New Cretaceous marsupial from Mongolia and the early radiation of Metatheria

Proc. Natl. Acad. Sci. USA Vol. 91, pp. 12569-12573, December 1994 Evolution New Cretaceous marsupial from Mongolia and the early radiation of Metatheria (vertebrate paleontology/therian mammals/biogeography) BORIS A. TROFIMOV* AND FREDERICK S. SZALAYt* *Paleontological Institute, Russian Academy of Sciences, Moscow, Russia; and thunter College, City University of New York, 695 Park Avenue, New York, NY 10021 Communicated by F. Clark Howell, September 8, 1994 ABSTRACT A marsupial mammal from Mongolia representing a previously unreported group, Asiadelphia, offers unequivocal evidence that metatherians were represented on the continent of Asia during the Late Cretaceous (Campanian), even in the northern arid interior. Asiatherium is known by skull, left and right mandibles, and most of its postcranial skeleton. Comparisons of the dentition, ear region, and aspects of the postcranium not only allow its diagnosis as an Asian group of metatherians but also permit a diagnosis of the Metatheria based on the unequivocally apomorphous condition of the dental formula and probable replacement pattern of the protometatherian, in contrast to the more primitive therian dental formula, represented by the few known Aegialodontidae and the first dental eutherians. It appears an inescapable conclusion that the first metatherians had a more derived dental formula (and probably replacement pattern) than the earliest dental eutherians, regardless of what specific derivation from a therian ancestry is contemplated. Such a diagnosis also supports the metatherian status of other Cretaceous Asian and American taxa such as the Deltatheroida and KokopeUia. The hypothesis that Metatheria originated in North America is largely dependent on the preponderance of Cretaceous forms from North America and negative evidence from Asia (i.e., on the lack of lower latitude Early Cretaceous tropical faunas). Conversely, the relative Cretaceous paucity of placentals from North America and their greater abundance in Asia suggests the earliest flowering of the Eutheria in an unspecified region of the Old World. The concept of a holophyletic Theria (=Tribosphenida) employed here, based on the biologically significant apomorphy of the molar form-function of the first therian, entails only the tribosphenic mammals (infraclasses Tribotheria, Metatheria, and Eutheria) but not the sundry quasi-triangular-toothed mammals and their derivatives in the Mesozoic (Eupantotheria, Peramura, Monotremata, etc.). It is suggested that the Asiadelphia and Deltatheroida, the protospecies of both, with the apomorphic therian dental formula, are ancient lineages of Metatheria independent from their North American relatives probably since the early part of the Late Cretaceous. Order Asiadelphia, nov. SYSTEMATICS "Diagnosis" and briefdiscussion. Because unequivocally derived attributes of the protometatherian or protoasiadelphian are nearly impossible to separate at present, this "diagnosis" by necessity also includes traits considered apomorphic for the first metatherian; some traits were prob- The publication costs of this article were defrayed in part by page charge payment. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. 1734 solely to indicate this fact. ably also present in the specific tribotherian ancestry as well: small Cretaceous therians with alisphenoid component to bulla and with basicranial attributes characteristic of metatherians, such as transversely elongated and elliptical fenestra rotunda and round fenestra ovale; postcanine dental formula premolars P1,2,3; molars M1,2,3,4, with a sharp break between the ultimate premolar and first molar (both conditions derived therian ones, primitive for all Metatheria); poorly developed stylar cusps (may represent primitive condition for Metatheria); protocone, in spite of its large appearance due to the large conules, of a quite small size comparable to the size of the occluding (lingual portion) hypoconid; carpus differing from known primitive Ameridelphia (last common ancestor of living Didelphimorphia and Paucituberculata) by lacking the prominent distolateral process of the scaphoid of the former; proximal fibula apomorphously slender (distinguishing it from other primitive metatherians)-a condition of fibula that is more derived than that of the Paucituberculata (the only comparable condition in ameridelphians and almost certainly independently evolved), but in articulation with the femur as in nontherian (i.e., atribosphenic) primitive mammals and primitive Theria, with a clear suggestion of terrestrial locomotor adaptation. Included taxa. Family Asiatheriidae, nov. Type genus. Asiatherium gen. nov. Type and only species. Asiatherium reshetovi, sp. nov. Etymology. For the memory of Yuri V. Reshetov, recently deceased member of the Paleontological Institute of the Russian Academy of Sciences, for his pioneering field work in Mongolia and other parts of Central Asia and for his seminal contributions to the understanding of early ungulate evolution. Holotype. Paleontological Institute of the Russian Academy of Sciences (PIN) no. 3907; crushed skull and both mandibles, nearly complete, with dentition represented from P1 to M4; skeleton lacking sternum, left hand, and both feet. Stratigraphic provenance and geographic range. Late Cretaceous Udan Sayr locality, 85 km NW of Bulgan Somon, Umuni Gobi Aimak (southern region), Mongolia; from beds of similar lithology and age to the Barun Goyot Formation (1). Udan Sayr, along with Shara Tsav, Khara Khutul, and Bambu Khuduk, among others, have been considered of Barungoyotian Age (2), approximately middle Campanian equivalent, perhaps between 80 and 76 million years before present (MYBP). Genus diagnosis. While dental formula is similar to other metatherians, the following differences diagnose this species (ofthe monotypic genus) from various Archimetatheria (3, 4), known primarily by dental remains from North America and from Asian Deltatheroida (5). P3 (lower premolar 3) not significantly taller than M1 (lower first molar) in contrast to the condition seen in the Albian metatherian Kokopellia (6) and in North American Cretaceous Metatheria (7). Paraconid tto whom reprint requests should be addressed. 12569

12570 Evolution: Trofimov and Szalay is the strongest lingual cusp on M4 only, metaconid taller than paraconid on M1_3, as in Kokopellia, but differs from the North American genus in having a deep U-shaped (rather than V-shaped) talonid notch (that between metaconid and entoconid), a relatively mesiodistally shorter talonid compared with trigonid, taller talonid cusps, but relatively less tall trigonid cusps compared with the talonid. The buccal and distal cristids of the talonid (the cristid obliqua and the posthypoconid cristid) form a more acute, smaller angle in Asiatherium than that in Kokopellia. The entoconid and hypoconulid are tall and acute cusps and are slightly more closely twinned than in Kokopellia, similar to those in Protalphadon (8) and Alphadon (9), but unlike those in deltatheroidans or known Cretaceous dental eutherians (7, 10-13). Asiatherium differs from other Cretaceous North American dentally primitive metatherians such as Anchistodelphys (9), Iqualadelphis (8), Protalphadon (8), or primitive and early Asian eutherians such as the otlestid Prokennalestes (10) in having its entoconid and hypoconulid taller and more acute and having a relatively narrow stylar shelf on the upper molars (or even none on M/1), in addition to having lost the connection of the preparacrista with the stylocone (cusp B). Unlike in some of the former genera, the stylar cusps are very faint as in Deltatheroida or Pediomys, and only their traces (incipient or reduced) are visible. As in Deltatheroida [e.g., Sulestes (5)], the metacone is relatively large compared with the paracone, larger than that in dentally primitive Eutheria, but it is nearly comparable in size to the paracone in other early metatherians. Upper molars have well-developed pre- and postcingulae. Unlike in all the former taxa listed, the conules and pre- and postprotocristae are strongly developed. A preliminary examination suggests that the enamel contains tubules, typical of most metatherians. DISCUSSION The type specimen of Asiatherium reshetovi is exceptionally complete for a Cretaceous therian mammal, particularly for a metatherian in that it has most of its skeleton preserved (Fig. 1). The crushed skull (Figs. 2 and 3) has been reconstructed and shows, in many respects, what may be considered primitive therian conditions in addition to a suite of apomorphies. The nasal structure contacts the lacrimal structure, and the jugal structure contacts the glenoid fossa, both being primitive, probably late synapsid traits [see particularly Proc. Natl. Acad. Sci. USA 91 (1994) Sinoconodon (3)]. The palate is fenestrated as in the probable primitive condition represented for the deltatheroidans by the Gurlin Tsav skull (5), but unlike in the ameridelphian borhyaenids. The basicranium shows continuity between the alisphenoid and a medially located, partially ossified bulla. The postcanine dental formula is unequivocally metatherian (Fig. 4). While the suggestion (14) that the Owen (15)- Lydekker (16) hypothesis for the homology of the metatherian Ml may prove to be correct [i.e., equivalent to the eutherian dp4 (fourth deciduous premolar)], at present we continue to use the M1-4 designation. As we and others discuss elsewhere (3, 4, 14), in any probable derivation of the metatherian dental formula, the reduction of the premolar number and the (putative) minimal replacement pattern of the antemolar dentition [only dp3 is replaced (14)] is unequivocally derived in therians (tribosphenic mammals), and therefore it represents a significant diagnostic trait (3, 4, 6, 14) that designates the first metatherian in a testable manner. Thorough reviews of molar features of early therians (7, 17-20) have attempted to sort out those topographic attributes that maximally postdict whether these rare taxa were either tribotherian, metatherian, or eutherian. In spite of a number of important new forms described during the past decade (8-10, 21-27), this has become increasingly difficult on the basis of molar teeth alone. As noted, the distolingual portion of the talonid (entoconid-hypoconulid) in Asiatherium is characteristically metatherian (Fig. 4), although this is not necessarily a diagnostic feature of the protometatherian. The near subequality of the metacone to the paracone signals the nearly ubiquitous metatherian "tendency" to hypertrophy the metacone (tied to entoconid-hypoconulid twinning functionally). Many living didelphids, Cretaceous and Paleogene archimetatherians, sudameridelphians (3), primitive deltatheroidans (in particular the Gurlin Tsav skull), Asiatherium, and Kokopellia have an increasing size gradient from M1-4 (except for the slightly reduced M3), where the first upper molar (Ml) is decidedly smaller than M2 and the latter is smaller than M3. This is not the conformation of the dentition in the dentally primitive eutherians such as the Early Cretaceous otlestids, where M1 is larger than M2 (10, 11) as in a number of Paleogene groups. By this criterion, in addition to the dental formula discussed below, not only asiadelphians but also deltatheroidans and Kokopellia are Metatheria. Protoconid is the tallest cusp in Asiatherium, a primitive therian condition, and the paraconid is taller than FIG. 1. Asiatherium reshetovi, new genus and species, type specimen. Paleontological Institute of the Russian Academy of Sciences no. 3907, Udan Sayr, Mongolia. Nearly complete skeleton, crushed skull, and mandibles (mandibles are not shown). 1 cm

Evolution: Trofimov and Szalay Proc. Natl. Acad. Sci. USA 91 (1994) 12571 FIG. 2. Asiatherium reshetovi, stereophotos in dorsal (Upper) and ventral (Lower) views. (Subdivisions on scale = 1 mm.) the metaconid only on M4; on the antecedent teeth, it is the metaconid that is the dominant cusp on the lingual side of the trigonid. The latter is a common didelphid condition, but significantly for the analysis of dental taxonomic properties (particularly for Deltatheroida), it was altered probably independently more in carnivorous (i.e., shearing) lineages such as Lutreolina, Didelphis, Chironectes, sparassocynids, sparassodontans, and dasyuromorphians. The scapula has a long and attenuated scapular spine, a primitive therian condition prominently displayed in living didelphids. A similarly ancient condition, not a metatherian diagnostic trait, is the presence of well-developed epipubic bones. The head of the radius is oval in outline, not round as in most didelphids, and the anterior edge of the olecranon process is twisted medially, in a manner characteristic of terrestrial didelphids and caenolestids. The knee is imperfectly preserved, but a number of attributes are clear, such as the articular contact of the fibula with the distal femur and the relatively reduced proximal fibula compared with didelphids, in which it is usually wide and flared (4). The derived condition (within Metatheria) of the distal femur is also characteristic in a number of terrestrial metatherian lineages such as caenolestids and peramelids. The lack of any sign of a patellar groove on the distal femur is a very strong corroboration that the animal is not a eutherian but a metatherian (4) Ṫhe tarsus has proved to be a singularly reliable and diagnostic attribute of metatherian higher taxa (3, 4, 28), and there are preliminary indications that the system of constraints that appears to govern the changes in the tarsus are somewhat similarly operational in the hand as well (4). There is enough preserved of the right hand of Asiatherium to rule out specific allocations to other higher taxa within Metatheria. The radius and ulna are closely pressed against one another, and the distal ends on both bones have relatively acute styloid processes, suggestive of the terrestrial adaptation seen in caenolestids. Only the scaphoid, trapezium, and trapezoid appear to be in their natural position, whereas the

12572 Evolution: Trofimov and Szalay Proc. Natl. Acad Sci. USA 91 (1994) 53 / B FIG. 3. Asiatherium reshetovi, unreconstructed drawing of palatal view of skull. (Scale bar = 5 mm.) three remaining carpals on the specimen appear to be displaced. The large scaphoid lacks, significantly, a welldeveloped distolateral process (ulnar, distally, in a quadrupedal mammal with a rotated radius) that is the inferred ancestral condition in known living metatherians (4). Phyletically three widely separated groups, Gondwanadelphia (Microbiotheria and Dasyuromorphia) (3, 4), Paucituberculata, and the Didelphidae show this ancestral (ameridelphian) condition. The process may be a centrale, a bone not known in metatherians, perhaps lost or ontogenetically fused (incorporated during phylogeny) into the scaphoid. The loss of the centrale is diagnostic for the last common ancestor of the living taxa. However, it may have been present in Asiatherium, as one assumes for the last common ancestor of living therians. Living metatherians which lack the distolateral process of the scaphoid (the syndactylans) have almost certainly secondarily reduced it (4) and thus this loss appears to be diagnostic in this regard. Notoryctes, which may be a syndactylan, has a complexly modified wrist. The magnum (capitate) of Asiatherium appears artificially displaced proximally, and the presence of the unciform (hamate), cuneiform (trapezium), or pisiform cannot be established. The bone in between the distal radius and ulna might be the lunate. These preliminary details of the dental morphology indicate without any doubt that Asiatherium was not a deltatheroidan (based on the well-preserved upper dentition of the primitive deltatheroidan cranium from the Maastrichtian Gurlin Tsav locality, Mongolia) and that it differed from what may be considered the last common ancestor of the carpally most primitive living families, the Didelphidae (suborder Didelphimorphia) and Caenolestidae [suborder Glirimetatheria (3, 4)]. The latter suggests that it also differed from the last common ancestor of all the South American forms, that ofthe order Didelphida (3, 4). A B FIG. 4. Asiatherium reshetovi, new genus and species, right postcanine dentition (composite drawing from both sides) of type specimen. (A) Occlusal view of upper teeth. (B) Occlusal view of lower teeth. (C) Buccal view of lower teeth. (D) Lingual view of lower teeth. (Scale bar = 1 mm.) The Asiadelphia, represented by Asiatherium, we speculate, is a major clade of Asian Cretaceous metatherians that probably flowered in the more tropical environments of that continent, and the genus described here is from a highlatitude semiarid environment, representative of this ancient and independent group. While several taxa of eutherians are very well known from this time level in Asia (29, 30), the rare occurrence of this metatherian suggests that its close relatives, like those of the many other mostly eutherian groups that appear in the latest Cretaceous-Paleocene, had a primarily southern distribution. While it is important to place Asiadelphia in perspective in relation to the Deltatheroida, this cannot be done satisfactorily here. Controversially (19, 28), the latter order was allocated to the Metatheria (5) after decades of suspected ties-an analysis not endorsed by all (6, 19) but supported here and elsewhere (3, 4). While primitive deltatheroidans (Gurlin Tsav skull, ref. 5 and personal observation) have the advanced therian dental formula of early and (in that respect) primitive metatherians and the molar size conformation noted above, they also share (we do not include the aegialodontan Kielantherium in Deltatheroida) with asiadelphians the slight hypertrophy of the metacone-a condition unlike that in dentally primitive eutherians and certainly unlike, judged from lower molars, that in aegialodontids. While upper molars of Aegialodonta are not known to us, the extreme brevity of the posthypoconid cristid argues against a hypertrophied metacone. The few dental similarities of Deltatheroida to Sparassodonta are not synapomorphies (they fail functional, adaptively conceived analytical tests) in the context of the total available evidence. The presence of a well-developed protocone in Deltatheroida strongly suggests that the width reduction of the talonid in that group, like in several metatherian and eutherian taxa, is secondary. This issue is analyzed and argued elsewhere that the sparasso- C D

Evolution: Trofimov and Szalay donts are unequivocally ameridelphians and not relatives of deltatheroidans (4). Figs. 1 and 3 were prepared by V. D. Kolganov. Travel for F.S.S. was supported by the National Geographic Society. Critical reading and constructive criticisms by Richard L. Cifelli, Zofia Kielan- Jaworowska, and Christian de Muizon are greatly appreciated; their thorough readings have improved the manuscript. 1. Ivakhnenko, M. F. & Kurzanov, S. M. (1988) Tr. Sovmestnaya Sov. Mong.-Paleotol. Eksped. 34, 100-105 (in Russian). 2. Jerzykiewicz, T. & Russell, D. A. (1991) Cretaceous Res. 12, 345-377. 3. Szalay, F. S. (1993) in Mammalian Phylogeny: Mesozoic Differentiation, Multituberculates, Monotremes, Early Therians and Marsupials, eds. Szalay, F. S., Novacek, M. J. & Mc- Kenna, M. C. (Springer, New York), pp. 216-241. 4. Szalay, F. S. (1994) Evolutionary History ofthe Marsupials and an Analysis of Osteological Characters (Cambridge Univ. Press, New York). 5. Kielan-Jaworowska, Z. & Nessov, L. A. (1990) Lethaia 23, 1-10. 6. Cifelli, R. L. (1993) Proc. Natl. Acad. Sci. USA 90,9413-9416. 7. Clemens, W. A. & Lillegraven, J. A. (1986) Contrib. Geol. Spec. Pap. 3, 55-85. 8. Cifelli, R. L. (1990) J. Vertebr. Paleontol. 10, 295-319. 9. Cifelli, R. L. (1990) J. Vertebr. Paleontol. 10, 320-331. 10. Kielan-Jaworowska, Z. & Dashzeveg, D. (1989) Zool. Scr. 18, 347-355. 11. Nessov, L. A. & Kielan-Jaworowska, Z. (1991) Contrib. Pal. Mus. Univ. Oslo 364, 51-52. 12. Clemens, W. A. (1979) in Mesozoic Mammals: The First Two- Thirds ofmammalian History, eds. Lillegraven, J. A., Kielan- Proc. Natl. Acad. Sci. USA 91 (1994) 12573 Jaworowska, Z. & Clemens, W. A. (Univ. of California Press, Berkeley), pp. 192-220. 13. Fox, R. C. (1984) Carnegie Mus. Nat. Hist. 9, 9-20. 14. Luckett, W. P. (1993) in Mammalian Phylogeny: Mesozoic Differentiation, Multituberculates, Monotremes, Early Therians and Marsupials, eds. Szalay, F. S., Novacek, M. J. & McKenna, M. C. (Springer, New York), pp. 182-204. 15. Owen, R. (1868) On the Anatomy of Vertebrates (Longmans, Green, London), Vol. 3. 16. Lydekker, R. (1887) Catalogue of the Fossil Mammalia in the British Museum (Natural History) (Trustees of the British Museum, Natural History, London). 17. Patterson, B. (1956) Fieldiana Geol. 13, 1-105. 18. Butler, P. M. (1978) Breviora Mus. Comp. Zool. 446, 1-27. 19. Cifelli, R. L. (1993) in Mammalian Phylogeny: Mesozoic Differentiation, Multituberculates, Monotremes, Early Therians and Marsupials, eds. Szalay, F. S., Novacek, M. J. & Mc- Kenna, M. C. (Springer, New York), pp. 208-215. 20. Fox, R. C. (1975) Can. J. Earth Sci. 12, 412-442. 21. Fox, R. C. (1971) Zool. J. Linn. Soc. 50 (Suppl. 1), 145-164. 22. Fox, R. C. (1972) Can. J. Earth Sci. 9, 1479-1494. 23. Fox, R. C. (1980) Can. J. Earth Sci. 17, 1489-1498. 24. Cifelli, R. L. (1990) J. Vertebr. Paleontol. 10, 330-331. 25. Cifelli, R. L. (1990) J. Vertebr. Paleontol. 10, 332-345. 26. Cifelli, R. L. (1990) J. Vertebr. Paleontol. 10, 346-360. 27. Cifelli, R. L. & Eaton, J. G. (1987) Nature (London) 325, 520-522. 28. Szalay, F. S. (1982) in Carnivorous Marsupials, ed. Archer, M. (Royal Zoological Society of New South Wales, Sydney), pp. 621-640. 29. Kielan-Jaworowska, Z. (1981) Paleontol. Pol. 42, 27-78. 30. Kielan-Jaworowska, Z. (1984) Paleontol. Pol. 46, 173-183.