North American includes near relatives of many if not most Mioclaeninae and that Didolodus would be orders of mammals (Szalay, 1969; Van Va- placed in the Phenacodontidae if found in len, 1978). In addition to the Didolodonti- North America, but summarized by observ- dae, the Condylarthra as usually construed is ing that, "the family considered as a whole comprised of the families Arctocyonidae, is an assemblage too advanced to be com- Periptychidae, Phenacodontidae, Mesony- pared with Paleocene Holarctic condylarths" chidae, Meniscotheriidae, and Hyopsodon- (1948, p. 97). On the basis of incomparably tidae. Several of these families as currently better material from the Riochican of Brazil, conceived may be artificial, although without Paula Couto (1952a) referred Asmithwood- exception they contain generic clusters, at wardia to the North American subfamily least, which appear to be monophyletic. The Hyopsodontinae, but later (1958, 1978b) re- mesonychids and several aberrant subfami- tracted this allocation and returned the genus lies dubiously placed in the Arctocyonidae to the Didolodontidae, in accordance with and Hyopsodontidae have never been cited arguments presented by McKenna (1956), in connection with South American ungu- who recently also reversed his opinion (1981). lates, and are not treated here. Van Valen Because of the relatively poor representa- (1969) placed the Arctocyonidae and Meson- tion of primitive South American forms, ychidae in a new suborder, Arctocyonia, comparisons with North American ungulates raised to ordinal rank by McKenna (1975), have heretofore been largely based on indi- who distributed the contents of the Condy- vidual genera such as Didolodus, for which larthra among various other ungulate ordinal unusually good materials are available. Such and subordinal taxa. Much ofthis rearrange- comparisons, like the familiar "three taxon ment remains to be documented, although in statements" of cladistic review papers, are at least one case, that of mesonychids and useful but not entirely representative and may whales, preliminary arguments have been be misleading because they are grossly over- presented (Van Valen, 1966; Szalay, 1969). simplified and do not take into account the Several synthetic but speculative ungulate possibilities of characters developed in par- phylogenies have appeared recently (e.g., allel. The close dental similarity of didolo- Sloan, 1969; Van Valen, 1978), but no com- donts and primitive litopterns as indicated prehensive and detailed systematic treatment in the systematics section above strongly sug- ofany condylarthran group has appeared since gests that they represent a monophyletic as- the now-classic memoirs of Matthew on the semblage with respect to other ungulates. This San Juan Basin Paleocene and Simpson on hypothesis remains to be demonstrated on the Crazy Mountain Field Paleocene, both of the basis of shared, derived morphological which were published in 1937. As a result, character states, but is corroborated by their recent advances, both of discovery and of geographic/temporal distribution and close change in conceptual perspective, have not patristic similarity, and provides a basis upon been incorporated into any general frame- 30 AMERICAN MUSEUM NOVITATES NO. 2772 work of ungulate phylogeny, and even the Maastrichtian genus Protungulatum (Sloan alpha level systematics ofseveral ofthe fam- and Van Valen, 1965) and several forms de- ilies involved is in such a state of disrepair scribed recently by Van Valen (1978) repre- that a synthesis is not possible at present. In sent the closest approximation to this prim- order to establish a basis for comparison with itive eutherian morphotype, both dentally and the South American forms, a brief outline of postcranially. Protungulatum lacks any the Condylarthra, with taxonomic notes on known feature not seen in primitive members the members ofits constituent families, must ofall condylarthran families and thus by im- now be given, with some prefacing remarks plication represents a reasonable primitive on the defining features of ungulates. ungulate morphotype. Compared with Ken- The description and discussion of numer- nalestes and other Cretaceous eutherians, ous Late Cretaceous eutherian mammals from Protungulatum exhibits a number of dental North America and Asia (Lillegraven, 1969; specializations, mostly related to increased Fox, 1970, 1979; Clemens, 1973, 1974; Be- importance of transverse "phase II" jaw ac- liajeva, Trofimov, and Reshetov, 1974; Sza- tion (Butler, 1977) and indicative of a shift lay and Decker, 1974; Kielan-Jaworowska, from insectivorous to omnivorous diet: up- 1975; see Lillegraven, Kielan-Jaworowska, per molars with pre- and postcingula, sub- and Clemens, 197, and references therein) equal para- and metacone, enlarged proto- provides a basis for postulation ofa primitive cone shelf with protocone placed somewhat eutherian morphotype and for comparison more posteriorly (i.e., in an isosceles trian- with arctocyonids and other presumed prim- gular arrangement with the paracone and itive ungulates. Although it is evident that a metacone); reduced stylar shelf, parastylar and major eutherian radiation was occurring in metastylar lobes, and cusp relief (with cor- the Late Cretaceous, relationships of and responding reduction in the shear surfaces among these early mammals remain highly connecting cusps); lower molars with reduced problematic. Aside from the "higher" euthe- height differential between trigonid and tal- rians which appear in the Maastrichtian onid, the latter buccolingually expanded; and (Sloan and Van Valen, 1965; Van Valen and P4 with paraconid and metaconid. It should Sloan, 1965), most of the unquestioned Cre- be borne in mind that such features seem to taceous Eutheria may be recognized in a broad be general trends in mammals undergoing sense as pertaining to three categories (Kie- such a change in diet, and many ofthese char- lan-Jaworowska, Bown, and Lillegraven, acters are evident also, for example, in the 1979): zalambdalestid, "palaeoryctoid," and leptictoid Gypsonictops, the paromomyid "leptictoid." The first group is wholly Asiatic primate Purgatorius (Van Valen and Sloan, in distribution, the second and third North 1965; Clemens, 1974), and in eutherians from American (unless Asioryctes and/or the the Milk River and Oldman formations de- "Champ-Garimond molar" (McKenna, scribed by Fox (1970, 1979) as erinaceoid 1969) be included with the "palaeoryctoids" lipotyphlans. The possibility thus remains and Kennalestes with the "leptictoids"). The that the Ungulata is a polyphyletic assem- greatest number of dental features common blage and that the South American forms are to these and other Cretaceous Eutheria is still more distantly related to Holarctic eu- found in the ?Campanian species Kennalestes therians than has been supposed. Szalay gobiensis, which thus seems to represent best (1977, p. 344) stated that, "specific derived a primitive eutherian morphotype (Cromp- features which unite the arctocyonids with ton and Kielan-Jaworowska, 1978; see also their many descendents are numerous and McKenna, 1969; Butler, 1977, 1980; Kielan- detailed on skulls, teeth, and postcrania," but Jaworowska, Bown, and Lillegraven, 1979). these characters remain to be listed and de- The presence of small pre- and postcingula tailed. (Those commonly cited, loss of clav- on upper molars of this species is the most icle and the presence of hooves, are untrue - significant deviation from a pattern from where known- of many condylarthrans. It which all other eutherians can be derived. might be added parenthetically that the clav- Among the diverse forms grouped within the icle was probably lost at least twice among Condylarthra, the arctocyonids, especially the the Notoungulata alone.) Nonetheless, this 1 983 CIFELLI: CONDYLARTHRA 31 somewhat non-exclusive dental concept ofthe (1978) has proposed a fourth subfamily, Lox- basic specializations uniting all ungulates olophinae, in addition to bringing new species provides a reasonable basis for phylogenetic to the literature. The aberrantly specialized analysis and is supported by the deep patristic triisodontines, possibly related to mesony- dental and postcranial similarity ofprimitive chids and perhaps ultimately to cetaceans, members of the various condylarthran fam- are not pertinent to the problem of origin of ilies and their presumed derivatives. As is the South American ungulates, and are not noted below, this concept does exclude from dealt with further here. the ungulates certain mammalian groups often Van Valen (1978) described the genera Ox- placed with them as condylarth descendents. yprimus and Thangorodrim and included with The heterogeneous family Arctocyonidae them in the Oxyclaeninae Protungulatum was long grouped with the Carnivora, al- Sloan and Van Valen, 1965, Chriacus Cope, though similarities between arctocyonids and 1883 (including Tricentes Cope,1884a,6 Spa- mammals generally placed in the Condylar- noxyodon Simpson, 1935c, and Metachria- thra had been noted (Scott, 1892; Matthew, cus Simpson, 1935c), Prothryptacodon Simp- 1937; Gazin, 1941a; Simpson, 1937, 1945). son, 1935c, Thryptacodon Matthew, 1915a, It is of some interest to note that it was Lantianius Chow, 1964, Oxyclaenus Cope, Ameghino who first placed them in the Con- 1884a (including Carcinodon Scott, 1892), dylarthra: and Deltatherium Cope, 188 lb. To this roster I "L'on croit generalement que les Ongules ont would append Ragnarok and Maiorana, du prendre origine des Cr6odontes primitifs, a placed by Van Valen (1978) in the Loxolo- cause de la transition apparente que dans la phinae and Periptychidae, respectively, and faune de Puerco et Torrejon on observe entre Oxytomodon (see below). Various workers les Ongules et les Creodontes de la meme for- have noted the strong resemblance between mation. Mais la raison de cette transition est Deltatherium and primitive pantodonts such tout autre; c'est que la plupart des soi-disants as Pantolambda on the one hand (Simons, Creodontes de la formation Puerco et Torrejon 1960; Sloan, 1969) and tillodonts such as Es- (Chriacus, Protochriacus, Oxyclaenus, Claeno- thonyx on the other (Gazin, 1953). Aside from don, Tricentes, Triisodon, etc.) n'ont pas de re- advanced features such as the reduction of lations avec les veritables Creodontes. Ces the lower molar entoconid and molar formes doivent etre considerees comme des On- upper gules primitifs que s'etaient adaptes a un regime conules, and hypertrophy of the upper molar carnassier" (1901, p. 352). parastylar and metastylar shear surfaces, there are strong patristic similarities as well. Del- The general acceptance in the mid 1960s of tatherium retains a wide upper molar stylar arctocyonids as basal ungulates (Patterson, in shelf with substantial parastylar and meta- Patterson and McGrew, 1962; Sloan and Van stylar lobes, anteriorly placed protocone, high Valen, 1965; Van Valen, 1966; but see Rus- cusps, and pronounced height differential be- sell and Sigogneau, 1965) represents a marked tween lower molar trigonid and talonid-all change in conceptual perspective. The pres- features in which the genus is more primitive ent arrangement ofthe family Arctocyonidae than any ungulate as that group is defined is largely due to the work of Matthew, who above. Deltatherium appears not to be an (1937) united as subfamilies the Oxyclaeni- arctocyonid but instead a large "insectivore." dae Scott, 1892, Triisodontidae Scott, 1892, It is of interest to note parenthetically that and Chriacidae Osbom and Earle, 1895, in the plausible derivation of both the herbiv- the Arctocyonidae Murray, 1866. Simpson orous pantodonts and tillodonts from a Del- (1937) proposed a trifold subdivision of the tatherium-like ancestor provides striking family, removing the contents of the Chria- contrast to the radiation of the great groups cinae (Chriacus and Deltatherium) to the Ox- of herbivorous ungulates, which passed yclaeninae. Since the classification of Simp- through a bunodont omnivorous stage first. son (1945), various new forms have been The Oxyclaeninae as thus composed is described (see, e.g., Patterson and McGrew, 1962; Chow, 1964; Sloan and Van Valen, 1965; Gingerich, 1978, 1979), and Van Valen 6 Listed by Simpson (1945, p. 106) as Cope, 1883. 32 AMERICAN MUSEUM NOVITATES NO. 2772 almost wholly primitive within the Condy- bertocyon Gingerich, 1979 (which I place here) larthra, although several of the genera are several advanced features which seem to jus- slightly specialized in different directions. tify their suprageneric distinction, such as the Prothryptacodon and Thryptacodon appear reduction to loss of P1, with concomitant de- to be closely allied (Simpson, 1937), sharing velopment ofdiastemata, twinned lower mo- such features as a "pinched" M3 trigonid (also lar entoconid and hypoconulid, and M3 para- true of loxolophines) and the loss of the P4 conid in a median position. paraconid (also seen in Oxyclaenus). Astrag- The nucleus of the family, the Arctocyon- ali of Chriacus and Thryptacodon are ad- inae, has gone through similarly confusing vanced in having a dorsoventrally com- rearrangements. Matthew (1937) based a then pressed head, broadly expanded flat articular new species ofArctocyoninae, Neoclaenodon surface for the medial malleolus of the tibia, procyonides, on Torrejonian material previ- superior margins of the digital flexor groove ously referred to the Puercan Mioclaenus prolonged into salient ridges, and in lacking protogonioides Cope, 1882a, placed by Scott the superior astragalar foramen. Among arc- (1892) in Claenodon. Simpson (1937) re- tocyonids, astragali are also known for Arc- moved the genotype of Neoclaenodon, N. tocyonides, ?Protungulatum, and Loxolo- montanensis Gidley, 1919, to Claenodon, and phus; these genera are primitive in lacking Van Valen (1978) synonymized Claenodon the above (mesonychid-like) characters. Scott, 1892 (including N. montanensis) with Van Valen (1978) synonymized Paradox- Arctocyonides Lemoine, 1891, and placed N. odon Scott, 1892 and most species previously procyonides in Colpoclaenus Patterson and referred to Protogonodon Scott, 1892, in- McGrew, 1962. I believe these referrals to be cluding the type, P. pentacus (Mioclaenus correct, and the last adds important new in- pentacus Cope, 1888) with Loxolophus Cope, formation to the concept of Colpoclaenus. 1885, and placed with it in the Loxolophinae Other genera included in the subfamily are Mimotricentes Simpson, 1937 (including the Arctocyon, Anacodon, and Mentoclaenodon. species subtrigonus, placed by Matthew, 1937, As thus recognized, the Arctocyoninae are in Tricentes and subsequently used as the easily distinguished by their very low crowned concept of that genus), Baioconodon Gazin, cheek teeth with crenulated enamel, upper 1941b, Deuterogonodon Simpson, 1935c, molar protocone anteriorly placed, lack of a Desmatoclaenus Gazin, 1941 a, and two new P3 protocone, and lower molar trigonids with genera, Platymastus and Ragnarok. Puercan three transverse crests and with paraconid material described by Cope, 1882a, as Mio- fused to metaconid. claenus protogonioides and referred by Mat- As originally proposed by Cope (1881a), thew (1937) to ?Protogonodon, was placed in the nucleus of the Condylarthra is the Phe- Desmatoclaenus by Van Valen. This transfer, nacodontidae. To the type late Paleocene and which I believe to be correct, is of special early Eocene genus Phenacodus Cope, 1873, interest because the type ofthe genus, D. her- was added the Torrejonian genus Tetraclae- maeus, represented by poorly preserved ma- nodon (whose type species was described ini- terial, has been included within the Phena- tially by Cope, 1881c, as a species of Phe- codontidae (Gazin, 1941 a; West, 1976) as a nacodus and referred incorrectly by Scott, transitional form. Both species lack clearly 1892, to the Arctocyonidae; see West and phenacodont specializations, and AMNH Baird, 1970). Ectocion Cope, 1882b (middle 27657, a lower jaw of D. protogonioides, in- Paleocene to early Eocene), has long been dicates that the characteristic loxolophine an- considered a phenacodont (Granger, 1915). terior premolar reduction had occurred in this Simpson (1935c) (see also Simpson, 1937; species, at least. Platymastus is poorly known, Gazin, 1956) added Gidleyina from the Tif- being based on only two isolated upper mo- fanian of the Crazy Mountain Field to the lars. Ragnarok is extremely primitive and its family, noting its similarity to Ectocion but pertinence to this subfamily is not demon- distinguishing it on the basis ofits somewhat strated; I refer it to the Oxyclaeninae. The more primitive premolars; West (1976) syn- better known forms, Loxolophus, Desmato- onymized them. Desmatoclaenus, from the claenus, and Mimotricentes share with Lam- late Puercan ofcentral Utah, was included in 1983 CIFELLI: CONDYLARTHRA 33 the family by Gazin (1941 a), who suggested work of Teilhard de Chardin (1922), most it to be intermediate between the arctocyonid students have considered the somewhat more "Protogonodon" and the phenacodont Tetra- primitive European genera Pleuraspidother- claenodon. In most respects, Desmatoclaenus ium and Orthaspidotherium as close allies of does represent a suitable morphotype from Meniscotherium (Simpson, 1929, 1945; Rus- which the Phenacodontidae may have arisen, sell, 1964). In analogy with artiodactyls, but as noted in connection with the arcto- Simpson (1929) observed that the posteroin- cyonids, it does not present any features di- ternal upper molar cusp of Pleuraspidother- agnostic of other phenacodonts, and a re- ium probably represents a displaced meta- ferred species, D. protogonioides, possesses conule or "pseudhypocone." This is not true several loxolophine specializations. Its inclu- of Meniscotherium, in which the posteroin- sion in the Phenacodontidae thus seems gra- temal cusp appears to be a derivative of the tuitous, and I place it in the Arctocyonidae cingulum (i.e., true hypocone). This, and nu- as suggested in other terms by West (1971). merous other differences noted by Gazin Prosthecion, an Ectocion-like form from the (1965), implies that Pleuraspidotherium (?and late Paleocene ofColorado, is the most recent Orthaspidotherium) may represent a far more North American addition to the Phenaco- remote lineage which independently acquired dontidae (Patterson and West, 1973). The dental specializations similar to those of North American members of the family are Meniscotherium, and is perhaps related to thus here recognized to include Tetraclaen- Protoselene, as Van Valen (1978) proposed. odon, Phenacodus, Ectocion, and Prosthe- Resemblance to hyracoids, especially some cion; Meniscotherium is probably a phena- of the Fay'um genera, is also close (see sum- codont derivative (Van Valen, 1978), as maries by Russell, 1964, and Gazin, 1965). outlined below. By comparison with a prim- Until the recent acceptance of the Arcto- itive ungulate morphotype, the primary den- cyonidae as primitive condylarthrans, the tal specializations of the Phenacodontidae Hyopsodontidae seem to have been a waste- appear to be in the molarization of the pre- basket for generally small, primitive ungu- molars. P4 bears a distinct metacone well sep- lates not clearly specialized enough to war- arated from the apex of the paracone, and rant allocation to other, more advanced has both protoconule and metaconule (?the suprageneric groups. This concept is implicit latter lost in Phenacodus); P4 has a bicuspid in Matthew's (1915b, p. 312; 1937, p. 194) talonid. The lower molars have a metastylid, characterization ofthe family, and it is in fact although this is variable in Tetraclaenodon. almost wholly primitive features which he Two generic subgroups of phenacodonts are used to unite the Hyopsodontidae and Mio- suggested (Patterson and West, 1973); Phe- claenidae into a single family. Such patristic nacodus and Tetraclaenodon are character- affinity does not deny the horizontal reality ized by a posteriorly shifted upper molar of the family as defined and indicates prob- metaconule, lingually placed lower molar able derivation from a fairly recent common paraconid connate with the metaconid, and ancestor, but it does not distinguish it as a a pre-entoconid lower molar cuspule. Ecto- monophyletic unit with respect to other cion and Prosthecion form the other subgroup, primitive ungulates and eutherians in gen- sharing features generally associated with their eral. This is not an adverse comment on the more lophodont dentition: upper molars with remarkably able and detailed work of Mat- mesostyle and with relatively high cusps and thew and Simpson (who made the most re- well-developed shear surfaces; P4 bilophate cent significant contributions to this group), and bearing a metastylid; lower molars with but an observation general to many primitive a paralophid and with talonid becoming lo- early Tertiary Eutheria. There simply are few phate (hypoconulid connected to hypoconid if any well defined and unvarying specializa- by a strong ridge, hypoconulid and entoconid tions characterizing these animals, although indistinct from each other); M3 elongate. certain lineages and generic clusters are evi- Many of these specializations are seen also dent. Based on their close overall resem- in the far more advanced late Paleocene and blance, I tentatively retain the mioclaenines early Eocene genus Meniscotherium. Since the and hyopsodontines as subfamilies of the 34 AMERICAN MUSEUM NOVITATES NO. 2772

Hyopsodontidae, with substantially the same is distinct from both Ellipsodon and Choe- concepts of these two groups as marshaled roclaenus, in Promioclaenus, but it differs by Simpson (1937). Other, even more enig- sharply from that genus, most notably in the matic groups such as the tricuspiodontines inflated P4 which lacks a metaconid. Van Va- (see Van Valen, 1978) and Apheliscus and len also synonymized Jepsenia Gazin, 1939, presumed allies (see Jepsen, 1930; Simpson, with Litaletes Simpson, 1935c, a synonymy 1937; Gazin, 1959; McKenna, 1960) have I believe to be correct. also been placed in the Hyopsodontidae, but As thus recognized, Choeroclaenus, Ellip- their pertinence to a group founded on a mio- sodon, Tiznatzinia, and Mioclaenus are dis- claenine-hyopsodontine nucleus has not been tinct but form a relatively cohesive group, established and they do not seem relevant to characterized mainly by the inflated premo- the problem at hand. Rigby (1980) has re- lars (extreme in Mioclaenus), with loss of ac- cently attempted a phylogeny of hyopsodont cessory cusps and cuspules, and by the re- genera. Because the definitions of many of duction of M3 (except in Choeroclaenus). As the genera included in the Hyopsodontidae indicated by Simpson (1937), Promioclaenus are so confused and readily susceptible to (then included with Ellipsodon) and Litaletes misunderstanding, hyopsodontid taxonomy may be tentatively allied with this group on must be briefly reviewed. the basis ofthe position and form ofthe lower The type genus of the Mioclaeninae, Mio- molar paraconid. Simpson's other major dis- claenus Cope, 1881 c, was early confused by tinguishing character ofthe Mioc/aenus group, a number of irrelevant additions, but is now fusion of the lower molar entoconid to hy- understood by the relatively well known type poconulid, applies equally to numerous other species, M. turgidus. Simpson (1945) includ- genera, including most arctocyonids and ed also Ellipsodon, Litaletes, Jepsenia, nearly all included by him in the Hyopso- Choeroclaenus, and Tiznatzinia in the dontinae, and is therefore not distinctive. subfamily. Matthew (1937) appended a num- Both Litaletes and Promioclaenus have a rel- ber of species to Ellipsodon, which is based atively molarized P4, with paraconid and on a poorly known type, E. inaequidens. Wil- metaconid, but comparison with other con- son (1956), following an observation by dylarth groups, especially oxyclaenine arc- Simpson (1937), restricted Ellipsodon to the tocyonids, indicates this to be primitive for type species, ?E. sternbergi Gazin, 1941a, and all ungulates, and that reduction and loss of a then new species, E. grangeri, and resur- such cusps in advanced mioclaenines is a spe- rected Trouessart's (1904) genus Promio- cialization presumably related to the inflation claenus for most of the remaining species. of the premolars. M3 of Promioc/aenus are Prior to publication of Matthew's memoir reduced as in advanced mioclaenines (except but with a copy ofit in hand, Simpson (1936b) Choeroclaenus). Pertinence of Dracoclaenus had in the meantime removed E. priscus and Protoselene to this group (Van Valen, Matthew, 1937, to his new genus Tiznatzinia. 1978) is less clear. The two genera are prob- It is apparent that Simpson's concept of El- ably closely related (Gazin, 1941a), sharing lipsodon was largely influenced by the species such advanced features as a P4 metacone and then referred to it, because the genoholotype upper molar mesostyles. Simpson (1937) is very inadequate, as already mentioned. Van placed Protoselene among the hyopsodon- Valen (1978) placed Tiznatzinia prisca in a tines, presumably because of the median po- new genus, Bomburia. I find Bomburia prisca sition of the lower molar paraconid, but the indistinguishable from Ellipsodon as that ge- condition in Dracoclaenus is rather more nus is currently understood, and therefore re- mioclaenine-like, perhaps indicating deri- turn to Matthew's placement of the species. vation ofthe former from the latter type with- Simpson (1 936b) also placed Mioclaenus tur- in this lineage. The posterior premolars of gidunculus Cope, 1888 (referred by Matthew Dracoclaenus and Protoselene are enlarged as to Oxyacodon) in Tiznatzinia, but later (1937) in mioclaenines, but the lengthened M3 and correctly removed it to a new genus, Choe- presence of a P4 metacone present several roclaenus. Van Valen (1978) placed the ge- character conflicts with other genera placed notype of Tiznatzinia, T. vanderhoofi, which in the subfamily. Conceivably, a P4 metacone 1983 CIFELLI: CONDYLARTHRA 35 was faintly present, about as in Litaletes, in the specialized P4, it belongs to the Mioclae- the most recent common ancestor of all in- ninae. The reduced M3, expanded P4 meta- cluded genera and was lost in Promioclaenus conule, and even the manner in which the and more advanced forms but further de- upper molar hypocone is attached to the pro- veloped in the divergent Protoselene group. tocone by a sweeping crest are features which Simpson (1945) included Oxyacodon, Pro- are adumbrated by Promioclaenus aquilo- toselene, Oxytomodon, Litomylus, Hapla- nius. The presence of a P4 metacone in Hap- letes, Dracoclaenus, Litolestes, Haplomylus, lomylus would accord also with that feature and Hyopsodus in the Hyopsodontinae. Dra- being primitively present in mioclaenines. coclaenus and Protoselene have been men- These-specializations are lacking in Hapla- tioned in connection with the Mioclaeninae. letes, which although generally primitive is Litolestes, considered a probable; insectivore in at least several respects (e.g., development by Jepsen (1930), was referred on the basis of a postprotocone cingulum on M3, P4 para- of better material to the Hyopsodontidae by conid placed lingually) more similar to Hyop- Simpson (1936c), a course followed by Gazin sodus. (1956). Krishtalka (1976) correctly, I believe, The typical and specialized members ofthe removed it to the Erinaceidae. Van Valen family Periptychidae were quickly recog- (1978) synonymized Oxytomodon Gazin, nized as a taxonomic unit by Cope (1882c), 1941 with Litomylus Simpson, 1935c, but who allied them to the Phenacodontidae in the former is distinct in the better developed, his order Condylarthra (1884b), but subse- lingually placed lower molar paraconid (as quently (1897) placed the periptychids in the Gazin observed) and in the distinct height "Amblypoda" based on pedal structure. This differential between trigonid and talonid. course was followed by Osborn (1898) and These primitive features suggest that Oxy- most other students, until Simpson (1937) tomodon is closer to the Oxyclaeninae, to returned them to the Condylarthra, noting which I refer it. Oxyacodon seems more that the vague tarsal resemblance of certain probably periptychid than hyopsodontid as large, advanced periptychines to primitive first suggested7 by Matthew (1914) and as re- pantodonts was lacking in smaller, more cently placed by Van Valen (1978). Aletodon primitive but undoubted members of the Gingerich, 1977, does appear to belong to the family (anisonchines) and that dental mor- Hyopsodontinae as a probable near relative phology argued decisively against pantodont of Litomylus. As here considered, the affinities. This view has not been seriously subfamily thus includes Litomylus, Hapla- challenged by later workers. Simpson (1945) letes, Aletodon, Hyopsodus, and with grave included the generally more primitive forms doubt, Haplomylus. In all, the lower molar Hemithlaeus, Conacodon, Anisonchus, Cori- paraconid, where present, is reduced and oc- phagus, and Haploconus in the subfamily cupies a median position, as Simpson (1937, Anisonchinae, and Ectoconus, Periptychus, p. 227) observed. In a number of respects, and Carsioptychus in the Periptychinae. such as the simple, unenlarged P4 protocone, Coriphagus Douglass, 1908 was referred by Litomylus seems to be the most primitive of Matthew and Granger (1921) to the Oxy- this assemblage, but it does have an ante- claeninae (under its junior synonym, Mixo- riorly expanded P4 trigonid, seen also in Hap- claenus) and by Simpson (1937) to the Per- lomylus and Aletodon. The latter genus ?aut- iptychidae; Van Valen (1967) removed it to apomorphously lacks the paraconid and the Pentacodontidae. Van Valen (1978) syn- metaconid on P4. The late Paleocene and ear- onymized Carsioptychus Simpson, 1936d, ly Eocene Haplomylus is curiously divergent with Periptychus Cope, 1881 b, rearranged the and it is distinctly possible that, despite the contents of the subfamilies, and proposed paraconid structure of the lower molars and several new genera and a new subfamily for Perutherium. As noted elsewhere, the famil- 7Periptychid affinities for this genus were actually ex- ial and even ordinal affinities of Perutherium pressed earlier, since a species referred to Oxyacodon by are highly uncertain. Of the new genera pro- Matthew ( 1937), 0. agapetillus, was originally described posed by Van Valen, Tinuviel (the type of by Cope (1884a) under Anisonchus. which is an M3, but the diagnosis mentions 36 AMERICAN MUSEUM NOVITATES NO. 2772 upper molars) and Fimbrethil (based on low- on lower dentitions and is otherwise un- er cheek teeth) are insufficiently known for known from upper teeth. Comparison of the present purposes and are disregarded here. type species Oxyacodon apiculatus Osborn The swollen but subconical P31 of Fimbreth- and Earle, 1895 and Conacodon entoconus il, lacking paraconid and metaconid, are rem- (Cope, 1882e) with A. cophater upholds Van iniscent of Hemithlaeus, and it may be allied Valen's view. Anisonchus cophater more to or synonymous with that genus (both de- closely resembles 0. apiculatus in the less rive from the same horizon in the Puercan inflated P34 with paraconid and metaconid ofthe San Juan Basin). The type ofEarendil, vestigially present; in the unreduced, un- E. undomiel Van Valen, 1978, is identical modified lower molar paraconids (primitive with a referred species ofMimatuta, M. min- features); and in the anteroposteriorly com- uial Van Valen, 1978, and both are from the pressed lower molar talonids, with the ento- same locality, Mantua lentil. Judging by the conids projecting lingually and with the hy- figures, the type species M. morgoth (from poconid, hypoconulid, and entoconid aligned Harbicht Hill) also is congeneric, though spe- transversely (derived features). cifically distinct. I consider these two genera With the inclusion in the family Perip- synonymous and elect Mimatuta as the se- tychidae of very primitive forms such as nior name. Maiorana noctiluca, another Mimatuta, distinction of periptychids from Mantua genotype proposed by Van Valen, is oxyclaenine arctocyonids and mioclaenine similar to Mimatuta but differs in lacking the hyopsodontids becomes less clear, although characteristically periptychid shallow lingual diagnostic and stereotyped specializations do slope on the upper molars, and is in other appear in more advanced anisonchines. Pe- respects so primitive that it is morejustifiably riptychids appear primitively to be charac- included in the Oxyclaeninae. Van Valen terized by an enlarged P4 protocone, upper (1978) also erected Mithrandir as a subgenus molars with a shallow lingual slope, low an- of Anisonchus, including in it the species terior and posterior cingula with the hypo- Haploconus gillianus Cope, 1882e, Anison- cone developing very lingually on the latter, chus onostus Gazin, 1939, and A. fortunatus upper molar trigon crests (postprotoconule Simpson, 1932b, as well as the type, A. oli- and premetaconule cristae) salient and di- gistus Gazin, 1941, and placed the species A. rected transversely, toward their respective sectorius Cope, 1881 c (genotype), A. dracus buccal cones, and slightly inflated posterior Gazin, 1939, and A. athelas and A. eowynae lower premolars. All except Oxyacodon and in the subgenus Anisonchus. Rigby (1981) Mimatuta have reduced upper molar styles rearranged this grouping and placed A. gil- and stylar shelves (redeveloped in Ectocon- lianus in a new genus Gillisonchus, based on us), strong hypocone, and conules indistinct comparison with A. willeyi and A. sectorius, except for their supporting crests. Inclusion considered by him representative of the sub- of "Anisonchus" cophater in Oxyacodon im- genera Mithrandir and Anisonchus, respec- plies a probable relationship of that genus tively. Anisonchus willeyi is not, however, the with Conacodon and Haploconus, based on type species of A. (Mithrandir), this is A. oli- the loss of the protocone on P3 and the ex- gistus. I find A. gillianus generically distinct tremely lingual position of the upper molar from A. sectorius but not from A. oligistus, hypocone, particularly on M3. The last two, and therefore use the prior name for the new in turn, have distinctive lower molar trigon- genus, Mithrandir (Van Valen, 1978), new ids (extreme in Haploconus), in which the Ml rank. As mentioned above, Oxyacodon, re- paraconid is shifted anteriorly and the para- ferred by Matthew (1937) to the Hyopso- conid is lost on M2-3. There are numerous dontidae, is now included in the Periptych- crosscutting specializations in these and es- idae. Van Valen (1978) referred Anisonchus pecially the remaining periptychid genera, and cophater Cope, 1884a (placed by Matthew, a number of convergences plague any phy- 1897, 1937, in Conacodon) to Oxyacodon. logenetic arrangement of them. The possi- This is a significant transfer because Anison- bility that the Periptychinae (Periptychus, in- chus cophater is represented by a rather com- cluding Carsioptychus, and Ectoconus) is plete dental series and Oxyacodon is based polyphyletic, indicated in other terms by Van 1983 CIFELLI: CONDYLARTHRA 37

Valen (1978, fig. 6) is suggested by certain somewhat labially from the P4 protoconid, resemblances between Anisonchus sectorius terminating in a simple, unicuspid heel or and Periptychus, especially the strongly cres- talonid (this is one ofthe few features in which centic protocones on P34. Ectoconus and Asmithwoodwardia, which has a more mo- Hemithlaeus, while lacking this develop- larized talonid, seems more advanced than ment, both have a multicusped P4 talonid, didolodonts such as Didolodus). As with the upper molar pericone (features also seen in upper molars, M1-3 are relatively bunoid; Periptychus), mesostyle, and arcuate cingula there is relatively little height differential be- on the anterior and posterior faces of the P4 tween trigonid and talonid. M1-3 are subequal protocone. Mithrandir bears resemblances to in size or increase slightly in length from first both Hemithlaeus and Anisonchus. to third, but the increase is not marked. The A broad comparison of the Didolodonti- paraconid, most distinct on M3, is in a lingual dae and primitive Litopterna, as recognized position and is closely appressed to but in- here, with reference to oxyclaenine arcto- completely fused with the metaconid (about cyonids such as Protungulatum, Ragnarok, as in Miguelsoria); the more complete fusion and Maiorana as outgroups for determina- in Asmithwoodwardia and Protolipterna on tion of relative primitiveness of morpholog- the one hand and the more distinct paraconid ical character states, allows certain aspects of with looping paralophid (in Lamegoia among a hypothetical didolodont/litoptern ancestor the didolodonts and Anisolambda among the to be reconstructed. This ungulate is small, Litopterna) on the other representing spe- hyopsodontid in size; the dental series would cializations attained in various lineages. The have been complete and similar in most re- cristid obliqua takes a median anterior at- spects to that of Asmithwoodwardia scotti. tachment to the trigonid; the talonid cusps The anterior teeth are unspecialized; the low- arejoined by low ridges, with entoconid close er canine may be somewhat differentiated to hypoconulid, but all are distinct. from I3 and P1 but no diastemata are present. Derivation of the Litopterna and South P3 bears a small metacone (?lost in Paula- American condylarths directly from the Arc- coutoia and some litoptems) on the posterior tocyonidae is possible, but this is merely per- flank ofthe paracone but close to the apex of missive in the sense that oxyclaenines such that cusp, and the protocone is low and pos- as Protungulatum are primitive enough in teriorly placed; the large, inflated P3 proto- known structure to be ancestral to all groups cone of Didolodus almost surely represents a of ungulates, and special relationship is not derived condition. P4 has both paracone and indicated by any advanced features shared in metacone, as in P3; the styles are reduced, the common which would exclude other primi- protocone is enlarged and transversely aligned tive ungulates. The South American primi- with the paracone, and a protoconule is pres- tive morphotype is more advanced in most ent. The upper molars have relatively low respects than either the loxolophines or arc- cusps and well-developed protostylar cingu- tocyonines and lacks the few known special- la, are less transverse than in Protungulatum, izations of either group. Van Valen (1978) and have a weak external cingulum and vir- referred the Late Cretaceous Perutherium al- tually no styles. A hypocone, completely tiplanense to the Periptychidae in a new lacking on M3, is well developed and almost subfamily, Perutheriinae, but as mentioned fully incorporated into the crown surface on above I consider remains currently assigned M1-2, although the protocone occupies its to this species indeterminate as to family and central position in the trigon and the molars order and not even demonstrably eutherian. are therefore not quadrate. A mesostyle and Ameghino (e.g., 1906, p. 467) placed forms similar elaborations are lacking. P3 is simple, such as Josepholeidya (now assigned to the uninflated, slightly compressed transversely, Litopterna) in the Periptychidae, but resem- and bears a faint heel; P4 has the full com- blance to the North American family is per- plement oftrigonid cusps with paraconid and haps closer for such advanced didolodonts metaconid placed relatively high, but the for- such as Didolodus or Lamegoia. The resem- mer is in a median (primitive), not lingual blances, such as the enlargement and infla- position. A ridge descends posteriorly and tion of the posterior upper premolars (ex- 38 AMERICAN MUSEUM NOVITATES NO. 2772 treme in P4, at least, of Lamegoia) and the small, that is to say, more primitive in this presence of an upper molar protostylar cusp respect than known phenacodonts. (pericone) and mesostyle are, however, su- Paula Couto (1952a) at first referred the perficial and imprecise (the upper premolars Itaborai species Asmithwoodwardia scotti to of didolodonts possess a metacone and are the Hyopsodontidae, subfamily Hyopsodon- differently specialized), and are not present tinae, on the basis of size and overall dental in the more primitive South American forms similarity to North American Paleocene and or in the ancestral morphotype postulated Eocene hyopsodontines. The upper molars of above. The basic periptychid dental features, hyopsodontines (as that group is considered as discussed in the preceding pages, are ab- here) are rather close to the primitive South sent in all known didolodonts and litopterns. American morphotype, although in Hyop- Students in the past have found the typical sodus the hypocone is less conical and more didolodonts most comparable to primitive fully incorporated into the crown surface, members of the Phenacodontidae (Ameghi- being partially united to the protocone. The no, 1901, 1904; Scott, 1937; Simpson, 1948) lower molars ofthe Hyopsodontinae have the such as Tetraclaenodon and Phenacodus. paraconid (completely lacking in Hyopsodus Overall resemblance of these genera to, for and Aletodon) in a median position, however, instance, Didolodus is close, but the basic ad- unlike the South American forms. P4 of the vanced features uniting the Phenacodontidae Litomylus group is somewhat specialized. (e.g., lower molar metastylid, molarized P4 With the removal of Dracoclaenus and Pro- talonid) are wanting. The primitive South toselene to the Mioclaeninae, only Haplo- American morphotype lacks the specialized mylus of known hyopsodontines has a meta- characters of either phenacodont subgroup cone developed on the posterior upper (posteriorly shifted upper molar metaconule, premolars, and the reference of this genus to lower molar anterior entoconid accessory the Hyopsodontinae is highly uncertain. cuspule of Phenacodus and Tetraclaenodon; Although not previously considered in de- lophate features of Ectocion and allies). The tail as possible allies of the South American development of the lower molar paraconid ungulates, the North American forms most in the South American hypothetical ancestor closely comparable to an ancestral didolo- is like that of Tetraclaenodon, but this seems dont/litoptern morphotype appear to lie to have developed within the Phenacodus within the Mioclaeninae, particularly genera subgroup, judged by comparison with Ecto- such as the Torrejonian Litaletes. The lower cion. The most fundamental differences lie in molars of the hypothetical South American a differential molarization of the posterior ancestor are possibly advanced in being premolars. Simpson (1948) noted that P3-4 slightly more bunoid, with the talonid cusps are almost equally molariform in Didolodus, somewhat more distinct, but in other essen- whereas in Tetraclaenodon P4 is definitely tials, especially the disposition of the para- more complex than P3. The condition in Di- conid, are very similar. P4 of Litaletes is as dolodus does not pertain, however, to the predicted for the South American morpho- primitive South American morphotype, ex- type, but P3 is somewhat more derived in emplified in this case by Asmithwoodwardia. bearing a small paraconid and metaconid and The distinction in the upper premolars seems a slightly better developed heel. The upper instead to be that the metacone arises more premolars are molarized in the same fashion, posteriorly in phenacodonts (along the me- in contrast to the phenacodont pattern, with tastylar ridge), with an anterior shift in the the protocone ofP4 large, transversely aligned paracone position, so that the three major with the paracone, and a small metacone near cusps are arranged in a trigonal pattern as in the apex of the latter cusp. As noted in the the molars. It is significant also that the phe- above discussion, the presence of a P3 meta- nacodonts are characterized by their large size cone in the primitive South American mor- with respect to the primitive ungulate con- photype, as in Litaletes and Asmithwood- dition, and that their presumed nearest allies, wardia is problematic. Also uncertain is the loxolophine arctocyonids, are also relatively significance of a differentiated lower canine large. The South American morphotype is in the protolipternids, which may be autapo- 1983 CIFELLI: CONDYLARTHRA 39 morphic or a condition more primitive than dont common ancestor and primitive mio- any hyopsodontid. claenines such as Litaletes is most corrobo- Because of South America's status as an rated by shared advanced features and least island continent throughout the Tertiary, with falsified by crosscutting specializations. only minor faunal exchange with the rest of Somewhat less probable but still plausible is the world, the history of its biota has been the alternative hypothesis, suggested by many heralded as the "greatest natural experiment previous workers, that didolodonts and li- in biogeography." Ameghino based his view topterns took origin from a common ancestor that Neotropical mammals were ancestral to with or within the Phenacodontidae. In either those of the rest of the world on his-belief case, the common ancestor would probably that the early South American faunas were have been a middle Paleocene (Torrejonian of Late Cretaceous age, significantly older North American Land Mammal Age) form. than beds from which comparable North Virtually none of the structural specializa- American faunas were known to derive. The tions-even regarded in a gradal and not phy- currently accepted island continent concept, logenetic sense-shared by the Holarctic and forwarded by Gaudry (1906), was developed South American ungulates are known in in detail by Simpson (e.g., 1950), who incor- North American condylarths prior to that porated the suggestion of Matthew (1915c) time, so that postulation of an earlier com- that the distinctive South American land mon ancestor implies convergence or paral- mammal fauna took its origin from North lelism in these features. The evidence as pre- American emigrants. This biogeographic sented here suggests a Torrejonian common model rests largely on the fact that proto- ancestor; any earlier form would be permis- therians and most of the eutherians so char- sive as an ancestral morphotype but would acteristic of Holarctic faunas are entirely be completely uncorroborated by supporting lacking at any time in South America's rich evidence of advanced features. By the Rio- Tertiary record. By this view, all South chican, or presumed late Paleocene of South American ungulates should with great prob- America (?Tiffanian North American equiv- ability have descended from a single late Cre- alent), considerable radiation of didolodonts taceous or early Tertiary ancestor, presum- and litopterns had already occurred, so that ably a mammal which would with convention this common ancestor could not have existed be placed in the Condylarthra, thus implying later than middle Paleocene. Because of the that South American ungulates should be aforementioned lack of Prototheria and typ- strictly monophyletic with respect to those of ical North American Eutheria in South other continents. This is expressed in the American faunas, such a common ancestor classification of McKenna (1975) in which is likely to have immigrated via sweepstakes the mirorder Meridungulata is erected to con- dispersal, probably but not necessarily from tain South America's native ungulate orders. north to south, rather than having a once Strict monophyly is not required by this Pan-American distribution disrupted by a model but is a maximum parsimony corre- "vicariance event." late: given the composition ofthe native Neo- Although detailed treatment of the other tropical mammal fauna with respect to that indigenous South American ungulates is not of Holarctica, it is unlikely that the South possible here, it is ofinterest to consider them American continent was originally populated briefly in this context. These mammals are by more than one kind of ungulate. generally grouped into five orders: Notoun- The morphological comparisons and hy- gulata, Astrapotheria, Trigonostylopoidea, pothesized relationships of and among the Xenungulata, and Pyrotheria, ofwhich some Didolodontidae and Litopterna, as detailed are known only by one or a few genera and and discussed above, merit some consider- may not be ordinally distinct. Notoungulates, ation with respect to this biogeographic mod- characterized by diagnostic specializations of el. Although distinctive and uniquely derived the dentition (Patterson, 1934) and ear region morphological character states are rare among (Patterson, 1936; Simpson, 1948; but see nearly all the mammals under consideration, Pascual, Vucetich, and Fernandez, 1978) are a relationship between the litoptern/didolo- already well differentiated by the Riochican, 40 AMERICAN MUSEUM NOVITATES NO. 2772 although the distinction between the major certain respects, as reflected by their place- groups at that time is not great and a fairly ment in Simpson's (1934) suborder Notio- recent common origin is suggested. As rep- progonia, but in many other respects (such resented by the Henricosborniidae, the most as the tremendously developed anteropos- primitive of the known Notoungulata, the terior vertical shearing surfaces on both up- dentition is already specialized and distinctly per and lower molars) they are aberrantly spe- advanced over the ancestral didolodont/li- cialized and unlike any South American toptern morphotype. Derivation of notoun- notoungulate. It should be noted that some gulates from this pattern is possible, and the of the hallmark specializations of notoun- primitive notoungulate dentition is in fact gulates, such as the transverse lower molar vaguely hyopsodont-like.8 The primitive no- entolophid and trigon complications (cristae, toungulate tarsus bears several advanced fea- crochet) of the upper molars are lacking in tures suggestive ofPromioclaenus and Hyop- primitive but undoubted arctostylopids. The sodus but not of other North American premolars are exceedingly simple (possibly condylarths (Cifelli, 1983) and work in prog- but not demonstrably a secondary reduction), ress suggests common specializations in the and in the most primitive species, Asiostylops manus of notoungulates and Hyopsodus as speciosus, the upper molars lack a hypocone, well. Tarsals of didolodonts and litopterns a condition more primitive than that in the are strongly derived and afford no evidence Henricosborniidae or in the hypothesized li- as to whether or not these characters were toptern/didolodont ancestor. Comparison present in their hypothetical ancestral mor- with more advanced arctostylopid genera such photype. The biogeographical history of no- as Anatostylops and Palaeostylops indicates toungulates is complicated by reference to the further that the posterointernal cusp of typ- order of the North American (Matthew, ical arctostylopid upper molars is not a true 1915d) and Asian (Matthew and Granger, hypocone, but a displaced metaconule which 1925) family Arctostylopidae. The auditory has become transversely expanded into a and pedal anatomy of arctostylopids has not metaloph. On this basis I would remove the yet been described (a partial skull of Asio- Arctostylopidae from the Notoungulata, but stylops is known), but numerous specializa- have no further suggestion as to their affini- tions of the molars are found only in no- ties. Closest resemblance seems to be with toungulates among mammals. As the name Lantianius, a somewhat atypical ?arctocyon- suggests, the cheek teeth are most comparable id from the ?Eocene of China, placed by its in structure to the Notostylopidae. Previ- describer in the Adapidae (Chow, 1964). ously known only from the early Eocene, arc- Pyrotheres, which appear in the Casama- tostylopids are now also known from the late yoran, and the single xenungulate genus Car- Tiffanian of North America (C. Schaff, per- odnia, known only from the Riochican, are sonal commun.) and presumably time equiv- so strongly modified at first appearance that alent beds of Mongolia (P. Gingerich, per- it is difficult to determine their relationships. sonal commun.), although the age ofthe newly Ameghino (e.g., 1906) believed pyrotheres to described Mongolian species is uncertain. be proboscideans, a course followed by Thus, northern (Matthew and Granger, 1925), Loomis (1914). Based on a restudy of the southern (Hoffstetter, 1970), and central auditory region of Pyrotherium, Patterson American (Gingerich and Rose, 1977) origins, (1977) considered the group as pertaining to with appropriate migration and dispersal, the Notoungulata, but this reference has not have been proposed for the Notoungulata. generally been followed because, as Simpson Central to this issue is the position of the (1978) pointed out, the differences in dental Arctostylopidae within the order, a problem specializations are irreconcilable. The dental which has never been addressed in detail. similarity of Carodnia to the Holarctic Dino- Arctostylopids are exceedingly primitive in cerata or uintatheres has been commented on since the description of the genus (Simpson, 8 Ameghino (e.g., 1906) originally placed "Selenoco- 1935b; Wheeler, 196 1; McKenna, 198 1), but nus" (=Henricosbornia) in the Hyopsodontidae, consid- the tarsus is differently specialized in the two ered at that time to be prosimian primates. groups (Cifelli, 1983). In this respect, Car- 1983 CIFELLI: CONDYLARTHRA 41 odnia seems allied to true pyrotheres, to which oping metaloph is a trend among other tri- I refer it. The manus of true pyrotheres is gonostylopids, such as Albertogaudrya and unfortunately unknown.9 Pyrotheres, includ- Tetragonostylops. Thus, ifthe foregoing anal- ing Carodnia, could plausibly be derived from ysis is correct, astrapotheres would be ex- the same South American ancestor which gave cluded from a common origin with didolo- rise to didolodonts and litopterns, although donts, litopterns, ?and notoungulates. such profound structural modifications Astrapotheres may ultimately be related to would, a priori, seem unlikely to have oc- pyrotheres, but there is at present little evi- curred in the relatively short span of time dence in support of or against this. Astra- between the hypothesized middle Paleocene potheres, like the Dinocerata, are character- age of the ancestral morphotype and the first ized by a strongly alternating tarsus, with great appearance (late Paleocene) of the Pyro- development of the medial malleolus of the theria, sensu lato. tibia as a weight supporting area; pyrotheres, Trigonostylops and closely allied forms like the Proboscidea and Embrithopoda, have have generally been placed within the a serial tarsus with modified calcaneofibular Astrapotheria, but Simpson (1967) recently contact (Cifelli, 1983). As with these non- erected the new order Trigonostylopoidea, Neotropical forms, the ankles of pyrotheres based on detailed comparisons of available and astrapotheres bear a number of "ambly- crania and dentitions. Other studies (Car- pod" characters associated partially with gra- bajal et al., 1977; Cifelli, 1983; Soria, per- viportal habitus but more closely with pen- sonal commun.) very strongly support a close tadactyl digitigrady. A remote but exclusive relationship between the Astrapotheriidae and ancestry for the pyrotheres and astrapotheres Trigonostylopidae, and the differences noted is vaguely suggested but not demanded by by Simpson seem to be cases ofrelative prim- these features. itiveness rather than drastically different spe- In sum, derivation of the litopterns and cializations indicative of a very remote and didolodonts from a Litaletes-like mioclae- non-exclusive common ancestry. The family nine hyopsodont implies that South Ameri- Trigonostylopidae may be returned, accord- can ungulates may not be a monophyletic ingly, to the Astrapotheria. Trigonostylops, unit. Astrapotheres and probably pyrotheres in most respects the most primitive astra- seem unlikely to have derived from this com- pothere, is characterized by a number of mon ancestry because too little time occurs modifications of the dentition, including ca- between the predicted age of this common nine tusks, molarized premolars (P3' have a ancestor and their first appearance in the rec- well-developed metacone in a triangular ar- ord as large, bizarrely modified animals and, rangement with the paracone and protocone), more importantly, because in at least one case upper molar ectoloph, and lower molars lack- (Trigonostylops) they appear to be more ing the paraconid and with the labial talonid primitive than this common ancestor. No- cusps joined by a crest continuous with the toungulates, excluding the Holarctic Arcto- cristid obliqua. However, the upper molar stylopidae, are reasonably derived from this hypocone is virtually or completely absent in unique common ancestor with didolodonts Trigonostylops, a condition more primitive and litopterns (in fact, the limited evidence than any other known ungulate from South available corroborates this arrangement). America. This seems not to represent a sec- An alternative hypothesis to the foregoing ondary reduction in this genus, because the is that the origin(s) ofthe native South Amer- development of a hypocone and its progres- ican ungulates is(are) far more remote - sive enlargement and union with the devel- whether or not they represent a monophyletic unit with respect to other mammals -and that between the di- 9 A manus described and figured by Ameghino (1897) the resemblances primitive as that of Pyrotherium was shown by Scott (1937) to be dolodont/litoptern morphotype and Holarc- astrapotherian (probably pertaining to Parastrapothe- tic forms are merely the result of adaptive rium), by analogy with a complete astrapothere skeleton convergence. By this interpretation, the one in the collections of the Field Museum of Natural His- or more forms which gave rise to South tory. American ungulates was(were) more primi- 42 AMERICAN MUSEUM NOVITATES NO. 2772 tive, perhaps far more primitive, than has of this model and favors the first hypothesis. previously been thought, and that the ances- This latter thus appears to be the more prob- tor(s) need not necessarily even have been able, although it implies that the South ungulate in the sense enjoined here. In de- American ungulate groups may not represent fense of this view it might be argued that the a monophyletic unit. Numerous other hy- discovery in South America ofprimitive ?un- potheses of the origin and biogeographic de- gulate mammals (such as Asmithwoodward- ployment of the South American ungulates, ia) which are similar to North American Con- especially more precise permutations of the dylarthra is quite unsurprising, considering possible relationships between specific Hol- the known range ofstructural diversity among arctic and Neotropical ungulate groups (such early Tertiary North American Eutheria. If as pyrotheres, astrapotheres, and the "am- monophyly of the Neotropical orders is as- blypod" orders) should be considered but sumed (that is, a single "landfall" of a prim- cannot now be evaluated. Present evidence itive eutherian in South America gave rise to suggests that didolodontids and litopterns all ungulates on the continent, an assumption form a monophyletic group and that they and of maximum parsimony), then acceptance of possibly also the Notoungulata are in turn the Late Cretaceous Perutherium as a euthe- most closely related to North American Mio- rian mammal (not necessarily condylarthran) claeninae; the other indigenous South Amer- related to the South American ungulates vir- ican ungulates possibly but not surely rep- tually demands this interpretation, because resenting radiations from a separate, more Perutherium antedates significantly the mid- remote eutherian (Perutherium, if at all rel- dle Paleocene radiation of North American evant, would thus be related to this group mammals which are structurally similar to and not to the Didolodontidae). Detailed the ancestral didolodont/litoptern morpho- comparative morphological analysis ofastra- type; Perutherium is in fact contemporane- potheres and pyrotheres would be of great ous with Protungulatum, the most ancient interest in this regard, as would be the dis- and primitive known ungulate. It is not pos- covery of pre-Riochican Paleocene mammal sible with the evidence available to recon- faunas in South America. struct hypothetically the morphology of such a remote and primitive common ancestor of LITERATURE CITED all South American ungulates and therefore Ameghino, F. it is difficult to determine when it arrived on 1885. Nuevos restos de mamiferos f6siles oli- the continent and from whence it came. Per- gocenos recogidos por el profesor Pedro haps it would represent a dispersal event be- Scalabrini y pertenecientes al Museo tween North and South America in the Late provincial de la ciudad del Parana. Bol. Cretaceous, but the possibility cannot be ruled Acad. Nac. Cien., C6rdoba, vol. 8, pp. out that it represents a distribution which 5-207. South Atlantic between Africa 1887. Enumeracion sistematica de las especias bridged the de mamiferos f6siles coleccionados por and South America. The record of early Ter- Carlos Ameghino en los terrenos eoce- tiary fossil mammals in Africa is very poor nos de la Patagonia austral y deposita- and offers no evidence on this point. dos en el Museo de La Plata. Bol. Mus. This second hypothesis, that the South La Plata, vol. 1, pp. 1-26. American ungulates arose from an early im- 1889. Contribuci6n al conocimiento de los migrant (i.e., before the middle Paleocene, mamiferos f6siles de la Repfiblica Ar- and possibly much earlier) is favorable in that gentina. Acta Acad. Nac. Cien. C6r- it is permissive of monophyly of the South doba, vol. 6, pp. 1-1027. American forms, but corroborative morpho- 1891. Caracteres diagn6sticas de cincuenta es- the pecias nuevas de mamiferos f6siles ar- logical evidence is entirely lacking. On gentinos. Rev. Argentina Hist. Nat., vol. other hand, the strong similarities (at least 1, pp. 129-167. some of which appear to be shared special- 1894. Enumeration synoptique des especies de izations) between the didolodont/litoptern mammiferes fossiles des formations primitive morphotype and North American eocenes de Patagonie. Bol. Acad. Nac. Mioclaeninae constitutes partial falsification Cien. C6rdoba, vol. 13, pp. 259-452. 1983 CIFELLI: CONDYLARTHRA 43

1897. Mammiferes cretaces de l'Argentine. of Cretaceous placentals. Acta Palaeon- Deuxieme contribution ai la connais- tol. Polonica, vol. 22, pp. 241-271. sance de la faune mammalogique des 1980. The tupaiid dentition. In, Luckett, W. couches a Pyrotherium. Bol. Inst. Geogr. P. (ed.), Comparative biology and evo- Argentino, vol. 18, pp. 406-521. lutionary relationships of tree shrews. 1901. Notices preliminaires sur des mammi- New York, Plenum Publ. Co., pp. 171- feres nouveaux des terrains cretaces de 204. Patagonie. Bol. Acad. Nac. Cien. C6r- Cabrera, A. doba, vol. 16, pp. 350-426. 1935. Estado actual de la cuestion del limite 1902. Premiere contribution a la connaissance cretaceo-terciario en la Argentina. Inst. de la faune mammalogique des couches Mus. Univ. Nac. La Plata, Obra del Cin- a Colpodon. Ibid.,vol. 17, pp. 71-140. cuentenario, pp. 1-22 (cited from Simp- 1903-1904. Nuevas especias de mamiferos cre- son, 1967; not seen by me). taceos y terciarios de la Repuiblica Ar- Carbajal, E., R. Pascual, R. Pinedo, J. A. Salfity, gentina. An. Soc. Cien. Argentina, vol. and M. G. Vucetich 56 (1903), pp. 193-208; vol. 57 (1904), 1977. Un nuevo mamifero de la Formaci6n pp. 162-175, 327-341; vol. 58 (1904), Lumbrera (Grupo Salta) de la Comarca pp. 35-41, 56-71, 182-192, 225-240, de Carahuasi (Salta, Argentina). Edad y 241-291. correlaciones. Publ. Mus. Municip. Cien. 1904. Rech&rches de morphologie phylogene- Nat. Mar del Plata "Lorenzo Scaglia," tique sur les molaires superieures des vol. 2, pp. 148-163. ongul6s. An. Mus. Nac. Buenos Aires, Chow, M. vol. 9 (ser. 3a, vol. 3), pp. 1-541. 1964. A lemuroid primate from the Eocene of 1906. Les formations sedimentaires du cr& Lantian, Shansi. Vert. Palasiatica, vol. tace superieur et du tertiare de Patago- 8, pp. 257-262. nie avec un parallele entre leurs faunes Cifelli, R. L. mammalogiques et celles de l'ancien 1983. Eutherian tarsals from the late Paleo- continent. Ibid., vol. 15, pp. 1-568. cene of Brazil. Amer. Museum Novi- 1913-1936. Obras completas y corresponden- tates, no. 2761, pp. 1-3 1, figs. 1-12. cia cientifica. Torcelli, A. J. (ed.). La Clemens, W. A. Plata:, vols. 1-24. 1973. Fossil mammals of the type Lance For- Beliajeva, E. I., B. A. Trofimov, and V. J. Reshetov mation, Wyoming. Part III. Eutheria and 1974. General stages in evolution of late Me- summary. Univ. California Publ. Geol. sozoic and early Tertiary mammalian Sci., vol. 94, pp. 1-102. faunas in Central Asia. In. Kramaren- 1974. Purgatorius, an early paromomyid pri- ko,-N. N. et al. (eds.), Cenozoic faunas mate (Mammalia). Science, vol. 184, pp. and biostratigraphy of Mongolia. Mos- 903-905. cow, Joint Soviet-Mongol. Paleont. Cope, E. D. Exped., pp. 19-45. 1873. Fourth notice ofextinct Vertebrata from Berg, C. the Bridger and Green River Tertiaries. 1899. Substituci6n de nombres genericos. III. Palaeontol. Bull., vol. 17, pp. 1-4. Comun. Mus. Nac. Buenos Aires, vol. 1881a. A new type of Perissodactyla. Amer. 1, pp. 77-80. Nat., vol. 15, pp. 1017-1018. Bonaparte, J. 1881b. Mammalia of the lower Eocene beds. 1978. El Mesoz6ico de America del Sur y sus Ibid., vol. 15, pp. 337-338. tetrapodos. Tucuman, Minist. Cult. 1881 c. On some Mammalia of the lowest Educ. Fund. Miguel Lillo, pp. 1-596. Eocene beds of New Mexico. Proc. Bordas, A. F. Amer. Phil. Soc., vol. 19, pp. 484-495. 1936. Un nuevo mamifero del Colpodon de 1882a. Some new forms from the Puerco Gaiman (Proheptaconus trelewense, gen. Eocene. Amer. Nat., vol. 16, pp. 833- et sp. nov.). Physis, vol. 12, pp. 110- 834. 112". 1882b. Notes on some Eocene Mammalia. Ibid., 1939. Diagnosis sobre algunos mamiferos de vol. 16, p. 522. las capas con Colpodon del valle del rio 1882c. The Periptychidae. Ibid., vol. 16, pp. Chubut (Repfublica Argentina). Ibid., 832-833. vol. 14, pp. 413-433. 1882d. Synopsis of the Puerco Eocene epoch. Butler, P. M. Proc. Amer. Phil. Soc., vol. 20, pp.461- 1977. Evolutionary radiation ofthe cheek teeth 471. 44 AMERICAN MUSEUM NOVITATES NO. 2772

1882e. New marsupials from the Puerco ology. Proc. U.S. Natl. Mus., vol. 91, Eocene. Amer. Nat., vol. 16, pp. 684- pp. 1-53. 686. 1941 b. Paleocene mammals from the Denver 1883. On the mutual relationships of the bu- Basin, Colorado. Jour. Wash. Acad. Sci., notherian Mammalia. Proc. Acad. Nat. vol. 31, no. 7, pp. 289-295. Sci. Philadelphia, vol. 35, pp. 77-83. 1953. The Tillodontia: an early Tertiary order 1884a. Second addition to knowledge of the of mammals. Smithsonian Misc. Coll., Puerco epoch. Proc. Amer. Phil. Soc., vol. 21, no. l0, pp. 1-110. vol. 21, pp. 309-324. 1956. Paleocene mammalian faunas ofthe Bi- 1884b. The Condylarthra. Amer. Nat., vol. 18, son Basin in southcentral Wyoming. pp. 790-805. Ibid., vol. 131, no. 6, pp. 1-57. 1885. The oldest Tertiary Mammalia. Ibid., 1959. Early Tertiary Apheliscus and Phenac- vol. 19, pp. 385-387. oadaptes as pantolestid insectivores. 1888. Synopsis of the vertebrate fauna of the Ibid., vol. 139, no. 7, pp. 1-7. Puerco series. Trans. Am. Phil. Soc., vol. 1965. A study of the early Tertiary condylar- 16, pp. 298-361. thran mammal Meniscotherium. Ibid., 1897. The position ofthe Periptychidae. Amer. vol. 149, no. 2, pp. 1-98. Nat., vol. 31, pp. 335-336. Gervais, P. Crompton, A. W., and Z. Kielan-Jaworowska 1855. Recherches sur les mammiferes fossiles 1978. Molar structure and occlusion in Cre- de l'Amerique meridionale. Paris, L. taceous therian mammals. In, Butler, P. Martinet, pp. 1-44. M., and K. A. Joysey (eds.), Studies in Gidley, J. W. the development, function, and evolu- 1919. New species of claenodonts from the tion of teeth. New York, Academic Fort Union (basal Eocene) of Montana. Press, pp. 249-287. Bull. Amer. Mus. Nat. Hist., vol. 41, Douglass, E. pp. 541-556. 1908. Vertebrate fossils from the Fort Union Gill, T. beds. An. Carnegie Mus., vol. 5, pp. 11- 1872. Arrangement of the families of mam- 26. mals. With analytical tables. Smithson- Fox, R. C. ian Misc. Coll., vol. I1, no. I, pp. 1-98. 1970. Eutherian mammal from the early Gingerich, P. D. Campanian (late Cretaceous) of Alber- 1977. Aletodon gunnelli, a new Clarkforkian ta, Canada. Nature, vol. 227, pp. 630- hyopsodontid (Mammalia, Condylar- 631. thra) from the early Eocene of Wyo- 1979. Mammalia from the upper Cretaceous ming. Contrib. Mus. Paleontol., Univ. Oldman Formation, Alberta. III. Euthe- Michigan, vol. 24, pp. 237-244. ria. Can. J. Earth Sci., vol. 16, no. 1, pp. 1978. New Condylarthra (Mammalia) from 114-125. the late Paleocene and early Eocene of Francisco, B. H. R., and F. L. de Souza Cunha North America. Ibid., vol. 24, pp. 245- 1978. Geologia e estratigraf'la de bacia de Sao 279. Jose, Municipio de Itaborai, RJ. An. 1979. Lambertocyon eximius, a new arcto- Acad. Brasil. Cien., vol. 50, pp. 381- cyonid from the late Paleocene of west- 416. ern North America. Jour. Paleontol., vol. Gaudry, A. 53,pp.524-529. 1904. Fossiles de Patagonie-dentition de Gingerich, P. D., and K. D. Rose quelques mammiferes. MWm. Soc. Geol. 1977. Preliminary report on the American France, vol.12, no. 1, m6m. 31, pp. 1- Clark Fork mammal faunas, and its cor- 26. relation with similar faunas in Europe 1906. Fossiles de Patagonie. Etude sur une and Asia. Geobios, Mem. Spec., vol. 1, portion du monde Antarctique. Ann. pp. 39-45. Paleont., vol. 2, pp. 101-143. Grambast, L., M. Martinez, M. Mattauer, and L. Gazin, C. L. Thaler 1939. A further contribution to the Dragon 1967. Perutherium altiplanense, nov. gen., Paleocene fauna of central Utah. Jour. nov. sp., premier mammifere meso- Wash. Acad. Sci., vol. 29, no. 7, pp. zoique d'Amerique du Sud. C. R. Acad. 273-286. Sci. Paris, ser. D, vol. 264, pp. 707-710. 1941 a. The mammalian fauna ofthe Paleocene Granger, W. of central Utah, with notes on the ge- 1915. A revision ofthe lower Eocene Wasatch 1983 CIFELLI: CONDYLARTHRA 45

and Wind River faunas. Part III Order Concord, N.H., The Rumford Press, pp. Condylarthra. Families Phenacodonti- 1-232. dae and Meniscotheriidae. Bull. Amer. Marshall, L. G. Mus. Nat. Hist., vol. 34, pp, 329-361. 1978. Evolution of the Borhyaenidae, extinct Gregory, W. K. South American predaceous marsupi- 1910. The orders of mammals. Bull. Amer. als. Univ. California Publ. Geol. Sci., Mus. Nat. Hist., vol. 27, pp. 1-524. vol. 117, pp. 1-89. Hoffstetter, R. Marshall, L. G., R. Hoffstetter, and R. Pascual 1970. Radiation initiale des mammiferes pla- [in press] Geochronology of the continental centaires et biogeographie. C. R. Acad. mammal bearing Tertiary of South Sci. Paris, ser. D, vol. 270, pp. 3027- America. 3030. Matthew, W. D. 1981. Historia biogeogriafica de los mamiferos 1897. A revision of the Puerco fauna. Bull. terrestres sudamericanos: problemas y Amer. Mus. Nat. Hist., vol. 9, pp. 259- enseiianzas. Acta Geol. Hisp., vol. 16, 323. pp. 7 1-88. 1914. Evidence of the Paleocene vertebrate Jepsen, G. L. fauna on the Cretaceous-Tertiary prob- 1930. Stratigraphy and paleontology of the lem. Bull. Geol. Soc. Amer., vol. 25, pp. Paleocene of northwestern Park Coun- 381-402. ty, Wyoming. Proc. Amer. Phil. Soc., 1915a. A revision ofthe lower Eocene Wasatch vol. 69, pp. 463-528. and Wind River faunas. Part I. Order Kielan-Jaworowska, Z. Ferae (Carnivora). Suborder Creodon- 1975. Preliminary description of two new eu- ta. Bull. Amer. Mus. Nat. Hist., vol. 34, therian genera from the late Cretaceous pp. 4-103. of Mongolia. Results Polish-Mongolian 1915b. A revision ofthe lower Eocene Wasatch Palaeontol. Exped. Pt. VI. Palaeontol. and Wind River faunas. Part II. Order Polon., vol. 33, pp. 5-16. Condylarthra, Family Hyopsondonti- Kielan-Jaworowska, Z., T. M. Bown, and J. A. dae. Ibid., vol. 34, pp. 311-328. Lillegraven 1915c. Climate and evolution. Ann. New York 1979. Eutheria. In, Lillegraven, J. A., Z. Kie- Acad. Sci., vol. 24, pp. 171-318. lan-Jaworowska, and W. A. Clemens 1915d. A revision ofthe lower Eocene Wasatch (eds.), Mesozoic mammals. Berkeley, and Wind River faunas. Part IV. En- Univ. California Press, pp. 221-258. telonychia, Primates, Insectivora (part). Krishtalka, L. Bull. Amer. Mus. Nat. Hist., vol. 34, 1976. Early Tertiary Adapisoricidae and Eri- pp. 429-483. naceidae (Mammalia, Insectivora) of 1937. Paleocene faunas ofthe San Juan Basin, North America. Bull. Carnegie Mus. New Mexico. Trans. Amer. Phil. Soc., Nat. Hist., vol. 1, pp. 1-40. n.s., vol. 30, pp. 1-5 10. Lemoine, M. V. Matthew, W. D., and W. Granger 1891. Etude d'ensemble sur les dents des 1921. New genera of Paleocene mammals. mammiferes fossiles des environs de Amer. Mus. Novitates, no. 13, pp. 1-7. Reims. Bull. Soc. Geol. France, ser. 3e, 1925. Fauna and correlation of the Gashato vol. 19, pp. 263-290. Formation of Mongolia. Ibid., no. 189, Lillegraven, J. A. pp. 1-12. 1969. Latest Cretaceous mammals of upper McKenna, M. C. part of Edmonton Formation of Alber- 1956. Survival ofprimitive notoungulates and ta, Canada, and review of marsupial- condylarths into the Miocene ofColom- placental dichotomy in mammalian bia. Amer. Jour. Sci., vol. 254, pp. 736- evolution. Univ. Kansas Paleontol. 743. Contrib., art. 50 (Vertebrata 12), pp. 1- 1960. Fossil Mammalia from the early Wa- 122. satchian Four Mile fauna, Eocene of Lillegraven, J. A., Z. Kielan-Jaworowska, and W. northwestern Colorado. Univ. Califor- A. Clemens (EDS.) nia Publ. Geol. Sci., vol. 37, pp. 1-130. 1979. Mesozoic mammals. The first two-thirds 1969. The origin and early differentiation of of mammalian history. Berkeley, Univ. therian mammals. Ann. New York California Press, pp. 1-311. Acad. Sci., vol. 167, pp. 217-240. Loomis, F. B. 1975. Toward a phylogenetic classification of 1914. The Deseado Formation of Patagonia. the Mammalia. In, Luckett, W. P., and 46 AMERICAN MUSEUM NOVITATES NO. 2772

F. S. Szalay (eds.), Phylogeny ofthe pri- Patterson, B., and P. 0. McGrew mates. New York, Plenum Publ. Co., 1962. A new arctocyonid from the Paleocene pp. 21-46. of Wyoming. Breviora, vol. 174, pp. 1- 1981. Early history and biogeography of 10. South America's extinct land mam- Patterson, B., and R. Pascual mals. In, Ciochon, R. L., and A. B. Chi- 1972. The fossil mammal fauna of South arelli (eds.), Evolutionary biology ofthe America. In, Keast, A., F. C. Erk, and New World monkeys and continental B. Glass (eds.), Evolution, mammals, drift. New York, Plenum Press, pp. 43- and southern continents. Albany State 77. Univ., New York Press, pp. 247-309. Murray, A. Patterson, B., and R. M. West 1866. The geographic distribution of mam- 1973. A new late Paleocene phenacodont mals. London, Day and Son, pp. 1-420. (Mammalia, Condylarthra) from west- Odreman Rivas, 0. E. ern Colorado. Breviora, vol. 403, pp. 1969. Los Polymorphinae, un diferente tipo 1-7. adaptivo de los Proterotheriidae (Mam- Patterson, C., and D. Rosen malia, Litopterna). Ameghiniana, vol. 1977. Review of ichthyodectiform and other 6, no. 1, pp. 57-64. Mesozoic teleost fishes and the theory Osborn, H. F. and practice of classifying fossils. Bull. 1898. Evolution of the Amblypoda. Part I. Amer. Mus. Nat. Hist., vol. 158, pp. 81 - Taligrada and Pantodonta. Bull. Amer. 172. Mus. Nat. Hist., vol. 10, pp. 169-218. Paula Couto, C. de 1910. The age of mammals in Europe, Asia, 1952a. Fossil mammals from the beginning of and North America. New York, Mac- the Cenozoic in Brazil. Condylarthra, millan Co., pp. 1-635. Litopterna, Xenungulata, and Astra- Osborn, H. F., and C. Earle potheria. Bull. Amer. Mus. Nat. Hist., 1895. Fossil mammals of the Puerco beds. vol. 99, pp. 355-394. Bull. Amer. Mus. Nat. Hist., vol. 7, pp. 1952b. Fossil mammals from the beginning of 1-70. the Cenozioc in Brazil. Notoungulata. Pascual, R. Amer. Mus. Novitates, no. 1568, pp. 1- 1965. Un nuevo Condylarthra (Mammalia) de 16. edad Casamayorense de Paso de los In- 1952c. Fossil mammals from the beginning of dios (Chubut, Argentina). Breves con- the Cenozoic in Brazil. Marsupialia: sideraciones sobre la edad Casamayor- Polydolopidae and Borhyaenidae. Ibid., ense. Ameghiniana, vol. 4, pp. 57-65. no. 1559, pp. 1-27. Pascual, R., M. G. Vucetich, and J. Fernandez 1952d. Fossil mammals from the beginning of 1978. Los primeros mamiferos (Notoungula- the Cenozoic in Brazil. Marsupialia: Di- ta, Henricosborniidae) de la Formacion delphidae. Ibid., no. 1567, pp. 1-26. Mealla (Grupo Salta, Subgrupo Santa 1954. On a notostylopid from the Paleocene Barbara). Sus implicancias filogeneti- of Itaborai, Brazil. Ibid., no. 1693, pp. cas, taxonomicas, y cronologicas. 1-5. Ameghiniana, vol. 15, pp. 366-390. 1958. Idade geologica das bacias Cenoz6icas Patterson, B. do vale do Paraibfa e de Itaborai. Bol. 1934. Upper premolar-molar structure in the Mus. Nac. Rio de Janeiro, vol. 25, pp. Notoungulata, with notes on taxonomy. 1-18. Geol. Ser. Field Mus. Nat. Hist., vol. 6, 1961. Marsupiais f6sseis do Paleoceno do no. 6, pp. 91-111. Brasil. An. Acad. Brasil. Cien., vol. 33. 1936. The internal structure ofthe ear in some pp. 321-333. notoungulates. Ibid., vol. 6, no. 15, pp. 1962. Didelfideos f6siles del Paleoceno de 199-227. Brasil. Rev. Mus. Argentino Cien. Nat. 1940. An adianthine litoptern from the De- "Bernardino Rivadavia" Zool., vol. 8, seado Formation ofPatagonia. Ibid., vol. pp. 135-166. 8, pp. 13-19. 1963. Um Trigonostylopidae do Paleoceno do 1977. A primitive pyrothere (Mammalia, No- Brasil. An. Acad. Brasil. Cien., vol. 35, toungulata) from the early Tertiary of pp. 339-351. northwestern Venezuela. Fieldiana 1970. Novo notoungulado no Riochiquense Geol., vol. 33, pp. 397-422. de Itaborai. Iheringia, vol. 3, pp. 77-86. 1983 CIFELLI: CONDYLARTHRA 47

1978a. Ungulados fosseis do Riochiquense de 1910. Mammalia ofthe Santa Cruz beds. Part Itaborai, RJ, Brasil. I-Xenungulata. I: Litopterna. Repts. Princeton Univ. An. Acad. Brasil. Cien., vol. 50, pp. 203- Exped. Patagonia, vol. 7, pp. 1-156. 207. 1913. A history ofland mammals in the West- 1978b. Ungulados fosseis do Riochiquense de ern Hemisphere. New York, Macmillan Itaborai, RJ, Brasil. II-Condylarthra e Co., pp. 1-693. Litopterna. Ibid., vol. 50, pp. 209-218. 1937. A history ofland mammals in the West- 1978c. Ungulados f6sseis do Riochiquense de ern Hemisphere. Second ed. New York, Itaborai, RJ, Brasil. III-Notoungulata, Macmillan Co., pp. 1-786. Trigonostylopoidea. Ibid., vol. 50, pp. Sige, B. 219-226. 1972. La faunule de mammiferes du Cretac6 1979. Ungulados f6sseis do Riochiquense de superieur de Laguna Umayo (Andes pe- Itaborai, RJ, Brasil. IV-Ratifica,co ruviennes). Bull. Mus. Nat. Hist. Nat. sobre os Notoungulata. Ibid., vol. 51, ser. 3, 99 Sci. Terr., vol. 19, pp. 375- pp. 345-348. 408. Rigby, J. K. Simons, E. L. 1980. Swain quarry of the Fort Union For- 1960. The Paleocene Pantodonta. Trans. mation, middle Paleocene (Torrejoni- Amer. Phil. Soc. n.s., vol. 50, pp. 1-8 1. an), Carbon County, Wyoming: geolog- Simpson, G. G. ic setting and mammalian fauna. Evol. 1929. Paleocene and lower Eocene mammals Monogr., vol. 3, pp. 1-178. of Europe. Amer. Mus. Novitates, no. 1981. A skeleton of Gillisonchus gillianus 354, pp. 1-17. (Mammalia: Condylarthra) from the 1932a. New or little known ungulates from the early Paleocene (Puercan) Ojo Alamo Pyrotherium and Colpodon beds ofPat- sandstone, San Juan Basin, New Mex- agonia. Ibid., no. 576, pp. 1-13. ico, with comments on the local stratig- 1932b. A new Paleocene mammal from a deep raphy ofBetonnie Tsosie Wash. In, Lu- well in Louisiana. Proc. U.S. Natl. Mus., cas, S. G., J. K. Rigby, and B. S. Kues vol. 82, no. 2, pp. 1-4. (eds.), Advances in San Juan Basin pa- 1934. Provisional classification of extinct leontology. Albuquerque, Univ. New South American hoofed mammals. Mexico Press, pp. 89-126. Amer. Mus. Novitates, no. 750, pp. 1- Roth, S. 21. 1899. Aviso preliminar sobre mamiferos me- 1935a. Occurrence and relationships ofthe Rio sozoicos encontrados en Patagonia. Rev. Chico fauna of Patagonia. Ibid., no. Mus. La Plata, vol. 9, pp. 381-388. 818, pp. 1-21. Russell, D. E. 1935b. Description of the oldest known South 1964. Les mammiferes paleocenes d'Europe. American mammals, from the Rio Chi- MWm. Mus. Nat. Hist. Natur. Paris ser. co Formation. Ibid., no. 793, pp. 1-25. c, vol. 13, pp. 1-324. 1935c. New Paleocene mammals from the Fort Russell, D. E., and D. Sigogneau Union of Montana. Proc. U.S. Natl. 1965. Etude de moulages endocraniens de Mus., vol. 83, pp. 221-244. mammiferes paleocenes. Mem. Mus. 1936a. Notas sobre los mamiferos mas anti- Nat. Hist. Natur. ser. c, vol. 16, pp. 1- guos de la colecci6n Roth. Inst. Mus. 35. Univ. Nac. La Plata, Obra del Cincuen- Schlosser, M. tenario, vol. 2, pp. 63-94. 1923. Mammalia. In, Zittel, K. A. von, 1936b. Additions to the Puerco fauna, lower Grundziige der Palaontologie (Palao- Paleocene. Amer. Mus. Novitates, no. zoologie). II. Abteilung: Vertebrata. 849, pp. 1-11. Neubearbeitet von F. Broili und M. 1936c. A new fauna from the Fort Union of Schlosser. Munich and Berlin, R. 01- Montana. Ibid., no. 873, pp. 1-27. denbourg, pp. 402-662. 1936d. Carsioptychus, new name for Plagiop- Scott, W. B. tychus Matthew, nec Matheron. Amer. 1892. A revision ofthe North American Creo- Jour. Sci., vol. 32, p. 234. donta with notes on some genera which 1937. The Fort Union of the Crazy Mountain have been referred to that group. Proc. Field, and its mammalian faunas. Bull. Acad. Nat. Sci. Philadelphia, vol. 44, U.S. Natl. Mus., vol. 169, pp. 1-287. pp. 291-323. 1945. The principles of classification and a 48 AMERICAN MUSEUM NOVITATES NO. 2772

classification of mammals. Bull. Amer. mesonychid condylarth feeding mech- Mus. Nat. Hist., vol. 85, pp. 1-350. anism. Evolution, vol. 23, pp. 703-720. 1948. The beginning of the age of mammals 1977. Phylogenetic relationships and a clas- in South America. Part 1. Ibid., vol. 91, sification of the eutherian Mammalia. pp. 1-232. In, Hecht, M. K., P. C. Goody, and B. 1950. History of the fauna of Latin America. M. Hecht (eds.), Major patterns in ver- Amer. Scientist, vol. 38, no. 3, pp. 361- tebrate evolution. New York, Plenum 389. Publ. Co., pp. 315-374. 1964. Los mamiferos Casamayorenses de la Szalay, F. S., and R. L. Decker coleccion Tournouer. Rev. Mus. Argen- 1974. Origin, evolution, and function of the tino Cien. Nat. "Bernardino Rivada- tarsus in Late Cretaceous eutherians and via" Paleontol., vol. 1, pp. 1-21. Paleocene primates. In, Jenkins, F. A. 1967. The beginning of the age of mammals Jr. (ed.), Primate locomotion. New York, in South America. Part 2. Bull. Amer. Academic Press, pp. 223-259. Mus. Nat. Hist., vol. 137, pp. 1-260. Tedford, R. H. 1970. Mammals from the early Cenozoic of 1974. Marsupials and the new paleogeogra- Chubut, Argentina. Breviora, vol. 360, phy. In, Ross, C. A. (ed.), Paleogeo- pp. 1-13. graphic provinces and provinciality. Soc. 1978. Early mammals in South America: fact, Econ. Paleontol. Mineral., spec. publ. controversy, and mystery. Proc. Amer. 21, pp. 109-126. Phil. Soc., vol. 122, pp. 318-328. Teilhard de Chardin, P. Simpson, G. G., and J. L. Minoprio 1922. Les mammiferes de l'eocene inferieur 1949. A new adianthine litoptern and associ- francais et leurs gisements. Ann. Pa- ated mammals from a Deseadan fau- leontol. Paris, vol. 11, pp. 1-116. nule in Mendoza, Argentina. Amer. Trouessart, E. L. Mus. Novitates, no. 1434, pp. 1-27. 1904. Catalogus mammalium tam viventium Simpson, G. G., J. L. Minoprio, and B. Patterson quam fossilium. Quinquennale supple- 1962. The mammalian fauna of the Divisa- mentum. Berolini, R. Friedlander und dero Largo Formation, Mendoza, Ar- Sohn, pp. 1-929. gentina. Bull. Mus. Comp. Zool., vol. Van Valen, L. 127, pp. 239-293. 1966. Deltatheridia, a new order ofmammals. Sloan, R. E. Bull. Amer. Mus. Nat. Hist., vol. 132, 1969. Cretaceous and Paleocene terrestrial pp. 1-126. communities of western North Ameri- 1967. New Paleocene insectivores and insec- ca. Proc. N. Amer. Paleontol. Conv., tivore classification. Ibid., vol. 135, pp. part E, pp. 427-453. 217-284. Sloan, R. E., and L. Van Valen 1969. The multiple origins of the placental 1965. Cretaceous mammals from Montana. carnivores. Evolution, vol. 23, pp. 118- Science, vol. 148, pp. 220-227. 130. Soria, M. F. 1978. The beginning of the age of mammals. 1980. Una nueva y problemfatica forma de un- Evol. Theory, vol. 4, pp. 45-80. gulado del Casamayorense. Actas II Congr. Argentino Paleontol. Bioestrat., Van Valen, L., and R. E. Sloan Congr. Latinoamericano Paleontol. 1965. The earliest primates. Science, vol. 150, Buenos Aires, vol. 2, pp. 193-203. pp. 743-745. 1981. Los Litopterna del Colhuehuapiense West, R. M. (Oligoceno tardio) de la Argentina. Rev. 1971. Deciduous dentition of the early Ter- Mus. Argentino Cien. Natur. "Bernar- tiary Phenacodontidae (Condylarthra, dino Rivadavia" Paleontol., vol. 3, pp. Mammalia). Amer. Mus. Novitates, no. 1-54. 2461, pp. 1-37. Stirton, R. A. 1976. The North American Phenacodontidae 1947. The first lower Oligocene vertebrate (Mammalia, Condylarthra). Milwaukee fauna from northern South America. Publ. Mus. Contrib. Biol. Geol., vol. 6, Compil. Estud. Geol. Ofic. Colombia, pp. 1-78. vol. 7, pp. 325-341. West, R. M., and D. Baird Szalay, F. S. 1970. Protogonia subquadrata Cope, 1881 1969. Origin and evolution of function of the (Mammalia): proposed suppression of 1983 CIFELLI: CONDYLARTHRA 49

generic and specific names under the body Mus. Nat. Hist., vol. 14, pp. 1- plenary powers. Bull. Zool. Nomencla- 93. ture, vol. 26, pp. 230-232. Wilson, R. W. Wheeler, W. H. 1956. ThecondylarthgenusEllipsodon. Univ. 1961. Revision of the uintatheres. Bull. Pea- Kansas Publ., vol. 9, pp. 105-116.