ASIA

Middle Asia

The vast area of Russia and adjacent countries formerly included in the Soviet Union was long a complete blank for Cretaceous mammals. A single specimen of a Late Cretaceous eutherian from Kazakhstan was described by Bazhanov (1972), but it remained for Lev A. Nessov to more fully exploit middle Asia for fossil mammals in a series of expeditions beginning in 1978. Nessov must be given the greatest credit for his remarkable contributions to knowledge of early mammals. His achievements are all the more impressive in light of the harsh field conditions and virtual lack of support he endured; among other difficulties, his team occasionally had to hitchhike to the field area and dig 3-m-deep holes to find drinking water (see, e.g., Averianov, 2000). Nessov described more than 25 species of Cretaceous mammals, and part of the collection has not yet been published; research is being continued by Alexander O. Averianov, J. David Archibald, and col leagues. The mammal assemblages of middle Asia (table 2.18) are noteworthy for several reasons. First, the fossils are reasonably complete, including many jaws and, in one case, a skull. Second, they sample an important time interval—the early part of the Late Cretaceous, which is otherwise poorly represented by mammalian fossils. Finally, the assemblages are important zoogeographically (see Nessov, 1992). The mammals are quite different from those to the east, in the Late Cretaceous of Mongolia, which are presumed to be arid-adapted and are in this sense specialized. Among others, the mammal assemblages of middle Asia appear to include primitive relatives of ungulates (Archibald, 1996b)—eutherians that would not flourish or radiate widely until the early Tertiary. Most of the Late Cretaceous mammals of middle Asia— Kazakhstan, Uzbekistan, Tadjikistan, and Kirghizia—are included in several recent summary papers (Nessov, Sigogneau-Russell, and Russell, 1994; Nessov, 1997; Nessov, Archibald, and Kielan-Jaworowska, 1998; Averianov, 2000). We review the occurrences in stratigraphic order.

Three mammal localities are known from the Khod-zhakul Formation, Karakalpakistan, Uzbekistan (figure 2.11). The mammal localities are in the upper part of the unit, which is believed to be lower Cenomanian based on correlation with a marine regressive phase and on some of the lower vertebrate fauna (Averianov, 2000). Undescribed specimens, consisting largely of edentulous jaws of unidentified therians, are known from the site of Khod-zhakulsay (figure 2.11), which is interpreted to represent a channel connecting two brackish reservoirs (Nessov, Si-gogneau-Russell, and Russell, 1994). Unidentified mammal remains have also been reported from the Chelpyk locality, which may be slightly higher in the section than the other sites. Sheikhdzheili, to the northeast of Khod-zhakulsay, is the source for three mammals; a large possible deltatheroidan and two eutherians, which are among the oldest representatives of this group.

A single specimen, belonging to the eutherian Sorlestes kara, was recovered from a drill core at the Ashikol "locality," east of the Ashikol lakes, Chimkent District, Kazakhstan (figure 2.11). The unit for the fossil horizon (500 m underground) is unnamed, but the specimen is believed to be at least as old as early Turonian, based on paly-nomorphs recovered from an overlying unit (Nessov, Sigogneau-Russell, and Russell, 1994).

The vast majority of the mammals known from the Late Cretaceous of middle Asia are from the Bissekty Formation, Kizylkum Desert, Uzbekistan (figure 2.11). Fossil mammals have been recovered from several horizons in this unit at Dzharakuduk (figure 2.11). The oldest mammal-yielding level may be late Turonian, based on sharks (Nessov, Sigogneau-Russell, and Russell, 1994). Six or table 2.18. Late Cretaceous Mammals of Russia and the Commonwealth of Independent States (see figure 2.11); locality numbers do not correspond between map and table). Localities: 1, Sheikhdzheili; 2, Khodzhakulsay; 3, Chelpyk (Karakalpakistan, Uzbekistan; Khodzhakul Formation, early Cenomanian); 4, Ashikol (Chimkent District, Kazakhstan; early Turonian); 5, Dzharakuduk (Kizylkum Desert, Uzbekistan; Bissekty Formation, Turonian); 6, Dzharakuduk (Bissekty Formation, Coniacian); 7, Dzharakuduk (Aitym Formation, Santonian); 8, Zhalmauz Well (Kyzyl-Orda District, Kazakhstan; Bostobe Formation, Santonian); 9, Kansay (Khodzant District, Tadzhikistan; Yalovach Formation, Santonian); 10, Yantardakh (Yakutia, Russia; Khets Formation, Santonian); 11, Grey Mesa (Chimkent District, Kazakhstan; Darbasa Formation, early Campanian)

CENOMANIAN Mammalia

Family incertae sedis

Family incertae sedis Oxlestes grandis (1) Eutheria Otlestidae

Otlestes meiman (1) "Zhelestidae"

Eozhelestes mangit (1) TURONIAN

Stem Trechnotheria ("symmetrodontans") Spalacotheriidae

?Shalbaatar bakht (5) Eutheria

Zalambdalestidae

Kulbeckia rara (5) "Zhelestidae"

Aspanlestes aptap (5) Sorlestes kara (4) Family incertae sedis

Daulestes kulbeckensis (5) Daulestes inobservabilis (5) CONIACIAN Multituberculata

Family incertae sedis

Uzbekbaatar kizylkumensis (6) Gen. et sp. indet. (6) Deltatheroida Deltatheridiidae

Deltatherus kizylkumensis (6) Sulestes karakshi (6) Sulestes sp. (6) Gen. et sp. indet. (6) Marsupialia Asiatheriidae

Marsasia aenigma (6) Marsasia sp. (6) Eutheria

Kennalestidae

?Kennalestes uzbekistanensis (6) Sailestes quadrans (6) "Nyctitheriidae"

Paranyctoides aralensis (6) Zalambdalestidae

Kulbeckia kulbecke (6)

"Zhelestidae"

Eoungulatum kudukensis (6) cf. Eoungulatum kudukensis (6) Kumsuperus avus (6) Ortalestes tostak (6) Parazhelestes minor (6) Parazhelestes robustus (6) Sorlestes budan (6) Zhelestes temirkazyk (6) Gen. et sp. indet. (6) Family incertae sedis Bulaklestes kezbe (6) Daulestes nessovi (6) Daulestes kulbeckensis (6) "Sazlestes tis" (6) SANTONIAN Mammalia

Family incertae sedis Gen. et sp. indet. (10) Multituberculata

Family incertae sedis

Uzbekbaatar wardi (7) Eutheria

Zalambdalestidae

Kulbeckia kansaica (9) ?Zalambdalestidae

Beleutinus orlovi (8) "Zhelestidae"

Gen. et sp. indet. (7) Family incertae sedis

Gen. et sp. indet. (8, 9) EARLY CAMPANIAN Multituberculata

Family incertae sedis Bulganbaatar sp. (11) Deltatheroida Deltatheridiidae

Deltatheridium nessovi (11) Eutheria

Kennalestidae

Gen. et sp. indet. (11) Zalambdalestidae

Alymlestes kielanae (11) ?Alymlestes sp. (11) "Zhelestidae"

?Aspanlestes sp. (11)

more mammals are known from this level. One of these, Shalbaatar bakht (named but not figured by Nessov, 1997), appears to be a "symmetrodontan"; the remainder are eutherians, including the "palaeoryctid"-like Daulestes and one of the oldest of the "Zhelestidae" (Aspanlestes), which are thought to be related to ungulates.

A dozen or more localities of Coniacian age have yielded mammals from the Bissekty Formation at Dzharakuduk (figure 2.11, table 2.18). For convenience, we treat them collectively here, though several separate levels are recognized (see Nessov, Sigogneau-Russell, and Russell, 1994; Averianov, 2000). Some 25 or more kinds of mammals have been reported from these sites. The multituber-culate Uzbekbaatar (see Kielan-Jaworowska and Nessov, 1992) is assigned to Cimolodonta incertae sedis (see chapter 8). Deltatheroida, a predominantly Old World group that may be related to Marsupialia (chapter 12), are represented by at least three taxa (Kielan-Jaworowska and Nessov, 1990; Nessov, 1997). Of special interest among the fauna is Marsasia, which appears to be a true marsupial and thus is the earliest representative of that group in Asia (Averianov and Kielan-Jaworowska, 1999). Marsupialia had long been thought to be of North American origin, not reaching the Old World until the Tertiary. Asia-therium, mentioned later, and Marsasia challenge this interpretation and suggest a much earlier presence in Asia than previously envisaged (see chapter 12). Some of the dentally primitive eutherians from the Coniacian of Dzharakuduk also have interesting biogeographic implications. One is possibly referable to Kennalestes, known from the Campanian of Mongolia; another, Paranyctoides aralensis, belongs to an insectivoran genus otherwise restricted to the Campanian-Maastrichtian of North America (Fox, 1979d, 1984b; Cifelli, 1990e; see detailed treatment by Archibald and Averianov, 2001b). Most of the eutherians from the Coniacian of Dzharakuduk are referred to the "Zhelestidae." These are especially noteworthy because they are believed to be related to ungulates (Archibald, 1996b; Nessov, Archibald, and Kielan-Jaworowska, 1998). As such, they represent the most compelling evidence for early Late Cretaceous radiation of a modern placental group (Archibald, 1999). Additionally, most eutherian groups from the Cretaceous of Asia lack obvious ties with more recent taxa, particularly those from North America (Novacek et al., 1997; Cifelli, 1999a). These ungulate-like "zhelestids" are the best known exception to this rule and thus provide a viable source for the later radiation of ungulates in North America. A final noteworthy occurrence in the Coniacian of Dzharakuduk is that of Daulestes nessovi (the genus is also known from the Turonian part of the Bissekty Formation), which is known by the oldest reasonably complete skull of a eutherian mammal (McKenna et al., 2000).

Fossil mammals have recently been reported from the Aitym Formation at Dzharakuduk (Averianov, 1999; Archibald and Averianov, 2001b). This unit, which overlies the Bissekty Formation, is of probable Santonian age and includes a fauna dominated by marine taxa, including chondrichthyans, plesiosaurs, turtles, and molluscs. Three mammals are known thus far from the Aitym Formation: a large eutherian represented by an edentulous jaw fragment, an insectivoran Paranyctoides sp., and the multi-tuberculate Uzbekbaatar wardi (known from a p4), which may be a descendant of U. kizylkumensis from the Bissekty Formation (Averianov, 1999).

Two additional sites of Santonian age are known from middle Asia. Zhalmauz Well, Kyzyl-Orda Province, Kazakhstan (figure 2.11), is in the lower Bostobe Formation, with a Santonian correlation based on turtles and chon-drichthyans (Averianov, 2000). Beleutinus orlovi, described from this site by Bazhanov (1972), may be related to a peculiar eutherian group known from Mongolia, the Zalambdalestidae (Nessov, Sigogneau-Russell, and Russell, 1994). The only other mammal specimen known from this site is a large, unidentifiable cervical vertebra (Nessov and Khisarova, 1988). The Kansay site, Khodzant District, Tadzhikistan (figure 2.11), has also yielded two mammal specimens. The occurrences are in the Yalovach Formation, believed to be early Santonian on the basis of fishes and turtles (Nessov, Sigogneau-Russell, and Russell, 1994). The only described mammal from Kansay is Kulbeckia kansaica (see Nessov, 1993); this genus is also represented in older horizons (Turonian and Coniacian, Bis-sekty Formation) of middle Asia.

Another Santonian mammal occurrence, well out of the region under discussion but mentioned here for convenience, is in the Khets Formation at Yantardakh, Siberia (not figured). The specimen consists of mammalian hair embedded in amber, and is notable for its high (70° N) paleolatitudinal occurrence (Nessov, Sigogneau-Russell, and Russell, 1994).

The youngest Late Cretaceous mammals of middle Asia are from Grey Mesa, near the Alymtau Range, Kazakhstan. The mammals, including some six taxa, come from the lower part of the Darbasa Formation, which is believed to be early Campanian (Nessov, 1993; Averianov and Nessov, 1995; Averianov, 2000). The fauna is thus important because it is close to the age of the oldest Late Cretaceous mammals known from Mongolia, which occur in the Djadokhta Formation and presumed equivalents. Two genera, the multituberculate Bulganbaatar and the deltatheroidan Deltatheridium, are also known from the Djadokhta Formation, and the placental families Ken-nalestidae and Zalambdalestidae are also represented in both the Grey Mesa and Mongolian faunas (see Averianov and Nessov, 1995; Averianov, 1997,2000; Archibald et al., 2001).

Mongolia

Without doubt, the best known of all Mesozoic mammals come from the Late Cretaceous of the Gobi Desert, Mongolia (table 2.19). Skulls and, commonly, articulated skeletons are the rule, and isolated fragments the exception. This is in stark contrast with the record from the Late Cretaceous of North America (and virtually everywhere else), where fossils generally consist of jaw fragments or isolated teeth, usually recovered through screenwashing. Indeed, though North America boasts many more Late

Cretaceous mammal sites and specimens, not a single reasonably complete skull of this age has yet been published from anywhere on the continent. The extraordinary preservation of the Mongolian specimens is due to an unusual depositional setting characterized by arid conditions and extensive dune systems (see later).

Initial discoveries in Late Cretaceous strata of the Gobi Desert were made in the 1920s by the Central Asiatic Expeditions of the American Museum of Natural History, led by Roy Chapman Andrews (see informative and readable account by Andrews, 1932). The major fossil area for Late Cretaceous mammals (e.g., Gregory and Simpson, 1926; Simpson, 1928d) and dinosaurs discovered by these expeditions is near the so-called Flaming Cliffs, at a place then called Shabarakh Usu and now termed Bayan Zag (spelled Bayn Dzak in most recent literature). The fossil-

ta b l e 2.19. Late Cretaceous Mammals of Mongolia, China, and Japan (see figures 2.4, 2.17; locality numbers do not correspond between maps and table). Localities or local faunas: 1, Bayan Zag (Djadokhta Formation, Campanian); 2, Togrog (Togrog Beds, equivalent of the Djadokhta Formation, Campanian); 3, Ukhaa Tolgod ("Ukhaa Tolgod Beds," not formally named , Campanian); 4, Khulsan (Baruungoyot Formation, Campanian); 5, Nemegt (Baruungoyot Formation, Campanian); 6, Hermiin Tsav (Red Beds of Hermiin Tsav, Campanian); 7, Udan Sayr (unnamed unit, Campanian); 8, Khaichin Uul (?Nemegt Formation, ?Maastrichtian); 9, Guriliin Tsav (?Nemegt Formation, ?Maastrichtian), Mongolia; 10, Tsondolein-Khuduk, Gansu, China (unit uncertain; ?Cenomanian); 11, Bayan Mandahu, Inner Mongolia, China (Bayan Mandahu Formation, Campanian); 12, Amagimi Dam, Kiyushu, Japan ("Upper Formation" of the Mifune Group, late Cenomanian or early Turonian)

Multituberculata

Family incertae sedis

Bulganbaatar nemegtbaataroides (1) Bulganbaatar cf. nemegtbaataroides (3) Chulsanbaatar vulgaris (4, 5, 6) Chulsanbaatar cf. vulgaris (3) Nemegtbaatargobiensis (3, 4, 5, 6) Gen. nov., sp. A (3) Gen. nov., sp. B (3) ?Cimolomyidae

Buginbaatar transaltaiensis (8) Djadochtatheriidae

Catopsbaatar catopsaloides ( 4, 6) Djadochtatherium matthewi (1, 2, 11) ?Djadochtatherium matthewi (3) Kryptobaatar dashzevegi (1, 2, 3, 6) Kryptobaatar mandahuensis (11) Tombaatar sabuli (3, ?11) Sloanbaataridae

Kamptobaatar kuczynskii (1) Kamptobaatar cf. kuczynskii (3) Nessovbaatar multicostatus (6) Sloanbaatar mirabilis (1) Sloanbaatar cf. mirabilis (3) Deltatheroida Deltatheridiidae

Deltatheridium pretrituberculare (1,3,6)

Deltatheroides cretacicus (1) Khuduklestes bohlini (10) Gen. et sp. nov. (9) Marsupialia Asiatheriidae

Asiatherium reshetovi (7) Eutheria

Family incertae sedis

Hyotheridium dobsoni (1) cf. Hyotheridium sp. (3) Asioryctidae

Asioryctes nemegetensis (4, 5, 6) cf. Asioryctes sp. (3) Ukhaatherium nessovi (3) Kennalestidae

Kennalestes gobiensis (1) Kennalestes cf. gobiensis (3) Kennalestes sp. (11) Zalambdalestidae

Barunlestes butleri (5, 6) Zalambdalestes lechei (1,2,3) Zalambdalestes sp. (11) "Zhelestidae"

Sorlestes mifunensis (12)

yielding unit is the Djadokhta Formation. This unit and many other geographic and stratigraphic terms of the region are variously spelled and referenced (see, e.g., Gradziriski et al., 1968; Gradziriski and Jerzykiewicz, 1974; Novacek et al., 1994). Polish-Mongolian Expeditions (1963-1971), led by Zofia Kielan-Jaworowska, worked these exposures and discovered many additional fossils at Bayan Zag and at some nearby newer sites. They also discovered a number of other exposures, most notably those to the west of the Nemegt Basin. These are significant in that they include mammals from younger horizons—the Baruungoyot Formation and its presumed equivalents (see, e.g., Jerzykiewicz and Russell, 1991). Fossil mammals collected by the Polish-Mongolian Expeditions have been described in a long list of publications, most of which are cited later. The region was also worked by Soviet-Mongolian Expeditions, mainly in the late 1960s and 1970s (see review by Kurochkin and Barsbold, 2000). New sites for Late Cretaceous mammals were discovered but, to our knowledge, specimens from only three of these have been published (see Kielan-Jaworowska and Sochava, 1969; Kielan-Jaworowska and Nessov, 1990; Szalay and Trofi-mov, 1996). Several other institutions have sent field parties to the Gobi Desert in search of Late Cretaceous mammals, the most notable of these being the current American Museum-Mongolian Academy Expeditions, which began in 1990. The most significant discovery of these expeditions to date is a new locality, Ukhaa Tolgod, in the Nemegt Basin, where literally hundreds of skulls and, in many cases, associated skeletons were discovered (Dashzeveg et al., 1995). The assemblage, which includes a few previously unknown taxa (Novacek et al., 1997; Rougier et al., 1997, see also Wible and Rougier, 2000), appears to be similar to that of the Djadokhta Formation, although it also contains the elements previously known only from the Baruungoyot Formation and its equivalent the Red Beds of Hermiin Tsav (see later).

The Gobi Basin includes three major subbasins. For the Upper Cretaceous rocks, syndepositional block faulting, discontinuous sedimentation, lack of laterally extensive marker horizons, and discontinuity of exposures, among other features, make lithostratigraphic correlation among geographically widespread fossil localities problematic. Three units are central to a discussion of the Late Cretaceous fossil vertebrate sequence in the Gobi Desert (lower to higher): the Djadokhta, Baruungoyot, and Nemegt formations. The type area of each unit includes one or more localities with significant assemblages of fossil vertebrates. Correlation of fossiliferous strata outside the type areas is based on similarity in lithology, inferred depositional facies, fauna, and other considerations. We refer readers to the excellent summaries of Gradzmski et al. (1977), Jerzykiewicz and Russell (1991), and Jerzykiewicz (1995).

Correlation of the Upper Cretaceous strata of the Gobi Desert with vertebrate sequences from elsewhere is also problematic. Marine strata are lacking, so that it is not possible to establish correlation on the basis of standard marine stages. Similarly, there are currently no radio-metric, paleomagnetic, or other data to establish a chrono-stratigraphic framework. Historically, age estimates have varied widely—with the Djadokhta mammals, for example, being considered as old as Cenomanian or as young as Tertiary (see the excellent summary and references cited by Lillegraven and McKenna, 1986). Comparisons based on the vertebrates themselves are fraught with difficulty because the Mongolian taxa are arid adapted and highly endemic (see, e.g., Osmolska, 1980). Based on the stage of evolution (e.g., Fox, 1978) and inferred climatological changes in the sequence (Jerzykiewicz and Russell, 1991), we follow Lillegraven and McKenna (1986) in considering the faunas of the Djadokhta and Baruungoyot formations to correlate with the Campanian. We also tentatively accept the early Campanian age for the Djadokhta and late Campanian for the Baruungoyot Formation. Similarity in the mammalian assemblages of the Djadokhta and Dar-basa (Kazakhstan, discussed earlier) formations led Averianov (1997) to propose a lower Campanian correlation for the Djadokhta Formation. This correlation would imply that the Baruungoyot Formation is also of early Campan-ian or, more probably, late Campanian age. The Nemegt Formation is generally considered to be Maastrichtian, perhaps middle Maastrichtian (Jerzykiewicz and Russell, 1991), in age.

The stratotype of the Djadokhta Formation is at Bayan Zag (see Gradzmski et al., 1977) in the Ulan Nur Basin. The unit consists mainly of fine poorly cemented and mostly aeolian sandstones with interbedded conglomerates and caliche horizons. The depositional environment is interpreted to have included extensive dune systems with sporadic and ephemeral lakes and the occasional development of arid paleosols (see, e.g., Loope et al., 1998). The Baruun-goyot Formation was defined on the basis of exposures at Khulsan in the Nemegt Basin, nearly 200 km west of Bayan Zag (Gradzmski and Jerzykiewicz, 1974). Its superposi-tional relationship with the Djadokhta Formation cannot be determined directly. The lower boundaries of both units are covered in their type areas; the Djadokhta is overlain unconformably by Tertiary rocks (Khashaat, or "Gashato" Formation), whereas the Cretaceous Nemegt Formation overlies the Baruungoyot. However, the Djadokhta Formation is generally believed to be lower, at least in part, than the Baruungoyot Formation (contra Dashzeveg et al.,

1995). Furthermore, the units are lithologically distinct, and Djadokhta-facies rocks are widespread in the Gobi Basin (Jerzykiewicz and Russell, 1991), including the Ne-megt Basin itself (Dashzeveg et al., 1995). Like the Djadokhta, the Baruungoyot Formation consists largely of poorly cemented sandstones, but it includes more ero-sional channels and interbedded claystones, and caliches are less common. While the depositional setting is mainly aeolian, the lack of paleosols and the greater abundance of lacustrine facies suggest a wetter, more humid environment than for the Djadokhta Formation.

The stratotype for the Nemegt Formation, also in the Nemegt Basin, is at the Nemegt locality, just west of Khul-san (Gradziiiski and Jerzykiewicz, 1974; Gradziiiski et al., 1977). The lithology is predominantly sandstone, like the underlying Baruungoyot Formation, but the Nemegt is interpreted as consisting predominantly of alluvial flood-plain and channel deposits, thereby indicating development of a perennial fluvial system and, hence, a wetter environment. Exposures of the Nemegt Formation and presumed equivalents are widespread.

The vertebrate faunas of the Djadokhta and Baruun-goyot formations, both of which include mammals, are rather similar. Both are included in the "Barungoytian land-vertebrate age" (Jerzykiewicz and Russell, 1991), although older "Djadokhta" and younger "Khulsan" subdivisions are sometimes recognized (e.g., Gradzuiski et al., 1977; Osmolska, 1980). Predictably, the differences seem to reflect environmental change, with aquatic taxa being rare or absent in the Djadokhta Formation. Mammals are less diverse in the Baruungoyot Formation, but the composition is rather similar; djadochtatheroidean multi-tuberculates dominate, with eutherians and deltatheroidans being represented by only a few species. Four or more species may be shared between the assemblages, depending on the correlation of the Ukhaa Tolgod locality and the identity of its fauna (see later). The fauna of the Nemegt Formation differs considerably from those of the Djadokhta or Baruungoyot formations and is assigned its own land-vertebrate "age," known as the Nemegtian (Jerzykiewicz and Russell, 1991). Aquatic vertebrates such as fish are present, reflecting the wetter, fluvially dominated deposi-tional setting (data for charophytes and ostracods are given by Szczechura and Blaszyk, 1970; Karczewska and Ziembinska-Tworzydlo, 1970, 1981; Szczechura, 1978). Large dinosaurs, such as the tyrannosaurid Tarbosaurus, are abundant, but microvertebrates are notoriously scarce. Mammals have not yet been reported from the Nemegt Formation in its type area, but two occurrences are in strata referred to the unit. In compiling mammal distributions from localities of the Gobi Desert, we have re ferred to our own unpublished observations, to existing compendia (Gradziiiski et al., 1977; Jerzykiewicz and Russell, 1991), and to the original literature (Kielan-Jaworowska and Sochava, 1969; Kielan-Jaworowska, 1970a, 1974a, 1975b,c, 1984b; Kielan-Jaworowska and Dashzeveg, 1978; Kielan-Jaworowska and Trofimov, 1980,1981; Trofi-mov and Szalay, 1994; Dashzeveg et al., 1995; Rougier, Wible, and Novacek, 1996; Szalay and Trofimov, 1996; Kielan-Jaworowska and Hurum, 1997; Novacek et al., 1997; Rougier et al., 1997,1998; Kielan-Jaworowska et al., 2003).

The classic locality for mammals from the Djadokhta Formation is Bayan Zag (figure 2.17), where 10 species are known. One of the multituberculates, Djadochtatherium matthewi, is similar to North American members of the Taeniolabididae and for a time was included in a North American genus (Kielan-Jaworowska and Sloan, 1979). Currently, all of the multituberculates from Bayan Zag and virtually all from the Late Cretaceous of Mongolia are referred to Djadochtatherioidea, which represent a largely Asiatic group (Kielan-Jaworowska and Hurum, 1997,2001; Rougier et al., 1997, see chapter 8). At least two multi-tuberculate families are represented at Bayan Zag, Djadochtatheriidae (Djadochtatherium, Kryptobaatar) and Sloanbaataridae (Sloanbaatar, Kamptobaatar); the affinities of the fifth genus, Bulganbaatar, are uncertain. Possibly two deltatheroidans, Deltatheridium and Deltatheroides, and Hyotheridium (of doubtful status) are present in the fauna. Interestingly, Deltatheridium, or something strikingly similar, is also known from the Late Cretaceous of North America (Fox, 1974a), suggesting Campanian dispersal between the two continents (Cifelli and Gordon, 1999). The affinities of the eutherians, Zalambdalestes and Kennalestes, are not settled; one or both may represent endemic Asiatic clades that left no descendants, or they may be related to groups that later achieved success either in Asia or elsewhere (e.g., Novacek et al., 1997; Rougier et al., 1998; Archibald et al., 2001; Fostowicz-Frelik and Kielan-Jaworowska, 2002, see chapter 13).

Two other mammal localities may be equivalent in age to Bayan Zag, though neither is formally placed in the Djadokhta Formation. The Tögrög (referred to also as Toogreeg, Toogreek, Tugrig, Tugrugeen Shireh) locality, some 40 km northwest of Bayan Zag (figure 2.17), is in the Tögrög Beds, which are similar to strata of the Djadokhta Formation and are thought to be equivalent (Gradzuiski et al., 1977). The occurrence of monospecific dinosaur accumulations at Tögrög and the fact that dinosaur skeletons commonly appear to be in a struggling death pose suggest that death occurred during episodic sandstorm events (Jerzykiewicz et al., 1993). Three mammals are figure 2.17. Late Cretaceous mammal localities of Mongolia and northern China. 1, Bayan Zag (Djadokhta Formation, ?early Cam-panian); 2, Togrog (?equivalent to Djadokhta Formation, ?early Campanian); 3, Udan Sayr (?equivalent to Baruungoyot Formation, ?late Campanian); 4, Ukhaa Tolgod (?equivalent to Djadokhta Formation, ?early Campanian); 5, Khulsan (Baruungoyot Formation, ?late Campanian); 6, Nemegt (Baruungoyot Formation, ?late Campanian); 7, Guriliin Tsav (?Nemegt Formation, ?Maastrichtian); 8, Bugiin Tsav (?Nemegt Formation, ?late Maastrichtian or early Tertiary); 9, Khaichin Uul (?Nemegt Formation, ?late Maastrichtian); 10, Hermiin Tsav (Hermiin Tsav Red Beds, ?equivalent to Baruungoyot Formation, ?late Campanian); 11, Bayan Mandahu (equivalent to Djadokhta Formation, ?early Campanian; Inner Mongolia, China); 12, Tsondolein-Khuduk (unknown unit, ?Cenomanian; Gansu, China).

Djadokhta Formation

figure 2.17. Late Cretaceous mammal localities of Mongolia and northern China. 1, Bayan Zag (Djadokhta Formation, ?early Cam-panian); 2, Togrog (?equivalent to Djadokhta Formation, ?early Campanian); 3, Udan Sayr (?equivalent to Baruungoyot Formation, ?late Campanian); 4, Ukhaa Tolgod (?equivalent to Djadokhta Formation, ?early Campanian); 5, Khulsan (Baruungoyot Formation, ?late Campanian); 6, Nemegt (Baruungoyot Formation, ?late Campanian); 7, Guriliin Tsav (?Nemegt Formation, ?Maastrichtian); 8, Bugiin Tsav (?Nemegt Formation, ?late Maastrichtian or early Tertiary); 9, Khaichin Uul (?Nemegt Formation, ?late Maastrichtian); 10, Hermiin Tsav (Hermiin Tsav Red Beds, ?equivalent to Baruungoyot Formation, ?late Campanian); 11, Bayan Mandahu (equivalent to Djadokhta Formation, ?early Campanian; Inner Mongolia, China); 12, Tsondolein-Khuduk (unknown unit, ?Cenomanian; Gansu, China).

known from this site and all three are present at Bayan Zag. (Kryptobaatar saichanensis, originally described as a separate species by Kielan-Jaworowska and Dashzeveg, 1978, is now regarded as a junior synonym of K. dash-zevegi, see Wible and Rougier, 2000, and chapter 8.)

A much more diverse assemblage is known from Ukhaa Tolgod, which (as noted) has yielded a virtual treasure trove of mammal skulls and skeletons. Lithologi-cally, the unnamed beds at Ukhaa Tolgod are quite similar to those of the Djadokhta Formation (e.g., Rougier et al., 1997) and are generally considered to be equivalent. However, in view of faunistic differences (see Kielan-Jaworowska et al., 2003) and the lack of geological studies demonstrating the continuity of beds that crop out at Ukhaa Tolgod with those of the Djadokhta Formation in other places (e.g., Bayan Zag), for the purpose of this book we refer to these beds provisionally as "Ukhaa Tolgod Beds." Ukhaa Tolgod is in the Nemegt Basin and a few tens of kilometers east of two mammal sites in the Baru-ungoyot Formation, Khulsan and Nemegt (figure 2.17). As at Togrog, death and burial of the fauna at Ukhaa Tolgod may have occurred during sandstorms, with the mammals perhaps being buried in their burrows (Dashzeveg et al., 1995). Somewhat surprisingly, fossorial adaptations have been suggested for only one of the mammalian taxa from the Djadokhta Formation, a multituberculate (Kielan-Jaworowska, 1989). To date, hundreds of mammal skulls, many with associated skeletons, have been collected at Ukhaa Tolgod, but the faunal list for the site (Dashzeveg et al., 1995) remains preliminary; thus far, only four of the mammal species from this locality have been described (Rougier, Wible, and Novacek, 1996; Rougier et al., 1997, 1998; Novacek et al., 1997; Kielan-Jaworowska, 1998; Wible and Rougier, 2000). Dashzeveg et al. (1995) make the interesting observation that multituberculate specimens far outnumber those of therians at Ukhaa Tolgod—

nearly 90% of the skulls belong to multituberculates. Notable among Ukhaa Tolgod mammals is the eutherian Ukhaatherium nessovi, represented by a skeleton (Novacek et al., 1997). Multituberculates of Ukhaa Tolgod include Sloanbaataridae (Sloanbaatar, Kamptobaatar), Djadoch-tatheriidae (Djadochtatherium, Kryptobaatar, and Tom-baatar) (see Rougier et al., 1997; Kielan-Jaworowska et al., 2003), and Chulsanbaatar, currently assigned to family incertae sedis.

Mammalian specimens have been published from five areas in the Baruungoyot Formation or strata thought to be equivalent. All but two (Udan Sayr and Hermiin Tsav) are in the Nemegt Basin. Khulsan, the stratotype area for the Baruungoyot Formation itself, has yielded three mammals, two multituberculates and one therian, the latter unknown from Ukhaa Tolgod. Each of the species at Khul-san is also known from the nearby site of Nemegt, which also includes the eutherian Barunlestes butleri. Hermiin Tsav (referred to also as Khermeen Tsav) is farther to the west and the fossiliferous strata cannot be directly correlated with named units. However, the Hermiin Tsav Red Beds are thought to be time equivalent with the Baruungoyot Formation (Gradziiiski et al., 1977); Jerzykiewicz and Russell (1991) simply refer the fossil strata at Hermiin Tsav to this unit. There are two mammal sites in the area, Hermiin Tsav I and II; because they are close geographically and stratigraphically, our list of seven mammals is a composite. The multituberculate Nessovbaatar multicosta-tus is known only from Hermiin Tsav (Kielan-Jaworowska and Hurum, 1997). Udan Sayr (Ivakhnenko and Kurzanov, 1988) lies about 60 km west-southwest of Bayan Zag. It occurs in rocks that are lithologically similar to those of the Baruungoyot Formation (Jerzykiewicz and Russell, 1991), and we provisionally consider the fauna to be time equivalent. One mammal, Asiatherium reshetovi, has been described from Udan Sayr (Trofimov and Szalay, 1994; Szalay and Trofimov, 1996). Asiatherium, represented by a partial skull and skeleton, is of particular interest because it is thought to be a marsupial (but see Fox, 1997a), suggesting a much earlier presence of Marsupialia in Asia than previously envisaged (Trofimov and Szalay, 1994; Szalay and Trofimov, 1996). Though the dentition is rather specialized, it is marsupial-like in certain respects (Cifelli and Muizon, 1997, see also chapter 12).

Two mammalian occurrences are known from either the Nemegt Formation or possibly equivalent strata (figure 2.17). The locality of Khaichin Uul (Khaicheen Ula) lies about 100 km west-northwest of the type Nemegt area and 20-25 km southwest of Bugiin Tsav. Initially, the age of the fossiliferous stratum (believed to be equivalent to the highest unit exposed in the badlands of Bugiin Tsav) was cautiously estimated as either latest Cretaceous or early Paleocene (Kielan-Jaworowska and Sochava, 1969). Later work suggested correlation with (Gradziniski et al., 1977) or placement in (Jerzykiewicz and Russell, 1991) the Nemegt Formation. The only mammal specimen collected at Khaichin Uul consists of the rostrum and den-taries of a multituberculate, Buginbaatar transaltaiensis (see Kielan-Jaworowska and Sochava, 1969; Trofimov, 1975). Buginbaatar has been referred to various families and continues to elude definitive taxonomic placement. Herein it is tentatively referred to the otherwise North American family Cimolomyidae (superfamily incertae sedis, see chapter 8), thus implying some Late Cretaceous interchange between North America and Asia. Ptilodon-toid multituberculates, which flourished during the Late Cretaceous of North America are not known from the Late Cretaceous of Mongolia (Kielan-Jaworowska, 1979, see also Kielan-Jaworowska and Hurum, 2001).

The second mammal site that may be in the Nemegt Formation is Guriliin Tsav, about 75 km northwest of the type area for the formation and near Bugiin Tsav. Like Khaichin Uul, the fossil horizon at Guriliin Tsav has been considered to be either correlative to the Nemegt Formation (Kielan-Jaworowska and Nessov, 1990) or within that unit (Jerzykiewicz and Russell, 1991). The skull of a large, dentally advanced metatherian is known from Guriliin Tsav. The specimen, which has not yet been described or named, is referred to simply as the "Guirliin Tsav skull."9 The specimen is of considerable importance in that it is the only known skull of a Cretaceous meta-therian that preserves substantial morphology of the basi-cranial region (see Szalay and Trofimov, 1996). One feature of the Guriliin Tsav skull, an auditory bulla with a contribution from the alisphenoid bone, is marsupial-like (Kielan-Jaworowska and Nessov, 1990), but the significance of this character is uncertain (Muizon, 1994).

China

The record of Late Cretaceous mammals in China (table 2.19) is rather limited. The oldest documented occurrence, both in terms of geological age and time of discovery, is at Tsondolein-Khuduk, Gansu Province (figure 2.17). An isolated mammalian axis vertebra from this site was reported by Bohlin (1953) and later designated the type of Khuduklestes bohlini by Nessov, Sigogneau-Russell, and Russell (1994), who follow Nessov et al. (1989) in considering Tsondolein-Khuduk to be of Cenomanian age, based on occurrence there of primitive protoceratopsian

9 The specimen was referred to "Eodeltatheridium kurzanovi" and "Neodeltatheridium" (no species designation) by Gambaryan (1989); each is a nomen nudum.

dinosaurs and other lower vertebrates. However, the rock unit and geographic position of the site are not known with certainty. The map location given by Bohlin (1953) appears to be close to at least three Early Cretaceous localities found by personnel of the Institute of Vertebrate Paleontology and Paleoanthropology, Beijing (see Dong, 1993: figure 1). Hence there is a possibility that the specimen of Khuduklestes—which in any case is not very informative—may be of Early Cretaceous age. We tentatively refer Khuduklestes to Deltatheroida,based mainly on its large size.

Unidentified mammalian fossils have been reported from the Quantou Formation, Jilin Province (Wood et al., 2001). Age for the occurrence is not well constrained, with estimates ranging from Aptian to Cenomanian. Comment must be deferred until the materials have been described, though the fossils promise to be of great interest.

More extensive finds of Late Cretaceous mammals have been made in the Gobi Desert, Inner Mongolia. The site of Bayan Mandahu (figure 2.17) has yielded about 50 mammal skulls, some with associated skeletons. The site has been worked by Chinese-Belgian Expeditions and much of the material is under study by Pascal Godefroit and Thierry Smith, with Chinese colleagues. The specimens come mainly from caliche nodules in a unit that is similar to the Djadokhta Formation (Dong, 1993); indeed, Bayan Mandahu was originally placed within this unit (Jerzykiewicz et al., 1989), but it occurs within a different depositional basin and is now simply considered a lateral equivalent (Jerzykiewicz et al., 1993). As at Bayan Zag, the type locality for the Djadokhta Formation, the most common fossils are protoceratopsian dinosaurs. Poses of the dinosaur skeletons (some ofwhich occur in monospecific bone beds) suggest that they died and were buried in situ—some struggling to free themselves—during sandstorm events. As noted, this is a common finding for Djadokhta or Djadokhta-equivalent sites, and the same appears to be the case for Togrog and Ukhaa Tolgod in Mongolia. The mammals from Bayan Mandahu remain largely unstudied; Wang et al. (2001a) list three taxa, the eutherians Ken-nalestes and Zalambdalestes, and the multituberculate Kryptobaatar (subsequently described as K. mandahuensis by Smith et al., 2001). Kielan-Jaworowska et al. (2003, based on a personal communication from T. Smith and Y.-M. Hu) listed also Djadochtatherium matthewi and ?Tombaatar sab-uli. One would guess that several more taxa are represented, and our compilation (table 2.19) thus almost certainly underrepresents mammalian diversity at Bayan Mandahu.

Japan

A Late Cretaceous mammal was recently reported from Japan (Setoguchi, Tsubamoto, et al., 1999). The locality is

Amagimi Dam near Mifune, Kumamoto Prefecture, Kiyushu (figure 2.4); the fossil horizon is in the upper part of the "Upper Formation" of the Mifune Group. Marine beds in the sequence permit age bracketing for the occurrence based on molluscs. The underlying "Middle Formation" is middle Cenomanian and the overlying Gan-kaizan Formation is lower Santonian; the Amagimi Dam site is thus believed to be of late Cenomanian to early Tur-onian age. The mammal specimen is a dentary fragment with one molar, referred to the new species Sorlestes mifu-nensis. This genus is otherwise known by two species from middle Asia, where it ranges from the early Turonian to the Coniacian (see Nessov, 1985a, 1993; Nessov, Sigogneau-Russell, and Russell, 1994). S. mifunensis is the oldest known member of "Zhelestidae," thought to be related to ungulates, and shows that the group has a greater antiquity and distribution than previously thought (Setoguchi, Tsubamoto, et al., 1999).

India

A diverse fauna and flora has been recovered from sedimentary rocks associated with the Deccan Basalts, Andhra Pradesh, India. The major fossiliferous horizon lies within the basalt sequence, in the so-called intertrappean beds. Long thought to be of Tertiary age, a diverse body of evidence indicates a Maastrichtian age, with radiometric dates indicating a 1- to 3-Ma eruptive phase for the basalts (Prasad et al., 1994). Thus far, the site of Naskal (figure 2.5, table 2.20) has yielded the bulk of mammalian fossils. Sed-imentologic and taphonomic study (Khajuria and Prasad, 1998) indicates that Naskal represents a drought-induced mass death assemblage situated at a distal floodplain lake. The bones and teeth are corroded, with some hydrologic sorting, but there is no evidence of digestion or scavenging; instead, the damage has been attributed to subaerial exposure. As many as four mammals are known from Naskal. Three are assigned to Deccanolestes (see Prasad and Sahni, 1988; Prasad et al., 1994). The affinities of this genus, which have significant bearing on eutherian bio-geography and faunal relations of India, are enigmatic (see chapter 13). Deccanolestes was described as a "palaeo-ryctid." However, Godinot and Prasad (1994) suggest that it may be related to Archonta—a major clade of eutheri-ans including primates, bats, and suspected allies. Interestingly, study of foot bones referred to Deccanolestes led these authors to conclude that it had some degree of arboreal specialization.

The fourth mammal from Naskal is as yet unnamed, but is of great significance because it appears to belong to the highly specialized Gondwanatheria, a group of bizarre mammals previously assigned to Multituberculata but now classified as Mammalia incertae sedis (see chapters 8

table 2.20. Late Cretaceous Mammals of India (see figure 2.5; locality numbers do not correspond between map and table). Localities: 1, Naskal; 2, Rangapur (both Intertrappean Beds; Maastrichtian ?and Paleocene, Andhra Pradesh)

Basal Boreosphenida ("tribotherians") Family incertae sedis Gen. et sp. indet. (2) Eutheria

Family incertae sedis

Deccanolestes hislopi (1) Deccanolestes cf. hislopi (1) Deccanolestes robustus (1) Deccanolestes sp. (2) Gondwanatheria Sudamericidae

and 14). This unnamed mammal appears to be closely related to taxa known from the Late Cretaceous of South America and Madagascar. Thus, it provides evidence of pangondwanan distribution among some southern taxa, at least, which may have dispersed between South America and Indo-Madagascar via Antarctica (Krause, Prasad, et al., 1997). A second mammal site in the intertrappean beds of Andhra Pradesh has been reported from the vicinity of Rangapur (Wilson and Rana, 2001). Fossils collected to date (some 20 mammalian teeth) evidently represent two taxa, Deccanolestes sp. and a primitive boreosphenidan. The age of the site is uncertain; it may be Late Cretaceous or Paleocene.

0 0

Post a comment