ISSN 0145-9058 BULLETIN OF CARNEGIE MUSEUM OF NATURAL HISTORY TAXONOMY AND EVOLUTION OF LATE CRETACEOUS LIZARDS (REPTILIA: SQUAMATA) FROM WESTERN CANADA GAO KEQIN and RICHARD C. FOX NUMBER 33 PITTSBURGH, 1996 BULLETIN of CARNEGIE MUSEUM OF NATURAL HISTORY MCZ , library DEC 1 8 1996 TAXONOMY AND EVOLUTION OF LATE CRETACEOUS LIZARDS (REPTILIA: SQUAMATA) FROM WESTERN CANADA GAO KEQIN Institute of Vertebrate Paleontology and Paleoanthropology, Academia Sinica, Beijing 100044, China; and Laboratory for Vertebrate Paleontology, Department of Biological Sciences, University of Alberta, Edmonton, Alberta, Canada T6G 2E9. Current address: Department of Vertebrate Paleontology, American Museum of Natural History, Central Park West at 79th Street, New York, New York 10024-5192. RICHARD C. FOX Research Associate, Section of Vertebrate Paleontology Laboratory for Vertebrate Paleontology, Department of Biological Sciences, University of Alberta, Edmonton, Alberta, Canada T6G 2E9. NUMBER 33 PITTSBURGH, 1996 BULLETIN OF CARNEGIE MUSEUM OF NATURAL HISTORY Number 33, pages 1-107, 40 figures Issued 16 December 1996 James B. Richardson III, Interim Director Editorial Staff: John L. Carter, Editor ; Bradley C. Livezey, Editor ; David R. Watters, Editor Mary Ann Schmidt, ELS, Assistant Editor Cover illustration: Glyptogenys ornata, Oldman Formation, Alberta: UALVP 29735 (holotype), incomplete left dentary, medial view (see Fig. 12, lower illustration). BULLETINS OF CARNEGIE MUSEUM OF NATURAL HISTORY are published at irregular intervals by Carnegie Museum of Natural History, 4400 Forbes Avenue, Pittsburgh, Pennsylvania 15213-4080, by the authority of the Board of Trustees of Carnegie Institute. © 1996 by Carnegie Institute, all rights reserved. ISSN 0145-9058 THE CARNEGIE MUSEUM OF NATURAL HISTORY Contents Abstract 1 Introduction Materials and Methods 3 Geological Setting 6 Dental Morphology and Terminology 10 Systematic Paleontology 13 Order Squamata Oppel, 1811 13 Suborder Lacertila Owen, 1842 13 Infraorder Iguania Cuvier, 1807 13 Family Iguanidae* Gray, 1827 13 Genus Cnephasaurus, new genus 13 Genus and species new (A) 16 Genus and species new (B) 17 Infraorder Scincomorpha Camp, 1923 18 Family Teiidae Gray, 1827 18 Genus Chamops Marsh, 1892 19 Genus Socognathus Gao and Fox, 1991 25 Genus Sphenosiagon Gao and Fox, 1991 28 Genus Glyptogenys Gao and Fox, 1991 31 Genus Gerontoseps Gao and Fox, 1991 33 Genus Leptochamops Estes, 1964 36 Genus Meniscognathus Estes, 1964 40 Genus Hciptosphenus Estes, 1964 42 Genus Stypodontosaurus, new genus 45 Family Scincidae Gray, 1825 46 Genus Penemabuya, new genus 46 Genus Orthrioscincus, new genus 49 Genus Aocnodromeus, new genus 51 Family ?Xantusiidae Baird, 1859 53 Genus and species new (unnamed) 54 Genus Contogenys Estes, 1969 55 Family ?Cordylidae Gray, 1837 58 Genus and species undetermined (A) 58 Genus and species undetermined (B) 60 Family incertae sedis 60 Genus and species new (A) 60 Genus and species new (B) 61 Genus and species undetermined 62 Infraorder Anguimorpha Furbringer, 1900 62 Family Xenosauridae Cope, 1886 62 Genus and species undetermined 66 Family Anguidae Gray, 1825 67 Genus Odaxosaurus Gilmore, 1928 67 Genus and species undetermined 73 Family Necrosauridae* Hoffstetter, 1943 74 Genus Colpodontosaurus Estes, 1964 74 Genus and species new (unnamed) 75 Genus Parasaniwa Gilmore, 1928 76 Family Helodermatidae Gray, 1837 81 Genus Paraderma Estes, 1964 82 Genus Labrodioctes, new genus 86 iii Genus and species undetermined 88 Family Varanidae Gray, 1827 89 Genus Palcieosaniwa Gilmore, 1928 90 Geological Distribution and Evolution of Late Cretaceous Lizards in Western Canada 94 Conclusions 99 Acknowledgments 100 Literature Cited 100 IV ABSTRACT Upper Cretaceous nonmarine deposits of western Canada have yielded fossil lizards representing some 40 species in some 30 genera and ten families. Hundreds of specimens from geological formations of Aquilan, Judithian, and Lancian age (Campanian- Maastrichtian) reveal previously unknown aspects of the evolu- tionary history of several lizard families during the last 18 Myr of the Cretaceous Period in western Canada. Among these liz- ards, iguanids are described for the first time from the Upper Cretaceous in North America, and primitive scincids, anguids, and possible cordylids are recorded there, as well. On the basis of taxonomic composition and geological age, three assemblages are recognizable: 1) An Aquilan (early Cam- panian) assemblage, known from the Milk River Formation, doc- uments an early diversification of the Iguanidae*, Scincidae, Xe- nosauridae, Anguidae, and Necrosauridae*, whereas the Teiidae are poorly recorded. 2) A Judithian (mid-Campanian) assem- blage, from the Oldman Formation, is composed predominantly of teiids; these are substantially different taxonomically from teiids of Lancian age and, hence, represent a distinct evolutionary stage of the family (this result is contrary to that of previous authors, but is comparable to mammalian patterns of evolution over the same interval). This assemblage also includes the ear- liest records of the Helodermatidae and Varanidae in North America. 3) A Lancian (late Maastrichtian) assemblage, from the Frenchman and Scollard formations, is virtually identical to that of the type Lance Formation, Wyoming, but includes specimens of several new taxa, and others that allow taxonomic revisions of several species previously known from outside the study area. These newly discovered lizards represent important components of the North American Late Cretaceous nonmammalian verte- brate fauna, and a taxonomic treatment of these lizards gives a more nearly complete picture and a better understanding of the evolutionary history of the relevant lizard families than was pos- sible before. 1 2 BULLETIN CARNEGIE MUSEUM OF NATURAL HISTORY NO. 33 INTRODUCTION Lizards are the most successful modem reptiles in terms of their great taxonomic and ecologic di- versities, and wide geographical distribution. Osteo- logically, the group can only be diagnosed as prim- itive squamates lacking the specializations of am- phisbaenians and snakes, but retaining such ple- siomorphic character states as loss of the lower temporal bar and development of a highly mobile joint between the quadrate and squamosal, a con- dition commonly termed streptostyly. The true liz- ards in the sense of Estes (1983a, 1983b) can be traced back to the Late Jurassic, but no lizards of this age can be referred to any modern lizard family, although infraordinal relationships with Recent groups can be determined. Early Cretaceous lizard fossils are poorly known, partly because of the worldwide transgression that submerged about one- third of the present land area of the Earth (Dott and Batten, 1988; also see Funnell, 1990). However, several recent discoveries have added significantly to the lizard fossil record of this age; these new discoveries include fossil materials from Central Asia (Nessov, 1985, 1988), Mongolia (Alifanov, 1989; M. A. Norell, personal communication, 1993), Texas (Murry et al., 1989; Winkler et al., 1990), and from southern Utah (R. L. Cifelli, per- sonal communication, 1991). Study of Late Creta- ceous lizard fossils is especially significant, since the Late Cretaceous appears to have been an im- portant period for the establishment and early di- versification of most modern lizard families (Estes, 1983b); furthermore, the widely exposed fossilifer- ous nonmarine deposits of this age in the Western Interior of North America and use of highly pro- ductive screen-washing techniques for recovery of small vertebrate fossils (Hatcher, 1896; McKenna, 1960; Clemens, 1963; Lillegraven, 1969) have pro- vided a unique advantage in undertaking this study. In North America, the first discovery of fossil lizards from the Upper Cretaceous was associated with paleontological exploration during the period 1889-1892 in Converse County (now Niobrara County), Wyoming. Marsh (1892) made the first re- port of Late Cretaceous terrestrial lizard fossils, naming Chamops segnis (based on an incomplete dentary) and Iguanavus teres (based on several ver- tebrae) from the “Laramie Formation” of Wyoming (the “Laramie Formation” was later renamed [Stan- ton, 1910] as the Lance Formation). Work since has shown that Chamops segnis is a common Late Cre- taceous teiid and Iguanavus teres is possibly the synonym of the former (Estes, 1964, 1983a), or a nomen dubium as treated in this paper. During this early period, although several thousand isolated teeth, jaws, and other materials of a variety of ver- tebrates had been found by J. B. Hatcher (chief col- lector for O. C. Marsh) and his assistants (based on these materials, for example. Marsh named 18 new genera and 32 species of mammals), no special at- tention was paid to lizards except for the two forms mentioned above (Gilmore, 1928). Gilmore (1928) made the pioneer contribution to the systematic study of Late Cretaceous lizards from North America. In his monograph “Fossil Lizards of North America,” Gilmore described 17 species in 12 genera from the Lance Formation of Wyo- ming and equivalent beds elsewhere; among these, 13 species in ten genera were new, and the names of seven genera and species are still valid today, with the others having been emended since (Estes, 1964, 1983a). Few subsequent papers in this field were published, until Estes’ (1964) comprehensive work on small nonmammalian vertebrates from the Lance Formation, Wyoming. Other North American Late Cretaceous lizards were described by Estes et al. (1969) from the Hell Creek Formation, Montana; Sahni (1972) from the Judith River Formation, Montana; and Armstrong-Ziegler (1978, 1980) and Sullivan (1981) from the Fruitland Formation, New Mexico. Recent discoveries also include unde- scribed materials from New Jersey (Parris and Grandstaff, 1989; Denton et al., 1991), Utah (R. L. Cifelli, personal communication, 1991; Eaton, 1993), and from Texas (Rowe et al., 1992). Estes (1983a) provided the first and the most recent tax- onomic codification of the data on all fossil lizards known at the time: his “Sauria terrestria, Amphis- baenia” contribution to the “Handbuch der Palao- herpetologie” included 18 species in 17 genera of Late Cretaceous lizards from North America. In western Canada, early discoveries of Late Cre- taceous terrestrial lizards were coincident with a pe- riod of the “Great Canadian Dinosaur Rush,” from 1912 to 1925 (see Koster et al., 1987). Gilmore (1923) marked the first report of Late Cretaceous lizards from western Canada by referring a nearly complete dorsal vertebra (UA 1 12) to a “lacertian” (= lacertilian). The specimen was found by G. F. Sternberg in 1921, at Sandy Point, South Saskatch- ewan River, near Empress, southeastern Alberta (Belly River Formation, now Oldman Formation; Russell and Landes, 1940). Later reports of Creta- 1996 GAO AND FOX— LATE CRETACEOUS LIZARDS 3 ceous lizards from western Canada were given in descriptive papers by C. M. Sternberg (1951), Waldman (1970), and Gao and Fox (1991); and in faunal lists by L. S. Russell (1935, 1964), Russell and Landes (1940), Fox (1972a, 1976, 1989), Cur- rie (1986), Eberth (1987), and Koster et al. (1987). This study focuses on fossil lizards from locali- ties in geologic formations of Aquilan, Judithian, and Lancian age (NALMA = North American Land Mammal Age, see L. S. Russell, 1964; Lillegraven and McKenna, 1986) in western Canada (see Fig. 1), covering a geologic span of approximately 18 Myr (Lillegraven and McKenna, 1986; Harland et al., 1990). However, the major concern of this study is the fossil lizards from the Oldman and Milk River formations. The former formation is about 1 1 Myr and the latter 1 8 Myr earlier than the Lance and Hell Creek formations (Lillegraven and McKenna, 1986; Harland et al., 1990). The fossil lizards from the Oldman Formation (Judithian) indicate that the Judithian lizard assemblage is substantially different from that of Lancian age, contrary to the conclu- MATERIALS This study is based mainly on fossils collected from the Upper Cretaceous Milk River, Oldman, Frenchman, and Scollard formations, western Can- ada; but some specimens from the Lance and Hell Creek formations in the United States are also in- cluded because they provide new information for a better understanding of the relevant taxa, and pro- vide materials for recognition of new taxa that have close relatives in the study area. The specimens from the Milk River Formation were collected by UALVP field parties from 1968 to 1970. The Old- man specimens are in the collections of both UALVP and RTMP, from several localites in south- ern Alberta, made from 1966 to the present. The specimens from the Frenchman Formation were col- lected by the SMNH and UALVP parties from 1979 to 1984. All the new materials from the Lance and Hell Creek formations used in this study are in the UALVP collections, obtained from field work in 1969-1976. Finally, the UALVP Scollard lizards were recently sorted from screened concentrate originally collected during 1964-1967 by Univer- sity of Kansas field parties (Lillegraven, 1969). The specimens used in this study are mostly tooth-bearing maxillaries and dentaries. These pro- vide materials for the recognition of a number of new taxa of lizards, as well as assistance in under- sions of previous authors (see Sahni, 1972; Arm- strong-Ziegler, 1978, 1980; Sullivan, 1981; Estes, 1983a). This study shows that most lizard speci- mens from the Oldman Formation are representives of new genera and new species. Some are referrable to genera occurring in Lancian horizons, but none can be definitely assigned to Lancian species. This result is consistent with the mammalian evidence that shows less than five percent of mammalian spe- cies of Judithian age are the same as those from Lancian horizons (see Lillegraven and McKenna, 1986). The Milk River (Aquilan) lizards document an early diversification of several modern families, especially Scincidae, and include the earliest known North American record of the Iguanidae*. These newly discovered lizards represent important com- ponents of the North American Late Cretaceous nonmammalian vertebrate fauna, and a study of these lizards gives a more nearly complete picture and a better understanding of the taxonomic and evolutionary history of the relevant lizard families than was previously possible. D METHODS standing the phylogenetic relationships of several previously known species. Other materials in the available collections include disarticulated skull el- ements, such as frontals, parietals, jugals, ptery- goids, premaxillaries (see Fig. 2 for anatomical po- sitions), and disarticulated vertebrae and limb bones. These latter elements in most cases are un- reliable for identification at generic and specific lev- els without reference to articulated skulls or skele- tons, with the exception of certain groups (e.g., Xe- nosauridae) in which the disarticulated skull ele- ments show a species-specific sculpture pattern. The identification of jaw material is based mainly on dental morphology (tooth form and cusp pattern) and general jaw configuration. Tooth form and cusp pattern are also important for ecological interpre- tations, as several studies (Lonnberg, 1903; Schmidt, 1919; Cowles, 1930; Hotton, 1955; Presch, 1974b; Greene, 1982; Estes and Williams, 1984) have shown the connections between dental morphology and food preference of lizards, and these studies have demonstrated that the tooth type and the particular cusp pattern as well as the pattern of wear provide indirect evidence as to the possible habit and diet of the lizards. Most of the specimens for this study have been obtained through screen washing in the field and in 4 BULLETIN CARNEGIE MUSEUM OF NATURAL HISTORY NO. 33 1996 GAO AND FOX— LATE CRETACEOUS LIZARDS 5 Fig. 2. — Skull and mandible of Eumeces obsoletus (Scincidae), showing anatomical structures of lizards (after Estes et al., 1988). A-C, skull in lateral, dorsal, and ventral views; D, E, mandible in lateral and medial views. Abbreviations for this and all sub- sequent figures: aif, anterior inferior alveolar foramen; amf, an- terior mylohyoid foramen; c, coronoid; d, dentary; f, frontal; iaf, inferior alveolar foramen; ims, intramandibular septum; j, jugal; lp, lateral parapet of dentary; m, maxillary; me, Meckelian canal; p, parietal; paf, parietal foramen; pm, premaxillary; pmp, pre- maxillary process; pt, pterygoid; saf, superior alveolar foramen; sbds, subdental shelf; sd, sulcus dentalis; sp, splenial; spds, su- pradental shelf; stp, supratemporal process of parietal. the laboratory, with some relatively large specimens collected by surface picking from outcrops. When necessary, specimens were prepared manually under a Wild M-7 microscope in the laboratory. All ho- lotypes and well-preserved specimens are illustrated in photoplates or outline drawings. The specimens were coated with magnesium chloride, and the pho- tographs were taken using an Olympus OM-2s cam- era with 50 mm macro lens and bellows. The text- figures were drawn with assistance of a camera lu- cida. All measurements of the specimens are in mil- limeters. Institutional abbreviations; AMNH — American Museum of Natural History, New York, New York; ANSP — Academy of Natural Sciences, Philadel- phia, Pennsylvania; CMN (= NMC) — Canadian Museum of Nature (= National Museum of Cana- da), Ottawa, Ontario; GSC — Geological Survey of Canada, Ottawa, Ontario; MCZ — Museum of Com- parative Zoology, Harvard University, Cambridge, Massachusetts; MNA — Museum of Northern Ari- zona, Flagstaff, Arizona; RTMP — Royal Tyrrell Museum of Palaeontology, Drumheller, Alberta; SMNH — Saskatchewan Museum of Natural Histo- ry, Regina, Saskatchewan; SMU — Southern Meth- odist University, Dallas, Texas; UALP — Laboratory of Paleontology, University of Arizona, Tucson, Ar- izona; UALVP (= UA) — Laboratory for Vertebrate Paleontology, Department of Biological Sciences, University of Alberta, Edmonton, Alberta; UCMP (= UC) — Museum of Paleontology, University of California, Berkeley, California; UMMZ — Museum of Zoology, University of Michigan, Ann Arbor, Michigan; UNM — University of New Mexico, Al- buquerque, New Mexico; USNM (= NMNH) — United States National Museum (= National Mu- seum of Natural History, Smithsonian Institution), Washington, D.C.; YPM (= YUM) — Yale Peabody Museum, New Haven, Connecticut; YPM-PU — “Princeton Collection” now housed in Yale Pea- body Museum; ZPAL — Institute of Paleobiology, Polish Academy of Sciences, Warsaw, Poland. Locality abbreviations: AB— Altman Blowout ( = UCMP loc. V-5616; Clemens, 1963), Lance For- mation, Lance Creek, Niobrara County, Wyoming; BB — Bonebed microfossil sites in Dinosaur Provin- cial Park (see Brinkman, 1990; Eberth, 1990/?); BBR — Blackbird Ridge site (NW 1/4, Sec. 6, Tp. 7, R 20), Judith River Formation, Golden Valley County, Montana; BCA — Bug Creek Anthills (Sloan and Van Valen, 1965; Estes et al., 1969), Hell Creek Formation, McCone County, eastern Montana; BTB — Bushy Tailed Blowout (= UCMP Fig. 1. — Geographical distribution of the Upper Cretaceous lizard localities (selected) in western Canada and nearby areas (Montana and Wyoming). Solid triangle indicates Aquilan locality in the Milk River Formation, Alberta; solid squares indicate Judithian localities of the Oldman and Judith River formations, Alberta and Montana; solid dots indicate Lancian localities of the Scollard (Alberta), Frenchman (Saskatchewan), Hell Creek (Montana), and Lance (Wyoming) formations. Abbreviations as listed in Materials and Methods. 6 BULLETIN CARNEGIE MUSEUM OF NATURAL HISTORY NO. 33 loc. V-5711; Clemens, 1963), Lance Formation, Lance Creek, Niobrara County, Wyoming; DPP — Dinosaur Provincial Park (see, e.g., Dodson, 1987; Eberth, 1990 b), Oldman Formation, Alberta; GR — Gryde locality (Tokaryk and James, 1989; Storer, 1991), Frenchman Formation, Frenchman River Valley, southwestern Saskatchewan; HC — Hell Creek Formation, McCone County, Montana (see BCA above); IRV — Irvine locality (Fox, 1968), Oldman Formation, southeastern Alberta; KUA — The University of Kansas, Alberta localities (Clem- ens and Russell, 1965; Lillegraven, 1969), Scollard Formation, Red Deer River Valley near Scollard, central Alberta; LC — Lance Creek Valley and en- virons (see Clemens, 1963), various localities of the Lance Formation, Niobrara County, Wyoming; MR— UALVP Milk River localities (Fox, 1972a), Milk River Formation, southern Alberta; WK — Wounded Knee locality (Fox, 1989), Frenchman Formation, Frenchman River Valley, southwestern Saskatchewan. GEOLOGICAL SETTING The specimens used in this study were collected from three horizons, chronologically from oldest to youngest: the upper member of the Milk River For- mation (Dowling, 1917; Russell and Landes, 1940), early Campanian (Williams and Burke, 1964; Fox, 1976; Lillegraven, 1991) or Aquilan (NALMA, L. S. Russell, 1964; Lillegraven and McKenna, 1986) in age; the Oldman Formation (Russell and Landes, 1940), mid-Campanian (L. S. Russell, 1975; Eberth, 1987) or Judithian (NALMA, L. S. Russell, 1964; Lillegraven and McKenna, 1986) in age; and the Frenchman Formation (Furnival, 1946) and its equivalent Scollard (Irish, 1970; Gibson, 1977; Ler- bekmo and Coulter, 1985), Lance (Stanton, 1910) and Hell Creek (Brown, 1907) formations, all late Maastrichtian or Lancian (Dorf, 1942; L. S. Russell, 1964; Lillegraven and McKenna, 1986) in age. These formations accumulated on the coastal plain adjacent to the Western Interior Seaway during Campanian-Maastrichtian time. The main source of these sediments was from the then-uplifting Rocky Mountains (Price and Mountjoy, 1970; Rahmani and Lerbekmo, 1975; Eberth, 1990a) west of the seaway. Paleogeography and Depositional History At the time when the upper member of the Milk River Formation was deposited (early Campanian), the North American continent was divided by the Western Interior Seaway, which extended from north to south along the eastern side of the Rocky Mountains and linked the Arctic Ocean with the Gulf of Mexico (see, e.g., Williams and Stelck, 1975; Funnell, 1990; Nicholls and Russell, 1990). The shallow epeiric sea covered most of what is now Alberta, southeastern Saskatchewan, eastern Montana, Wyoming, Colorado, northeastern New Mexico, and Texas. In western Canada, continental deposits of that age are exposed only in southern Alberta. By mid-Campanian time, the rising Rocky Moun- tains caused the retreat of the Interior Sea to the east and the deposition of detritus in the Mackenzie River valley of the Northwest Territories, in Alberta and Saskatchewan, and in Montana, Wyoming, and New Mexico (Williams and Stelck, 1975:text-fig. 7; Funnell, 1990:fig. 11). The Oldman Formation and the upper part of the Judith River Formation, Mon- tana, were formed at about the same time. Lizard fossil-bearing deposits of the same age are also known from New Jersey, one of the few occur- rences of nonmarine vertebrates on the eastern side of the Late Cretaceous epeiric seaway (Parris and Grandstaff, 1989; Denton et al., 1991). By late Campanian and early Maastrichtian time, the last major transgression of the Interior Sea (Bearpaw Sea; see Caldwell et al., 1978) covered a large area including Alberta north of the Peace Riv- er Arch (Wall, 1975) and west to the edge of the foothills in southern Alberta and Montana (Williams and Stelck, 1975). Continental deposits were devel- oped in central and western Alberta, and in central Saskatchewan (Edmonton Group and St. Mary Riv- er Formation); however, lizard fossils are rarely re- covered from these deposits, except for a Chamops- like teiid from the Wapiti Formation, northwestern Alberta (see later discussion). In northwestern New Mexico and southwestern Colorado, the Fruitland Formation was probably deposited at about this time: the age of the formation has been determined as Judithian, based on pollen (Fassett and Hinds, 1971), ammonites, and radiometric dating (Cobban, 1973; Fassett, 1987; Brookins and Rigby, 1987), and vertebrate faunal comparison with Lancian and Judithian assemblages from Wyoming and Montana (L. S. Russell, 1975; Fox, 1978; Armstrong-Ziegler, 1996 GAO AND FOX— LATE CRETACEOUS LIZARDS 7 1980; Lillegraven and McKenna, 1986; Lucas et al., 1987; Rigby and Wolberg, 1987); but more recently as Edmontonian on mammalian fossils (Lillegraven and Ostresh, 1990). In the latest Cretaceous, overfilling of sediments from the west accompanying uplift in the Macken- zie Mountains region caused the regression of the late Maastrichtian Sea from the Western Interior (Williams and Stelck, 1975). At the same time, ex- tensive clastic deposits were developed in Alberta (Scollard Formation), Saskatchewan (Frenchman Formation), Wyoming (Lance Formation), and Montana (Hell Creek Formation). A rich fossil ver- tebrate fauna, including a large number of lizard taxa, has been found in these deposits. Geological Formations Within the study area. Upper Cretaceous non- marine deposits are composed mainly of sandstones, siltstones, and shales, and are extensively exposed along the river valleys and occasionally in badlands. Various authors have given detailed descriptions of these deposits; the lizard fossil-bearing sediments include the following formations (see Fig. 3 for cor- relations): Milk River Formation. — Exposed along the Milk River and its tributaries in southern Alberta, the Milk River Formation is approximately equivalent to the Eagle Formation of Montana. Having a max- imum thickness of about 77 m (Russell and Landes, 1940), the Milk River Formation can be subdivided into upper and lower members. The lower member (maximum thickness of 41 m) is dominantly marine in origin, characterized by dark shales grading up- ward into medium-grained sandstones with concre- tions. The upper member is composed of nonmari- ne, drab gray, brown, and dark gray argillaceous sandstones and sandy shale with streaks of impure lignite (Russell and Landes, 1940; Crockford and Clow, 1965). The upper member is about 36 m in maximum thickness (Russell and Landes, 1940) and is Aquilan (early Campanian) in age (L. S. Russell, 1964; Fox, 1976; Lillegraven, 1991). L. S. Russell (1935) recorded the discovery of fossil molluscs and vertebrates from the upper member of the Milk River Formation, from which a fragmentary dentary (GSC 8721) was listed as “Lacertilia incertae sed- is.” Recently, Fox (1968, 1969, 1970, 1971, 1972a, 1972 b, 1984, 1985) reported a series of discoveries of mammalian fossils from the upper member of the formation exposed in Verdigris Coulee, including the geologically youngest triconodonts and sym- metrodonts, primitive lipotyphlan insectivores, and diverse early marsupials. Fox (1972a) also pub- lished a faunal list, which included six lizard taxa associated with those of elasmobranchs, actinopter- ygian fishes, salamanders, turtles, champsosaurs, crocodiles, dinosaurs, and mammals. The lizards are described here for the first time. Oldman Formation. — The Oldman Formation is composed of nonmarine deposits in Alberta that correlate with the upper part of the Judith River Formation of Montana (L. S. Russell, 1964; Eberth, 1987; Thomas et al., 1990). The name “Oldman Formation” was proposed by Russell and Landes (1940) as the substitute for the commonly called “Pale beds” and the upper division of Dawson’s (1881) Belly River series. McLean (1971, 1977) recommended to drop the “Oldman” from the no- menclature and extend the use of “Judith River” into the Alberta area. This motion has been fol- lowed by some authors but rejected by others (e.g., Rahmani and Lerbekmo, 1975; Fox, 1976, 1978; see also Eberth and Hamblin, 1993). In the current relevant literature, both “Oldman” and “upper Ju- dith River” are used for the same geologic forma- tion in Alberta. The name “Oldman Formation” remains in this paper as Russell and Landes (1940) defined it, for its convenient use as a geographically extensive, lithologically mappable unit (J. F. Lerbekmo, per- sonal communication, 1989). The characteristic rocks of this formation are pale gray bentonitic sandstones, with interbedded gray and brown shales; coal seams are seen only at the top of the formation (L. S. Russell, 1964). The outcrops of the formation are widely exposed along the Milk River and the South Saskatchewan River north of Medi- cine Hat, and are best exposed in the Dinosaur Pro- vincial Park along the Red Deer River north of Brooks. The thickness of the formation as known from drilling is about 127 m in the Irvine area, in- creasing southward to about 180 m in the Many- berries area (Crockford and Clow, 1965). The for- mation has yielded radiometric dates of about 76 Myr (Thomas et al., 1990) and a large vertebrate fauna (summarized in L. S. Russell, 1964; Lang- ston, 1965; Fox, 1976; Currie, 1986; Eberth, 1987; and Koster et al., 1987). Within the uppermost 100 m of the formation, a recently recognized discon- formity reflects a major Campanian uplift event in the Cordillera (Eberth, 1990a, 1990/?), and the de- posits above this disconformity have been recently named as the Dinosaur Park Formation (Eberth and Hamblin, 1993). Recent studies of the 24 microfos- sil localities above and below the disconformity 8 BULLETIN CARNEGIE MUSEUM OF NATURAL HISTORY NO. 33 u w> £ Chart showing correlation of the Lpper Cretaceous (Campanian-Maastrichtian) lizard fossil-bearing formations (marked with an asterisk) in North America and Asia 1996 GAO AND FOX— LATE CRETACEOUS LIZARDS 9 (Dodson, 1983, 1987; Brinkman, 1990; Eberth, 1990 b) have yielded important results relevant to the sedimentology and the paleoecology of the for- mations. Gilmore (1932) named Polyodontosaurus gran- dis, a supposed lizard, on the basis of a small den- tary (NMC 8540) from the Belly River Formation (= Oldman and Foremost formations, see Lerbek- mo, 1989); but later work by C. M. Sternberg (1951) and D. A. Russell (1969) indicated that the specimen actually belongs to a troodontid dinosaur. D. A. Russell (1969) regarded Polyodontosaurus grandis as a synonym of Stenonychosaurus ine- qualis, which is now commonly agreed to be a syn- onym of Troodon formosus Leidy, 1856 (see Currie, 1987; D. A. Russell, 1989; Osmolska and Barsbold, 1990). The Wapiti Formation (Dawson, 1881) in north- western Alberta may include nonmarine deposits that are younger than the Oldman Formation (Allan and Carr, 1946; Langston, 1965). C. M. Sternberg (1951) described a lizard ( Chamops cf. [C.] segnis ) from member B of the Wapiti Formation, Kleskun Hills, near Grande Prairie. The age of the formation ranges from Campanian to Maastrichtian (Stott, 1975), whereas member B of the formation is prob- ably late Campanian or somewhat younger in age. Recent discovery of the ceratopsian P achy rhino sau- rus in the same formation at Pipestone Creek by RTMP field parties may indicate that at least part of the formation is equivalent to the Horseshoe Can- yon Formation of central Alberta, and hence is Ed- montonian in age (Tanke, 1988; Dodson and Currie, 1990). Lance Formation and Equivalents. — The Lance Formation in Wyoming and its equivalent Hell Creek Formation in Montana are somewhat outside the region for this study, and the lizard fossils from these formations have been previously described by various authors (e.g., Marsh, 1892; Gilmore, 1928; Estes, 1964, 1969/?, 1983a; and Estes et al., 1969). However, new specimens collected by UALVP field parties during the 1969-1976 period yield new in- formation requiring a review of certain known taxa, and more importantly provide the specimens for recognition of several new taxa, which are closely related to those in the study area; therefore, these specimens are also included in this paper. Exposed mostly in the Powder River Basin of eastern Wyoming, the Lance Formation consists of a succession of interbedded siltstones, channel sand- stones, shales, and lignites (Clemens, 1960, 1963; Estes, 1964). The earliest discovery of vertebrate fossils from this formation was reported by Marsh (1889), who referred to the deposits as the “Cera- tops beds.” Since then, a series of names has been used for the deposits (“Lance Creek beds,” “Con- verse County beds,” “Laramie beds,” and “Lower Fort Union”; for earlier citations, see Clemens, 1963; Estes, 1964) until Stanton (1910), who named the unit as the Lance Formation. The formation is overlain by the Fort Union Formation or Group and underlain by the brackish water sediments of the Fox Hills Formation. The maximum thickness of the formation is about 750 m in the type area (Clemens, 1963). The lizard fossils previously re- ported from this formation represent about 13 spe- cies in 13 genera (Estes et al., 1969; Estes, 1983a). The Hell Creek Formation consists of nonmarine deposits overlying the Fox Hills Formation (brack- ish water deposits) in eastern Montana. The for- mation is composed mainly of gray to brown sand- stones and shales (L. S. Russell, 1964) of fluvial origin (Fastovsky, 1987). The maximum thickness of the formation is about 100 m along Hell Creek (Brown, 1907), but varies in adjacent areas (Archi- bald, 1982). The Bug Creek Anthills (BCA) and nearby localities in the Hell Creek Formation have yielded a vertebrate fauna (Sloan and Van Valen, 1965; Estes et al., 1969; Estes and Berberian. 1970) that has figured importantly in recent debates con- cerning the events at the K-T boundary in North America and the timing of aspects of mammalian evolution. As Archibald (1982) pointed out, the ma- jor argument centers on the age of the fauna: wheth- er the BCA fauna is younger than or essentially contemporaneous with the Lancian fauna. Estes et al. (1969) and Estes and Berberian (1970) stressed the overall similarities in the components of the two faunas, suggesting the “time of deposition of the upper parts of both the Lance Formation and the Hell Creek Formation was essentially synchronous” (Estes and Berberian, 1970:9), whereas Archibald (1987a, 1987/?) and Sloan (1987) proposed a Bu- gcreekian Age (Stage) in recognizing the age of the BCA local fauna as younger than Lancian. Since the evidence is not decisive, the latter proposal has not been widely accepted, although the uppermost parts of the Hell Creek Formation otherwise include Paleocene deposits (Archibald, 1982). In central Alberta, Lancian rocks are called the Scollard Formation, formerly “the upper part of the Edmonton Formation,” “the Upper Edmonton For- mation” (see Lillegraven, 1969), or “the Scollard Member” of the Paskapoo Formation (Irish, 1970). The Scollard Formation, as Gibson (1977) defined 10 BULLETIN CARNEGIE MUSEUM OF NATURAL HISTORY NO. 33 it, includes Paleocene deposits at its top. Litholog- ically, the formation “consists of an interbedded, interfingering sequence of argillaceous sandstone, siltstone, mudstone, and shale. Minor amounts of coal (Seams Nos. 13-14), bentonite, and tuff occur in the formation" (Gibson, 1977:7). The maximum thickness of the formation is about 85 m in the Hux- ley-Big Valley area (Gibson, 1977), and Coal Seam No. 14 (Paleocene part, about 18 m above the K- T boundary) has recently yielded a radiometric date of 63 Myr (Lerbekmo and Coulter, 1985). The Scol- lard Formation rests unconformably on the Battle Formation, which contains a layer of volcanic de- positon called the Kneehills Tuff, dated as 65-66 Myr (Folinsbee et al., 1965; also see Lerbekmo and Coulter, 1985). Fossil lizards were previously un- known from the Scollard Formation, although a rich mammalian fauna from several localities along the Red Deer River valley has been reported and de- scribed (see Clemens and Russell, 1965; Lillegrav- en, 1969). DENTAL MORPHOLO In this section, we define dental terms used sub- sequently and explain their use as appropriate, not- ing ambiguities and even errors in the previous lit- erature on lizard tooth morphology, including the use of mammalian nomenclature for the dental anat- omy of lizards. Tooth Form and Implantation It is common in lizards that the primitive reptilian homodont dentition has been modified by a differ- entiation of the marginal teeth, resulting in some degree of heterodonty. The anterior teeth of some lizards are enlarged and canine-like, and the lateral or cheek teeth may be conical, bicuspid, tricuspid, multicuspid, crested, or even enlarged to form crushing plates. Estes and Williams (1984) have compiled data on the development of heterodonty in various groups of lizards. The evolution of heterodonty in lizards is asso- ciated largely with the variety of feeding habits and different dietary demands. For example, the com- bination of sharp anterior conical teeth with bicus- pid or tricuspid cheek teeth seems best adapted for capturing and eating insects with chitinous exo- skeletons (Hotton, 1955); while blunt anterior con- ical teeth combined with enlarged posterior crush- ing teeth seems designed for feeding on hard- shelled beetles or even molluscs (Conant, 1955). In southwestern Saskatchewan, the northward ex- tension of the Hell Creek Formation of eastern Montana is called the Frenchman Formation, which was previously known as the “lower Ravenscrag" (see Russell and Landes, 1940; Furnival, 1946; Johnston and Fox, 1984). The formation is mainly composed of coarse-grained greenish-brown sand- stones, silty sandstones, and shales (Furnival, 1946), and rests on the Battle and Whitemud formations at an unconformable contact. The maximum thickness of the formation is about 100 m (L. S. Russell, 1964), representing the terrestrial sediments of the last 1.5 Myr of Late Cretaceous time (Lerbekmo, 1987). The earliest discovery of lizards from the Frenchman Formation occurred in 1921 by C. M. Sternberg (see Gilmore, 1928:18; GSC 2912, a cau- dal vertebra, originally cited as from the Lance For- mation, Rocky Creek, Saskatchewan). Later reports of vertebrate fossils from the formation were made by L. S. Russell (1964), Fox (1989), Tokaryk and James (1989), and Storer (1991). AND TERMINOLOGY Except for the acrodont dentition of adults in agam- ids and chamaeleonids, the marginal teeth of most lizards are replaced a number of times during their life history (see, e.g., Edmund, 1969). Thus, the teeth of most lizards are polyphyodont. However, some teiids show replacement suppression as adults (MacLean, 1974). It is generally considered that there are basically three types of tooth attachment in reptiles; namely, acrodont, in which the tooth is ankylosed to the summit of the jaw (also see discussion in Gauthier et al., 1988; Estes et al., 1988); pleurodont, in which the tooth is attached to the inner surface of the lat- eral wall of the jaw; and thecodont, in which the tooth root is sheathed in a deep socket in the jaw. However, as Romer (1956) has pointed out, the dis- tinctions between these classic types have tended to break down owing to increasing knowledge of fossil and living reptiles. Besides the three terms above, at least the following have been used in the litera- ture to describe intergrades between the basic types of tooth implantation: protothecodont, subthecodont (Camp, 1923; Edmund, 1969; Evans, 1990); suba- crodont (Estes, 1964), hyperacrodont (Camp, 1923; Malan, 1963), pseudoacrodont (Sulimski, 1978); subpleurodont (Estes, 1964), pleurothecodont (Gil- more, 1928); ankylothecodont (Chatterjee, 1974; Benton, 1983). These terms have certainly enriched 1996 GAO AND FOX— LATE CRETACEOUS LIZARDS I I our vocabularies in the recognition of intermediate conditions between the three classic types of tooth attachment, but they also cause confusion. Using different terms for the same mode of tooth attachment results in ambiguities, since the differ- ences between “protothecodont” and “subtheco- dont,” “subpleurodont” and “pleurothecodont,” and “subacrodont,” “pseudoacrodont,” and “hy- peracrodont” are indistinguishable. We prefer to use four of these terms as supplementary terms for the intergrades between the three classic types of tooth implantation. Specifically, protothecodont is appro- priate for the condition in some early (stem) rep- tiles, such as cotylosaurs; subthecodont for some diapsids, such as choristoderes; subpleurodont (sen- su Estes, 1964) for the condition between pleuro- dont and subacrodont, as is commonly seen in teiids; and subacrodont for certain aberrant teiids, such as Hciptosphenus (Estes, 1964) and Adamisau- rus (Sulimski, 1978). Another term, ankylotheco- dont, has been clearly defined for the specialized tooth implantation seen in rhynchosaurs, in which the marginal teeth are fused in the sockets as a com- bination of acrodonty and thecodonty. This type of tooth attachment is unknown in any lizards. Voro- byeva and Chugunova (1986) use “hyperpleurodon- tal attachment” with respect to the dentition of geckos, a condition termed as pleurodont by most other authors. Using the same term for different conditions of tooth attachment also engenders confusion. For ex- ample, Camp (1923) used the term “hyperacro- dont” for the pattern in Sphenodon, Agamidae, and Chamaeleonidae (a mode of implantation now com- monly termed acrodont), while Malan (1963) used the same term “hyperacrodont” to describe the con- dition seen in Howesia, a South African rhyncho- saur, which is more appropriately termed ankyloth- ecodont (Chatterjee, 1974; Benton, 1983). Another example is that of Camp (1923), who used the term “subthecodont” for Chcimops (more appropriately termed subpleurodont by Estes in 1964), while the term “subthecodont” has been used more recently as a substitute for “protothecodont” in most stem reptiles (see Edmund, 1969) and in some diapsids (see Evans, 1990). Terms Used in Text In this study, technical terms for osteological and dental features of lizards are adopted, for the most part, from previous authors, including Goodrich (1930), Estes (1964, 1983a), Romer (1956), Ed- mund (1969), and Estes et al. (1988). Since some terms are not commonly used in the description of fossil reptiles other than lizards, it is necessary to define them explicitly as follows: Inferior alveolar foramina: external mental fo- ramina in the lower jaw through which pass fibers of the inferior alveolar branch of cranial nerve V3 (Estes, 1964; Northcutt, 1979). Interior alveolar canal: a canal for blood vessels and nerves supplying the teeth and lateral surface of the jaws; includes the superior alveolar canal in the maxillary and the inferior alveolar canal in the dentary (Estes, 1964; Northcutt, 1979). Interior alveolar foramen: internal opening of the superior or inferior alveolar canal within the upper or lower jaws, including an anterior interior alveolar foramen and a posterior interior alveolar foramen (Estes, 1964). Internal Meckelian foramina: internal mandibular foramina (Goodrich, 1930), including anterior Meckelian foramen, anterior alveolar foramen, an- terior mylohyoid foramen, and the posterior man- dibular foramen (fossa) in the lower jaw, for the passage of mandibular nerve branches and blood vessels supplying the floor of the mouth and the tongue (Romer, 1956). Intramandibular septum: a partition posteriorly located in the Meckelian canal, functioning in sep- arating Meckel’s cartilage from the lateral nerves and blood vessels (Estes, 1964; Estes et al., 1988). Subacrodont: attachment of the marginal teeth mostly to the jaw summit, but also to the remaining lateral parapet, which is one-third or less of the tooth height (see Gao and Fox, 1991); this condition is usually associated with loss of the sulcus dentalis. Subdental shelf: a shelf-like medial extension of the dentary below the tooth row (Estes et al., 1988), often contributing to the sulcus dentalis and the spleniodentary articulation in the lower jaw. The shelf may carry a subdental ridge in some scinco- morphans. Subpleurodont: attachment of the marginal teeth to the medial surface of the lateral parapet of the jaw, which remains low, between one-half and one- third of tooth height (Estes, 1964). Superior alveolar foramina (= maxillary forami- na; du Bois, 1943): superior labial foramina on the maxillary, through which pass fibers of the superior alveolar branch of cranial nerve V2. Supradental shelf: a shelf-like medial extension of the maxillary above the tooth row, often bearing an anterior vomerine process and a posterior pala- tine process; when a sharp ridge is developed on 12 BULLETIN CARNEGIE MUSEUM OF NATURAL HISTORY NO. 33 the shelf, it is termed the supradental ridge (Estes, 1964). Sulcus dentalis: dental gutter separating the tooth bases from the supradental ridge of the maxillary and from the subdental ridge of the dentary (Estes, 1964). Terms Unsuitable for Dental Description in Lizards Certain terms have been used inappropriately in the literature on the dental morphology of lizards. The relevant publications are scattered; the terms listed below are only those that we have encoun- tered in an incomplete survey of the literature. “Triconodont” and “Symmetrodont” . — Some authors (Camp, 1923; Gilmore, 1928; Twente, 1952; Savage, 1963; Presch, 1970, 1974a, 19746; Estes, 1983a; Estes and Williams, 1984; Winkler et al., 1990) apply the terms “triconodont,” “trilo- bate” (Camp, 1923), “tri tubercular,” or “symme- trodont” when referring to lizard teeth having three cusps. Contrary to Estes and Williams (1984), who favor the term “triconodont,” we strongly suggest using “tricuspid” to replace this and other terms applied to teeth having three cusps. “Triconodont,” “symmetrodont,” and “tritubercular” are terms that have been used conventionally as the basis for tax- onomic names for certain groups of fossil mammals or as descriptive terms for mammalian dentitions, so that use of these terms for lizards is inappropriate and can render only confusion. “Trilobate” is so close in spelling to Trilobites (the synonym of Jan- assa, a chondrichthyan fish, see Romer, 1966:351) and “trilobites” (a group of arthropod invertebrates) that use of this term for lizards should be avoided in order to designate clearly the dentition of lizards. Similarly, “bicuspid” is considered to be a better term than “biconodont” for those teeth having two cusps. “ Molar ” and “Molar Hypertrophy" . — “Mo- lars” are the unreplaced postcanine teeth of mam- mals (Romer, 1966:191); therefore, this term is best restricted to mammals, and should not be used for other vertebrates, even for the ancestral stock of mammals (mammal-like reptiles). Smith (1946:7) stated that “in the Agamidae the tooth modification is carried still further, to such a point that molars, incisors, and canines may be distinguished; the par- allelism with the mammalian condition is empha- sized by the fact that the molars are not replaceable whereas the other teeth are.” This statement is in- correct in its use of mammalian terminology: some agamids indeed develop anterior, enlarged, canine- like teeth (but not the canine as defined in mam- mals, as teeth situated between the lateral incisors and the first premolar), and posterior cheek teeth that are not replaced because they are acrodont. There is no differentiation of incisors, canines, pre- molars, and molars in lizards, although some lizards are heterodont; for example, some teiids display a combination of canine-like anterior teeth and en- larged, crushing cheek teeth (see Estes and Wil- liams, 1984). These teeth are better termed canine- like (caniniform) and molar-like (molariform), rath- er than canines and molars. Similar use of terms as Smith’s is seen in Gilmore (1943a: “incisors,” “ca- nines,” and “last molar”), Harris (1963, 1964: “in- cisor of rainbow lizard”), and Estes and Williams (1984: “molar hypertrophy” and “molar expan- sion”). Applying these terms to the dentition of liz- ards is incorrect and should be avoided. “Dental Formula". — Some authors (Gilmore, 1926, 1928, 19426; Presch, 1970, 19746; Sulimski, 1975) inappropriately refer to a dental formula in systematic descriptions, or even diagnoses, of lizard taxa. A dental formula is the formula that expresses mammalian heterodonty; that is, the numbers of each category of teeth for a given taxon of mam- mals (Barghusen and Hopson, 1979; Carroll, 1988). Accordingly, a dental formula is used for mammals because, generally speaking, both tooth number and tooth form for a mammalian species are invariably stable, with a fixed number of tooth types (e.g., in- cisors, canines) and a fixed number of positions or loci for each type along the tooth row. The lizard dentition is not the same. Since the marginal teeth of lizards are not differentiated into incisors, ca- nines, premolars, and molars, use of a dental for- mula for lizards is inappropriate. Furthermore, the dentition of lizards is not static ontogenetically as in mammals: owing to rapid (and sometimes irregular) tooth replacement along with ontogenetic changes in the length of tooth row, the number of marginal teeth in lizards is individually variable in the same species. Ray (1965) and Greer (1991) pointed out that the number of tooth posi- tions in lizards is correlated largely with size, and size is mainly an ontogenetic or sex character. Ed- mund (1969:tables I— VIII ) demonstrated that tooth number is quite variable in different individuals of the same species (e.g., dentary teeth of Iguana igua- na vary from 17 to 29, depending on differences of size and age). MacLean (1974) provided compara- ble data for dentary tooth counts in gymnophthal- mids (Ptychoglossus, 15-35; Alopoglossus, 18-32; Placosoma, 17-18; Argalia, 18 — 19). Thus, tooth 1996 GAO AND FOX— LATE CRETACEOUS LIZARDS 13 count alone is nondiagnostic for lizard taxa in most Nevertheless, unless the ontogenetic change is well cases. However, in combination with tooth form and understood, using tooth count to diagnose lizard other characters, tooth count is broadly useful in taxa should normally be avoided, comparison of closely related species of lizards. SYSTEMATIC PALEONTOLOGY Class Reptilia Linnaeus, 1758 Subclass Diapsida Osborn, 1903 Infraclass Lepidosauromorpha Benton, 1983 Superorder Lepidosauria Haeckel, 1866 Order Squamata Oppel, 1811 Suborder Lacertilia Owen, 1842 The suborder “Lacertilia” is a paraphyletic group by cladistic definition as it includes lizards but not amphisbaenians and snakes, which are highly spe- cialized squamates sharing a common ancestor with lizards. Because of the cladistic paraphyly of the “Lacertilia,” several authors (Estes et al., 1988; Rieppel, 1988; Estes, 1991) recommend abandon- ment of this name. The term “Sauria” is used as a synonym of “Lacertilia” by some authors (e.g., Evans, 1988; Rieppel, 1988), but is defined by oth- ers (e.g., Gauthier et al., 1988; Laurin, 1991) in a very different way, as a higher taxon including both Lepidosauromorpha and Archosauromorpha. “Lac- ertilia” is retained in this paper as a taxonomic term for “lizards” as traditionally employed (see, e.g.. Camp, 1923; Romer, 1956; Robinson, 1967; Car- roll, 1988), since consensus on squamate classifi- cation has not been achieved. There have been sug- gestions (Estes et al., 1988; Rieppel, 1988; Estes, 1991) to classify Iguania and Scleroglossa (noni- guanian squamates including amphisbaenians and snakes) as equal taxa; however, the higher classifi- cation of squamates is beyond the scope of this pa- per and is not considered further here. Infraorder Iguania Cuvier, 1807 Family Iguanidae* Gray, 1827 The family Iguanidae* is a large group, contain- ing more than 600 species in some 60 genera (e.g.. Frost and Etheridge, 1989). Cladistically, the family is a metataxon indicated by an asterisk (Gauthier et al., 1988; Archibald, 1994: ambitaxon), which can only be defined provisionally as primitive iguanians that lack acrodontan synapomorphies (Estes et al., 1988) . To clarify the phylogenetic status of the group, a thorough study of the interrelationships of its former subfamilies or subgroups is needed badly (an attempt was made by Frost and Etheridge, 1989) . Extant iguanids are largely New World (including the West Indies and Galapagos) in distribution, with the exception of three genera that occur in Mada- gascar, and in Fiji and the Tonga Islands, respec- tively; they do not inhabit Eurasia, Africa, and Aus- tralia. Such a puzzling distribution has been known as “one of the paradoxes in classic biogeography, “ and “a biogeographic enigma” (Blanc, 1982). Both the classic view (e.g., Darlington, 1957) and the current South American origin hypothesis (see, e.g., Estes and Price, 1973; Cracraft, 1973; Blanc, 1982; Estes, 1983b) suggested a southern continent origin for the Iguanidae*; however, more recent discov- eries of Late Cretaceous fossils of the group from Spain (see Astibia et al., 1990), the Gobi Desert (Borsuk-Bialynicka and Alifanov, 1991; Gao and Hou, 1992, 1995), and North America (this paper) argue for a possible northern continent origin of the group. Genus Cnephasaurus, new genus Etymology. — knephas + sauros (Greek, masculine), meaning “twilight lizard”; this lizard may have been active at twilight hunting for grasshoppers, by analogy with its presumed relative, extant Crotaphytus. Type and Only Known Species. — Cnephasaurus locustivorus, new species. Range. — Upper Cretaceous, North America. Diagnosis. — As for the type and only known spe- cies. Cnephasaurus locustivorus, new species (Fig. 4, 5A-C) Etymology. — locusta + voro (Latin), meaning “grasshopper- eating”; the diet of this lizard may have consisted largely of grasshoppers, judging from its dentition. Holotype. — UALVP 29921, a right maxillary having 1 1 well-preserved teeth and broken bases of three others. Type Locality and Horizon. — MR-6 locality, Ver- digris Coulee, approximately 30 km east of the vil- lage of Milk River, southern Alberta; upper member of the Milk River Formation (Aquilan), Upper Cre- taceous. 14 BULLETIN CARNEGIE MUSEUM OF NATURAL HISTORY NO. 33 Fig. 4. — Cnephasaurus locustivorus, new genus and species, Milk River Formation, Alberta: UALVP 29921 (holotype), in- complete right maxillary, lateral (above) and medial (below) views. Scale = 3 mm. Referred Specimen. — UALVP 33386, topotypic left maxillary with four teeth and broken bases of two others. Known Distribution. — Known only from the type locality and horizon. Diagnosis. — An early Campanian small iguanid, closely resembling extant Crotaphytus Holbrook, 1843 in having a highly heterodont dentition, but differing from the latter as follows: premaxillary process much shorter; nasal process unretracted; palatal process on supradental shelf lacking; poste- rior interior alveolar foramen retaining position close to posterior end of tooth row. Description. — The holotype maxillary (UALVP 29921, Fig. 4, 5A, B) is small and delicately built, having a total length of about 6 mm. The specimen is anteriorly nearly complete, but broken dorsally and posteriorly; the nasal process (dorsal process or fa- cial process of authors) is missing and the tooth row is incom- plete. Anteriorly, both the lateral and the medial premaxillary processes are short and delicate, and not clearly divergent from each other. The anterior interior alveolar foramen is as small as the lateral superior alveolar foramina, and opens anteriorly above the base of the premaxillary processes. Medially, the holotype bears a slender but clearly defined su- pradental shelf, the medial border of which is straight in dorsal view, owing to the lack of a palatal process. No definite supra- dental gutter can be recognized. Anteriorly above the supradental shelf (at the level of the sixth tooth position) is a small pocket- like notch that probably received the lateral process of the pal- atine. Posteriorly, a relatively large posterior interior alveolar fo- ramen opens above the 12th tooth position. This condition is different from that in Crotaphytus, in which the foramen is shift- ed to a more anterior position, close behind the anterior interior alveolar foramen. UALVP 29921 has 14 tooth positions preserved, including 1 1 teeth and the broken bases of three others. With probably the two posteriormost teeth missing, a complete count is estimated at about 16. The maxillary tooth row is strongly heterodont: the first two teeth are broken, but were apparently slender and pro- cumbent; the following three are clearly cylindrical and unicus- pid; the sixth tooth is much reduced in size but still remains unicuspid and as cylindrical as the more anterior teeth. This tooth is followed by five even shorter and stouter middle teeth, the crowns of which are laterally compressed and medially concave. Behind these are three posterior teeth, which are strongly en- larged. Both the middle and posterior teeth are broadly based and tricuspid (or incipiently tricuspid), but the posterior ones are much larger and more robust, and separated from the smaller middle ones by an abrupt increase in size along the tooth row. Corresponding to the heterodonty of the maxillary teeth, the spacing between the tooth positions is also variable along the tooth row: the anterior unicuspid teeth are closely spaced and the middle and posterior tricuspid teeth are more widely spaced from one another. These teeth are pleurodont, with half or more than half of their height attached to the lateral parapet of the jaw. No replacement pits can be seen in this specimen. The lateral surface of the maxillary is smooth, with no osteo- dermal incrustation. Six superior alveolar foramina, all small and rounded, are spaced equally except for the anteriormost two, which are closer together. The posteriormost foramen, which is usually associated with the posterior interior alveolar foramen, is located close to the posterior end of the maxillary, in this respect differing from that in Crotaphytus, in which the foramen is lo- cated lateral to the nasal process and anterior to the midpoint of the tooth row. The nasal process of the maxillary is mostly bro- ken off, but the dorsal border posterior to the nasal process in- dicates that the nasal process probably descended sharply anterior to the midpoint of the tooth row; therefore, the nasal process is located anteriorly in front of the midpoint of the tooth row. An incomplete survey of this character state among extant lizards indicates that this condition is probably unique to iguanians with- in the Squamata. Discussion. — The new genus and species Cne- phasaurus locustivorus is based principally on a small maxillary, UALVP 29921, which is designat- ed as the holotype. In addition, UALVP 33386 (Fig. 5C), a topotypic maxillary fragment of similar size and having the same tooth form, is referrable to the same species. This new lizard from the upper mem- ber of the Milk River Formation is classified in the family Iguanidae* on the basis of its close resem- blance in dentition to extant Crotaphytus. Perhaps the most unusual characteristic of this lizard is the regional differentiation of the tooth form accompanied by a sharp change of tooth size. As described above, the tooth row of this species consists of six slender, high-crowned, unicuspid an- terior teeth, followed by five or so “pygmy” and incipiently tricuspid middle teeth, and then the en- 1996 GAO AND FOX— LATE CRETACEOUS LIZARDS 15 Fig. 5. — Cnephasaurus locustivorus, new genus and species. Milk River Formation, Alberta: A, B. UALVP 29921 (holotype), incomplete right maxillary, lateral and medial views; C, UALVP 33386, incomplete left maxillary, medial view. Iguanidae*, new genus and species (A), Milk River Formation, Alberta: D, E, UALVP 29908, left maxillary, lateral and medial views; F, UALVP 29909, left dentary, medial view. Iguanidae*, new genus and species (B), Frenchman Formation, Saskatchewan: G. SMNH P1927.916. left dentary, medial view; H, SMNH P1927.885, left dentary, medial view; I, SMNH P1927.998, right dentary, medial view; J, SMNH P1927.980, left dentary, medial view. Scale = 1 mm. 16 BULLETIN CARNEGIE MUSEUM OF NATURAL HISTORY NO. 33 larged posterior teeth. Both the holotype and re- ferred maxillary have the same pattern of size change with no replacement pits shown on these specimens, suggesting that this pattern of hetero- donty is taxonomically significant. An alternative explanation might be that both specimens are from young individuals that have not reached the stage for tooth replacement, and the small size of the mid- dle teeth reflects retarded development at an early postembryonic stage. Nevertheless, since the two specimens show the same condition, the first alter- native seems preferable. A similar condition is seen in extant Crotaphytus, the dentition of which, ac- cording to Hotton (1955), is adapted for feeding on highly active grasshoppers having an integument of intermediate weight. Although the homology of this similarity cannot be tested at present, Crotaphytus is indeed the only group that shows this feature. This similarity may indicate a close affinity of the new Milk River iguanid to Crotaphytinae Smith and Brodie, 1982. However, this possibility cannot be fully verified until more extensive evidence is found. Recognition of Cnephasaurus as an early iguanid is both taxonomically and biogeographically signif- icant. Taxonomically, this new form documents an early fossil record of the family Iguanidae*, as the Milk River Formation is of early Campanian age (L. S. Russell, 1975; Fox, 1976; Lillegraven and McKenna, 1986; Lillegraven, 1991), some 18 Myr (Harland et al., 1990) earlier than the Maastrichtian Pristiguana Estes and Price, 1973 from Brazil. Bio- geographically, it marks a rare discovery of Meso- zoic iguanids from North America, which was part of the northern landmass during the Cretaceous. It has been suggested that the Iguanidae* originated in South America (e.g., Estes and Price, 1973; Es- tes, 1983 b\ Estes and Baez, 1985) and did not reach North America until middle Paleocene or early Eo- cene time (Sullivan, 1982; Estes and Baez, 1985). The new discovery of early Campanian iguanids from North America, coupled with recent discov- eries of a slightly younger fossil record of the same group from Spain (Astibia et al., 1990; S. E. Evans, personal communication, 1995) and the Gobi Desert (Borsuk-Bialynicka and Alifanov, 1991; Gao and Hou, 1992, 1995), indicates the presence of iguan- ids in North America, Europe, and East Asia well before they occurred in South America in the latest Cretaceous or Paleocene (Estes and Price, 1973; de Muizon et al., 1983; Rage, 1991). The early geo- logical age and primitive morphology of these ig- uanids challenge the notion of a South American origin, and argue for a possible northern continental origin of the family. Genus and Species New (A) (Fig. 5D-F) Specimens. — UALVP 29908, left maxillary fragment with three well-preserved teeth and the base of another; UALVP 29909, left dentary fragment with two teeth and the space for another. Locality and Horizon. — MR-6 locality, Verdigris Coulee, approximately 30 km east of the village of Milk River, southern Alberta; upper member of the Milk River Formation (Aquilan), Upper Cretaceous. Description. — UALVP 29908 (Fig. 5D, E) is a left maxillary fragment that is broken both anteriorly and posteriorly. The facial process is largely missing, but the preserved posteroventral part is vertically straight, indicating that the process in this species is high. The posterodorsal edge of the specimen is a natural border (not a broken surface), representing the descent of the maxillary posterior to the facial process. This specimen also shows that the lateral surface of the maxillary is smooth, bearing relatively large superior alveolar foramina (one shown on this specimen). Medially, UALVP 29908 shows a thin and slender supradental shelf, which has no dental gutter. Dorsal to the shelf is the narrow opening of the posterior interior alveolar foramen. This foramen usually opens at the posterior part of the tooth row, and the inclined dorsal edge above it clearly indicates that the specimen represents the posterior part of the maxillary. UALVP 29908 bears three well-preserved teeth, which are high crowned and pleurodont. with half of their height attached to the lateral par- apet of the jaw. The teeth are basically cylindrical, but are con- cave medially with the crowns slightly bent lingually. The crowns are tricuspid, having a strongly prominent central cusp and much weaker but clearly defined anterior and posterior side cusps. The crowns are slightly twisted, so that the three cusps are aligned at an angle of approximately 45 degrees to the tooth row, and the central cusp is curved posteromedially. The anterior accessory cusp is slightly stronger than the posterior one. A dentary fragment (UALVP 29909, Fig. 5F) from the same locality has two teeth preserved, one complete and the other with the crown broken off. The central and anterior cusps are similar to those on the maxillary (UALVP 29908). but the tooth lacks a clearly defined posterior cusp, a condition that can be termed either bicuspid or incipiently tricuspid. It seems likely that the differences in the development of these cusps are relevant to tooth position, with the posterior teeth having strongly tricuspid crowns. However, this interpretation cannot be verified until more nearly complete specimens of this species are recovered. Because of the fragmentary nature of the specimens, the possi- bility of the second specimen representing another species cannot be ruled out. Discussion. — The new genus and species (A) is recognized principally on UALVP 29908, while the other specimen (UALVP 29909) is referred to the same species with some uncertainty. Although frag- mentary, UALVP 29908 clearly shows that at least the middle and posterior teeth of this lizard are def- initely tricuspid, with the central cusp being much 1996 GAO AND FOX— LATE CRETACEOUS LIZARDS 17 higher than the lateral ones, and the crowns twisted at a 45-degree angle to the straight line of the tooth row. This particular cusp pattern, in association with other characters such as crown height and mode of tooth attachment, provides useful information con- cerning its taxonomic placement at a familial level. Tricuspid teeth are commonly seen in Teiidae and Iguanidae*, and also occur in some genera of Xan- tusiidae (Savage, 1963). In tricuspid teeth of teiids, the three cusps are characteristically more or less equal in height, a pattern that is often closely as- sociated with other characters, such as subpleuro- donty and a well-developed dental shelf and sulcus dentalis (see Estes, 1964, 1983a; Presch, 1970; Gao and Fox, 1991). All of these features are different from the condition seen in UALVP 29908, exclud- ing the possibility of its affinity to the family Teiidae, which is highly diverse in the North Amer- ican Upper Cretaceous. Two genera of Xantusiidae ( Klauberina and Lepidophymd) are known to have tricuspid teeth (Savage, 1963), but the tricuspid condition in these genera is obviously different from that of iguanids: in xantusiids, the two side cusps are more lingually located than the central cusp (“anterior and posterior cusps on inner surface of teeth”; Savage, 1963:30). UALVP 29908 has an interesting combination of high-crowned, tricuspid, and highly pleurodont teeth, while lacking a dental gutter. This combina- tion of character states indicates that the specimen is referrable to the family Iguanidae*. The teeth of iguanids are generally high crowned and pleuro- dont, and with more or less flared crowns that can be either tricuspid or polycuspid (having four or more cusps), with the exception of having simply peg-like teeth in some highly derived groups. The tricuspid condition in the Iguanidae* is commonly characterized by having “a large apical cusp, small- er anterior and posterior cusps, and more or less parallel sides” (Etheridge and de Queiroz, 1988: 297). Recent cladistic analysis of iguanid phylogeny indicates that tricuspid tooth form is a primitive condition in the family Iguanidae* (Etheridge and de Queiroz, 1988). Thus, UALVP 29908 may rep- resent another primitive early iguanid from the Milk River Formation. However, the specimen is too fragmentary to show the nature of the premaxillary and facial processes, making a comparison with Re- cent specimens impossible. Genus and Species New (B) (Fig. 5G-J) Specimens. — SMNH PI 927.9 16, dentary fragment bearing five teeth and the base of another; SMNH P 1927.885, left dentary fragment bearing four teeth and the bases of two others; SMNH PI 927.998, right dentary fragment with three teeth and the base of another; SMNH PI 927.980, deniary fragment bearing four well-preserved teeth. Locality and Horizon. — Gryde locality, in Sec. 19, Tp. 4, R 18, W 3, Frenchman Valley, south- western Saskatchewan; Upper Cretaceous French- man Formation (Lancian). Description. — SMNH PI 927.9 16 (Fig. 5G) is a fragmentary dentary, which is obviously from a small individual, as shown by the size of the specimen. Although fragmentary, the specimen shows that the subdental shelf is poorly defined and has no dental gutter (sulcus dentalis). The lower portion of the jaw fragment is missing, providing no information about the splenial bone, spleniodentary articulation, and the Meckelian canal. The den- tary fragment bears five teeth and the base of another. The five teeth are complete or nearly complete; only the last on the spec- imen is badly broken. These teeth are high crowned and pleu- rodont, with the lower half of the teeth attached to the lateral parapet of the jaw. The tooth crowns are flared, slightly recurved, and tricuspid, having a highly prominent main cusp and two much lower side cusps. The side cusps are clearly separated from the central main cusp, and each has a vertically running ridge on the medial surface of the crown. The anterior side cusp is slightly stronger than the posterior one on each tooth. The tooth shafts are slenderly cylindrical and are closely packed along the tooth row. The tooth bases are slightly thicker than the shafts and show no sign of tooth replacement on the specimen. SMNH P1927.885 (Fig. 5H), a left dentary fragment, has four teeth and spaces for two others preserved. The teeth are high crowned, pleurodont, and the cylindrical tooth shafts are slightly curved. The tooth crowns are not well preserved, but apparently tricuspid. The subdental shelf is slightly stronger than on the other three specimens, and it seems to bear a poorly defined sulcus dentalis. SMNH PI 927.998 (Fig. 51) is the fragment of a small right dentary, which bears three teeth (one complete and the other two with the crowns broken off) and the space for another. The tooth form of this specimen is similar to that of SMNH PI 927.9 16; that is, slender, cylindrical, and tricuspid, and the subdental shelf is weak with no sulcus dentalis. Each of the two teeth with a broken crown has a resorption pit at the medial side of the tooth base. The Meckelian canal is extremely restricted, opening nar- rowly ventromedially, indicating a greatly reduced splenial. The lateral surface of the jaw fragment is smooth, relatively flat, and bears two mental foramina that are widely separated from each other by four tooth positions. SMNH PI 927.980 (Fig. 5 J) is a left dentary fragment bearing four well-preserved teeth. Like SMNH PI 927.9 16, this fragment shows a weakly developed subdental shelf, and high-crowned and pleurodont teeth closely spaced from one another along the tooth row. The tooth shafts are slender, cylindrical, and the crowns are flared and slightly recurved. However, the tooth crowns of this specimen are clearly bicuspid, differing from the tricuspid condition in SMNH PI 927.9 16. Owing to the incom- pleteness of the two specimens, we are unable to determine whether the difference of the crown pattern is taxonomically sig- nificant, or simply that the two specimens are from a different portion of the dentary tooth row. Therefore, its referral with the other three specimens to the same unnamed taxon is tentative. Discussion. — Whereas the four specimens de- 18 BULLETIN CARNEGIE MUSEUM OF NATURAL HISTORY NO. 33 scribed above are too fragmentary to show the gen- eral jaw configuration, the tooth form of these spec- imens indicates the presence of at least one new genus and species previously unknown from Lan- cian horizons. However, the new taxon is unnamed in this study because the available specimens are inadequate to show diagnostic features other than tooth form; jaw structure, such as the morphology of the splenial, is important in determining taxo- nomic position. Although this new genus and new species re- mains unnamed (thus no holotype is designated), SMNH PI 927.9 16 is the specimen on which the new taxon is recognized. It clearly shows tall teeth having flared and tricuspid crowns, a poorly defined subdental shelf, and no sulcus dentalis. These are derived character states that in combination strongly indicate the affinity of this lizard to the family Igua- nidae*. Within the Iguanidae*, the closest resem- blance in terms of tooth morphology of this Lancian form is to Ctenosaura (e.g., C. similis Gray, 1831, see Ray, 1 965 : fig. 5), which has a geographical dis- tribution in Mexico and Central America (Estes, 1983a). According to Hotton (1955), C. similis is chiefly herbivorous, but its diet also includes insects at times. SMNH PI 927.998 is an important specimen as it shows strong evidence of iguanid affinities: the total loss of the sulcus dentalis, greatly reduced splenial as indicated by the extremely narrow Meckelian ca- nal, highly pleurodont dentary teeth with tricuspid crowns, and direct tooth replacement as suggested by the resorption pits at the medial side of the tooth bases. However, we are unable to determine if this specimen can be grouped with PI 927.9 16 and PI 927.885 in the same species as the teeth of this jaw fragment are less high crowned and less flared than those of the two specimens mentioned above. We tentatively put them together on the basis of their tricuspid crown pattern, regarding the differ- ences in tooth height as related to regional variation along the tooth row. SMNH P1927.980 is similar to P1927.916 in having high-crowned, cylindrical teeth with flared crowns, but differs from the latter in having a bi- cuspid rather than tricuspid crown pattern. It is closely similar to PI927.916 in every aspect but lacks a posterior side cusp on the crowns. At this stage, we are unable to determine whether this dif- ference is taxonomically significant, or simply re- flects differences in cusp pattern along the different parts of the tooth row. From the above discussion, the four specimens from the Upper Cretaceous Frenchman Formation represent at least one (and maybe more than one) new taxon that is referrable to the Iguanidae* based on tooth form and implantation. They are referred to the same unnamed taxon from their general sim- ilarity in tooth form and origin from the same lo- cality. SMNH P1927.916 and P1927.885 are clearly the same kind, whereas P1927.980 and P1927.998 seem to resemble less clearly the former two spec- imens in terms of crown pattern (number of cusps), but cannot be determined to be a separate taxon at the present time. Infraorder Scincomorpha Camp, 1923 Family Teiidae Gray, 1827 (sensu Estes, 1983a; Presch, 1983) Boulenger (1885) subdivided the Teiidae (sensu lato) into four major groups (I— IV). His group I was later informally known as “macroteiids” and group II as “microteiids” (Ruibal, 1952). MacLean (1974) recognized the two groups as separate subfamilies (Teiinae and Gymnophthalminae). More recently, Estes (1983a) and Presch (1983) independently ranked the two groups as families. “Teiidae” (sensu stricto) as used in this paper refers to Boulenger’s group I (macroteiids) in the sense of Estes’ (1983a) and Presch’s (1983) definition. In the most recent study of the phylogenetic relationships of the lizard families, Estes et al. (1988) listed 14 synapomor- phies for the family Teiidae, among which the fol- lowing are unambiguous and reliable character states for fossil teiid studies: replacement teeth de- veloping in deep subcircular cavities at tooth bases (Romer, 1956); cementum deposited extensively on tooth bases (Presch, I974Z?); vomer elongated, ap- proaching pterygoid (Romer, 1956); splenial hyper- trophied, extending anteriorly to the symphysis leaving only a small symphyseal foramen (Mac- Lean, 1974). So far, the earliest known fossil record of this family is of Albian age, from the Coman- chean of central Texas (Winkler et al., 1990). Genus Chamops Marsh, 1892 Type Species. — Chamops segnis Marsh, 1892. Range. — Upper Cretaceous, North America. Diagnosis. — As for the type species (see below). Chamops segnis Marsh, 1892 (emend. Gilmore, 1928; Estes, 1964, 1983a) (Fig. 6, 7A-E) Holotype. — YUM 1036, a fragmentary left den- tary with five well-preserved teeth, from the Lance Formation, Niobrara County, Wyoming. The holo- 1996 GAO AND FOX— LATE CRETACEOUS LIZARDS 19 type was originally identified by Marsh ( 1892) as a right maxillary; the identification was erroneous and was later corrected by Estes (1964). Referred Specimens. — Frenchman Formation, Gryde locality: SMNH P1927. 824-825, P 1 927.83 1 , P1927.876, P1927.882, P1927.899, PI 927.966, P1927.969, P1927.993, P2004.86 (total: ten), incomplete maxillaries; P1927.827, P1927.874, P1927.883, P1927.890, P1927.920, P1927.991 (total: six), fragmentary den- taries. Wounded Knee locality: UALVP 29731, incomplete max- illary; UALVP 29728, 29811-29812, fragmentary dentaries. Scollard Formation, KUA-1 locality: UALVP 29841, fragmen- tary maxillary; KUA-3 locality: UALVP 29842-29843, incom- plete dentaries. Known Distribution. — Upper Cretaceous Lance Formation, eastern Wyoming; Hell Creek Forma- tion, eastern Montana; Scollard Formation, central Alberta; and Frenchman Formation, southwestern Saskatchewan (all Lancian). Diagnosis ( Revised from Estes, 1983a). — North American Late Cretaceous teiid, differing from oth- er teiids in the following combination of character states: marginal teeth not crowded along tooth row, oval to subcircular in cross dimensions, except at base where moderately transverse, increasingly swollen posteriorly and becoming distinctly barrel- like in shape; dentary and maxillary tooth row weakly heterodont, with relatively tall anterior teeth unicuspid to weakly bicuspid or tricuspid, some- what recurved towards tip, becoming more strongly tricuspid and erect posteriorly; accessory cusps set off from main cusp by prominent groove lingually, short groove labially. Description. — More than 20 maxillaries and dentaries collect- ed in the study area can be confidently identified as Chumops segnis on the basis of tooth form and jaw structure. The follow- ing description is based on several relatively well-preserved specimens in the collections available for this study. In addition to the specimens enumerated above, several maxillaries and den- taries of the same species, but from the Lance and Hell Creek formations, are also included in the study because they show individual variation in certain features of the tooth row. These include: UALVP 29727, 29729, 29813 (maxillaries), and 29814 (dentary) from the BTB locality of the Lance Formation, Wyo- ming (the BTB locality is assumed to be near the type locality of C. segnis [“Lance Creek” in Gilmore (1928)], and hence, the UALVP specimens are considered topotypic); and UALVP 29724-29726, 29730, 29737-29738 (maxillaries), 29721, 29723, and 29815 (dentaries) from the BCA locality. Hell Creek For- mation, Montana. Maxillary. SMNH P2004.86 (Fig. 6, 7A) is the left maxillary of a young individual, judging from its small size (about half size of UCMP 46033, see Estes, 1964:fig. 47). The maxillary has part of its dorsal process broken away; otherwise it is complete. It has a short premaxillary process, behind which the nasal pro- cess rises abruptly as a vertical lateral wall, as in UCMP 46033, a maxillary from the Lance Formation (Estes, 1964:fig. 47). Me- dially, the maxillary bears a well-developed supradental shelf. i i i Fig. 6. — Chamops segnis, Frenchman Formation, Saskatchewan: SMNH P2004.86, left maxillary, lateral (above) and medial (be- low) views. Scale = 3 mm. the medial edge of which is nearly straight, and lacks a palatine process. The posterior interior alveolar foramen opens above the shelf at a position between the fifth and the sixth posteriormost teeth. The full maxillary tooth row is preserved on this specimen and contains 16 short-crowned teeth with swollen shafts. The first four anterior teeth are conical (or weakly bicuspid) and are slight- ly recurved; the next three are weakly tricuspid, while the rest are broadly tricuspid. The crown pattern of the maxillary teeth is best shown on SMNH P1927.824 (Fig. 7B), an incomplete left maxillary of the same size as SMNH P2004.86 described above. The specimen has 14 tooth positions (12 teeth and bases of two others), with probably the two anteriormost teeth broken away. In SMNH PI 927.824, the tricuspid condition of the maxillary teeth starts from the fifth anterior tooth, and the middle and posterior teeth are short, broadly tricuspid, and slightly concave medially. The third posteriormost tooth on this specimen is medially offset from the tooth row, and the crown is bent laterally; this seems to be the result of pathologic development of the tooth. Tooth implan- tation is subpleurodont. A shallow sulcus separates the tooth row from the supradental ridge. As in other specimens of the species, the maxillary teeth are moderately spaced from one another, and are slightly concave on the lingual surface. The teeth have a basal deposit of cementum, but it is much less heavy than on larger specimens in the collection. This feature, plus its small size, im- plies that the maxillary is from a young individual. In comparing smaller and larger maxillaries of C. segnis in our collection, the accessory cusps are stronger and better defined in the larger and presumably older individuals, an observation that contrasts with that of Denton and O'Neill (1995) for this species. Anteriorly, the internal surface of the maxillary of C. segnis shows two depressions, one smaller, more anterior and ventral in position, the second and larger one posterodorsal to the first. These depressions probably were associated with soft tissues of 20 BULLETIN CARNEGIE MUSEUM OF NATURAL HISTORY NO. 33 pig 7 Chamops segnis, Frenchman Formation, Saskatchewan: A, SMNH P2004.86, left maxillary, medial view; B. SMNH PI 927.824, left maxillary, medial view; C, SMNH PI 927.874, left dentary, medial view; Scollard Formation, Alberta: D, UALVP 29841, left maxillary, lateral view; E, UALVP 29842, right dentary, medial view. Chamops sp.. Milk River Formation, Alberta: F, UALVP 29816, left maxillary, medial view; G, UALVP 29826, left maxillary, medial view; H, UALVP 29817, right maxillary, medial view; I, UALVP 29819, right dentary, medial view; J, UALVP 29820, left maxillary, medial view. Scale = 1.5 mm. 1996 GAO AND FOX— LATE CRETACEOUS LIZARDS 21 Jacobson’s organ (Hechl, 1951). The configuration of these de- pressions in the maxillaries from the Frenchman Formation matches that in the topotypic maxillaries from Wyoming in the UALVP collection, corroborating the identification of these spec- imens as based on their dental morphology. Dentary. In this study, 1 1 dentaries from the Frenchman and Scollard formations are referred to Chcunops segnis on the basis of their tooth form and jaw structure; the best preserved of these is described as follows: SMNH P1927.874 (Fig. 7C), from the Gryde locality of the Frenchman Formation, is an incomplete left dentary bearing 1 1 tooth positions (eight teeth and the bases of three others). The dentary is from a young individual, as ev- idenced by its small size. Although most of the lateral wall of the Meckelian canal is broken off, the tooth form and the parts of the dentary that remain provide reliable characters for iden- tification. Medially, the dentary bears a strong subdental shelf, a well-developed sulcus dentalis, and a weak symphyseal surface. The first five teeth are slender, conical, and slightly procumbent; and the remainder of the teeth strongly increase in size posteri- orly. Most of these teeth are poorly preserved, but the tricuspid crown pattern is shown at the seventh and the ninth tooth posi- tions. In spite of the small size of the specimen, tooth replace- ment is shown at the sixth and the tenth positions, where the remaining bases of the replaced teeth still can be seen. In dorsal view, the tooth row is straight, lacking the curvature seen in Leptochamops Estes, 1964, another teiid lizard from the same horizon. Comparison of the specimens from the study area with those from the Lance and Hell Creek formations allows a better un- derstanding of the jaw construction and tooth morphology of this teiid. Generally speaking, the dentary of Chamops segnis is short and robust, but it has a weak symphysis; it differs from the den- tary of Leptochamops in having a much wider Meckelian sulcus, much deeper boat-shaped ventral curvature, straight subdental shelf, and a posterior interior alveolar foramen that opens under the 12th tooth position. In addition, the dentary external surface of Chamops is smoothly convex dorsoventrally, lacking the labial concavity seen in Leptochamops denticulatus (Gilmore, 1928). The tooth row is weakly heterodont; the anterior teeth (four or five) are conical or weakly bicuspid, as shown on the two maxillaries described above and UALVP 29841 (Fig. 7D) from the Scollard Formation; these are followed by a few weakly tri- cuspid teeth, and then by widely tricuspid posterior teeth with short and swollen shafts. Estes (1964) stated that the incipiently tricuspid anterior teeth are separated from the posterior ones by a “step" (see discussion below), and that the dentary tooth count of Chamops segnis is 18 on the basis of UCMP 49881, which shows the complete tooth row. Among the specimens used in this study, no dentary contains the full tooth row; the highest number of preserved tooth positions is 11, as shown on SMNH PI 927.874 from the Frenchman Formation, Saskatchewan, and the posterior tricuspid teeth are best shown on UALVP 29842 (Fig. 7E) from the Scollard Formation, central Alberta. The UALVP collections include two specimens of interest from the Oldman Formation that are best considered here. UALVP 3391 1 is a left maxillary fragment containing four teeth, UALVP 33912 is a right dentary fragment with three complete teeth and a fourth that is broken; both are from the Irvine locality, southeastern Alberta. The teeth on both specimens are tricuspid, with swollen, barrel-like shafts, unique features of the dentition of Lancian Chamops segnis. Pending the collection of better pre- served material, we identify these specimens as Chamops sp., cf. C. segnis ; they are readily distinguished from all other teiids known from Judithian horizons. Denton and O’Neill (1995) cite longitudinal (= vertical) stri- ations of the crown enamel as a synapomorphy uniting Chamops segnis and their new teiid Prototeius stage ri. The occurrence of striae on the crown of C. segnis is, however, individually variable and, hence, not a characteristic feature of the species, irrespective of its occurrence in P. stageri. Similar striations are widespread in scincomorphan lizards and may be primitive for teiids in any case. Discussion. — Chamops segnis was previously known from the Lance Formation, Wyoming (Marsh, 1892; Gilmore, 1928; Estes, 1964, 1983a), Hell Creek Formation, Montana (Estes et al., 1969), Frenchman Formation, Saskatchewan (L. S. Rus- sell, 1964; Fox, 1989), and now from the Scollard Formation, Alberta (this paper). There have been other reports of Chamops segnis from geologically older horizons (for citations see Estes, 1983a:95), but a careful review of the fossil evidence indicates that these reports are either unconvincing or based on misidentified specimens (see discussion below). Problems bearing on the taxonomic history, diag- nosis, synonyms, and misidentified specimens for this Cretaceous lizard are treated in the following discussion. Taxonomic History and Relationships of Chamops segnis. Chamops segnis was originally named by O. C. Marsh (1892), based on several specimens collected by J. B. Hatcher in 1889 from the Lance Formation, Wyoming. Marsh (1892) des- ignated a left dentary (misidentified as a maxillary) as the type specimen, but provided no specimen number, no description, and no diagnosis for the new genus and species. It was not until 1928, when Gilmore gave a detailed description of Marsh’s type specimen and pointed out its diagnostic characters, that Chamops segnis became a fully established tax- onomic name. Gilmore (1928) followed Marsh’s in- correct identification and described the type speci- men (YUM 1036) as a maxillary. Estes (1964) was the first to note that the type specimen of Chamops segnis is in fact a dentary. Estes (1964:pl. 2) pro- vided a clear photograph of the type specimen, which shows five (rather than ten) complete dentary teeth and the tooth bases of five others. This con- dition of the specimen is consistent with Gilmore’s (1928:24) description, but different from Marsh’s (1892:fig. 2, 3) original figures that show nine com- plete teeth and the base of another. Therefore, it may be that between 1892 and 1928 the type spec- imen was damaged, with four teeth broken and lost. Marsh (1892) offered no opinion regarding the relationships of Chamops segnis. Subsequently, the 22 BULLETIN CARNEGIE MUSEUM OF NATURAL HISTORY NO. 33 species was referred to the Teiidae by Boulenger (1892) and Nopcsa (1908) without comment, and was placed in the Iguanidae* by Hay (1902), Camp (1923), and Gilmore (1928). Estes (1964:105) noted that " Chamops differs in many ways from known iguanids, principally in the broadly open Meckelian fossa, the concomitantly large splenial, and less pleurodont implantation of teeth.” Recognizing the significant similarities to the extant teiids Crocodi- lurus Spix, 1825 and Tupinambis Daudin, 1802, Es- tes (1964) placed Chamops in the Teiidae. More recently, Estes (1969a, 1983a) further recognized that the closest relationship of Chamops is with Cal- lopistes Gravenhorst, 1838, from similarity of max- illary, nasal, quadrate, and parietal bones. The max- illary that Estes used for comparison (UCMP 46033) is restored dorsally from UCMP 46094, and “the parietals and quadrates are referred to Cham- ops on the basis of both size and frequency, as well as on their generally teiid appearance” (Estes, 1969a:5). New specimens used in this study add no further information as to the reliability of the as- sociation of these disarticulated skull elements; however, the relationship of Chamops segnis to the Teiidae can be determined without reference to skull elements, because teiid synapomorphies are clearly shown on the dentary specimens. These synapo- morphies include: (1) large splenial, as indicated by the widely open Meckelian canal; (2) extensive de- posit of cementum on the tooth bases; and (3) re- placement teeth developing in deep subcircular ba- sal cavities (see Estes et al., 1988, for citations). In addition, other characters, such as the subpleurodont tooth attachment, well-developed subdental shelf, and sulcus dentalis, are consistent with the family Teiidae, and are not found in nonteiids having sim- ilar tooth form (e.g., extant Crotaphytus collaris). As Estes (1964, 1969a) noted, a character state that Chamops shares with Callopistes, but not with Crocodilurus and Tupinambis, is the relatively short face. In Callopistes, the upper jaw is shorter than one-half of the total length of skull, while it is half, or often longer than half, the length of the skull in other teiids (see Presch, 1970). The short face is also consistent with a higher nasal process of the maxillary in extant teiids (personal observation). Facial elongation is prominent in advanced teiids, such as Cnemidophorus Wagler, 1 830 and Ameiva Meyer, 1795 (MacLean, 1974: “long-snouted spe- cies”). Thus, the deep maxillary and short face are probably primitive conditions in teiids (Estes, 1964). Diagnostic Characters of Chamops segnis. As mentioned above, Marsh (1892) provided no diag- nosis for Chamops segnis when he named the spe- cies. Gilmore (1928:24) was the first to diagnose the lizard as having “teeth pleurothecodont [sub- pleurodont]; internal and external surfaces of teeth grooved longitudinally; tricuspid; subequal in size; 23 teeth in complete dentary series.” However, his count of 23 dentary teeth was based on YUM 1062, a specimen that pertains to another genus and spe- cies ( Leptochamops denticulatus\ see Estes, 1964). Actually, the full dentary tooth count of Chamops segnis is about 18, based on the evidence from UCMP 49881 and UCMP 49871 (see Estes, 1964: fig. 45). Based on additional collections from the Lance Formation, Estes (1964:102) revised the diagnosis for the species as “a teiid lizard closely related to the living Crocodilurus lacertinus and Tupinambis nigropunctatus, differing principally from these in the less pronounced heterodonty, and less swollen tooth crowns; if the parietal is correctly referred, this species has a large parietal foramen, unlike all Recent teiids.” More recently, Estes (1983a:95) provided the lat- est version of the revised diagnosis for Chamops segnis : “teiid lizards apparently with a parietal fo- ramen; tooth row very weakly heterodont, anterior teeth bicuspid or weakly triconodont, separated by an area of reduced tooth size from posterior tricon- odont teeth that have swollen bases and wrinkled crown surfaces; maxilla deep dorsoventrally; splen- ial very extensive, almost completely filling the widely open Meckelian sulcus.” About this diag- nosis, we offer the following comments. First, the nature of the “area of reduced tooth size” (Estes, 1964: “step”) is still unclear; however, it is individually variable within Chamops segnis, as the change in tooth size is not shown on the holotype (YUM 1036, see Marsh, 1892:fig. 2, 3; Estes, 1964:pl. 2) nor on various other specimens. In the two best-preserved maxillaries (young indi- viduals) from the Frenchman Formation, the “step” is present on SMNH PI 927.824 but absent on the other (SMNH P2004.86). This variation is also shown in those large maxillaries from the Lance and Hell Creek formations collected by UALVP field parties: the “step” is present on UALVP 29729 (from BTB), but absent on UALVP 29727 (BTB) and 29738 (BCA). Furthermore, this “step” in the tooth row is also developed in extant Crocodilurus and Tupinambis (see Estes, 1964:fig. 48), and at least some of the fossil Leptochamops (see Estes, I964:fig. 50). Therefore, the “area of reduced tooth 1996 GAO AND FOX— LATE CRETACEOUS LIZARDS 23 size” or the “step” is not unique for Chamops seg- nis, and, thus, cannot be regarded as a diagnostic character. It is difficult to evaluate this character without a complete survey of the dentition of teiids and related groups; however, it seems to us that the “step” is functionally significant, as it may corre- spond to the biting mechanism of the jaws; that is, it may be the area frequently used to hold prey (in- sects). Estes et al. (1988:162) regarded “a step in the maxillary tooth row to be a synapomorphy of Lacertiformes.” Second, the referral of several parietals to Cham- ops segnis, including UCMP 57147 (see Estes, 1964:fig. 49) is uncertain, because it is mainly “on the basis of size and frequency of occurrence” (Es- tes, 1964:104). There are other teiid species known from the same formation, and the specimens are dis- articulated, being collected mostly by screen wash- ing. Moreover, the size difference between Cham- ops segnis and other contemporaneous teiids is not great enough to separate the fragmentary skull el- ements one from another taxonomically. Therefore, it is difficult to recognize the natural association of the disarticulated parietals with the jaw material of Chamops segnis, unless other evidence, such as a more or less articulated specimen, is discovered. Furthermore, even if a complete skull of Chamops could be found and proved to have a parietal fora- men, the presence of the foramen as a diagnostic character still is not warranted: ontogenetic evi- dence shows that loss of the foramen in most living species is a derived condition and must have been achieved quite recently in evolution (Presch, 1970). Therefore, the presence of a parietal foramen in teiids is a primitive condition, and, thus, is not a diagnostic feature for fossil species of teiids. Synonyms of Chamops segnis. Estes (1964, 1983a) placed Iguanavus teres Marsh, 1892, Lan- ceosaurus hatcheri Gilmore, 1928, Lanceosaurus compressus Gilmore, 1928, and Alethesaurus quad- ratic Gilmore, 1928 in synonymy with Chamops segnis', we have compared the published data on these taxa (description, figures, and discussion), and offer the following comments: Marsh (1892:451) founded Iguanavus teres on several unnumbered dorsal vertebrae (his “type specimens”) from the Lance Formation, Wyoming. Gilmore (1928:17-18) designated and described the type specimen (YUM 530), and apparently con- firmed the taxonomic status of the genus Iguanavus. Even later, Estes (1964, 1983a) synonymized I. ter- es with Chamops segnis on the basis of size and frequency of occurrence of the specimens, although no vertebrae of Chamops have ever been described. Marsh’s naming of Iguanavus teres based on only vertebrae is unfortunate, because vertebrae in most cases provide little information for species identi- fication. In other words, the nominal taxon Iguana- vus teres is based on specimens that are inadequate, probably nondiagnostic at a lower taxon level; the type specimen (YUM 530) was never figured, and the species was never diagnosed. Thus, Iguanavus teres is a “nomen dubium” of ICZN (1985) or a “nomen vanum” of Simpson (1948) and Mones (1989); probably the type species Iguanavus exilis (see Marsh, 1872; Gilmore, 1928) is as well (Estes, 1983a). A “nomen dubium” or a “nomen vanum” is an invalid name that has no standing in zoology (ICZN, 1985; Mones, 1989), although each stands as a nominal species in nomenclature (see Simpson, 1948; Mones, 1989). Gilmore’s (1928) Lanceosaurus includes the type species L. hatcheri, and L. compressus, both later placed as synonyms of Chamops segnis by Estes (1964). The type specimen of L. hatcheri (USNM 10706) is a left dentary containing partly dissolved teeth. Estes (1964) stressed the eroded nature of the type and argued that a similar kind of erosion in some specimens of Chamops indicates they belong to the same species. However, to synonymize Lan- ceosaurus with Chamops is probably not valid, be- cause USNM 10706 obviously differs from the den- tary of Chamops segnis in having a more slender configuration, a much deeper subdental shelf, and a more restricted Meckelian canal. Therefore, USNM 10706 should not be referred to Chamops segnis, and is so poorly preserved that it shows no char- acters adequate to establish a species. In view of this fact, Lanceosaurus hatcheri should be regarded as a “nomen vanum”. The other species, L. com- pressus, displays a similar condition, and should be treated the same as the type species. Gilmore (1928) named Alethesaurus quadratus and placed it in the Teiidae. The holotype (USNM 10802) is a poorly preserved maxillary fragment “bearing four perfect teeth” as described, but there are actually only three and these are badly dissolved as photographed (see Gilmore, 1928:162, pi. 26, fig. 13). Estes (1964) reexamined the holotype and con- cluded that the specimen was a well-worn dentary fragment of Chamops segnis, which bears the “last few teeth.” We agree with Gilmore’s original iden- tification of the specimen as a maxillary, as the den- tal shelf on the specimen is thin and plate-like, in- dicating a supradental rather than a subdental shelf; however, Alethesaurus quadratus is in fact founded 24 BULLETIN CARNEGIE MUSEUM OF NATURAL HISTORY NO. 33 on a specimen that is inadequate for a generic or specific identification. Thus, in this study, we can neither revalidate the name, nor agree with placing it as a synonym of Chamops segnis. The posterior teeth of Chamops are widely tricuspid, and the shafts are strongly swollen or barrel-shaped; none of these features are shown on USNM 10802, al- though Gilmore (1928:162) mentioned “a vestigial denticle" at the base of the cutting edge of the teeth. In this paper, Alethesaurus quadratus is treated as a “nomen vanum,” since it is “based on types which are inadequate for definitive diagnosis” (Mo- nes, 1989:233). Problematic Reports and Misidentified Speci- mens. C. M. Sternberg (1951) identified a jaw frag- ment (NMC 8891) as “ Chamops cf. [C.] segnis ” (incorrect syntax in using “cf.”; for comments see Lucas, 1986; Estes, 1987). The specimen was col- lected from the Upper Cretaceous Wapiti Forma- tion, which is partially correlative with the Horse- shoe Canyon Formation of central Alberta (Tanke, 1988; Dodson and Currie, 1990), in the Kleskun Hills, northwestern Alberta. Although the specimen was unavailable for this study, the original descrip- tion and the illustration (C. M. Sternberg, 1951:256) show that NMC 8891 differs from Chamops segnis in having relatively high-crowned and much less swollen teeth, as well as having a subdental shelf that sharply decreases in depth posteriorly. There- fore, the specimen NMC 8891 is probably not a jaw of Chamops segnis, and a restudy of the specimen is needed. Waldman (1970) described a lower jaw ramus (NMC 13563) from the Oldman Formation as “ Chamops sp.” Estes (1983a) subsequently re- ferred the specimen to Chamops segnis, and re- stored the mandible of C. segnis based on this spec- imen (Estes, 1983a:fig. 22; but see discussion on Sphenosiagon below). But a recent restudy of this specimen resulted in its removal from Chamops, and referral to Sphenosiagon (Gao and Fox, 1991). The specimen NMC 13563 shows a lightly built and wedge-shaped dentary, narrow subdental shelf, and poorly defined sulcus dentalis. All of these are di- agnostic features for Sphenosiagon, differing from those of Chamops and other teiids. The teeth of NMC 13563 are partly dissolved, showing no cusp pattern, but another specimen (UALVP 29742) clearly shows that the crowns of this lizard are un- icuspid, rather than tricuspid as in Chamops (see later discussion on Sphenosiagon). Sahni (1972) referred three specimens (AMNH 8486-8488) from the Judith River Formation, Mon- tana, to Chamops segnis, thereby extending the geo- logic record of this particular species from Lancian to Judithian. However, these specimens (see Sahni, 1972:fig. 8A-D) are too poorly preserved to justify a definite referral at species level to Chamops seg- nis, although they are probably referrable to Cham- ops at the generic level. AMNH 8486 (Sahni, 1972: fig. 8 A, B) is a robust dentary having only tooth bases preserved. The specimen shows no indication of having short-crowned, broadly tricuspid, and weakly heterodont teeth, which are characteristic of the Lancian Chamops segnis. Another specimen, AMNH 8488 (Sahni, 1972:fig. 8C, D), is a maxil- lary fragment with four anterior teeth. Once again, the specimen shows none of the diagnostic charac- ters for Chamops segnis, although short and robust anterior teeth, and robustness of the maxillary may count as indication of Chamops affinity broadly. Through personal observation of these specimens, it seems to us that they can be more appropriately referred as to “Teiidae, gen. et sp. indet.”. If the boat-shaped dentary configuration of AMNH 8486 can be regarded broadly as a significant similarity to Chamops, the specimen can be identified in no further detail than “ Chamops sp.” Armstrong-Ziegler (1980 ) reported the occur- rence of Chamops segnis in the Fruitland Forma- tion, New Mexico. The specimen on which this re- port is based is a dentary fragment having one tooth and the base of another (MNA PI. 1613). It shows a subpleurodont dentition that is indicative of the family Teiidae, but shows no diagnostic features at either generic or specific levels. The specimen is too fragmentary to be identifiable as Chamops segnis or even Chamops, and may be more appropriately re- ferred to as “Teiidae, gen. et sp. indet.” Sullivan (1981) identified another specimen (UNM FKK-038a) from the Fruitland Formation as “cf. Chamops segnis," and stated that “this speci- men, although somewhat larger than previously de- scribed specimens (Estes, 1964), agrees in every re- spect with the species Chamops segnis, where het- erodonty is little pronounced and tooth crowns are not swollen” (Sullivan, 1981:77). The specimen was originally described as “posteriad teeth are tri- cuspid; anteriad teeth are pointed and slightly re- curved” (Sullivan, 1981 :77). Probably based on this description, Estes (1983a) referred this specimen to Chamops segnis. However, the above-mentioned statement and description are not in agreement with the illustrations of the specimen (see Sullivan, 1981: text fig. 4.1). The drawings show that the anterior teeth on the specimen are either partially dissolved 1996 GAO AND FOX— LATE CRETACEOUS LIZARDS 25 or broken, and the posterior teeth show no indica- tion of having tricuspid or barrel-shaped crowns. From the illustrations, no evidence is available to show affinities with Chamops segnis, but the ex- tremely heavy deposit of cementum around the tooth bases and the large replacement pits show re- semblance to some Tertiary tupinambines (e.g., Tu- pinambis huilensis\ see Estes, 1983tf:fig. 23). There- fore, although there is not much doubt that UNM FKK-038a represents a teiid, its referral to Chamops segnis is questionable. When more specimens from the same formation become available, this specimen should be restudied. Chamops sp. (Fig. 7F-J) Referred Specimens.— UALVP 29816-29818, 29820, 29826 (total: five), fragmentary maxillaries; UALVP 29819, fragmen- tary dentary. Locality and Horizon. — MR-6 and MR-20 local- ities, Verdigris Coulee, approximately 30 km east of the village of Milk River, southern Alberta; upper member of the Milk River Formation (Aquilan), Upper Cretaceous. Description. — The specimens UALVP 29816 and 29826 (Fig. 7F, G) are both the posterior part of left maxillaries, each having five teeth preserved. The maxillary teeth are short, stout, and tricuspid. Although swollen in lateral view, the tooth bases are slightly compressed anteroposteriorly. These teeth are evidently subpleurodont, being attached to a low lateral parapet for about one-third of the tooth height. Dorsal and medial to the tooth row is a well-developed supradental shelf on UALVP 29826 (broken on 29816), which carries a shallow but well-defined sulcus den- talis medial to the tooth bases. The third fragmentary maxillary (UALVP 29817, Fig. 7H) has the posteriormost five teeth and the broken base of another pre- served. The specimen shows the same tooth form as that of the two maxillaries described above, but has replacement pits de- veloped at the bases of the last and the fourth posteriormost teeth. The posterior interior alveolar foramen, as in the Lancian species Chamops segnis, opens above the fifth tooth position from the back. Another maxillary, UALVP 29818, shows a short external pre- maxillary process and the more or less plate-like base for the medial process. A notch between the two processes is clearly recognizable on the specimen. The nasal process of the bone rises abruptly behind the notch for the naris, forming a steep lateral wall of the maxillary. Although broken dorsally and posteriorly, the preserved part shows that the dorsal process of this species is higher and more steep anteriorly than that of the Lancian Chamops segnis (see Estes, 1964:fig. 47). The maxillary frag- ment shows the anteriormost six tooth positions (four teeth and the broken base of two others). The teeth are conical but have swollen bases, and the tips of the crowns are slightly recurved. The tooth form of this species is better shown on UALVP 29819 and 29820 (Fig. 71, J). The former is a fragmentary right dentary having three teeth and the latter a left maxillary fragment bearing four teeth. The teeth are short and subpleurodont, and the tooth shafts are swollen, having a heavy deposit of cementum at the tooth bases. Having a “bottle neck” constriction, the crowns are slightly narrower than the swollen shafts, and the middle and posterior parts of the tooth row are broadly tricuspid. The main cusp is much stronger than the side cusps, and the anterior side cusp is, in most cases, more prominent than the posterior one. The unworn teeth on UALVP 29819 show that the crown surface has faint wrinkling or striations, which are shown more clearly lingually than labially. Discussion. — The six specimens described above represent the geologically oldest records of the ge- nus Chamops, as the specimens were collected from the upper member of the Milk River Formation, which is Aquilan in age (see the section “Geolog- ical Setting” above). The teeth of these specimens resemble those of Chamops segnis (from all Lan- cian sites) more closely than those of any other ear- ly teiids in already showing a tendency towards the apomorphic swollen crowns which characterize that species, and the more posterior teeth are strongly tricuspid, as in C. segnis, although the side cusps are smaller relative to the central cusp than in the Lancian form. In addition, the vertically directed an- terior edge and the height of the nasal process (shown in UALVP 29818) differ from those of Chamops segnis (see Estes, 1964:fig. 47). Impor- tantly in the Milk River specimens, the crowns are taller and less swollen than in C. segnis specimens of comparable size, as if representing membership in a transformation series between ancestral states in which the tooth shafts are columnar and straight- sided (see, e.g., Winkler et al., 1990, for Early Cre- taceous teiids) and the barrel-like shafts of C. seg- nis. In the light of these comparisons, we accept as a working hypothesis that the Milk River specimens represent a species of Chamops more primitive than C. segnis, with the morphological differences cor- roborated by the 18 Myr chronological gap that sep- arates Milk River from Lancian horizons; in the ab- sence of more extensively preserved material, this species is best designated “ Chamops sp.” Genus Socognathus Gao and Fox, 1991 Type Species. — Socognathus unicuspis Gao and Fox, 1991. Range. — Upper Cretaceous, North America. Diagnosis. — As for the type and only known spe- cies. Socognathus unicuspis Gao and Fox, 1991 (Fig. 8, 9) Holotype. — UALVP 29739, an incomplete left dentary having 1 1 well-preserved teeth and the bas- es of six others. 26 BULLETIN CARNEGIE MUSEUM OF NATURAL HISTORY NO. 33 Type Locality and Horizon. — Railway Grade lo- cality, in Sec. 29, Tp. 21, R 12, W 4, southeastern Alberta; Upper Cretaceous Oldman Formation (Ju- dithian). Referred Specimens. — UALVP 29910-29911, incomplete tooth-bearing maxillaries; UALVP 29732, 29736. 29740, 29743- -9745, and RTMP 82.24.57, incomplete tooth-bearing dentaries; all the referred specimens were collected from the Irvine locality (legal description below). Known Distribution. — Upper Cretaceous Oldman Formation (Judithian), southeastern Alberta. Possi- ble occurrence of the genus in the Hell Creek For- mation is indicated by an unnumbered jaw fragment in the UALVP collections from the BCA locality, but identity cannot be confirmed owing to the poor preservation of the specimen. Diagnosis (Revised from Gao and Fox, 1991). — Relatively large Late Cretaceous teiid differing from other teiids in having a unique combination of the following character states: dentary greatly elon- gate and robust; mandibular symphysis strongly en- hanced by ventral bony buttress; marginal teeth not crowded along tooth row, but variable in spacing and orientation; teeth tall, with straight anterior and posterior sides, somewhat compressed anteroposte- riorly and recurved, set at an oblique angle to long axis of jaw; tooth crowns unicuspid, with cusp pointed, inclined somewhat posterolingually, and having moderately strong anterior ridge and weaker posterior ridge curving lingually from apical cusp; tooth attachment subpleurodont, with lateral parapet low, about one-third of tooth height. Description. — The holotype (Fig. 8, 9A, B) is a left dentary having 1 1 complete teeth and the bases of six others, with prob- ably the posteriormost one or two teeth missing on account of breakage. The dentary is elongated but robust, and is relatively straight when viewed from above. Medially, the dentary bears a sturdy subdental shelf, the dorsal (subdental) ridge of which bor- ders a well-developed sulcus dentalis along the medial side of the tooth row. The shelf gently narrows posteriorly, but anteriorly turns slightly dorsomedially to the mandibular symphysis. The ventral side of the shelf bears a narrow but clearly defined groove for the dorsal spleniodentary articulation, which terminates an- teriorly below the sixth tooth position, and extends posteriorly to the end of the dental shelf. Anteriorly, the dentary has a strong ventral bony buttress, which would greatly strengthen the man- dibular symphysis of this teiid in life. Posteriorly, the inferior alveolar canal opens into the Meckelian canal through the pos- terior interior alveolar foramen below the 13th tooth position. The widely open Meckelian canal on the specimen indicates a significant hypertrophy of the splenial. The dentary teeth are basically unicuspid, with no tendency towards developing tricuspid crowns as is characteristic of Chamops. although the posterior teeth have stronger anterior and weaker posterior ridges curving ventrolingually; faint striations are present lingually on the apical cusp in some specimens. The Fig. 8. — Socognathus unicuspis, Oldman Formation, Alberta: UALVP 29739 (holotype), incomplete left dentary, lateral (above) and medial (below) views. Scale = 5 mm. crowns are generally bent lingually but variably so; the teeth also vary in size and spacing posteriorly along the tooth row (best shown on UALVP 29736, Fig. 9C), in contrast to the more reg- ular spacing, size, and orientation of the teeth in Chamops. The tooth bases are slightly expanded laterally; a deposit of cement is clearly shown around the bases. The lateral parapet of the jaw is low, about one-third of the tooth height; thus, the dentary teeth are subpleurodont in mode of attachment. The holotype (from a large, evidently mature, individual) shows a suppression of tooth replacement, but resorption pits can be seen on smaller speci- mens (e.g., UALVP 29740). The tooth row of the holotype is straight, but variation of this character is seen in UALVP 29736, which shows a strong curvature as in Leptochamops. The number of teeth anterior to the posterior interior alveolar foramen is about 13, posterior to the foramen five or six; therefore, the complete tooth row is estimated to include 18-20 teeth. The posterior teeth are enlarged, resulting in a weakly developed heterodonty along the tooth row. The external surface of the dentary is strongly convex dorso- ventrally, and strongly swollen laterally as a consequence of the posterior expansion of the Meckelian canal. On the holotype, the inferior alveolar foramina on this surface increase posteriorly in size, and the spaces between the foramina lengthen as well, but these attributes seem to be somewhat irregular on other speci- mens, such as UALVP 29736. The lower and posterior part of the holotype, as in most other specimens, is missing as a con- sequence of breakage; however, the well-preserved anterior part suggests that the lower jaw is deeper than that of Leptochamops. but significantly shallower than that of the contemporaneous teiid Glyptogenys Gao and Fox, 1991. Except as noted above, these characters do not change across the range in size of the dentaries in the present collection, and the larger specimens no more re- semble dentaries of other species described here than do the smaller ones. Denton and O’Neill (1995) cited significant onto- genetic changes in coronal morphology in Prototeius, their new teiid from the Campanian of New Jersey. Comparable changes 1996 GAO AND FOX— LATE CRETACEOUS LIZARDS 27 Fig. 9. — Socognathus unicuspis, Oldman Formation, Alberta: A, B, UALVP 29739 (holotype), incomplete left dentary, lateral and medial views; C, UALVP 29736, left dentary, medial view; D, UALVP 29910, left maxillary, medial view; E, UALVP 29911, right maxillary, medial view; F, UALVP 29743, right dentary, medial view. G, UALVP 29744, right dentary, medial view; H. RTMP 82.24.57, right dentary, medial view. Scale = 2 mm. are not present in the dentaries referred to Socognathus unicuspis, even though the smallest (UALVP 29743) is less than half the size of the largest (UALVP 29739, holotype). UALVP 29910 (Fig. 9D) is an incomplete left maxillary bear- ing nine teeth and the bases of three others. The specimen is smaller than the holotype dentary, indicating that it is from a younger individual. The maxillary is broken above the superior alveolar foramina and behind the 12th tooth position, but the anterior end is nearly complete. Medially, the specimen bears a strong supradental shelf and a well-developed sulcus dentalis. The posterior interior alveolar foramen is located above the 12th position in the maxillary tooth row (the posteriormost tooth re- maining in the specimen). The first three teeth are broken off, with only the bases preserved, but the remaining teeth are com- plete, except for the two in the middle of the tooth row, which have the apex of the crown broken off. As in the dentary spec- imens, the maxillary teeth are unicuspid, subpleurodont, and variable in size, spacing, and curvature along the tooth row. In lateral view, UALVP 29910 has a very short premaxillary pro- cess. The superior alveolar foramina positioned laterally above the tooth row are small and are more or less equally spaced from one another, but the facial process above the foramina is not preserved. The other specimen in the collection (UALVP 2991 1, Fig. 9E) is a right maxillary fragment with seven teeth and the broken bases of nine others. The specimen is larger and more robust than UALVP 29910, and shows a nearly complete tooth count of the maxillary tooth row. The first tooth is broken, followed by seven complete teeth, and tooth bases of nine others. The posteriorly narrowed dental groove indicates that the tooth row is nearly complete, with probably one, at most two, positions missing; therefore, the maxillary probably had about 18 tooth positions. The well-preserved teeth on this specimen show the same crown pattern and tooth attachment as the holotype dentary. 28 BULLETIN CARNEGIE MUSEUM OF NATURAL HISTORY NO. 33 but have a “step' of the tooth row at the fifth through the eighth positions, where the teeth are significantly smaller than those anterior to them (also in UALVP 29910). This kind of “step" is also seen in Chcimops and other lizards (see above discussion on Chamops). Unlike UALVP 29910, the ventral view of this spec- imen shows that the maxillary tooth row is strongly curved, cor- responding to the facial constriction in front of the orbits, and it is strongly flexed dorsally as well, in contrast to the maxillary of Chamops segnis (e.g., UALVP 29729), in which these surfaces are relatively straight. As in UALVP 29910, the posterior interior alveolar foramen is open above the supradental shelf at the level of the 12th tooth position. Posterior to the foramen, the supra- dental shelf bears a well-developed groove along the lateral bor- der of the maxillary. This groove is a clear indication of a tongue-and-groove articulation of the maxillary with the anterior ventral process of the jugal, which is unknown. The internal surfaces of the maxillary of Socognathus differ significantly from those of Chamops and Leptochamops, which are virtually identical in these regards. For example, the posterior interior alveolar foramen opens more anteriorly in Socognathus and the articulation surface for the palatine is significantly deeper and anteriorly more extensive — anatomical differences that sup- plement those of the dentition in distinguishing Socognathus from Chamops and Leptochamops. Discussion. — The referral of Socognathus to the Teiidae is basded mainly on two character states: hypertrophy of the splenial (as indicated by the widely open Meckelian canal), and a heavy basal deposit of cementum; both are recognized synapo- morphies of the family Teiidae (MacLean, 1974; Presch, 1974b; Estes, 1983a; Estes et al., 1988). Some other character states, such as strong subden- tal shelf and well-developed sulcus dentalis, are also indicative of the Teiidae. Socognathus unicuspis is now known from sev- eral well-preserved, tooth-bearing dentaries and maxillaries, as enumerated above. In addition, sev- eral uncatalogued fragmentary specimens from both the Railway Grade and Irvine localities are also re- ferrable to this species. A comparison of the holo- type dentary with the referred specimens allows re- construction of the dentary proportions of this teiid, and shows that Socognathus differs in both tooth form and structure of the dentary from other teiids of the same age, including Glyptogenys, Sphenosi- agon, and Gerontoseps Gao and Fox, 1991, known from the same formation. A strong mandibular sym- physis enhanced by a ventral bony buttress seems to be unique for Socognathus unicuspis, and hence offers no information about its relationship to other teiids. However, the large size of this lizard, variable spacing of the marginal teeth, and the tendency of the teeth to increase in size posteriorly (resulting in a certain degree of heterodonty) are unique resem- blances to extant Tupinambis, a large omnivorous teiid distributed in tropical South America. There- fore, this Judithian teiid is probably referrable to the subfamily Tupinambinae (sensu Presch, 1983), which includes the extant genera Tupinambis, Cal- lopistes, Crocodilurus, Dracaena, and the Late Cre- taceous fossil genus Chamops (Estes, 1964, 1983a; Presch, 1983). Many large specimens of Socogna- thus show no resorption pits, indicating suppression of tooth replacement in adults, which has also been recognized in Tupinambis (MacLean, 1974). UALVP 29743-29744 (Fig. 9F, G) and RTMP 82.24.57 (Fig. 9H) from the Irvine locality are three small dentaries referred to Socognathus unicuspis with some uncertainty. The three specimens show a similar tooth form (cusp pattern) to the holotype, but the dentaries are slightly more slender, and the anterior teeth are packed closely together. Further- more, RTMP 82.24.57 shows that the subdental shelf sharply decreases in thickness posteriorly, dif- fering from other specimens mentioned above. These three dentaries are tentatively referred to So- cognathus unicuspis, awaiting further study of more and better preserved specimens to clarify their tax- onomic position. Genus Sphenosiagon Gao and Fox, 1991 Type Species. — Sphenosiagon simplex Gao and Fox, 1991. Range. — Upper Cretaceous, North America. Diagnosis. — As for the type and only known spe- cies. Sphenosiagon simplex Gao and Fox, 1991 (Fig. 10, 11A-D) Chamops sp. Waldman, 1970:546, fig. 2-3. Holotype. — UALVP 29742, incomplete left man- dible having the splenial firmly articulated with the dentary, and having eight posterior teeth and the base of another. Type Locality and Horizon. — Outcrop of the Old- man Formation near Irvine, in Sec. 31, Tp. 11, R 2, W 4, about 40 km east of Medicine Hat, south- eastern Alberta; Upper Cretaceous Oldman Forma- tion (Judithian). Referred Specimens. — NMC 13563, a tooth-bearing right man- dibular ramus collected by C. M. Sternberg from a locality near Steveville, Alberta (Waldman, 1970). UALVP 29836-29837, fragmentary tooth-bearing dentaries from MR-6 locality of the Milk River Formation. Known Distribution. — Upper Cretaceous Oldman Formation (Judithian), and the upper member of the Milk River Formation (Aquilan), southern Alberta. Diagnosis (Revised from Gao and Fox, 1991). — 1996 GAO AND FOX— LATE CRETACEOUS LIZARDS 29 Fig. 10. — Sphenosiagon simplex, Oldman Formation, Alberta: UALVP 29742 (holotype), incomplete left mandible, lateral (above) and medial (below) views. Scale = 3 mm. Abbreviations as in Fig. 2. A Late Cretaceous teiid differing from other teiids in having in combination the following character states: dentary shallow, strongly bilaterally com- pressed and wedge-shaped, with straight dorsal and ventral margins along tooth-bearing parts; subdental shelf narrow; sulcus dentalis poorly defined; den- tary teeth relatively short and regularly arranged; tooth shafts straight, pillar-like, little expanded transversely even at base, not set obliquely across jaw; tooth crowns unicuspid, but not recurved, lack- ing lingual striae, with anterior apical crest stronger than posterior crest; tooth attachment tending to be subacrodont, with lateral parapet one-third or less of tooth height. Description. — The holotype UALVP 29742 (Fig. 10, 1 1A, B) is an incomplete tooth-bearing left dentary with a firmly articu- lated splenial. The mandibular symphysis and the postdentary part of the jaw are missing. Sharply tapering anteriorly, the spec- imen clearly shows that the lower jaw was wedge-shaped, lack- ing the boat-shaped curvature along the ventral border that is seen in some other teiids (e.g., Chamops, Socognathus). As shown as well on the referred specimens, the mandible is later- ally compressed so that its lateral surface is almost flat and not expanded posteriorly, differing from Socognathus and other con- temporaneous teiids that have a much more convex lateral sur- face of the dentary. The small inferior alveolar foramina are un- evenly spaced, and the posteriormost foramen is located anterior to the midpoint of the tooth row. The same arrangment of these foramina is also shown on NMC 13563, a right mandibular ra- mus that preserves a more nearly complete configuration of the jaw. The medial surface of the jaw is concave, and is occupied largely by the hypertrophied and wedge-shaped splenial. Two Meckelian foramina (anterior inferior alveolar foramen and an- terior mylohyoid foramen) are close to each other and penetrate the splenial to open into the Meckelian canal close to the mid- point of the tooth row. The splenial is firmly attached, but not fused, to the dentary, having clearly recognizable dorsal and ven- tral spleniodentary sutures. The ventral suture is developed along the ventral edge of the jaw, not crossing laterally beyond the ventral midline. The slender subdental shelf is distinctive: the anterior maximum depth of the shelf is no more than half of the tooth height, and is greatly diminished posteriorly. The dorsal ridge of the shelf is so poorly developed and so close to the tooth row that a sulcus dentalis is barely recognizable, in contrast to the well-developed sulcus in Chamops and Socognathus. The holotype has eight posterior teeth and the base of another preserved. A comparison with NMC 13563 (referred specimen with a complete posterior end of the tooth row) suggests that the holotype may have the posteriormost one or two teeth missing (both specimens have the anterior end broken off). Among the preserved teeth, three are complete, and the remaining ones have the crowns partly broken. The crowns are unicuspid with a point- ed cusp forming an apex and a faint ridge on both the mesial and distal sides of the cusp. The tooth shafts are straight and are thickened toward the tooth bases, giving a pillar-like configura- tion. An extensive, heavy basal deposit of cementum is clearly visible on the holotype and referred specimens. The dentary teeth are moderately spaced, forming a relatively straight tooth row. The attachment of these teeth can be called subpleurodont, but the lateral parapet is so low (no more than one-third of the tooth height) and the dental gutter is so poorly defined, that the term subacrodont is probably more appropriate. One of the tooth po- sitions on the holotype shows a new tooth erupting posteromedial to the base of the tooth to be replaced (see Fig. 10). The replaced tooth, still in the functional tooth row, has been slightly pushed away laterally from the midline of the tooth row. showing a kind of irregular replacement compared to the “standard” teiid pat- tern, in which the replacement teeth are developed posterome- dially, in subcircular resorption pits (Romer, 1956). In the UALVP collections, two specimens from the upper member of the Milk River Formation document the Aquilan oc- currence of Sphenosiagon simplex. One of the two (UALVP 29836, Fig. 1 1C) is the posterior part of a right dentary, with five complete teeth and the broken base of another. The dentary is larger than the holotype, but clearly resembles the latter in tooth form, spacing, and the mode of tooth attachment. The splenial is missing, leaving the Meckelian canal widely open, showing that a large posterior interior alveolar foramen opens under approximately the fourth posteriormost tooth position; the dorsal and ventral borders of the dentary that define the Meck- elian canal are straight, and in the context of the overall shape of the fragment, indicate that the dentary was shallow and wedge-shaped, as in Sphenosiagon. The other specimen (UALVP 29837, Fig. 1 1 D) is the dentary fragment of an old individual, as evidenced from its size and robustness, and evident suppres- sion of tooth replacement. Although fragmentary, this specimen is clearly referrable to Sphenosiagon simplex based on its char- 30 BULLETIN CARNEGIE MUSEUM OF NATURAL HISTORY NO. 33 Fig. 1 1 . — Sphenosiagon simplex, Oldman Formation, Alberta: A, B, UALVP 29742, holotype, incomplete left mandible, lateral and medial views; Milk River Formation, Alberta: C, UALVP 29836, right dentary fragment, medial view; D, UALVP 29837. left dentary fragment, medial view. Glyptogenys ornata, Oldman Formation, Alberta: E, F, UALVP 29735 (holotype). left dentary, lateral and medial views. Scale = 2 mm. acteristic crown pattern and a distinct groove below the subdental shelf for the spleniodentary articulation. Discussion. — As described above, Sphenosiagon simplex from the Oldman and Milk River forma- tions has a large splenial covering almost the entire medial side of the dentary and a heavy deposit of cementum around the tooth bases; therefore, the liz- ard is referrable to the family Teiidae. Taxonomic Assignment of NMC 13563. Wald- man (1970) described a right mandible (NMC 13563) from the Oldman Formation, southeastern Alberta, which he identified as Chamops sp.; this specimen was subsequently referred to Chamops segnis by Estes (1983a:95). A reconstruction of the mandible of Chamops segnis given by Estes (1983a:fig. 22) is also based on NMC 13563. How- ever, restudy of this specimen strongly indicates that NMC 13563 does not belong to Chamops segnis and should be allocated to Sphenosiagon simplex, as Gao and Fox (1991) demonstrated. First, a comparison of NMC 13563 with the ho- lotype of Sphenosiagon simplex indicates that the two specimens belong to the same species; they show the same unique shape of the jaw, the same development of the splenial that is firmly attached to the dentary, a much more narrow subdental shelf than in Chamops segnis, a poorly defined sulcus dentalis, and the same kind of tooth attachment. Second, although the tooth crowns on NMC 13563 have been partially dissolved diagenetically (only at the tips, not more ventrally), those on the holotype (UALVP 29742) are unambiguously unicuspid, not tricuspid as in Chamops segnis, nor do the tooth shafts display the barrel-like proportions that were illustrated in Estes’ (1983a:fig. 22) figure for NMC 13563; in fact, Estes’ figure differs substantially 1996 GAO AND FOX— LATE CRETACEOUS LIZARDS 31 from that of Waldman (1970:fig. 2), which is the more accurate. For example, no “surangular win- dow” is preserved on NMC 13563 (see Estes, 1983a:94), and the so-called “coronoid,” which both Waldman (1970) and Estes (1983a) have shown, is actually a piece of unidentifiable postden- tary bone glued into the position of a coronoid. Therefore, future study of better preserved speci- mens may well change the reconstruction of the jaw. Waldman (1970) estimated that the full tooth count of NMC 13563 is about 15, which is correct. His claim, however, that the “dental gutters are present on both sides of the tooth-row” (Waldman, 1970:546) is incorrect and unlikely for any lizard taxon. In fact, in NMC 13563, only the eighth tooth from the back is located more medially than the others, leaving a short space from the low lateral parapet of the dentary. This tooth is not fully func- tional, but is still developing and is not yet attached to the lateral parapet of the jaw. All of the remain- ing teeth are attached to the low lateral parapet, and there is no dental gutter along the labial side of the tooth row. Comments on Several Characters of Spheno- siagon simplex. The Judithian teiid Sphenosiagon simplex clearly shows several osteological charac- ters that are worth special notice. One of these char- acters is the tendency towards subacrodonty. The dental gutter in this species is so poorly developed, and the lateral parapet of the jaw is so low, that the term subacrodont is almost appropriate. The tooth attachment in teiids can be either subpleurodont, as in extant Kentropyx Spix, 1 825 and fossil Lepto- chamops; or subacrodont, as in extant Teius Mer- rem, 1820 and fossil Haptosphenus (Estes, 1964; see also later description). There is no true acrodont condition known in the Teiidae; however, a suba- crodont pattern has independently evolved in all three subgroups of the family (Polyglyphanodonti- nae, Teiinae, and Tupinambinae; sensu Presch, 1983) as a more derived condition than subpleuro- dont. The subacrodonty is known from fossil po- lyglyphanodontines, such as the Campanian Adam- isciurus Sulimski, 1972, and the Maastrichtian Hap- tosphenus and Polyglyphanodon Gilmore, 1940 (probably including Paraglyphanodon Gilmore, 1940); and from fossil and extant teiinines, such as Peneteius Estes, 1969a, and Dicrodon Dumeril and Bibron, 1839. It is also known from extant tupinam- binines, such as Dracaena Daudin, 1802. As far as is known among Late Cretaceous teiids, Adamisau- rus and Haptosphenus demonstrate the most derived condition of subacrodonty in the family Teiidae. Another character worth noting is the articulation of the splenial with the dentary. Unlike most fossil teiids, in which the splenial is often lost before buri- al, two specimens (holotype and NMC 13563) of Sphenosiagon have the splenial firmly attached to the dentary, even though the lower jaw is lightly built. A strong interlocking articulation between the splenial and the dentary may play a key role in the retention of the splenial. Two more fragmentary specimens (UALVP 29736, 29837) show that a deep groove is developed posteriorly on the ventral side of the subdental shelf. This groove differs from that of other teiids in having a ventromedial tongue- like structure that prevents the splenial from sliding ventrally. It could be that this type of articulation has prevented the loss of the splenial in Sphenosi- agon simplex. The functional meaning of this struc- ture is unclear, but might be related to the feeding habits of this lizard, coupled with the nearly suba- crodont dentition. It is possible that preservation of the relatively fragile dentary with the splenial in articulation re- flects a taphonomic factor: the Irvine locality, where the holotype of Sphenosiagon simplex was obtained, may represent an “in-channel locality” (Brinkman. 1990), one containing a locally derived concentra- tion of microfossils (Eberth, 1990£>). However, Waldman (1970:542) thought that NMC 13563 had formed “part of the faeces of a crocodile of the genus Leidyosuchus as he stated that the specimen was found “between the first sacral rib and the an- terior margin of the left ischium, [and] together with the curiously eroded teeth, is, to me, highly sug- gestive of digested material” (Waldman, 1970:546). This inteipretation seems to be supported by recent experiments on crocodilian digestion, which show that teeth defecated by crocodiles lack enamel, but are often complete in other respects (Fisher, 1981). If NMC 13563 is indeed coprolitic in origin, the holotype (UALVP 29742) certainly represents an- other type of preservation under quite different con- ditions, because no “digestive demineralization” is shown on the specimen. The two specimens from the Milk River Formation are fragmentary and do not retain the splenial. This type of preservation may reflect the fact that the MR-6 locality repre- sents a paleofacies that is slightly different from the Irvine locality. Genus Glyptogenys Gao and Fox, 1991 Type Species. — Glyptogenys oniata Gao and Fox, 1991. 32 BULLETIN CARNEGIE MUSEUM OF NATURAL HISTORY NO. 33 i 1 1 i i i Fig. 12. — Glyptogenys orncita, Oldman Formation, Alberta: UALVP 29735 (holotype), incomplete left dentary, lateral (above) and medial (below) views. Scale = 5 mm. Range. — Upper Cretaceous, North America. Diagnosis. — As for the type and only known spe- cies. Glyptogenys ornata Gao and Fox, 1991 ' (Fig. HE, F, 12) Holotype. — UALVP 29735, incomplete left den- tary bearing 12 well-preserved teeth and bases for two others. Type Locality and Horizon. — Outcrop of the Old- man Formation near Irvine, in Sec. 31, Tp. 11, R 2, W 4, about 40 km east of Medicine Hat, south- eastern Alberta; Upper Cretaceous Oldman Forma- tion (Judithian). Referred Speciemens. — UALVP 2036, 29756, 29758—29759, 29913, 29767, 29769, all tooth-bearing dentaries, and all topo- typic but UALVP 29758, from the Railway Grade locality of the same formation. Known Distribution. — Upper Cretaceous Oldman Formation (Judithian), southeastern Alberta. Several unnumbered dentary fragments indicate possible oc- currence of this lizard in the Milk River Formation (Aquilan), but this needs support from better pre- served specimens. Diagnosis (Revised from Gao and Fox, 1991). — A relatively large Late Cretaceous teiid that differs from other fossil and extant teiids in having the fol- lowing combination of character states; dentary deep, massively built, and, in at least adult individ- uals, heavily ornamented below inferior alveolar fo- ramina; subdental shelf significantly deeper than in other primitive teiids, maximum depth of which al- most equals height of anterior dentary teeth; sulcus dentalis deep, but narrow, with medial wall ap- pressed close to tooth bases; mandibular symphysis weak, lacking robust ventral buttress; dentary teeth heterodont, closely spaced, anteroposteriorly com- pressed, standing parallel to each other to form comb-like structure, and set obliquely to long axis of jaw; crowns slightly recurved, unicuspid anteri- orly, bicuspid posteriorly, with main cusp in pos- terior position and anterior cusp smaller but prom- inent; tooth bases strongly widened transversely, having thick, ring-like deposit of cementum; tooth attachment subpleurodont, with lateral parapet about half of tooth height. Description. — The holotype UALVP 29735 (Fig. 11E, F, 12) is an incomplete but well-preserved left dentary, broken imme- diately behind the 14th tooth position. The dentary is deep, mas- sively built, laterally compressed, and probably proportionally shorter than that of Socognathus. It has a deep boat-shaped cur- vature along the ventral border, but tapers anteriorly and termi- nates with an extremely weak symphysis in relation to the size and robustness of the jaw. The symphysis lacks a ventral bony buttress, and is sharply different in this aspect from the jaw of Socognathus, besides having other differences in general pro- portions of the jaw and tooth form. Medially, the dentary bears a deep and strong subdental shelf, the maximum depth of which is almost equal to the height of the anterior dentary teeth. The subdental ridge of the shelf forms the medial border of a narrow but well-developed sulcus dentalis along the lingual side of the tooth row; this border is appressed close to the tooth bases, and is shallowly concave opposite sev- eral teeth. The shelf ventromedially bears a deep groove for the dorsal edge of the splenial. This groove extends anteriorly and terminates at a point below the sixth tooth position. The anterior end of the shelf bears a blunt but prominent dorsomedial process, forming part of the symphysis. Below the process is a short, fissure-like anterior Meckelian foramen developed under the first six tooth positions. Posteriorly below the deep and robust sub- dental shelf, the Meckelian canal is open widely, and no splenial is preserved; the Meckelian canal occupies over half of the total depth of the dentary. The lower border of the dentary is a sharp edge, which anteriorly turns medially and upward to the sym- physis. On the medial side of the ventral border, a large surface area for the ventral spleniodentary articulation is clearly visible on the specimen, indicating a strong articulation of the two bones along the ventromedial side of the jaw. The lateral surface of the dentary is flattened, but weakly con- vex posteriorly. A row of inferior alveolar foramina perforates its dorsal external surface, and the spaces between the foramina obviously increase in length posteriorly. These foramina are 1996 GAO AND FOX— LATE CRETACEOUS LIZARDS 33 rounded and extremely small in relation to the size of the jaw. The external surface of the dentary is smooth above the foram- ina, but is heavily ornamented below, with irregular bony rugae. Other specimens show that the external ornamentation is variable in extent, ranging from the ventral edge of the jaw to above the line of mental foramina dorsally. Generally, larger individuals show more extensive ornamentation than smaller ones, implying an increase of ornamentation with age. The dentaries of no other teiids in this study show comparable ornamentation at any size. The preserved part of the holotype dentary bears 12 teeth and the bases of two others. These teeth are robust and closely spaced, but with slightly recurved, transversely expanded, par- allel shafts forming a comb-like dental apparatus; they are sub- pleurodont, with the parapet of the jaw slightly lower than half of the tooth height (on two well-preserved, smaller specimens, UALVP 29756 and 29913, the parapet is about half the height of the teeth). On the holotype, the first five or six teeth are slight- ly procumbent, unicuspid, more closely spaced, and more an- teroposteriorly compressed than the posterior ones. The teeth posterior to the sixth are bicuspid, with a moderately-sized, rounded accessory cusp developed anterior to the main cusp and set off from it by a clearly-defined groove labially and lingually. In UALVP 29756, the anterior cusp first appears on the ninth tooth; the more anterior teeth are unicuspid with a robust anterior crest running ventrally from the main cusp, and a weaker, shorter apical crest present posteriorly; the crowns are faintly striated lingually. UALVP 29913, with teeth present through the 12th tooth position, shows a distinct anterior crest, but no cusp, on each tooth, although its other dental and mandibular features in- dicate it pertains to Glyptogenys. We interpret these differences to mean that in Glyptogenys, the cuspation of the dentary teeth becomes more pronounced with age, as it does in the other teiids included in this study, the reverse of the pattern hypothesizied for Prototeius and other teiids by Denton and O’Neill (1995). On the holotype, the crowns of the posteriormost teeth lean medioposteriorly, and the shafts become increasingly robust pos- teriorly. The tooth bases are transversely expanded and oriented slightly obliquely across the jaw; in lingual view, they arise from the bone of attachment at a constant height, in contrast to the irregular pattern in Socognathus. A ring of cementum is devel- oped around the base of each tooth, and an irregular collar of cementum is developed externally, at the junction between the parapet edge and tooth shafts. The dentary tooth row is incom- plete, with probably five to seven posteriormost teeth missing. It is possible when complete specimens are discovered that the pos- teriormost teeth will prove to be tricuspid or incipiently tricuspid, as the combination of anterior conical with middle bicuspid and posterior tricuspid teeth is common in both extant (e.g., Crocod- ilurus and Tupinambis) and fossil teiids (e.g., Meniscognathus). Discussion. — The familial assignment of Glyp- togenys ornatci to the Teiidae is based mainly on the conspicuous deposit of cementum at the tooth bases, and the enlarged splenial, as indicated by the widely open Meckelian canal in the holotype. These are well-recognized synapomorphies of the family Teiidae (Mac Lean, 1974; Presch, 1974/9; Estes, 1983a; Estes et al., 1988). Another synapomorphy often recognizable on jaw fragments is that the re- placement teeth develop posteromedially within subcircular basal cavities (Romer, 1956; Estes et al., 1988). However, this character is in some cases, as on the holotype specimen, unidentifiable as a result of suppression of tooth replacement in adult indi- viduals (MacLean, 1974). One of the most obvious characters of the holo- type is the sculpture on its lateral surface. This sculpture is so pronounced that Gao and Fox (1991) regarded it as taxonomically significant, since sculp- ture on the lower jaw is not common among modern teiids. In a specimen of Cnemidophorus (UMMZ 182042), the lower jaw is ornamented; however, the sculpture on this specimen is developed mostly on the angular bone and is much less rugose than that in Glyptogenys. A similar condition is also seen in another extant teiid, Ameiva (personal observation). Cnemidophorus and Ameiva are more closely relat- ed to each other than to any of the other genera of the family (Gorman, 1970; Presch, 1974a), while Glyptogenys shows no feature resembling the two extant genera in terms of jaw shape and tooth form: Ameiva and Cnemidophorus have a slender mandi- ble with pointed, bicuspid or tricuspid teeth, suitable for feeding on soft-shelled insects, while the deep and robust jaw with thick and bluntly bicuspid teeth in Glyptogenys is best designed for prey on hard- shelled beetles or other resistant terrestrial inverte- brates. It seems that this character (presence of sculpture on the lateral surface of dentary) has been independently evolved in Glyptogenys and in the extant Cnemidophorus-Ameiva group. The osteodermal sculpture on the lower jaw may be ontogenetically variable in extent in lizards. Eth- eridge (1964) showed ontogenetic development of sculpture on a series of dentaries in Anolis iguanids. However, it seems that Glyptogenys may represent a different case, because the holotype was probably from a mature but not an old individual (as evi- denced by the unworn tooth crowns), but shows ex- tensive ornamentation on the jaw nonetheless. In addition, two small jaw fragments (unnumbered specimens) in the UALVP collection show heavy sculpture, while a larger specimen (UALVP 29756) shows less heavy sculpture. Perhaps the ornamen- tation in this lizard is sexually dimorphic, but this interpretation is more problematic because of the small sample size of the specimens for this lizard presently available; there is no evidence from mod- ern lizards supporting this interpretation. Genus Gerontoseps Gao and Fox, 1991 Type Species. — Gerontoseps irvinensis Gao and Fox, 1991. Range. — Upper Cretaceous, North America. 34 BULLETIN CARNEGIE MUSEUM OF NATURAL HISTORY NO. 33 Diagnosis. — As for the type and only known spe- cies. Gerontoseps irx’inensis Gao and Fox, 1991 (Fig. 13, 14A-D) Holotype. — UALVP 29754, an anterior part of a right mandible having four teeth and the broken bases of nine others. Type Locality and Horizon. — Outcrop of the Old- man Formation near Irvine, in Sec. 31, Tp. 11, R 2, W 4, about 40 km east of Medicine Hat, south- eastern Alberta; Upper Cretaceous Oldman Forma- tion (Judithian). Referred Specimens.— UALVP 29757, 29760-29762, 29765- 29766, 29768, all incomplete tooth-bearing dentaries. In addition, several unnumbered dentary fragments from the same horizon are also referrable to this species. Known Distribution. — Upper Cretaceous Oldman Formation (Judithian), southeastern Alberta. Diagnosis (Revised from Gao and Fox, 1991). — A small Late Cretaceous teiid differing from other fossil and extant teiids in having the following com- bination of character states: mandible shallow but wide, with convex lateral and medial sides; splenial partially fused to dentary in some individuals; sul- cus dentalis narrow and conspicuously deep; den- tary teeth slightly expanded transversely, with straight-sided shafts; crowns unicuspid, nearly con- ical, but broad apical cusp having symmetrical an- terior and posterior crest; tooth attachment subpleurodont, with lateral parapet about one-third of tooth height. Description. — The holotype UALVP 29754 (Fig. 13, 14A, B) is an incomplete right mandible having the splenial partially fused to the dentary. The jaw is relatively shallow, but strong and convex both laterally and medially. The lateral convexity is probably characteristic for this species, since the lateral surface is greatly swollen at a 45-degree angle, and is roughly “>- shaped” in cross section. The medial surface has a similar but much weaker convexity along the dorsal spleniodentary articu- lation. The Meckelian canal is restricted in comparison to that of GIvptogenys and Socognathus (see above), but has a wide and horizontal “roof” below the subdental shelf; this wide roof, to- gether with the peculiar type of convexity mentioned above, gives the Meckelian canal an inverted triangular shape in cross section. In medial view, the subdental shelf is deep anteriorly but sharply reduced posteriorly. The shelf dorsally bears a sharp sub- dental ridge, which forms the medial border of a narrow but deep sulcus dentalis. Below the shelf, the splenial slightly deepens posteriorly but greatly tapers anteriorly and terminates below the sixth or the seventh anterior tooth. The splenial is dorsally fused to the dentary, along the spleniodentary suture, which is partly recognizable, but is lost below the seventh and eighth teeth as well as posterior to the 12th tooth position. The ventral splen- Fig. 13. — Gerontoseps irx’inensis, Oldman Formation, Alberta: UALVP 29754 (holotype), incomplete right mandible, lateral (above) and medial (below) views. Scale = 5 mm. iodentary articulation remains unfused, and is located medially along the ventral border of the jaw. The holotype shows the anteriormost 13 tooth positions of the jaw, including four teeth (three complete) and the broken bases of nine others. The anterior six or seven teeth are slightly com- pressed anteroposteriorly, but the middle and posterior teeth are increasingly columnar, swollen, and widely spaced. Tooth attach- ment is subpleurodont, with the lateral parapet as low as one- third of the tooth height. The crowns are bluntly unicuspid and nearly conical, but have symmetrical anterior and posterior ridg- es. The basal deposit of cementum is less heavy than in Glyp- togenys and Socognathus. Replacement pits are developed pos- teromedially at the base of the eighth and tenth teeth. Among the referred specimens, UALVP 29760-29761 (Fig. 14C, D) are better preserved than the others. These specimens clearly show the same kind of tooth form as the holotype, but the fusion of the splenial to the dentary is not developed. The two specimens have 13 or 14 dentary teeth preserved, but are broken at or in front of the posterior interior alveolar foramen. This gives an indication that the total tooth count of Gerontoseps is approximately 18 on the dentary, since the number of tooth positions posterior to the foramen is usually four to five in teiids (personal observation). Discussion. — Perhaps the most obvious feature of the holotype (UALVP 29754) is the fusion of the splenial with the dentary along the dorsal spleniod- entary suture. Gao and Fox (1991) interpreted this fusion on the holotype as taxonomically significant. 1996 GAO AND FOX— LATE CRETACEOUS LIZARDS 35 Fig. 14. — Gerontoseps ir\’inensis, Oldman Formation, Alberta: A, B, UALVP 29754 (holotype), incomplete right mandible, lateral and medial views; C, UALVP 29760, left dentary, medial view; D, UALVP 29761, left dentary, medial view. Leptochamops denticulatus, Frenchman Formation, Saskatchewan: E, SMNH P1927.231, fragmentary left maxillary, medial view; F. SMNH P1927.893, fragmentary left dentary, medial view; G, H, SMNH P2004.89, left maxillary, lateral and medial views. Scale = 1 mm. owing to the small size of the specimen and the low probability that such extensive fusion would occur in such a young individual. Now with more speci- mens in the collection, the fusion appears to be in- dividually variable. The holotype probably repre- sents a mature individual, in spite of its small size and unworn tooth crowns. This type of fusion is obviously different from the homoplastic enclosure of the Meckelian canal in iguanids (Etheridge and de Queiroz, 1988), gymnophthalmids (MacLean, 1974), some scincids (Greer, 1970), xantusiids, and amphisbaenians (Gans, 1978), because the condi- tions in the latter groups are all associated with splenial reduction, while the splenial in Gerontoseps is large. The deep sulcus dentalis, well-defined sub- dental shelf, large splenial, and subpleurodont den- 36 BULLETIN CARNEGIE MUSEUM OF NATURAL HISTORY NO. 33 tition are all consistent with classification of Ger- ontoseps in the Teiidae. Spleniodentary fusion is also known in the Lan- cian teiid Haptosphenus (see Estes, 1964; see also description below), but Gerontoseps is substantially different from the latter in other aspects of jaw con- struction and tooth form. The dentary teeth of Ger- ontoseps are basically conical (although having symmetrical side ridges), and show no tendency to- wards developing a tricuspid or incipiently tricuspid crown pattern. Instead, the bluntly unicuspid teeth of Gerontoseps increase in size posteriorly, showing a tendency to develop into enlarged cheek teeth. These features, combined with the shortness of the jaw and partial spleniodentary fusion, indicate a crushing function. Perhaps this lizard had a food preference for small hard-shelled insects. Genus Leptochamops Estes, 1964 Type Species. — Chamops denticulatus Gilmore, 1928. Range. — Upper Cretaceous, North America. Diagnosis (Revised from Estes , 1983 a). — A Late Cretaceous teiid broadly related to the extant Ken- tropyx—Cnemidophorus—Ameiva group, sharing with the group derived character states such as hav- ing the dentary tooth row curved corresponding to the lateral curvature of the dentary; differing from the group in having higher-crowned teeth, which are uniformly tricuspid or incipiently tricuspid, except for anterior teeth, which are unicuspid; probably more closely related to Meniscognathus Estes, 1964, but differing from the latter genus in having a combination of the following character states: sub- dental shelf narrow and strongly dorsoventrally con- cave; marginal teeth high-crowned, having cylin- drical shaft and tricuspid or incipiently tricuspid crown pattern; tooth attachment subpleurodont, hav- ing lateral parapet about one-half to one-third of tooth height. Leptochamops denticulatus (Gilmore, 1928) Estes, 1964 (Fig. 14E-H) Holotype. — YUM 1062, nearly complete left dentary with 23 teeth, collected by J. B. Hatcher in 1889 (see Gilmore, 1928). The same specimen was later cited as USNM 16514 (see Estes, 1964). Type Locality and Horizon. — “Peterson’s quar- ry,” Lance Creek, Niobrara County, eastern Wyo- ming; Upper Cretaceous Lance Formation (Lan- cian). Referred Specimens. — SMNH PI 927.23 I, PI 927.871, P1927.908, P2004.89, fragmentary maxillaries; SMNH PI 927.872, PI 927.893, P2004.129, fragmentary dentaries. Locality and Horizon. — Gryde locality, in Sec. 19, Tp. 14, R 18, W 3, Frenchman River valley, southwestern Saskatchewan; Upper Cretaceous Frenchman Formation (Lancian). Known Distribution. — Upper Cretaceous Lance Formation, eastern Wyoming; Hell Creek Forma- tion, eastern Montana; Frenchman Formation, southwestern Saskatchewan (all Lancian). Diagnosis. — A species of Leptochamops differ- ing from L. thrinax Gao and Fox, 1991 in having the middle and posterior dentary teeth weakly or only incipiently tricuspid, more closely spaced; marginal tooth row strongly curved, corresponding to stronger facial constriction; subdental shelf more slender and dorsoventrally concave. Description. — The referred specimens include four maxillaries and three dentaries, all collected from the Gryde locality (see Storer, 1991) in the Frenchman Formation. Among these speci- mens, two (SMNH P1927.231, P1927.893) are better preserved than the others in terms of showing tooth morphology. SMNH PI 927.231 (Fig. 14E) is the posterior part of a left maxillary, with six complete teeth and the base of another. Although frag- mentary, the well-preserved teeth are characteristically of Lep- tochamops denticulatus type: cylindrical, high crowned, slender, and tricuspid posteriorly in the tooth row. Tooth attachment is subpleurodont, with a low lateral parapet of the maxillary being less than one-half of the tooth height. In medial view, the supra- dental shelf is much more slender than in Chamops, but similar to that of Meniscognathus from the same horizon. In ventral view, the posterior narrowing of the shelf indicates that the spec- imen is broken close to the end of tooth row, with probably only one posteriormost tooth missing. The other better preserved specimen (SMNH PI 927.893, Fig. 14F) consists of a fragmentary left dentary containing six pos- terior teeth. The subdental shelf is slender, differing from that in Chamops. This specimen is too fragmentary to show the dentary structure of this lizard, but the teeth preserved resemble those in SMNH P1927.231 in cusp pattern and the mode of attachment. Discussion. — Gilmore (1928) founded “ Cham- ops denticulatus ” on several jaws from the Lance Formation. The holotype (YUM 1062), a complete left dentary containing 23 teeth, was the specimen originally assigned to Chamops segnis as one of the “dentary bones with precisely similar teeth, and corresponding in size with the jaws figured” (Marsh, 1892:450). Gilmore (1928) placed “ Cham- ops denticulatus ” in the Iguanidae*. This familial assignment was based largely on Camp’s (1923) comments on form and attachment of the teeth in Chamops segnis, which included the specimens of the then-unrecognized Chamops denticulatus. Estes (1964) revised Gilmore’s “ Chamops denticulatus ” by referring it to the new genus Leptochamops. The 1996 GAO AND FOX— LATE CRETACEOUS LIZARDS 37 holotype that Estes cited as USNM 16514 is the very same specimen as Gilmore’s YUM 1062. The change of the specimen number was apparently a result of an exchange of specimens between Yale and the Smithsonian Institution in the early 1920s (M. A. Turner, personal communication, 1989). Es- tes (1964) confirmed the affiliation of L. denticu- latus with the Teiidae, and later (Estes, 1983a) pointed out its possible relationships with the living Kentropyx—Cnemidophorus—Ameiva group, on the resemblance in curvature of the tooth row. Leptochamops denticulatus is definitely known only from the Lance, Hell Creek, and the French- man formations (all Lancian in age). The same spe- cies has been reported from pre-Lancian horizons in Montana and New Mexico, but these reports are either unconvincing or simply based on misidenti- fied specimens, as discussed below: 1) Sahni (1972) identified two specimens (AMNH 8490-8491) as Leptochamops denticulatus and stated that “the Judith River material of Lep- tochamops denticulatus is identical to that from the Lance Formation” (Sahni, 1972:353). AMNH 8490 is a dentary fragment with two teeth preserved (per- sonal observation). The teeth are high crowned, cy- lindrical, and appear to be incipiently tricuspid; however, lack of information on jaw structure pre- vents referral of the specimen to this particular spe- cies, although there seems to be little doubt about its association with Leptochamops. The other spec- imen (AMNH 8491; see Sahni, 1972:fig. 8K) is a maxillary fragment with seven teeth. These teeth are low crowned and slightly recurved rather than pro- cumbent; comparison should be made to those of Meniscognathus. Until this study, the maxillary structure of Lep- tochamops was poorly known, but SMNH P2004.89 (Fig. 14G, H) and UALVP 29746 (a comparative specimen from the Lance Formation) show a slen- der premaxillary process and relatively low facial process, as in the extant Cnemidophorus (see Presch, 1970). The teeth preserved on these max- illaries are clearly of the L. denticulatus type, as they are high crowned, cylindrical, and have an in- cipiently tricuspid crown pattern. AMNH 8491 lacks these features (personal observation) and shows no significant resemblance to Lancian Lep- tochamops denticulatus in terms of tooth form or jaw configuration. Therefore, AMNH 8491 would be better referred to as “Teiidae, gen. et sp. indet.,” if its affiliation with Meniscognathus cannot be con- firmed. 2) Armstrong-Ziegler (1978), in a faunal list, re- ported the occurrence of Leptochamops denticula- tus, together with Chamops segnis, from the Fruit- land Formation, New Mexico. Subsequently, the same author (1980), in a descriptive paper, referred six jaw fragments (UALP 75 1 37D, E, K-N) to Lep- tochamops denticulatus ; and this identification was accepted in Estes’ (1983a) review of fossil lizards. Unfortunately, some of these specimens appear to have been anatomically and taxonomically misiden- tified. A “dentary” (UALP 75137D, see Arm- strong-Ziegler, 1980:pl. 2d) is neither a dentary nor from a lizard; instead, the photograph of the spec- imen clearly shows a premaxillary of Albanerpeton, an aberrant amphibian previously thought to be a prosirenid salamander (see Fox and Naylor, 1982; McGowan and Evans, 1995). The original descrip- tion of UALP 75137K stated that “many of the tooth crowns have a black shaft capped by a sharply contrasting light brown occlusal surface” (Arm- strong-Ziegler, 1980:18). This description and the photograph of UALP 75137D give a clear picture of the teeth of Albanerpeton rather than those of a lizard. Two other specimens (UALP 75137M, N), have “bicolored teeth that extend about one-half their height over the parapet of the deep dorsoven- tral border of the bone” (Armstrong-Ziegler, 1980: 20); these may be also referrable to Albanerpeton, although some specimens of L. denticulatus retain a band of black pigment between the shaft and crown (Estes, 1964:112). The specimen UALP 75137E was neither described nor figured in Arm- strong-Ziegler’s paper. Consequently, at least one of the specimens from the Fruitland Formation has been misidentified. and various doubts attend the others, even “the most complete and diagnostic dentary” (UALP 75137K. Armstrong-Ziegler, 1980:18), if only because the latter has not been figured. The identification of AMNH 8490-8491 as Leptochamops denticulatus is unconvincing; and hence, the occurrence of the species in the Judith River Formation, Montana, is questionable. Several specimens of pre-Lancian Leptochamops are known from the Oldman For- mation, but cannot be assigned to the Lancian spe- cies L. denticulatus with certainty (see later discus- sion). Leptochamops thrinax Gao and Fox, 1991 (Fig. 15, 16A-D) Holotype. — UALVP 29749, incomplete left den- tary having five teeth and bases of ten others. Type Locality and Horizon. — Outcrop of the Old- man Formation near Irvine, in Sec. 31, Tp. 11, R 38 BULLETIN CARNEGIE MUSEUM OF NATURAL HISTORY NO. 33 i 1 1 i i i Fig. 15. — Leptochamops thrinax, Oldman Formation, Alberta: UALVP 29749 (holotype), incomplete left dentary, lateral (above) and medial (below) views. Scale = 5 mm. 2, W 4, about 40 km east of Medicine Hat, south- eastern Alberta; Upper Cretaceous Oldman Forma- tion (Judithian). Referred Specimens.— UALVP 29748, 29750, 29753; all to- potypic dentary fragments having several teeth preserved. Known Distribution. — Upper Cretaceous Oldman Formation (Judithian), southeastern Alberta. Diagnosis ( after Gao and Fox, 1991). — A Judi- thian species of Leptochamops differing from the Lancian Leptochamops denticulatus in having the dentary teeth broadly tricuspid, widely spaced, more robust, and fewer in number; subdental shelf relatively straight and stronger; and Meckelian ca- nal more restricted. Description.— The holotype, UALVP 29749 (Fig. 15, 16A, B), is long and slender, gently tapering anteriorly and having a point- ed anterior end. Medially, the dentary bears a relatively strong subdental shelf, which abruptly diminishes in depth posteriorly, in contrast to the gentle reduction in depth seen in Leptochamops denticulatus. The shelf is relatively straight by comparison, curv- ing only weakly dorsally at its anterior end. The dorsal ridge of the shelf is low, medially bordering a narrow and shallow sulcus dentalis along the tooth bases. The Meckelian canal takes up about half the depth of the dentary, and thus is more restricted than in L. denticulatus, indicating a more narrow splenial. The dorsal groove and the ventral surface for the spleniodentary ar- ticulation imply that the splenial terminates under the seventh anteriormost tooth position, leaving a long, fissure-like anterior Meckelian foramen. The mandibular symphysis is weakly built. Laterally, the external surface of the dentary is smooth, and is more convex posteriorly than anteriorly. Seven inferior alveolar foramina can be seen on the holotype, but other specimens, such as UALVP 29748 and 29753 (Fig. 16C, D), show that the num- ber and arrangment of these foramina are individually variable. The holotype bears five teeth and the bases of ten others. The teeth are high-crowned, cylindrical, widely spaced, and subpleu- rodont in their mode of attachment. The tooth crowns, slightly tipped lingually, are broadly tricuspid, having a prominent central cusp and two much less prominent accessory cusps; the apices of some teeth are heavily striated labially and lingually. The an- terior cusp is stronger and more clearly separated from the main cusp than the posterior accessory cusp. The complete tooth count for this species is estimated as 16-17, on the basis of the ter- mination of the subdental shelf (the tooth row usually ends slightly anterior to where the subdental shelf meets the coronoid). In dorsal view, the tooth row is relatively straight, lacking the characteristic curvature seen in the Lancian species L. denticu- latus and Meniscognathus altmani Estes, 1964. The tooth bases are surrounded with a deposit of cementum, as commonly seen in other fossil teiids. Discussion. — The holotype, UALVP 29749, shows several taxonomically significant characters, including long and slender dentary, weak symphy- sis, restricted Meckelian canal, and straight, cylin- drical, high-crowned teeth, that distinguish this liz- ard from other teiids from the Oldman Formation, including Socognathus and Gerontoseps. On the ba- sis of these same features, Gao and Fox (1991) re- ferred this lizard to Leptochamops, distinguishing it from the Lancian L. denticulatus. When compared to the latter species, its nearly straight subdental shelf, strongly tricuspid and widely spaced dentary teeth, and nearly straight tooth row indicate differ- ences at the specific level. Several other specimens from the same formation show character states that more closely resemble L. denticulatus than L. thrinax (see discussion below), suggesting that the above-mentioned differences may be generically significant. Indeed, if the prim- itive condition for the dentition of Leptochamops and the closely related Kentropyx—Cnemidophorus — Ameiva group is to have incipiently tricuspid teeth, the Judithian L. thrinax with broadly tricuspid teeth may be less closely related to L. denticulatus than Gao and Fox (1991) thought. However, current un- derstanding of L. thrinax is not yet sufficient to sep- arate it from the Lancian L. denticulatus at the ge- neric level. Leptochamops sp., cf. L. denticulatus (Fig. 16E, F) Referred Specimens. — UALVP 29741, 29772, right dentaries each having 16 tooth positions preserved. Locality and Horizon. — Outcrop of the Oldman 1996 GAO AND FOX— LATE CRETACEOUS LIZARDS 39 Fig. 16. — Leptochamops thrinax, Oldman Formation, Alberta: A, B, UALVP 29749 (holotype), lateral and medial views; C, UALVP 29748, left dentary, medial view; D, UALVP 29753, left dentary, medial view. Leptochamops sp., cf. L. denticulatus, Oldman Formation, Alberta: E, UALVP 29772, nearly complete right dentary, medial view; F, UALVP 29741, right dentary with partial splenial, medial view. Scale = 2 mm. Formation near Irvine (for legal description, see above). Description. — The specimen UALVP 29772 (Fig. 16E) is a nearly complete right dentary having 14 complete teeth and the bases of two others. Medially, as in the Lancian Leptochamops denticulatus, the dentary bears a slender but well-defined sub- dental shelf. The shelf curves upwards anteriorly and gently di- minishes posteriorly, differing from L. thrinax, in which the shelf is straight and abruptly decreases posteriorly. The splenial is missing as a consequence of preservation, leaving the Meckelian canal open proportionally as widely as that in L. denticulatus. On the ventral side of the subdental shelf, the groove for the dorsal spleniodentary articulation indicates that the splenial ter- minated anteriorly below the sixth anteriormost tooth, leaving a narrow opening for the Meckelian canal facing downwards and medially. The splenial does not cross over the ventral midline, and thus is located entirely on the medial side of the jaw. The lateral surface of the dentary is more convex posteriorly than anteriorly, but is not strongly swollen near the dentary- coronoid articulation. The seven interior alveolar foramina on UALVP 29772 represent the total number for this lizard. The first six foramina are small, closely spaced, and located below the first ten teeth; a gap then separates these foramina from the seventh, under the 13- 14th teeth. In dorsal view, the lateral sur- face of the dentary is curved as in L. denticulatus, with the mid- dle part of the jaw being slightly concave. The tooth row on UALVP 29772 is incomplete, with probably two or three posterior teeth missing. The first two teeth are bro- ken, with only the tooth bases preserved. The next 14 teeth are complete, and are subpleurodont in their mode of attachment. The first seven are conical and their bases are more anteropos- teriorly compressed than those that follow. All of the other teeth are incipiently tricuspid, having a blunt central main cusp, which is slightly tipped medially, with poorly defined side “cusps" or “ridges.” The posterior teeth are basically cylindrical, but are slightly concave lingually. No replacement pits are evident on this specimen. The basal deposit of cementum is much less heavy than that of Chamops, Socognathus, and the holotype of Lepto- chamops denticulatus. A sulcus dentalis is well developed along the medial side of the tooth row, and is deeper anteriorly than posteriorly. The tooth row is curved in accordance with the cur- vature of the dentary. Another specimen, UALVP 29741 (Fig. 16F), is a tooth-bear- ing right dentary broken directly behind the 16th tooth position. The splenial on this specimen has its anteriormost part still at- tached to the dentary, but the posterior part is broken off, as in most fossil teiid specimens. Although the crowns of most of the teeth are partially dissolved, the anterior teeth closely resemble those of UALVP 29772. This character, plus the high-crowned teeth and curvature of the tooth row, indicate that this specimen is also referrable to “ Leptochamops sp., cf. L. denticulatus Discussion. — The specimens UALVP 29741 and 40 BULLETIN CARNEGIE MUSEUM OF NATURAL HISTORY NO. 33 Fig. 17. — Meniscognathus altmani, Frenchman Formation, Saskatchewan: A, SMNH P1927.900. left maxillary fragment, medial view; B, SMNFI PI 927.903, right dentary fragment, medial view; C, SMNH PI 927.887, left dentary fragment, medial view; D, UALVP 29833, left maxillary, medial view; Lance Formation, Wyoming: E, UALVP 29774, nearly complete left dentary, medial view; Hell Creek Formation, Montana: F, UALVP 29775, right maxillary fragment, medial view. Scale = 1 mm. 29772 represent a Judithian teiid that is closely re- lated to the Lancian Leptochamops denticulatus. The curved dentary and tooth row, dorsally concave subdental shelf, and incipiently tricuspid crown pat- tern all are different from Leptochamops thrinax from the same horizon, but resemble the Lancian L. denticulatus as revised by Estes (1964). On the ba- sis of these character states, these two specimens are referred to Leptochamops, close to the type spe- cies. However, the differences, such as less high- crowned and fewer dentary teeth, much weaker cus- pidation, and less heavy basal deposit of cementum on UALVP 29741 and 29772, may indicate sepa- ration at a species level from L. denticulatus. It ap- pears that the two specimens from the Oldman For- mation represent a species that is more closely re- lated to L. denticulatus than to L. thrinax, which has broadly tricuspid teeth and a straighter jaw. Genus Meniscognathus Estes, 1964 Type Species. — Meniscognathus altmani Estes, 1964. Range. — Upper Cretaceous, North America. Diagnosis. — As for the type and only known spe- cies. Meniscognathus altmani Estes, 1964 (Fig. 17) Holotype. — UCMP 46610, complete left dentary with 13 teeth and the broken bases of 15 others. Type Locality and Horizon. — Lull 2 quarry (UCMP loc. V-5620), Lance Formation, Niobrara County, eastern Wyoming (Clemens, 1963; Estes, 1964). Referred Specimens. — Frenchman Formation, Gryde locality: SMNH PI 927.900, PI 927.9 18, PI 927.960, PI 927.999, P2004.91 1996 GAO AND FOX— LATE CRETACEOUS LIZARDS 41 (total: five), incomplete maxillaries; SMNH PI 927.887, P1927.895, P1927.903, P1927.915 (total: four), fragmentary den- taries. Wounded Knee locality: UALVP 29833, fragmentary maxillary. Lance Formation, BTB locality (UCMP loc. V-571 1): UALVP 29832, maxillary; UALVP 29774, 29777, 29834, den- taries. Hell Creek Formation, BCA locality: UALVP 29775, 29779, maxillaries; UALVP 29776, 29831, dentaries. The BTB and BCA specimens from outside of the study area are included here, because the BCA specimens document the first occurrence of this species in the Hell Creek Formation, and the BTB specimens show significant variation in tooth form that was not clearly known before. Known Distribution. — Upper Cretaceous Lance Formation, eastern Wyoming; Hell Creek Forma- tion, eastern Montana; Frenchman Formation, southwestern Saskatchewan (all Lancian). Diagnosis (Revised from Estes, 1983 a). — A Late Cretaceous teiid differing from the closely related Leptochamops in having a dentary more slender and more delicately built; dentary teeth short, laterally compressed, and medially concave; tooth crowns broadly bicuspid or weakly tricuspid; tooth attach- ment pleurodont, with about two-thirds of tooth height attached to lateral parapet of jaw. Differing from extant Kentropyx and related forms in having the marginal teeth low crowned, less lingually con- cave, with blunter cusps. Description. — Most of the referred specimens from the Frenchman Formation are too fragmentary to show details of jaw structure, but preserved teeth show taxonomically significant fea- tures on which the referral of these specimens is based. SMNH P1927.900 (Fig. 17A) is a fragmentary left maxillary from the Gryde locality. Although fragmentary, the specimen has four teeth preserved, and these show diagnostic features of Menis- cogncithus altmcini, being low crowned, medially concave, slight- ly recurved, and closely spaced. The four teeth are bicuspid and are not strongly compressed labiolingually, as they represent the middle part of a maxillary tooth row (comparable with those on the middle part of the holotype dentary, see Estes, 1964:fig. 53). These teeth are pleurodont, with about two-thirds of their height attached to the lateral parapet of the maxillary. SMNH PI 927.903 (Fig. 17B) is a dentary fragment, with four teeth preserved. Differing from the fragmentary maxillary de- scribed above, the teeth on this specimen are tricuspid, laterally compressed, and less recurved. These features, in keeping with the slender subdental shelf, indicate that the four teeth are from the posterior part of tooth row. Teeth on another dentary frag- ment, SMNH PI 927.887 (Fig. 17C), also a posterior part, show a similar morphology. Six other specimens (SMNH PI 927.895, P1927.915, P1927.918, P1927.960, P1927.999, and P2004.91) from the Gryde locality are more fragmentary than the three described above, but from their similar tooth form, are clearly referrable to the same species. UALVP 29833 (Fig. 17D) is the posterior part of a left max- illary from the Wounded Knee locality of the Frenchman For- mation. The maxillary fragment has nine tooth positions, includ- ing five teeth and the broken bases of four others. The five teeth are not well preserved, but still show the distinctive features of Meniscognathus, as they are low crowned, laterally compressed, medially concave, and tricuspid. The specimen UALVP 2977 4 (Fig. I7E) from the BTB local- ity is a nearly complete left dentary, which displays details of the jaw structure and retains the complete tooth count of this teiid (see below). As in the holotype (UCMP 46610), the dentary is more delicately built than that of Leptochamops denticulatus. The teeth on this dentary are pleurodont, having about one-third of their height projecting above the lateral parapet. These teeth are low crowned, closely spaced, and the tooth row is slightly curved corresponding to the lateral curvature of the dentary. In contrast to the holotype, which shows 28 tooth positions, there are a total of 21 in this specimen, including 16 teeth and tooth bases, as well as spaces for five others. The anterior seven teeth are conical, followed by six bicuspid and then eight posterior weakly tricuspid teeth. The bicuspid teeth in the middle part of the tooth row have the crowns bearing a prominent posterior main cusp and a smaller but clearly defined anterior accessory cusp. The posterior teeth have a weak posterior accessory cusp, leaving the crowns only weakly tricuspid. Unlike the correspond- ing teeth in the holotype, these teeth are mostly straight, although some (8-10, 12, and 14) are slightly recurved. The short tooth shafts are anteroposteriorly compressed for the anterior conical teeth, but increasingly laterally compressed and lingually con- cave from the middle bicuspid to the posterior tricuspid teeth. The basal deposit of cementum for the dentary teeth is much less heavy than in Chamops and Leptochamops, as well as other Lan- cian teiids. A narrow sulcus dentalis is clearly recognizable above the subdental shelf and medial to the tooth row. In the UALVP collections, four specimens document the first record of Meniscognathus altmani from the Hell Creek Forma- tion, eastern Montana. These specimens are: UALVP 29775 (Fig. 17F), a right maxillary fragment with six well-preserved teeth; UALVP 29776, anterior part of a right dentary with four teeth and bases of six others; UALVP 29779. an incomplete left max- illary with 12 badly worn teeth; and UALVP 29831. an incom- plete right dentary having five teeth plus bases and spaces for ten others. All of these specimens show a tooth morphology com- parable to those from the type Lance Formation (see Estes, 1964: fig. 53), and are clearly referrable to M. altmani. One of the specimens (UALVP 29779) has heavily worn and eroded crowns, but still shows significant features, such as low-crowned, pleu- rodont, medially concave teeth forming a curved tooth row. Discussion. — Meniscognathus altmani has been previously reported only from the Upper Cretaceous Lance Formation of eastern Wyoming (Estes, 1964. 1983a), but not from the Hell Creek Formation of eastern Montana (Estes et al., 1969; Estes and Ber- berian, 1970). The new specimens reported here from the Hell Creek Formation (BCA locality) and the Frenchman Formation (Gryde and Wounded Knee localities) represent the first discoveries of this teiid from Late Cretaceous deposits other than the Lance Formation. These discoveries indicate that Meniscognathus altmani is a widely distributed Lancian teiid, like Leptochamops denticulatus, in the Western Interior of North America. A problem with Meniscognathus altmani as de- scribed by Estes (1964) concerns the number of 42 BULLETIN CARNEGIE MUSEUM OF NATURAL HISTORY NO. 33 marginal teeth. The species was originally diag- nosed as “a teiid lizard with from twenty-one to twenty-eight dentary and about twenty-two (in the only complete specimen) maxillary teeth. . .” (Es- tes, 1964:113). Probably realizing that the tooth count alone is a nondiagnostic feature for most liz- ard species, Estes (1983a) eliminated the above-cit- ed character from the revised diagnosis, but it re- mained in the description. The tooth count of this lizard ranges from 21 to 28, unexpectedly wide for a teiid from present understanding of variation in the living species of the group. The normal range of individual variation in extant teiid species is no more than five positions (Edmund, 1969), mostly around three. Perhaps Meniscognathus altmani is exceptional among teiids with the tooth count vary- ing more than in other species. The nearly complete dentary described above (UALVP 29774) has 21 tooth positions in a full tooth row. Its low number of teeth, in keeping with the relatively straight tooth shafts and more delicately built dentary, is sugges- tive of an alternative interpretation: the subdivision of the “21-phenon” (term “phenon” sensu Mayr and Ashlock, 1991) with straight teeth and the “28- phenon” with recurved teeth may indicate that this lizard is sexually dimorphic in terms of tooth count and curvature. However, current sample size and poor preservation of most specimens known for this taxon are not sufficient to demonstrate this alter- native interpretation. Another problem with Meniscognathus altmani concerns the morphology of the crowns of the mar- ginal teeth. The species was first diagnosed as hav- ing broadly tricuspid crowns (Estes, 1964), but later revised as having “broadly triconodont or bicuspid crowns’’ (Estes, 1983a:91). The holotype (UCMP 46610) shows that only the posterior teeth are weak- ly tricuspid, the middle teeth are bicuspid, while the anterior ones have been broken away. Estes’ “broadly tricuspid” characterization was based on a fragmentary maxillary (UCMP 46081) bearing six posterior teeth. The newly discovered dentary, UALVP 29774, in comparison with two other well- preserved specimens known for this teiid (UCMP 46610, 54266), indicates that the anterior teeth of this species are conical, the middle bicuspid, and the posteriormost weakly tricuspid. The diagnosis for this lizard has been revised accordingly. Among Lancian teiids, Meniscognathus is closely related to Leptochamops (Estes, 1964, 1983a), and the two together can be placed in the Teiinae (sensu Presch, 1983), as evidenced by their resemblances to the extant Kentropyx-Cnemidophorus-Ameiva group in tooth morphology and jaw structure (Estes, 1983a). In regard to jaw shape and tooth form, this group is generally characterized by having a shal- low, slender, and elongated dentary in comparison with the closely related Tupinambinae, and in hav- ing the combination of anterior conical + middle bicuspid + posterior tricuspid teeth, as described here. Genus Haptosphenus Estes, 1964 Type Species. — Haptosphenus placodon Estes, 1964. Range. — Upper Cretaceous, North America. Diagnosis. — As for the type and only known spe- cies. Haptosphenus placodon Estes, 1964 (Fig. 18) Holotype. — UCMP 46035, a nearly complete right dentary with 16 tooth positions (the specimen number is misprinted as 46305 in Estes, 1964:1 16). Type Locality and Horizon. — Lull 2 quarry (UCMP loc. V-5620), “near the head of a tributary of Buck Creek,” Lance Formation, Niobrara Coun- ty, eastern Wyoming (Clemens, 1963:23; Estes, 1964). Referred Specimens.— Frenchman Formation, Gryde locality: SMNH P1920.23, left mandible; Wounded Knee locality: UALVP 29771, 29780, incomplete mandibles. Scollard Forma- tion. KUA-1 locality: UALVP 33921, incomplete mandible. Known Distribution. — Upper Cretaceous Lance Formation, eastern Wyoming; Hell Creek Forma- tion, eastern Montana; Frenchman Formation, southwestern Saskatchewan; and Scollard Forma- tion, central Alberta (all Lancian). Diagnosis (Revised from Estes, 1964). — A Late Cretaceous teiid characterized by the following combination of character states: mandible short and robust, with extensive fusion of splenial, coronoid, dentary and surangular bones; splenial strongly notched anteriorly; dentary teeth low crowned, lat- erally compressed, and bluntly tricuspid; sulcus dentalis lost; tooth attachment subacrodont, with lat- eral parapet remaining less than one-third of tooth height. Description. — SMNH PI 920.23 (Fig. 18 A) from the Gryde locality. Frenchman Formation, consists of an incomplete left mandible (most of the dentary with splenial), which is broken just behind the 13th tooth position. Tapering upward anteriorly, the jaw is short, robustly built, and has the same configuration as the holotype (UCMP 46035). Medially, the dentary bears a weak process at its anterior end, forming the main part of the weak mandibular symphysis. The 1996 GAO AND FOX— LATE CRETACEOUS LIZARDS 43 Fig. 18. — Haptosphenus placodon, Frenchman and Scollard formations, Saskatchewan and Alberta: A, SMNH PI 920.23, left mandible, medial view; B, UALVP 33921, fragment of left mandible, medial view; C, UALVP 29771, fragment of right mandible, medial view; D, UALVP 29780, fragment of right mandible, medial view. Scale divisions for A = 1 mm; scale for B-D = 2 mm. large splenial, covering the extensive internal aspect of the jaw, is fused to the dentary along both the dorsal and ventral splen- iodentary sutures, but the remnant suture lines are still recogniz- able. The anterior end of the splenial is strongly notched, and a short anterior opening for the Meckelian canal is developed un- der the first three dentary teeth. Anteriorly, the ventral spleniod- entary suture is medial in position, but posterior to the midpoint of the tooth row it is ventrally and then laterally located; hence, the posterior half of the ventral suture is not shown in medial view. The splenial has a depression at its central part where two (contra Estes, 1964:fig. 54) Meckelian foramina (anterior inferior alveolar foramen and anterior mylohyoid foramen) open; these are the passageways for fibers of cranial nerve V3. The anterior inferior alveolar foramen is smaller and more ventrally located than the anterior mylohyoid foramen (UALVP 33921, a left man- dible from the Scollard Formation also shows the same feature, see Fig. 18B). The external surface of the dentary is slightly convex; it is mostly smooth, but the ventral edge is slightly scarred, reflecting the attachment of hyoid musculature. The six external inferior alveolar foramina shown on the specimen are small, rounded, increase slightly in size posteriorly, and are un- equally spaced from each other. The posteriormost foramen is probably located under the 12th tooth position, where the spec- imen is broken. The dentary has 13 teeth preserved, but they are all partially dissolved, showing no pattern of cusps. The total tooth count for this specimen is estimated as no more than 16, as only about four positions can be added posterior to the anterior inferior al- veolar foramen (see Estes, 1964). The tooth row is straight in dorsal view, lacking the curvature seen in Leptochcimops and Meniscognathus. The sulcus dentalis is lost entirely, and the rem- nant lateral parapet is so low (less than one-third of the tooth height) that the term “subpleurondont” is incongruous for this condition; we believe that the term “subacrodont" (sensu Estes, 1964) is more appropriate. Two other referred specimens (UALVP 29771, 29780). both from the Wounded Knee locality of the Frenchman Formation, probably pertain to H. placodoiv, they are mandible fragments bearing water-worn teeth (Fig. 18C, D). Both of the specimens show spleniodentary fusion, with more extensive fusion seen on UALVP 29771, in which the suture is barely recognizable. Discussion. — Haptosphenus placodon was origi- nally named and described by Estes (1964) on the basis of several dentary specimens from the Lance Formation, Wyoming. This species was later re- ported from the Hell Creek Formation (Estes et al., 1969; Bryant, 1985), Montana, and now from the Scollard and Frenchman formations, Alberta and Saskatchewan. The newly discovered specimen SMNH PI 920.23 resembles the holotype (UCMP 46035) in general jaw configuration, in having a subacrodont dentition and extensive spleniodentary fusion; therefore, it is clearly referrable to the same species. One difference between the two specimens seems to be that the holotype shows one foramen penetrating the splenial for cranial nerve V3, while SMNH PI 920.23 has two; however, the holotype is broken at the location of the anterior mylohyoid fo- 44 BULLETIN CARNEGIE MUSEUM OF NATURAL HISTORY NO. 33 ramen, and the breakage may have caused the “ab- sence" of this foramen on the specimen. Accord- ingly, the Frenchman specimen, with two internal Meckelian foramina, represents the natural condi- tion of the jaw of this lizard. Estes (1964) referred Haptosphenus placodon to the family Teiidae, and hinted at a tupinambine re- lationship by stating that “the closest resemblances of Haptosphenus seem to be with Chamops, differ- ing from the latter both in bone fusion and in having almost acrodont teeth; it may be an aberrant teiid derived from a Chamops-Uke ancestor” (Estes et al., 1969:19). But more recently, Estes (1983a) re- ferred Haptosphenus to the extinct Polyglyphano- dontinae, while noticing that “it is possible that when skull material is known that it will prove re- ferable to an extinct family” (Estes, 1983a:83). It is difficult to clarify further the phylogenetic rela- tionships of this lizard until more complete mate- rials are found. The Lancian Haptosphenus is indeed highly spe- cialized, with extensive fusion of the mandibular elements and a subacrodont dentition (Estes, 1964, 1983a; Estes et ah, 1969), and its relationship with modern teiids is much less clear than that of other teiids from the same horizon (Estes and Baez, 1985). The fusion of some mandibular elements oc- curs independently in several modern lizard families (e.g.. Iguanidae*, Gekkonidae, Xantusiidae) that are obviously remote from Haptosphenus. This feature has not been reported in any modern teiids (see later discussion) to which Haptosphenus is possibly re- lated; however, some Cretaceous teiids show the po- tential development of this feature: in some individ- uals of Sphenosiagon, the large splenial is firmly articulated with the dentary, although the mandible is lightly built; and in Gerontoseps, the splenial can be partially fused to the dentary along the dorsal spleniodentary suture (see Gao and Fox, 1991, and relevant discussion in this paper). In spite of these similarities, the relationships of Haptosphenus with the above-mentioned two teiids are still obscure, as the phylogenetic value of these characters is still poorly understood. Another unusual character of this lizard, the su- bacrodont dentition, is associated with the loss of the sulcus dentalis and a great reduction of the lat- eral parapet of the dentary. Among squamates, in- cluding amphisbaenians, the true acrodont condition is independently evolved in acrodontan iguanians, and trogonophid and oligodontosaurid amphisba- enians (Zangerl, 1944; Gans, 1958, 1978; Estes, 1975). It seems clear that acrodonty in the Acro- donta Cope, 1864 (= Chamaeleonoidea Moody, 1980; Chamaeleonidae sensu Frost and Etheridge, 1989) is unlikely to be associated with the subacro- donty of certain teiids, such as Haptosphenus, as the relationships of the two groups are remote, as reflected in their classification in two different in- fraorders. Comparing the subacrodonty of Haptosphenus with the acrodonty of amphisbaenians raises another problem. A sister-group relationship of amphisba- enians with Teiidae (sensu lato; = Teiioidea Estes et al., 1988) has been suggested by Boulenger (1884), Bogert (1964), and Bohme (1981), and sup- ported by several other authors (Saint-Girons, 1968; Lecuru, 1969; Presch, 1975; Schwenk, 1988). How- ever, the subacrodont condition in Haptosphenus is evidently nonhomologous to the acrodonty in those oligodontosaurid and trogonophid amphisbaenians, because the more primitive pleurodont condition oc- curs in both Teiioidea and Amphisbaenia (see Estes, 1983a), and the monophyly of each of the two groups is well supported by a suite of synapomor- phies (Estes et al., 1988). Based on the available evidence from jaw shape and tooth form, it seems more reasonable to accept Estes’ (1964, 1983a, 1985) assignment of Haptos- phenus as a member of Teiidae than of any other lizard family. In particular, Estes (1983a, 1985) sug- gested a possible relationship of this lizard to the Polyglyphanodontinae (sensu Estes, 1983a), which includes two genera ( Polyglyphanodon Gilmore, 1940 and Adamisaurus Sulimski, 1972) also having a subacrodont dentition. Polyglyphanodon is known from the Lancian North Horn Formation, Utah (Gil- more, 1940, 19427>, 1 9437?), and Adamisaurus is from the mid-Campanian Djadochta Formation of the Gobi Desert (Sulimski, 1972, 1978). The latter two lizards have greatly expanded crushing teeth, while those in Haptosphenus are tricuspid. In ad- dition, the teeth of Adamisaurus have well-devel- oped resorption pits, while the subacrodont teeth of Haptosphenus are apparently not replaced. These differences throw doubt on the inclusion of Hap- tosphenus in the Polyglyphanodontinae. Here, we draw attention to the close resemblance of this ab- errant form to the extant Draceana, as these two lizards show the following similarities: 1) dentary short, robust, and strongly tapering upward anteri- orly; 2) splenial extends posteriorly to the lateral surface of the jaw; 3) location and number of in- ternal mandibular foramina comparable; 4) mandib- ular elements fused or tend to be fused; 5) subacro- dont tooth attachment. If these similarities are phy- 1996 GAO AND FOX— LATE CRETACEOUS LIZARDS 45 logenetically significant, Haptosphenus may be a member of the Teiinae (sensu Estes, 1983c/). Genus Stypodontosaurus, new genus Etymology. — stypos + odontos + sauros (Greek, masculine), meaning “stumpy-toothed lizard,” in reference to the short and stout teeth of this lizard. Type Species. — Stypodontosaurus melletes, new species. Range. — Upper Cretaceous, North America. Diagnosis. — As for the type and only known spe- cies. Stypodontosaurus melletes, new species (Fig. 19) Etymology. — melletes (Greek, masculine), mean- ing “delayer,” in reference to its late geologic oc- currence in the Late Cretaceous, and its delayed re- covery in 1991 from the screen-washed concentrate collected in the 1960s. Holotype. — UALVP 29844, a nearly complete right dentary having 16 teeth and the base of an- other. Type Locality and Horizon. — KUA- 1 locality (Lillegraven, 1969), Red Deer River valley near the village of Scollard, central Alberta; Upper Creta- ceous Scollard Formation (Lancian). Known Distribution. — Known only from the type locality and horizon. Diagnosis. — Relatively large Late Cretaceous teiid; dentary greatly elongate and robust, very shal- low anteriorly, with no ventral buttress developed; subdental shelf deep and strong, maximum depth about equal to tooth height; sulcus dentalis narrow and deep; dentary teeth short, stumpy, and well sep- arated from each other; tooth bases circular to slightly compressed laterally in cross section; tooth crowns unicuspid, but with well-defined crest en- circling horseshoe-shaped depression lingually; api- cal cusp small, and cuspation not changing signifi- cantly along tooth row; tooth attachment subpleu- rodont, with over half of tooth height projecting above lateral parapet; basal cementum little devel- oped lingually. Description. — The holotype and only known specimen (UALVP 29844, Fig. 19) is a well-preserved right dentary, lack- ing only its posteroventral part. The dentary is basically straight in dorsal view, but is curved anteriorly at the mandibular sym- physis. The shape of the dentary indicates that the mandible of this lizard is shallow, elongated, but robustly built. Medially, the dentary bears a sturdy subdental shelf, which is greatly deepened over the anterior two-thirds of the specimen, but abruptly reduced over the posterior one-third. The maximum depth of the shelf is Fig. 19. — Stypodontosaurus melletes. new genus and species, Scollard Formation, Alberta: UALVP 29844 (holotype), incom- plete right dentary, lateral (above) and medial (below) views. Scale = 10 mm. about the same as the tooth height. The anterior end of the shelf bears a robust medial process, which forms the main part of the strong mandibular symphysis, although the anterior tip of the jaw is very shallow. The ventral surface of the shelf bears a well- developed groove for the spleniodentary articulation. The groove shows that the splenial (missing on the specimen) extended well anteriorly, to beneath the sixth tooth position. The dorsal edge of the subdental shelf, which is straight in medial view, is de- veloped as a sharp and strong ridge along the medial border of the narrow, deep sulcus dentalis. The Meckelian canal is strongly restricted under the first five tooth positions, where the fissure- like anterior opening of the canal is located, but abruptly increas- es in depth and width posteriorly from the sixth tooth. The dentary teeth are short, stout, pillar-like, and widely spaced from each other. With a faint central cusp that is slightly tipped medially, the tooth crowns can be technically termed un- icuspid, but characteristically have well-defined side ridges (crests) running laterally from the cusp, then down and medially, enclosing a more or less horseshoe-shaped medial surface of the crown. The tooth attachment is subpleurodont, with the lateral parapet of the dentary remaining less than half of the tooth height. The basal deposit of cementum is extensive as in other teiids, but appears to be absent lingually. The size of the dentary and the absence of replacement pits indicate that the holotype represents a mature individual in which tooth replacement has been suppressed. The dentary has 17 teeth (16 complete teeth and the base of another) for the complete tooth row of the ho- lotype. A dorsal view of the specimen shows that the tooth row is straight for the 12 posterior teeth, but slightly curved towards the symphysis for the first five teeth. The lateral surface of the dentary is slightly convex for the anterior two-thirds of the bone, but the posterolateral part bulges strongly as shown on the specimen, a result of the posterior wid- ening of the Meckelian canal. The posteroventral part of the den- 46 BULLETIN CARNEGIE MUSEUM OF NATURAL HISTORY NO. 33 tary is missing, but the remainder shows that the ventral border turns strongly towards the medial side of the jaw; accordingly, the ventral spleniodentary suture faces ventromedially. The lat- eral surface has seven inferior alveolar foramina, which are small, rounded, and more or less evenly spaced in a horizontal row. Discussion. — Collected from the Scollard For- mation, UALVP 29844 is the holotype and the only known specimen of this new genus and species. The specimen shows a conspicuous deposit of cemen- tum around the tooth bases except lingually, and a large splenial indicated by the widely open Meck- elian canal (see Estes et al., 1988, for evaluation of these character states). On the basis of these fea- tures, Stypodontosciurus can be referred to the Teiidae. Other character states, such as a well-de- veloped sulcus dentalis, a subpleurodont dentition, and suppression of tooth replacement, are also in- dicative of its affinity with the family Teiidae. How- ever, this referral is tentative, as the crowns of the dentary teeth are in some degree similar to those of the African Cordylus breyeri (Cordylidae), and the depth and robustness of the subdental shelf also re- call the cordylids. Therefore, it is possible, when more and better preserved specimens are recovered, that Stypodontosciurus may be assigned to another family. Compared with other fossil and extant teiids, the most conspicuous features of this new lizard are the extremely deep subdental shelf, and the short, stout teeth with a unique crown pattern as described above. A strong subdental shelf makes the jaw more solidly built, and hence strengthens the biting mech- anism. This type of crushing jaw mechanism is of- ten associated with low-crowned, stout teeth in oth- er teiids, such as Chamops, Tupinambis, and Dra- caena. The tooth crowns of Stypodontosaurus are laterally compressed, having a faint central cusp that tips medially, and lateral ridges that curve down and medially. This pattern is clearly different from those of Judithian teiids, such as Gerontoseps, in which a narrow crown bears a rounded cusp with weakly developed lateral ridges. It is difficult, at this stage, to determine whether this crown pattern is derived from a tricuspid condition in which ridges linked the side cusps with the main cusp, or is a more primitive condition in which there were no side cusps originally. The dentary teeth show no replacement pits at the tooth bases, perhaps as a consequence of age, in which tooth replacement was suppressed. In general, the referral of Stypodontosaurus to the Teiidae is tentative. The jaw configuration and the tooth form of this new lizard are so different com- pared to other known teiids, that we find no evi- dence on which to make a reasonable interpretation about its relationships below the familial level. However, it is possible to include this form in the problematic subfamily Polyglyphanodontinae, which includes several Late Cretaceous teiids hav- ing aberrant tooth forms. Family Scincidae Gray, 1825 The family Scincidae is probably the most suc- cessful modern lizard family, as evidenced by its large numbers of species and individuals, as well as its worldwide distribution. The family is estimated as having 600-1275 living species (probably around 1000; Mattison, 1989), and is widespread not only in Australia, Southeast Asia, and Africa, but also in Europe and the Western Hemisphere. The Scincidae are usually subdivided into four sub- families (Greer, 1970; Estes, 1983a): Feyliniinae Camp, 1923 and Acontiinae Gray, 1845, both of which are presently restricted to Africa; Scincinae Gray, 1925, which are presently Old World exclu- sive of Australia in distribution; and Lygosominae Mittelman, 1952, which are worldwide and are the only subfamily occurring in both Australia and South America (data from Greer, 1970; Estes, 1983a; Mattison, 1989). Despite their great diversity and wide geograph- ical distribution today, the Scincidae have a poor fossil record (Estes, 1983a, 1983b). There is no pre- Miocene record from Asia, Africa, or Australia, where extant scincids are widespread; and as dis- cussed later in this paper, several supposed early scincids may have relationships with other groups rather than the Scincidae: Late Cretaceous Conto- genys Estes, 1969b may be more closely related to Xantusiidae than to the Scincidae; Sauriscus Estes, 1964 may have a cordylid affinity (Estes, 1983a, 1983b), and Upper Jurassic Mimobecklesisaurus Li, 1985 should be relocated to the Paramacellodidae. However, Rowe et al. (1992) have recently de- scribed a scincid with a durophagous dentition from the late Campanian (Judithian) of Texas, accompa- nied there by a second, Sauriscus-Yike skink. The oldest record for the Scincidae is provided by re- cently described incomplete jaws with teeth from the Lower Cretaceous of Spain (Richter, 1994). Genus Penemabuya, new genus Etymology. — pene (Latin), meaning “near, almost”; mabuya (Latin), a kind of skink. 1996 GAO AND FOX— LATE CRETACEOUS LIZARDS 47 Type Species. — Penemabuya antecessor, new species. Range. — Upper Cretaceous, North America. Diagnosis. — As for the type and only known spe- cies. Penemabuya antecessor, new species (Fig. 20, 21A-C) Etymology. — antecessor (Latin), meaning "forerunner,” in ref- erence to the Mabuya-\ike dentary morphology of this ancient skink. Holotype. — UALVP 29789, incomplete right dentary having six teeth and the broken bases, as well as vacant spaces, of 16 others. Type Locality and Horizon. — MR-6 locality, Ver- digris Coulee, approximately 30 km east of the vil- lage of Milk River, southern Alberta; upper member of the Milk River Formation (Aquilan), Upper Cre- taceous. Referred Specimens. — UALVP 29788, 29827—29828, and sev- eral unnumbered jaw fragments; all topotypic specimens. Known Distribution. — Upper Cretaceous Milk River Formation (Aquilan), southern Alberta. Diagnosis. — A small Late Cretaceous scincid re- sembling in dentary structure and dentition primi- tive extant scincids, such as Mabuya', dentary shal- low, slender, and elongate; subdental shelf narrow, extending ventrally nearly to level of ventralmost margin of dentary external to Meckelian canal; sul- cus dentalis poorly developed; Meckelian canal nar- row, anterior part opening ventrally; dentary teeth tall, slender, somewhat recurved, especially poste- riorly along tooth row; tooth crowns unicuspid, with weak anterior and posterior crest, and faint medial striations; tooth attachment pleurodont, with one- half to three-quarters of tooth height attached to lat- eral parapet, depending on position in tooth row. Description. — Two specimens of this new skink, UALVP 29789 (the holotype) and UALVP 29788, are well preserved. The holotype (Fig. 20, 21 A, B) is a nearly complete right dentary, with only the very posterior end missing. It is shallow, slender, and greatly elongated. The lateral surface of the dentary is smooth, showing no sculpture. Five inferior alveolar foramina are unequally spaced from one another, and each shows a certain degree of anterior elongation. Below the foramina, the ventral edge of the dentary turns sharply medially, tending to enclose the Meckelian canal, and forming a more or less flat surface for the anterior two-thirds of the dentary ventral border. Medially, the specimen bears a slender but prominent subdental shelf, which gently curves upward and is diminished posteriorly. The anterior tip of the shelf turns dorsomedially, forming most of the delicate mandibular symphysis. Below the shelf, the Meckelian canal is extremely narrow along its anterior half but slightly in- creases in depth posteriorly. Judging from the articulation sur- i i i j Fig. 20. — Penemabuya antecessor, new genus and species. Milk River Formation, Alberta: UALVP 29789 (holotype), incomplete right dentary, lateral (above) and medial (below) views. Scale = 3 mm. face, the missing splenial seems to have been a shallow and slender bone that ventromedially covered the posterior two-thirds of the Meckelian canal, leaving the anterior third open ventrally. Posteriorly within the Meckelian canal, the posterior interior al- veolar foramen opens below the 20th tooth position, and no in- tramandibular septum is developed in the canal. The dentary teeth are pleurodont, with about two-thirds of the tooth height attached to the well-developed lateral parapet of the jaw. There are 22 tooth positions on the preserved part of the holotype; possibly only the two posteriormost teeth are missing (judging from the posterior extremity of the subdental shelf that corresponds to the termination of the tooth row). Four teeth (sev- enth through the ninth, and the 17th) are completely preserved, and two others (the 20th and 21st) are nearly complete with just the tips broken; the rest of the tooth row consists of either broken bases or teeth that are missing entirely. The tooth shafts are slen- der and the crowns are unicuspid and pointed, with weak medial striations. There is no notable change of tooth structure along the tooth row, except that the anterior teeth have slightly re- curved crowns. The basal part of the eighth tooth has been large- ly resorbed, but its upper half is still functional and weakly at- tached to the parapet. Another well-preserved specimen. UALVP 29788 (Fig. 21C), is also a right dentary but shows all of the positions in the tooth row. On this specimen, seven teeth are complete; the other 17 are tooth spaces or broken bases, making the full tooth count 24 for this dentary. The specimen shows close resemblances to the holotype in shape and structure of dentary, and in crown pattern of the teeth. However, UALVP 29788 has cementum around the tooth bases, and shows an unusual pattern of tooth replacement that is different from that in the holotype. In this specimen, the sixth and the seventh teeth from the back are new replacement teeth that have developed posteromedially to the replaced teeth and have already reached the functional height of the tooth row, but the old teeth (being replaced) are not excavated and are still 48 BULLETIN CARNEGIE MUSEUM OF NATURAL HISTORY NO. 33 Fig. 21. — Penemabuya antecessor, new genus and species. Milk River Formation, Alberta: A, B, UALVP 29789 (holotype), incomplete right dentary, lateral and medial views; C, UALVP 29788, right dentary, medial view. Orthrioscincus mixtus, new genus and species, Oldman Formation, Alberta: D, E. UALVP 29141 (holotype), incomplete left dentary, lateral and medial views; F, UALVP 29904, right dentary, medial view. Scale = 2 mm. functional together with the new teeth. This type of tooth re- placement is not only different from the “iguanid method” in scincids (Edmund, 1969) but also the “varanid method” in most anguimorphans (Edmund, 1969), and apparently represents an abnormal condition. Discussion. — Osteologically, the family Scinci- dae is characterized by several synapomorphies, in- cluding a jugal-squamosal contact, a closure of the supratemporal fenestra, and a medial extension of the palatines to form a bony secondary palate (see Estes et al., 1988). Although these characters are unavailable for the screen-washed materials in this study, a combination of the following character states on the jaws of Penemabuya is strongly indic- ative of its affiliation with primitive scincids: a slen- der, shallow, strongly elongated dentary; a narrow Meckelian canal, which tends towards complete clo- sure anteriorly; a Mabuya- like pleurodont dentition in which the shafts are slender and the crowns un- icuspid and pointed, bearing medial striae; and a posterolingual (intermediate) replacement pattern. Among modern scincids, Mabuya Fitzinger, 1826 and Eumeces Wiegmann, 1834 (Estes, 1983a: the most primitive living skink) are the two most wide- spread genera geographically (Greer, 1970; Peters et al., 1986; Mattison, 1989). Besides the character states discussed by Greer (1970), we here draw at- tention to differences between the two genera in jaw structure and tooth form. These differences are: 1) the dentary of Mabuya is generally much more slen- der and elongated than that of Eumeces ; 2) the spen- ial is much more reduced in the former, and does not extend anteriorly beyond the midpoint of the tooth row; 3) the Meckelian canal of Mabuya is often closed anteriorly by fusion of the dentary, but is commonly open in Eumeces ; 4) the marginal teeth of Mabuya are cylindrical and pointed, while in Eumeces , the tooth shafts are more or less an- teroposteriorly compressed, and the crowns are wid- er and have lateral crests with strong medial stria- tions; 5) tooth replacement is more frequent in Ma- buya than in Eumeces ; correspondingly, the former has a basal foramen at each tooth base, while the latter does not. A comparison of the new skink from the Milk 1996 GAO AND FOX— LATE CRETACEOUS LIZARDS 49 River Formation with available specimens of extant skinks indicates that the new species is probably more closely related to Mabuya than to Eumeces: it shares with Mabuya character states such as having a slender and elongate dentary, a greatly reduced splenial (indicated by a narrow Meckelian canal), and slenderly cylindrical and pointed teeth. As brief- ly discussed above, these are probably derived char- acter states in the Scincidae. In addition, Penema- buya clearly shows a tendency towards closure of the Meckelian canal by medial extension of the ven- trolateral edge of the dentary. The biogeographical origin of the New World skinks (in particular, the Mabuya group) is an open question, although Dunn (1936) noted the close re- lationship of the New World species with the Af- rican species of the genus; Darlington (1957:194) believed that the New World Mabuya “may have come from Africa on drift”; and Greer (1970:180) noted the difficulty of determining “whether the group arrived from Asia via the Bering Land Bridge or from Africa by over-water rafting,” but favored the latter possibility. Greer (1970:180) also stated that “the lack of diversity of the New World species may indicate that the group has not been in the New World very long.” Discovery of the new skink from the Upper Cre- taceous Milk River Formation indicates that the Mabuya- like scincids have inhabited the North American continent since early Campanian time (some 83 Myr ago, Harland et al., 1990), and with the somewhat younger occurrences cited by Rowe et al. (1992), along with the species described be- low, may count in support of a possible northern continental origin for the New World skinks. Fur- ther, the Campanian radiation of North American scincids implies a prior history on the continent of some unknown length, one that is in agreement with European evidence of the establishment of the Scin- cidae in the eastern parts of the Euramerican con- tinent by at least the Early Cretaceous (Richter, 1994). Genus Orthrioscincus, new genus Etymology. — Orthrios + skinkos (Greek, masculine), meaning “early skink.” Type Species. — Orthrioscincus mixtus, new spe- cies. Range. — Upper Cretaceous, North America. Diagnosis. — As for the type and only known spe- cies. Orthrioscincus mixtus, new species (Fig. 21D-F, 22) Etymology. — Mixtus (Latin), referring to the combination of Eumeces- like characters of the dentary and Gerrhosaurus- like teeth of this lizard. Holotype. — UALVP 29747, incomplete left den- tary having seven teeth and tooth bases as well as spaces for ten others. Type Locality and Horizon. — Outcrop of the Old- man Formation near Irvine, in Sec. 31, Tp. 1 1, R 2, W 4, about 40 km east of Medicine Hat, south- eastern Alberta; Upper Cretaceous Oldman Forma- tion (Judithian). Referred Specimen. — UALVP 29904, topotypic, nearly com- plete right dentary bearing 19 tooth positions. Known Distribution. — Upper Cretaceous Oldman Formation (Judithian), southeastern Alberta. Diagnosis. — A small Late Cretaceous scincoid having a dentary structure broadly resembling those of primitive extant Eumeces- like skinks, but with tooth morphology more similar to that of cordylid gerrhosaurs. Dentary slender, elongate, and deli- cately built; differing from Penemabuya in having a deeper dentary anteriorly, slightly narrower sub- dental shelf, and weaker sulcus dentalis; dentary teeth strongly bicuspid, with somewhat rounded main cusp and prominent anterior cusp, set off from main cusp by lingual and more prominent labial groove; enamel of main cusp with striae lingually and faint striae labially; tooth attachment pleuro- dont, with approximately two-thirds of tooth height attached to lateral parapet. Differing from Gerrho- saurus Wiegmann, 1828, in having much more slen- der and curved subdental shelf, anterior cusp less clearly separated from the posterior main cusp, and main cusp not recurved. Description. — The holotype UALVP 29747 (Fig. 2 ID, E; 22) is a tooth-bearing left dentary, which has only the posteroventral part missing. The specimen is designated as the holotype because it clearly shows the tooth form of this new species, while the referred specimen, UALVP 29904 (Fig. 2 IF), a right dentary, has less well-preserved teeth but shows the nearly complete con- figuration of the dentary. The holotype dentary is shallow, straight, elongated, and light- ly built, as in most extant scincoids. The lateral surface is incom- plete, but was apparently smoothly convex like that in UALVP 29904. Four small inferior alveolar foramina clearly shown on the specimen are unevenly spaced from one another in a hori- zontal row. Another (probably the seventh) foramen is recogniz- able by its dorsal border at the broken edge of the jaw and is located lateral to the 13th tooth position, close to the posterior interior alveolar foramen within the Meckelian canal. Medially, the dentary bears a slender subdental shelf, the maximum depth of which is much less than one-third of the tooth height, and the 50 BULLETIN CARNEGIE MUSEUM OF NATURAL HISTORY NO. 33 i i i i i i Fig. 22. — Orthrioscincus mixtus, new genus and species, Oldman Formation, Alberta: UALVP 29747 (holotype), incomplete left dentary, lateral (above) and medial (below) views. Scale = 5 mm. shelf strongly curves dorsally at its anterior end. These propor- tions closely resemble those of the extant scincid Eumeces in comparable parts. Above the subdental shelf and medial to the tooth row is a weakly developed sulcus dentalis, which forms a shallow trough. Below the shelf, the Meckelian canal posteriorly has a maximum depth about equal to the tooth height, but the canal becomes increasingly narrow and faces ventromedially to- wards the anterior end of the jaw. The posterior interior alveolar foramen opens into the Meckelian canal under the 14th tooth position (between the 13th and the 14th on the holotype). Both the holotype and UALVP 29904 document the complete dentary tooth row; the two specimens together indicate that the dentary tooth count of this lizard is variable, ranging at least between 17 (on the holotype) and 19 (on UALVP 29904). The tooth form is clearly shown on the holotype, while the crowns of the UALVP 29904 are partially dissolved, showing no detail about the cusp pattern. The dentary teeth are pleurodont, with about one-third of the tooth height projecting above the lateral parapet of the dentary. Several broken tooth shafts show that the anterior and the middle teeth are slightly compressed anteropos- teriorly, but the posterior ones (the last four or five) are cylin- drical. In general, the teeth are close to one another along the tooth row, but the spacing is variable as the result of tooth re- placement. The first five or six teeth are slightly procumbent, conical, and slightly shorter than those midway along the tooth row. The posterior teeth are slightly stronger than the middle teeth. The crowns of the middle and posterior teeth are bicuspid, having a small, pointed, low anterior cusp, and a high and blunt posterior main cusp, which is weakly striated on its medial sur- face. This particular cusp pattern in a certain degree resembles that of the extant Gerrhosaurus (Cordylidae), but conspicuous differences in jaw structure including the robustness of the sub- dental shelf imply that this similarity may be convergent. Both the holotype and UALVP 29904 show that several teeth have developed an excavation posteromedially at the tooth bases. This indicates an intermediate replacement pattern (Edmund, 1969), which differs from the “iguanid pattern” in having the replacement tooth developed posteromedial to (rather than di- rectly within) the old tooth, and from the “varanid pattern” in having a resorption cavity, although this is shallow. On the ho- lotype, the ninth tooth is loosely attached to the dentary, with the tooth base being slightly off the main line of the tooth row. Behind this tooth is a large space for the tenth tooth, which has already fallen from the tooth row. This abnormal condition of the ninth tooth is possibly the result of mechanical damage, be- cause this tooth is large enough to have been fully functional, but would have been attached to the lateral parapet in the normal condition. UALVP 29904 is a nearly complete right dentary from the type locality. This specimen resembles the holotype in overall size and shape, number and spacing of external foramina, length of tooth row, development of subdental shelf and sulcus dentalis, and configuration of the Meckelian canal. This specimen is re- ferred to the same species as the holotype based on these resem- blances; however, all of the teeth (19 in total) have the crowns partially dissolved, and the poor preservation of the crowns makes this referral somewhat uncertain. Discussion. — The new lizard Orthrioscincus shows a combination of a primitive scincid config- uration of the dentary with a Gerrhosaurus- like tooth form (e.g., compare UA 382, Gerrhosaurus flavigularis with UALVP unnumbered specimen of Eumeces inexpectatus). This mosaic of characters makes the referral of this form at a familial level uncertain, since the Scincidae and Cordylidae are primarily distinguished from each other by skull and external structures (see Estes et al., 1988) that are not preserved in the Oldman specimens. How- ever, this lizard could be referred to the Scincidae based on its Eumeces- like shape of dentary, with its tooth form convergent on that of Gerrhosaurus. Discovery in the Oldman Formation of Orthrios- cincus with a Gerrhosaurus-\ike dentition reveals the following possibilities: 1) ancestral gerrhosaur- ines once had a much wider geographical distribu- tion than they have today. Orthrioscincus may rep- resent one of the primitive members of the group that occupied similar ecological niches in North America as does Gerrhosaurus in Africa and Mad- agascar today. Indeed, with reference to tooth form only, we would place Orthrioscincus in the Cor- dylidae rather than Scincidae. However, the jaw configuration of this new form strongly indicates it is a primitive skink. From current understanding of scincoid evolution, the slenderness and the dorsal curvature of the subdental shelf anteriorly are likely more-derived character states than the stronger and straight subdental shelf in extant gerrhosaurine cor- dylids. Accordingly, if Orthrioscincus is regarded as a primitive gerrhosaur, it would be difficult to explain why such an early member of the group (Judithian in age) has such a derived morphology 1996 GAO AND FOX— LATE CRETACEOUS LIZARDS 51 in the dentary. 2) Orthrioscincus may be a primitive member of the Scincidae. In this interpretation, the similar jaw configuration, especially its slenderness and the curvature of the subdental shelf, is the guide to relationship; in that case, the similarity in cusp pattern to Gerrhosaurus indicates a similar type of food preference. According to Mattison (1989:164), extant Gerrhosaurus lizards “eat a substantial amount of plant material, but are otherwise insec- tivorous.” Nevertheless, Orthrioscincus is a Judithian mem- ber of the Scincoidea, as it possesses the combina- tion of a primitive scincid, Eumeces- like dentary structure and GerrhosaurusAike dentition. Here we tentatively place this genus in the Scincidae, re- garding the similarity in dentary structure as a re- liable indication of relationship, and the Gerrhosau- rusAike crown pattern as convergent. Genus Aocnodromeus , new genus Etymology. — Aoknos + dromeus (Greek, masculine), meaning “restless runner.” Type Species. — Aocnodromeus corrugatus, new species. Range. — Upper Cretaceous, North America. Diagnosis. — As for the type and only known spe- cies. Aocnodromeus corrugatus, new species (Fig. 23, 24A-D) Etymology. — Corrugatus (Latin), meaning “wrinkled,” in ref- erence to the wrinkled enamel on the tooth crown of this lizard. Holotype. — UALVP 29905, a nearly complete left dentary bearing 1 1 teeth and spaces for five others. Type Locality and Horizon. — MR-6 locality, Ver- digris Coulee, approximately 30 km east of the vil- lage of Milk River, southern Alberta; upper member of the Milk River Formation (Aquilan), Upper Cre- taceous. Referred Specimen. — UALVP 29906, topotypic left dentary fragment with nine anterior tooth positions. Known Distribution. — Currently known only from the type locality and horizon. Diagnosis. — A Late Cretaceous scincid distin- guished from all other species by its possession of the following combination of character states: den- tary robustly built, having strong subdental shelf and well-defined sulcus dentalis; Meckelian canal shallow, narrow, and opening ventrally anteriorly; dentary teeth solidly subcolumnar, widely spaced; Fig. 23. — Aocnodromeus corrugatus, new genus and species. Milk River Formation, Alberta: UALVP 29905 (holotype), nearly complete left dentary, lateral (above) and medial (below) views. Scale = 5 mm. tooth attachment pleurodont, with lateral parapet high, about two-thirds of tooth height; tooth crowns broad, unicuspid, with cusp apex posterior to mid- point of crown, medially concave and striated; pos- terior teeth molariform, with laterally compressed and squared-off crowns. Description. — The holotype UALVP 29905 (Fig. 23; 24A. B) is well preserved, showing most of the dentary and 16 tooth positions. The dentary is broken posteriorly at the 16th tooth position, but with no evidence of the posterior interior alveolar foramen that normally opens close to the posterior end of the dentary tooth row. The proportions of the dentary indicate that the jaw was shallow, elongated, but quite robustly built. The lateral surface of the dentary is smooth, and gently convex both dorsoventrally and anteroposteriorly. This surface has six small and round inferior alveolar foramina that are widely spaced, with the space between them increasing gradually posteriorly. The last foramen on the specimen is located lateral to the 14th tooth po- sition. In medial view, the dentary shows a straight and robust sub- dental shelf, which gently decreases its depth posteriorly. The maximum depth of the shelf is slightly less than half of the tooth height, and the anterior end of the shelf forms a relatively strong mandibular symphysis. Dorsally on the shelf is a well-defined sulcus dentalis medial to the tooth row, but the sulcus lacks a sharp subdental ridge for its medial border. Below the subdental shelf, the Meckelian canal is restricted; its maximum depth is shallower than the lateral parapet of the tooth row. The specimen shows no indication of an intramandibular septum within the posterior part of the canal. The restricted Meckelian canal indi- cates that the splenial (not preserved) was probably narrow and 52 BULLETIN CARNEGIE MUSEUM OF NATURAL HISTORY NO. 33 Pig 24. Aocnodromeus corrugatus, new genus and species. Milk River Formation, Alberta: A, B, UALVP 29905 (holotype), left dentary. lateral and medial views; C, D, UALVP 29906, left dentary fragment, lateral and medial views. ?Xantusiidae, new genus and species (unnamed). Hell Creek Formation, Montana: E, F, UALVP 29840, fragment of right mandible, lateral and medial views; Contogenys sloani, Hell Creek Formation, Montana: G. H, UALVP 29839, right mandible, lateral and medial views. Scale = 2 mm. elongated, and the spleniodentary articulation terminated anteri- orly below the seventh and eighth tooth positions. The referred specimen UALVP 29906 (Fig. 24C, D) is compatible in every respect with the holotype, except for its larger size. The dentary teeth are widely spaced from one another and are pleurodont, with about two-thirds of their height attached to the lateral parapet of the dentary. The tooth shafts are solidly built and subcolumnar; even the anteriormost ones show little anter- oposterior compression. Each tooth has a small basal foramen, while at least three (the eighth, the lith, and the 14th) have developed a basal excavation for the replacing tooth postero- medial to the tooth base. The posteromedial location of these basal excavations indicates an “intermediate method” (Edmund, 1969) of tooth replacement for this lizard; only scincids combine this replacement pattern with the dentary and tooth morphology seen in Aocnodromeus. All of the dentary teeth that are preserved have their bases cemented to the dentary. The tooth crowns are medially concave and striated (see Fig. 23), with fainter striae labially. The anterior and middle teeth (the first 14) are unicus- pid, with the tooth apex curved posteromedially to form an “in- 1996 GAO AND FOX— LATE CRETACEOUS LIZARDS 53 wardly pointing V” (Estes, 1964:125) in dorsal view; but the posterior teeth (the last five to seven) tend to be labiolingually compressed, and more or less molariform with truncate crowns. The lingual striations on these posterior teeth are also stronger than those on the anterior and middle teeth. This difference in crown pattern of the posterior teeth is clearly shown on the com- pletely preserved 15th tooth on the holotype. Discussion. — From the characters described above, the specimens UALVP 29905-29906 from the upper member of the Milk River Formation rep- resent a new genus and species that can be placed in the Scincomorpha; this new lizard has a well- defined subdental shelf and a sulcus dentalis, and no indication of a vertically directed intramandi- bular septum. The familial affinities of this lizard are difficult to determine, because there are no re- liable character states (synapomorphies at the fa- milial level) shown on the holotype (the coronoid, an important structure, is broken off). Character states from the tooth morphology of this lizard (un- icuspid crowns curved to form an “inwardly point- ing V,” medial striations, and small basal foramen) indicate affinity to the family Scincidae (see, e.g., Rowe et al., 1992, and Richter, 1994, for Cretaceous scincids with similar tooth morphology). But the ro- bustness of the dentary and strong mandibular sym- physis, especially the robust subdental shelf, are in- dicative of possible cordylid affinities. This new liz- ard is tentatively referred to the Scincidae, pending further study on more specimens to clarify its tax- onomic affiliation. Estes et al. (1988:226) cited Gauthier’s (1982) recognition of medial striations on the tooth crowns as an anguid synapomorphy; in fact, Gauthier’s (1982:19) original statement cited the combination of the “teeth with twisted, chisel-shaped, striated, unicuspid and obtusely-pointed crowns” as the syn- apomorphy. Nevertheless, the presence of medial striations on the tooth crowns also occurs in several scincomorphan families: Scincidae (e.g., Eumeces ), Cordylidae (e.g., Cordylus), and in Xantusiidae (e.g., Palaeoxantusia). Presence of this character in the early scincoid Aocnodromeus, corroborated with other evidence from Cretaceous scincomorphans (this paper; Rowe et al., 1992; Richter, 1994), sug- gests that this character is symplesiomorphic for scincomorphans. Of course, the medial striations in Anguidae, especially in glyptosaurines, are much stronger than those in the other groups and are fre- quently associated with an obtuse, chisel-shaped cutting edge of the crown. The latter combination is obviously best designed for durophagous herbiv- orous feeding adaptations (Estes and Williams, 1984), functionally different from that in insectiv- orous or omnivorous scincids and others. Whether the presence of medial striations in the Anguidae is homologous to that in scincomorphans is still an open question. The “inwardly pointing V” crown pattern rec- ognized by Estes (1964) in some anguids is com- parable to the problem concerning the medial stri- ations: this morphology also occurs in some scin- cids, particularly those in the Mabuya group (per- sonal observation). The occurrence of a similar crown pattern in the two phylogenetically remote groups is probably correlated with a similar food preference, and is thus convergent; it can be easily distinguished in the two groups by reference to oth- er characters, such as jaw construction. Estes et al. (1988) argued that this kind of similarity does not invalidate the synapomorphic status of the “inward- ly pointing V” crown pattern for the Anguidae, be- cause the “inwardly pointing V” in the Anguidae and that in the Scincidae are not the same character in terms of origin, although they are the same in terms of morphology. Family ?Xantusiidae Baird, 1859 The Xantusiidae or night lizards are small rock dwellers, which in many ways convergently resem- ble the geckos: most extant xantusiids are nocturnal insectivores with large eyes and fused eyelids, and their flattened body enables them to squeeze into narrow crevices (e.g., Mattison, 1989). The family has extant representatives of about 16 species in four genera (Savage, 1963), and a geographical dis- tribution in North and Central America, including the West Indies. The phylogenetic relationships of the family have been controversial, as the group has been allied with scincomorphs by some authors (e.g.. Camp, 1923; Romer, 1956; Moffat, 1973; Es- tes, 1983a) but with gekkotans by others (e.g., Mc- Dowell and Bogert, 1954; Savage, 1963; Northcutt, 1978). The most recent phylogenetic analysis strongly suggests a sister-group relationship with the Lacertiformes (Estes et al., 1988). The evolutionary history of the Xantusiidae can be traced back to the Paleocene, as Palaeoxantusia Hecht, 1956. is closely related to the extant Xan- tusia (Hecht, 1956; Estes, 1983a). Three genera from the Upper Cretaceous of the Asian Gobi Des- ert ( Slavoia Sulimski, 1984; Eoxanta and Globaura Borsuk-Bialynicka, 1988) show xantusiid affinities, but their relationships within the Scincomorpha have not been clarified. As discussed below, the North American fossil record includes possible xan- 54 BULLETIN CARNEGIE MUSEUM OF NATURAL HISTORY NO. 33 tusiids that are Late Cretaceous in age. Contogenys, previously considered as an early scincid, is ques- tionably assigned to the Xantusiidae, although a more definite referral to that family needs further study on better preserved specimens. Genus and species new (unnamed) (Fig. 24E, F) Specimen. — UALVP 29840, fragmentary right mandible consisting of partially fused splenioden- tary bearing six teeth and the bases for three others. Locality and Horizon. — BCA locality, SW 1/4, Sec. 9, Tp. 22, R 43, McCone County, eastern Mon- tana; Upper Cretaceous Hell Creek Formation (Lan- cian). Description. — The specimen UALVP 29840 (Fig. 24E, F) is a small right spleniodentary having nine tooth positions preserved. Although incomplete, the dentary is short, probably bearing no more than 15 teeth originally. The lateral surface of the jaw is smooth, and shows three tiny lateral inferior alveolar foramina that are equally spaced from one another. The ventral border of the dentary is strongly turned medially, so that the ventral splen- iodentary articulation (partially fused) is entirely medial to the ventral midline of the jaw. In medial view, the well-defined sub- dental shelf is at such a low position (similar to that in Conto- genys) that the Meckel ian canal is extremely restricted dorsoven- trally. In cross section, exposed posteriorly where the dentary is broken, the Meckelian canal is just slightly wider than the more laterally located inferior alveolar canal. The splenial is greatly narrowed, but not retracted, is medioventrally located, and par- tially fused to the dentary along the ventral spleniodentary artic- ulation below the anterior inferior alveolar foramen, which pen- etrates the splenial below probably the fourth or the fifth tooth position from the rear. Although the general proportions of this specimen are similar to those of the dentary of Contogenys, the teeth are obviously different. The preserved part of the dentary bears nine tooth po- sitions, including six teeth and the tooth bases for three others. These teeth are so closely spaced along the tooth row that little free space intervenes between successive teeth. The tooth attach- ment is pleurodont, with one-third of the tooth height projecting above the lateral parapet of dentary. The tooth shafts are cylin- drical, lacking either anteroposterior compression or medial ex- pansion. The crowns are nonstriated, and bluntly subconical, be- ing somewhat compressed bilaterally and slightly recurved, with faint anterior and posterior lateral ridges. Two teeth on the jaw have developed deep resorption pits medially at the base, sug- gesting a direct replacement (Rieppel, 1978) or the “iguanid method” of tooth replacement (Edmund, 1969), in which the replacement teeth are developed directly at the bases of old teeth. Discussion. — UALVP 29840 from the Hell Creek Formation represents a previously unknown lizard, which may be closely related to a Contogenys-Uke form occurring in the study area (see later discus- sion), and for this reason, this new specimen is in- cluded in this study. In terms of dentary structure and tooth form, the new specimen shows the fol- lowing character states that may be taxonomically significant: short spleniodentary with the splenial greatly narrowed and partially fused to the dentary; a strong tendency to enclose the dentary tube by spleniodentary fusion; and pleurodont teeth having cylindrical shafts, with subconical and nonstriated crowns. However, UALVP 29840 is not complete enough to be the name-bearer of a new species; therefore, it is appropriate to leave the new genus and species unnamed until more, better preserved specimens are found. The specimen UALVP 29840 shows a strong subdental shelf, a well-defined sulcus dentalis, and an indication of direct tooth replacement. These to- gether are indicative of a scincomorphan relation- ship for this lizard. However, the familial affinities of this species are somewhat uncertain, as the splen- iodentary is too incomplete to show reliable syna- pomorphies with a scincomorphan family. Both the shape of the dentary and the tooth form of this spe- cies show close similarities to those of Palaeoxan- tusia (see discussion below), indicating a possible affinity with the Xantusiidae. The partial fusion with the dentary of the greatly narrowed splenial shows a tendency to develop an enclosed dentary tube, which is characteristic of the Xantusiidae (see Estes et al., 1988). On the basis of these similarities, we tentatively classify this species in the Xantusi- idae, until further study on more nearly complete specimens can clarify its taxonomic position. The Tertiary Palaeoxantusia was originally named by Hecht (1956:4) as an Eocene “xantusiid lizard distinguished from the smaller members of the genus Xantusia by its smaller size, more robust dentary, its distinct splenio-dentary depression, and groove for the adductor mandibulae externus super- ficialis.” At present, the genus has four species and a geologic range from middle Paleocene to lower Oligocene (see Schatzinger, 1975, 1980; Estes, 1983a). The type species Palaeoxantusia fera (Hecht, 1956) was founded on a nearly complete spleniodentary (holotype AMNH 3815) from the mid-Eocene Bridger Formation, Sublette County, Wyoming. The spleniodentary is short and robust, with 13 tooth positions. “The teeth are pleurodont, conical, robust, and rounded at the apex. Below the teeth on the internal surface there is a distinct shelf with a groove” (Hecht, 1956:5). The specimen UALVP 29840 from the Hell Creek Formation shares several character states in jaw construction and tooth form with Palaeoxan- tusia, indicating its possible relationship to the Xan- tusiidae. However, UALVP 29840 differs from Pa- laeoxantusia in having a comparatively larger 1996 GAO AND FOX— LATE CRETACEOUS LIZARDS 55 splenial that is less extensively fused to the dentary, and in having the Meckelian canal open anteriorly. Our current knowledge on xantusiid evolution in- dicates that these are primitive character states with- in the family (Estes et al., 1988). According to Sav- age (1963), Palaeoxantusia is more primitive than Xantusia Baird, 1859, in morphology of the splen- iodentary, and may represent the ancestral stock from which the latter form is derived. From the available evidence, the newly discov- ered Hell Creek specimen is structurally more prim- itive than Palaeoxantusia. The ventral splenioden- tary articulation below the anterior inferior alveolar foramen is clearly fused without a suture, and this fusion can be regarded as the initial step from a more primitive condition towards the complete fu- sion of spleniodentary and the enclosure of the den- tary tube in the early evolution of the Xantusiidae. Therefore, UALVP 29840 may represent a basal taxon of the Xantusiidae, which in jaw structure is more primitive than Palaeoxantusia and Xantusia. Genus Contogenys Estes, 1969 b Type Species. — Contogenys sloani Estes, 1969 b. Range. — Upper Cretaceous, North America. Diagnosis. — As for the type species. Contogenys sloani Estes, 1969 b (Fig. 24G, H) Holotype. — MCZ 3681, nearly complete left den- tary having five teeth and the spaces for ten others (see Estes, 1969 b). Pciratypes. — MCZ 3682-3685, fragmentary maxillary and dentaries (Estes, 1969 b). Type Locality and Horizon. — BCA locality, SW 1/4, Sec. 9, Tp. 22, R 43, McCone County, eastern Montana; Upper Cretaceous Hell Creek Formation (Lancian). Included Specimen. — UALVP 29839, topotypic right dentary collected by a UALVP field party in 1969. This specimen is included here because it provides new information for a taxo- nomic revision of Contogenys sloani ; C. sloani appears to be phyletically close to a poorly known taxon from the study area (family incertae sedis, genus and species undetermined, see be- low). Known Distribution. — Upper Cretaceous Hell Creek Formation, eastern Montana. The referrals of several Paleocene specimens (YPM-PU and AMNH specimens) to this taxon are unconvincing (see later discussion). There is also a recent report of speci- mens of this genus from Judithian beds of New Jer- sey (Denton et al., 1991), but a taxonomic descrip- tion is as yet unpublished. Diagnosis (Revised from Estes, 1983-d). — A pos- sible xantusiid of Late Cretaceous age, character- ized by the following combination of character states: dentary short, with 16 or fewer teeth; tooth shafts anteroposteriorly compressed and medially expanded; tooth crowns truncate, having weak an- teroposterior apical groove flanked by crests; infe- rior alveolar canal distinctively wider than Mecke- lian canal; dentary coronoid process strongly pro- jecting posterodorsally. Description. — Estes (1969b) has provided a sufficient descrip- tion of the holotype (MCZ 3681) and the several Paleocene spec- imens. However, the newly collected topotypic specimen (UALVP 29839) reveals new information and provides the basis for emending the diagnosis of this taxon. Therefore, a description of the new specimen is necessary, in comparison with the orig- inal description of the holotype. The dentary UALVP 29839 (Fig. 24G, H) has only the anterior tip and the posteroventral corner missing; otherwise it is com- plete. Like the holotype, the new specimen is extremely short and relatively deep, having 15 positions in the complete tooth row, including three complete teeth, bases for two more, and spaces for ten others. In medial view, the subdental shelf is at a position close to the ventral border of the jaw; the ventral posi- tion of the shelf, combined with the widening of the inferior alveolar canal, results in a great restriction of the Meckelian ca- nal into an extremely narrow tube. The intramandibular septum is vertically directed but poorly developed, and has no free ven- tral border, differing from that in anguimorphans. The splenial, which is firmly articulated with the dentary, is greatly reduced dorsoventrally and terminates anteriorly below the sixth tooth position. The anterior inferior alveolar foramen penetrates the splenial below the fourth posteriormost tooth. The subdental shelf curves posterodorsally, ending at the last dentary tooth; immediately behind this tooth and medial to the small but prominent coronoid process of the dentary is a distinct facet that receives the anteromedial process of the coronoid bone (Estes, 1969b), which is not preserved. The subdental shelf lacks a dorsal ridge, and the sulcus dentalis is poorly defined (contra Estes, 1969b). The dentary teeth are pleurodont, having one-third of their height projecting above the high lateral parapet of the jaw. The tooth shafts are strongly compressed anteroposteriorly, with a weak, shoulder-like, medial expansion. Small basal foram- ina that open medially are clearly shown at the tooth bases. The crowns are squared-off, more or less chisel-like, with slight lat- eral compression. There are no medial striations, but faint ante- rior and posterior ridges are developed. The lateral surface of the dentary is strongly convex, reflecting the bulge of the inferior alveolar canal internally, which is wider than the greatly reduced Meckelian canal in posterior view. The coronoid process of the dentary is small, but prominent, and strongly projecting posterodorsally. Posterolateral to and below this process is a depression for the attachment of the adductor muscle, as in other xantusiids (Estes, 1969b). The posteroventral (Meckelian) process is broken off, but its large base indicates that the process is much stronger than, and extends posteriorly well beyond, the dorsal process. The lateral surface has a row of inferior alveolar foramina as seen in other lizards, but as a minor 56 BULLETIN CARNEGIE MUSEUM OF NATURAL HISTORY NO. 33 individual variant, at least two extra foramina are developed an- teriorly below the main row (the holotype MCZ 3681 shows a similar condition, see Estes, 1969£>:fig. la). Discussion. — Estes (19696) founded Contogenys sloani on several specimens (maxillary, dentary, and parietal materials) from the BCA locality, and assigned the species to the family Scincidae. Estes (19696) also referred several specimens (YPM-PU 17035, 17036a, 170366) from the Paleocene Tongue River Formation, Montana, to this species. The results of this study show that both the classi- fication of C. sloani in the Scincidae and the referral of the Paleocene YPM-PU specimens to this species are questionable. Sullivan’s (1982) referral of a maxillary from the ‘'Fort Union Formation” to this species is questionable as well, as it is based on an unreliable character. Familial Assignment of Contogenys. The orig- inal classification of Contogenys in the Scincidae was determined largely “on the basis of the Eu- meces-Y\ke wedge on the posterior border of the maxilla and on the general appearance of the teeth” (Estes, 19696:5). However, the following observa- tions indicate that the character states on which this placement is based are unreliable: first, the posterior extremity of the maxillary in primitive extant scin- cids (e.g., Eumeces inexpectatus, Mabuya sp.; per- sonal observation, and see Estes et al., 1988:fig. 25, Eumeces obsoletus) is marked by an anteriorly di- rected triangular notch defined by two posteriorly directed processes, one dorsal, one ventral; the dor- sal process fits into a triangular depression on the lateral surface of the jugal, so that the maxillary- jugal suture is actually Z-shaped in lateral view (personal observation). Contrary to Estes (19696), there is no triangular wedge of the maxillary that inserts into the jugal in primitive skinks; instead, the jugal inserts into the maxillary notch. Second, while the enlarged “triangular wedge” on the max- illary in Contogenys (MCZ 3682) appears to be the homologue of the more dorsal of the two processes in scincids, the ventral process, defining the lower border of the notch in scincids (see above), is “ab- sent” (Estes, 1983a: 106). Lacking well-preserved specimens, we do not know the pattern of articu- lation between the maxillary and jugal in Conto- genys other than it is not like that in primitive scin- cids; hence, the articulation between the maxillary and jugal cannot be invoked as evidence of the scin- cid affinities of Contogenys. Third, Contogenys has a short jaw with fewer, more or less chisel-shaped teeth, showing no resemblance to extant scincids (e.g., Eumeces, Mabuya), which have elongated jaws with numerous unicuspid teeth. In sum, there- fore, Estes’ (19836:370) il Eumeces- like aspect” of Contogenys is questionable, and cannot be used to support referral of this lizard to the Scincidae. We also have discovered that at least some cordylids (e.g., UALVP unnumbered specimen, Cordylus breyeri ) have a scincid-like articulation between the maxillary and jugal, effectively removing the basis for Estes’ (19696, 1983a, 19836) argument. In this study, we have compared Contogenys with the Tertiary Palaeoxantusia (see Hecht, 1956; Estes, 1983a); similarities between the two suggest that Contogenys is a xantusiid. They seem to share a unique jaw configuration and tooth form with extant xantusiids. They both have a short dentary with rel- atively few, more or less chisel-shaped teeth, each with a small basal foramen, and a small, but prom- inent, coronoid process of the dentary with a dis- tinct posterodorsal facet (see Estes, 1983a:fig. 3 IB). In both Contogenys and Palaeoxantusia, the den- tary has a posteroventral process (Meckelian pro- cess of Estes, 19696) that is much stronger than, and extends posteriorly well beyond, the coronoid process of the dentary. This condition is in contrast with that of scincids, in which the two processes are more or less equally developed, but is similar to that of extant Xantusia (see Estes et al., 1988:fig. 21) and a possible early xantusiid, Eoxanta Borsuk-Bi- alynicka, 1988, from the Upper Cretaceous of the Gobi Desert. A large posteroventral process is also seen in Gekkonidae, Pygopodidae, Cordylidae, Par- amacellodidae, and Amphisbaenia (personal obser- vation). If this is a primitive feature, it is retained in the Xantusiidae but lost in the Scincidae. The only notable difference in dentary construction be- tween Contogenys and Palaeoxantusia is the en- closed dentary tube in the latter; however, its Cre- taceous relative might be expected to have an un- fused dentary tube, but one showing the tendency towards fusion, as is the case with Contogenys. Fur- thermore, MCZ 3682, a paratype of Contogenys (Estes, 19696:fig. 2; 1983a:fig. 26D), shows a dis- tinct ventrally directed curvature at the posterior end of the supradental shelf. A similar condition is seen in extant Xantusia, but not in scincids Eumeces or Mabuya (personal observation). This adds anoth- er character state indicating the affinity of Conto- genys to the Xantusiidae, not to the Scincidae. As indicated in the above discussion, we have found no solid evidence that indicates association of Contogenys with the Scincidae, but have noted some resemblances to Palaeoxantusia and Xantusia. We believe that an enlarged posteroventral process 1996 GAO AND FOX— LATE CRETACEOUS LIZARDS 57 of the dentary and a posteroventral curvature of the supradental shelf are diagnostic features for a taxon that includes extant Xantusia and fossil Palcieox- antusia and C ontogeny s, that is, the Xantusiidae or the family plus its sister group Contogenys. There- fore, Contogenys is tentatively placed in the Xan- tusiidae in this paper, pending verification by a more extensive study of fossil and Recent materials. The Problem of Paleocene Contogenys. The YPM-PU specimens referred to Contogenys sloani by Estes (1969b) are several dentaries from the mid- Paleocene Tongue River Formation, Montana. These specimens, as Estes (1969b:fig. 4) figured, seemingly differ from the holotype (MCZ 3681) and other Late Cretaceous specimens of Contogenys in having columnar teeth with large basal excava- tion, striated crowns, a more prominent subdental shelf, and a deeper Meckelian canal that indicates a larger splenial. All of these differences are beyond the range of individual variation from current un- derstanding of the species. The short jaw with fewer teeth, and the nature of the coronoid and Meckelian processes of the dentary may suggest affinity with Contogenys at or above the generic level, but these specimens seem not referrable to C. sloani. Sullivan (1982) identified AMNH 12069 (from the mid-Paleocene “Fort Union Formation,” Wyo- ming) as a right maxillary, and referred it to “cf. Contogenys sloani , ” which Estes later (1983a) list- ed as “cf. Contogenys sp.” According to Sullivan (1982:999), the referral of this specimen to Conto- genys sloani is based “on the evidence of an ex- pansion of the maxilla posterolaterally that indicates the probable presence of a triangular wedge (as re- ported by Estes, 1969) as well as the presence of strongly pleurodont teeth with squared-off tooth crowns.” The so-called “triangular wedge” is not a diagnostic feature for Contogenys, nor for the Scincidae, as discussed before. Another feature on which Sullivan’s referral is based, “strongly pleu- rodont teeth with squared-off tooth crowns,” is not unique to Contogenys, but shared with taxa in other families (e.g., Xantusiidae, Anguidae). Sullivan’s (1982:text fig. 4) figure shows that the teeth on AMNH 12069 are not strongly anteroposteriorly compressed and lack a basal foramen, which is well developed in Contogenys sloani. The fragmentary nature of the specimen makes further comments dif- ficult; however, in our view the specimen is, in both size and tooth form, indistinguishable from Pa- laeoxantusia from the same locality. The Status of Paracontogenys Schatzinger ex Estes, 1983a. Schatzinger (1975) named Paracon- togenys estesi in his unpublished thesis, based on several specimens from the upper Eocene Mission Valley Formation, San Diego, California. Golz and Lillegraven (1977) cited this name in a faunal list, although the description of the taxon had not yet been published (hence the name was not valid). Es- tes ( 1983a: 1 10-1 1 1 ) figured the holotype (UCMP 113228), provided a diagnosis, description, and comments on the taxon, but cited the taxon name as “ Paracontogenys estesi Schatzinger 1975 in Golz and Lillegraven 1977.” Since Estes (1983a) is the first who made the name available and valid (ICZN, 1985), a correct citation of this name should be “ Paracontogenys estesi Schatzinger ex Estes, 1983a.” Paracontogenys is diagnosed as a scincid differ- ing from Contogenys in having a greater number of dentary teeth, a closed Meckelian canal, and medial striations on the tooth crowns (Schatzinger, 1975; Estes, 1983a). As Schatzinger (1975) noted, this Eocene lizard is closely similar to the Late Creta- ceous Contogenys', however, as is the case with Contogenys, its familial placement in the Scincidae is questionable, as no diagnostic characters for Scin- cidae are shown on the specimens; it may pertain to the Xantusiidae. Probably in respect to the faint medial striations on the tooth crowns as a taxonomically significant character, Estes (1983a: 111) commented that “a reference [of Paracontogenys ] to Xantusiidae is possible but improbable as tooth crown structure is more like scincids than xantusiids.” Now it is clear that this character is shared by several scincomor- phan families, and is also seen in some xantusiids (e.g., Schatzinger, 1980). This wide distribution of this character state, and its occurrence in Jurassic paramacellodids (see Estes, 1983a), suggests that possession of striations on the crown surface is probably symplesiomorphic for scincomorphans, but is not unique for the Scincidae. It seems to us that Paracontogenys is best referred to the Xantu- siidae. The most indicative character of Paracon- togenys in respect to its familial affinities is prob- ably its short jaw, with the fusion of the dentary tube extending posteriorly close to the posterior end of tooth row. This is characteristically the xantusiid condition, but is different from that in the Scincidae, in which the dentary is elongate and only the an- terior half of the tube is closed, with the posterior half covered by the splenial (personal observation). In sum, neither Contogenys nor Paracontogenys are likely members of the Scincidae, but are prob- ably referrable to the Xantusiidae. An incomplete 58 BULLETIN CARNEGIE MUSEUM OF NATURAL HISTORY NO. 33 survey of dentary structure indicates that the two genera share with the Xantusiidae the following character states: (1) the posteroventral process (or Meckelian process) of the dentary is larger overall than the coronoid process of the dentary, (2) the posteroventral process of the dentary extends pos- teriorly well beyond the level of the dorsal coronoid process, (3) the coronoid process of the dentary me- dially has a distinctive facet behind the tooth row (Hecht, 1956). The first two character states are probably primitive at a higher taxon level, while the third is unambigiuously unique for the Xantusiidae. Family ?Cordylidae Gray, 1837 The family Cordylidae, as defined by Romer (1956), consists of the Cordylinae and the Gerrho- saurinae, subgroups that were previously placed in two separate families (e.g.. Camp, 1923). The Cor- dylidae include approximately 70 extant species in about ten genera, which are confined to Africa and Madagascar in their geographic distribution. The monophyly of the Cordylidae by Romer’s definition is supported by evidence from chromo- somal analysis (Olmo and Odierna, 1980) and sev- eral anatomical synapomorphies (Estes et al., 1988), but the relationships of this family with the Jurassic Paramacellodidae Estes, 1983 927.880 (Fig. 25D) is the anterior half of a right dentary, which is broken immediately behind the tenth tooth, and also posteroventrally at the Meckelian canal. The specimen shows no indication of erosion through transportation, but displays several tooth marks, probably from a sharp-toothed predatory lizard or a small mammal. Medially, the dentary bears a robust subdental shelf, which is anteriorly nearly as deep as the height of the teeth but sharply diminishes posteriorly. The shelf has a sharp subdental ridge, which medially borders a narrow and deep sulcus dentalis along the tooth row. The mandibular symphysis is strong, with a robust symphysial process. The Meckelian canal is greatly narrowed anteriorly, to a fissure-like opening under the first four tooth po- sitions, and is widened posteriorly, corresponding to the deep- ening of the jaw and the diminishing of the subdental shelf; how- ever, the canal is not as widely open as in teiids, which have an hypertrophied splenial. The specimen has eight well-preserved teeth and the broken bases of two others, representing the first ten positions of the complete dentary tooth row. The teeth are pleurodont, and the lateral parapet is well developed to about half of the tooth height. Most of the teeth show a deep basal excavation, with part of the tooth base already having been resorbed. This type of basal ex- cavation implies a direct tooth replacement pattern (Edmund, 1969). The tooth shafts are anteroposteriorly compressed for the first three or four teeth, but become increasingly pillar-like to- wards the posterior end of the tooth row. The tooth crowns are bluntly conical, with a weak crest (or ridge) encircling the crown surface, and are faintly striated on the medial surface. In lateral view, the dentary is more or less wedge-shaped, hav- ing the horizontal dorsal border diverging posteriorly from the ventral border at an angle of 40-45 degrees. This particular jaw configuration to some extent resembles that of some Jurassic par- amacellodids (e.g., Becklesius, see Estes, 1 983zz: fig. 28). The lat- eral surface has five inferior alveolar foramina that are small and rounded. The first four are equally spaced from one another, but the fifth is separated from the fourth by a space twice the length of those between the more anterior foramina. Discussion. — As described above, SMNH PI 927.880 has a well-developed subdental shelf and sulcus dentalis, and a relatively large splenial as in- 1996 GAO AND FOX— LATE CRETACEOUS LIZARDS 61 dicated by the open Meckelian canal. On the basis of these characters, this lizard can be placed confi- dently in the infraorder Scincomorpha; however, its familial relationships cannot be determined at pres- ent, because no synapomorphic character states can be clearly recognized in this specimen. Neverthe- less, this new Lancian lizard differs from members of the Teiidae (one of the most common North American Late Cretaceous scincomorphan groups) in having a less heavy deposit of basal cementum, and a different tooth replacement pattern, as shown by deep and large basal excavations (compared to subcircular replacement pits in teiids). Instead, both the general jaw construction and the tooth form of this new lizard closely resemble those of Becklesius hoffstetteri (see Estes, 1983a:fig. 28), which is cur- rently classified in the family Paramacellodidae (Es- tes, 1983a). Paramacellodids are known only from the Upper Jurassic of Europe, particularly from En- gland and Portugal, and are possibly a paraphyletic group with exclusion of cordylids (Estes, 1983a). Both the geologic and geographic distributions of paramacellodids seem to hinder the referral of this species to the family. However, the above-men- tioned similarities shed some light on the possible affinity of SMNH PI 927.880 to the family Cordy- lidae, since the close relationship of Paramacellod- idae with Cordylidae has been suggested by resem- blances in mandibular construction and osteodermal morphology (Estes, 1983a). These similarities are so strong that Estes (1983a) predicted that “it is possible that the Paramacellodidae will prove a syn- onym of the Cordylidae” (Estes, 1983a: 115). Nevertheless, it appears that SMNH PI 927.880 represents a new taxon, which may have a relation- ship with the Cordylidae. However, we here leave the taxon unnamed and classify it as “family incer- tae sedis,” since the specimen is too incomplete to be designated as a holotype and its affiliation to the Cordylidae is less certain than the two questionable cordylid taxa described above. Genus and species new (B) (Fig. 25E) Specimen. — SMNH PI 927.886, P2004.90, both fragmentary dentaries having several teeth and bro- ken bases preserved. Locality and Horizon. — Gryde locality, in Sec. 19, Tp. 14, R 18, W 3, Frenchman River Valley, southwestern Saskatchewan; Upper Cretaceous Frenchman Formation (Lancian). Description. — SMNH PI 927.886 (Fig. 25E) is a fragmentary left dentary, which was probably originally short and deep, but quite robustly built in relation to its small size. The subdental shelf is mediolaterally narrow but dorsoventrally deep below the first eight tooth positions. This part of the subdental shelf is so deep that the ventral edge of the shelf is brought close to the ventral border of the jaw, depressing the anterior part of the Meckelian canal to form an extremely shallow and narrow ven- tral groove. The shelf posterior to this part is abruptly reduced in depth above what must have been a small splenial. The shelf has a subdental ridge that forms the medial border of the narrow but deep sulcus dentalis. The dentary is anteriorly pointed, in- dicating a weak mandibular symphysis. SMNH PI 927.886 has four teeth and the broken bases of six others preserved, representing the first ten positions of the com- plete tooth row. The teeth are high-crowned, closely spaced, and pleurodont, with the lower half of the tooth height attaching to the lateral parapet of the jaw. The tooth shafts are slender and cylindrical, neither compressed nor expanded. The tooth bases are clearly exposed and lack the heavy deposit of cementum seen in teiids. The bases also show no resorption pits or excavation. The crowns can be described as “incipiently tricuspid,” as the crown of the ninth tooth has a prominent central cusp and faint side cusps. The posterior cusp is slightly stronger than the an- terior one. The lateral surface of the dentary shows five inferior alveolar foramina, and these foramina are about equal in size and in spac- ing from one another. Like SMNH PI 927.880 (unnamed taxon A, described above), this dentary also shows tooth marks on both the medial and lateral sides of the jaw. Another specimen (SMNH P2004.90) is a left dentary fragment with four complete teeth. It is more fragmentary than PI 927.886, but is referrable to the same species on the basis of its close resemblances in crown pattern, tooth attachment, and the robustness of the subdental shelf. Discussion. — The two specimens described above represent a new genus and species, which is possibly referrable to the Cordylidae. Owing to the poor preservation of the specimens, the new taxon is unnamed, and hence, no holotype is designated and no diagnosis is presented. However, the unna- med taxon B seems to be different from taxon A above primarily in having high-crowned, closely compact, and incipiently tricuspid teeth, and in hav- ing a narrower Meckelian canal, besides its sub- stantially smaller size. These differences may be counted as character states that separate the two liz- ards at the generic level. However, similarities in general shape of the jaw, robustness of the subden- tal shelf, and the nature of the sulcus dentalis sug- gest that the two taxa should be placed in the same family, either Cordylidae or Paramacellodidae. As does SMNH PI 927.880 (unnamed taxon A, see above), SMNH PI 927.886 also shows a paleoe- cologically significant feature of tooth marks, prob- ably indicative of predation or post-mortem scav- enging in a subaerial environment. In addition, the breakage of the first six teeth on the jaw seems to 62 BULLETIN CARNEGIE MUSEUM OF NATURAL HISTORY NO. 33 have happened before fossilization, also implying action by a predator or scavenger. Genus and species undetermined (Fig. 25F, G) Specimens. — Milk River Formation, MR-4 local- ity: UALVP 29830, incomplete right maxillary hav- ing ten teeth preserved; Oldman Formation, BB 78 (DPP): RTMP 85.36.275, fragmentary left dentary having four teeth preserved. Description . — UALVP 29830 (Fig. 25F) consists of an in- complete right maxillary having ten teeth preserved. Although the dorsal process of the maxillary is broken, it shows a sharply descending posterior border, indicating a much shallower poste- rior part of the bone. Medially above the tooth row, the supra- dental shelf is obviously wider posteriorly than anteriorly, indi- cating a prominent palatine process. A large posterior interior alveolar foramen opens close to the midpoint of the tooth row. The maxillary teeth are pleurodont, with the upper half of the teeth attached to the lateral parapet of the jaw. The tooth shafts are strongly compressed anteroposteriorly, giving them a trian- gular shape in anterior or posterior view. Although strongly com- pressed, the tooth shafts lack a shoulder-like medial expansion. There are no foramina at the tooth bases; instead, several teeth have developed a basal excavation for tooth replacement. The tooth crowns are partially dissolved on this specimen, but enough remains to show that the crowns were more or less chisel-shaped originally, with faint lateral crests. The anterior maxillary teeth are slightly shorter and more slender than the posterior ones. RTMP 85.36.275 (Fig. 25G) is the posteriormost part of a left dentary, as indicated by its greatly diminished posterior end of the subdental shelf. The dentary fragment has four teeth pre- served, and the teeth are similar to those on UALVP 29830: they are slightly compressed laterally, more or less chisel-shaped, and have faint anterior and posterior ridges running vertically on the medial side of the crowns. There are no striations on the crown surface. Differences in crown height and other characters suggest that the two specimens may not pertain to the same species; however, this possibility cannot be verified at this stage, owing to the fragmentary nature of the specimens. Discussion. — The two specimens described above were collected from two localities at different geological horizons. They are lumped together as “genus and species undetermined” on the basis of their general resemblances in having anteroposteri- orly compressed tooth shafts, and more or less chis- el-like crowns without striations. These features are Contogenys- like, but poor preservation of the two specimens does not allow a reliable taxonomic re- ferral, even at a familial level. However, the two specimens may not pertain to the same taxon, as RTMP 85.36.275 has taller, and more strongly com- pressed teeth than does UALVP 29830. They also differ from each other in tooth attachment: RTMP 85.36.275 has half, while UALVP 29830 has one- third, of its tooth height projecting over the parapet of the jaw. Nevertheless, the two poorly preserved specimens are inadequate for us to clearly distin- guish them from one another, and to make a rea- sonable taxonomic assignment. Here, we place them together as “genus and species undetermined,” pending further study on more and better preserved specimens to clarify their taxonomic position. Infraorder Anguimorpha Ftirbringer, 1900 Family Xenosauridae Cope, 1886 The family Xenosauridae includes the living spe- cies of Central American Xenosaurus and Asian Shinisaurus (McDowell and Bogert, 1954; Estes et al., 1988). Both the skull osteology and the natural history of the two genera are still poorly known; and the monophyly of the family, although sup- ported by 12 synapomorphies (Estes et al., 1988), is still obscured by the wide geographic separation and distinct adaptations of the two genera (Gauthier, 1982; Pregill et al., 1986). The Xenosauridae have a poor fossil record com- pared with other anguimorphan families. With the exclusion of Nordenosaurus Holman, 1973, from Squamata (Gauthier, 1982), there are only three spe- cies in two genera of fossil xenosaurids known from North America (Estes, 1983a). In addition, unde- scribed materials were recently reported from New Jersey (Denton et al., 1991). Several fossil taxa from the Gobi Desert also show xenosaurid affini- ties (Borsuk-Bialynicka, 1985, 1986), but further study is needed to clarify their relationships. In this paper, new specimens from the Lance and Hell Creek formations provide important materials for a better understanding of Exostinus Cope, 1873; sev- eral specimens from the Milk River Formation rep- resent by far the earliest North American fossil rec- ord of this family, and may provide materials for future recognition of a new taxon. Exostinus lancensis Gilmore, 1928 (Fig. 26, 27A-F) Exostinus ? lancensis Gilmore, 1928:23, pi. 26, fig. 2. Harpagosaurus parvus Gilmore, 1928:156, fig. 98. Prionosaurus regularis Gilmore, 1928:159, pi. 26, fig. 1. Included Specimens. — Lance Formation, BTB locality: UALVP 29847, parietal; UALVP 29849-29850, incomplete maxillaries; UALVP 29838, dentary; UALVP 29920, jugal. AB locality: UALVP 29848, parietal. Hell Creek Formation, BCA locality: UALVP 29919, frontal. Although collected from outside the study area, these specimens are included in this paper because they represent an important but poorly known North American xenosaurid, which is closely related to the Oldman and Milk River forms known from the study area (see later description). Known Distribution. — Upper Cretaceous Lance 1996 GAO AND FOX— LATE CRETACEOUS LIZARDS 63 i i i i i i Fig. 26. — Exostinus lancensis, Lance Formation, Wyoming: UALVP 29838, nearly complete left dentary, lateral (above) and medial (below) views. Scale = 5 mm. Abbreviations as in Fig. 2. Formation, eastern Wyoming; Hell Creek Forma- tion, eastern Montana (all Lancian). Diagnosis (Revised from Estes, 1983a). — A Late Cretaceous xenosaurid from North America, differ- ing from the Oligocene Exostinus serratus by the following combination of character states: prefron- tal incision of frontal not extending to midpoint of orbit; subdental shelf well defined, sulcus dentalis present; intramandibular septum less extensive pos- teriorly; posterior interior alveolar foramen more anteriorly located. Description. — Gilmore (1928) described the holotype maxil- lary (USNM 10689), and Estes (1964) briefly described the den- tary specimens of Exostinus lancensis. Newly discovered speci- mens from the Lance and Hell Creek formations include a fron- tal, a jugal, and the anterior end of a maxillary, parts that pre- viously were unknown for this species. These specimens provide important information for a better understanding of this fossil xenosaurid, and hence are described below. Maxillary. Two maxillaries (UALVP 29849, 29850), both from the Lance Formation, are referrable to Exostinus lancensis on the basis of their tooth form and the sculpture on the lateral surface. UALVP 29849 is the anterior part of a right maxillary, having six tooth positions preserved. The lateral surface of the maxillary is extensively ornamented with flattened patches of vermiculate sculpture that are separated by shallow grooves (Fig. 27A). At the lower border of the sculpture, the superior alveolar foramina are small and circular, as in the holotype (USNM 10689; see Estes, 1964). The maxillary has a prominent premax- illary process, behind which the vertical nasal process is slightly notched for the posterior border of the external narial opening; it then gently turns posterodorsally. In medial view, the vomerine process of the maxillary is stronger than the premaxillary process and is somewhat plate-like, with its widened base as the anterior extension of the supradental shelf. This medial process diverges from the lateral premaxillary process at a 45-degree angle. The supradental shelf is clearly defined, and a slight supradental gut- ter can be seen. The six tooth positions preserved consist of five teeth and the space for another. Most of the teeth are broken, but the third is complete, showing that the crown is unicuspid, slightly recurved, and has a longer and trenchant anterior cutting edge; no striations are present. The tooth shafts are subcylindrical and thick-walled, and the tooth bases are firmly cemented to the lateral parapet. These teeth are pleurodont, having about two-thirds of their height attached to the parapet. Of the five teeth preserved, three clearly show a resorption cavity. The third tooth has almost half of its base resorbed, indicating a direct replacement mode (in agreement with Estes, 1964; but in conflict with McDowell and Bogert, 1954, who stated that xenosaurs have interdental replace- ment). Another specimen, UALVP 29850 (Fig. 27B), is the posterior end of a left maxillary, which is broken at the posterior interior alveolar foramen in front of the fifth tooth position from the back. Although fragmentary, this specimen shows the same tooth mor- phology and osteodermal ornamentation as the holotype maxil- lary described and figured in Gilmore (1928). This fragment also preserves a structure unknown from other specimens: the pos- teriormost part of the supradental shelf develops a sigmoid twist, and the last two maxillary teeth are sharply reduced in size cor- responding to the twist of the shelf (see Fig. 27B). This down- ward twist of the shelf seems to have a function in enhancing the maxillary-jugal articulation. Frontal. UALVP 29919, a fragmentary frontal (Fig. 27C) from the Hell Creek Formation, is clearly referrable to Exostinus lancensis. as it is fused along the midline and hourglass-shaped, and is ornamented with vermiculate osteoderms. This frontal dif- fers from that of Restes rugosus (Gilmore, 1942a) in having a prominent crust of tuberculate osteoderms (rather than flattened tesserae), and differs from that of Exostinus serratus Cope, 1873 (type species) in having larger and fewer tuberculate osteoderms, with a less complex surface texture. The ventral view of the specimen shows that the frontal fusion is complete, with the mid- line suture barely recognizable. The right subolfactory process is entirely broken off, but the left side indicates that the two pro- cesses tend to meet at the ventral midline and hold a trough-like subolfactory tract. In lateral view, the left anterior corner of the bone has a clearly defined prefrontal incision, the posterior ex- tension of which does not reach the midpoint of the orbit. Parietal. Two parietals (UALVP 29847 from the BTB locality and UALVP 29848 from the AB locality) are referred to Exos- tinus lancensis. both having osteodermal sculpture that is similar to the holotype maxillary; presence of “vermiculate sculpture” is a synapomorphy of the Xenosauridae (Estes et al., 1988). UALVP 29847 (Fig. 27D) shows a widened lateral border, im- plying suppression of the supratemporal opening and a short su- pratemporal process. UALVP 29848 shows that the parietal is a fused, single unit. Both specimens have osteodermal ornamen- tation that is heavy anteriorly but diminished posteriorly. Jugal. UALVP 29920 (Fig. 27E) from the Lance Formation is the only jugal known that can be referred to Exostinus lan- censis. The large size of this jugal indicates an individual larger than those represented by the other elements referred to this spe- cies, but it is of a characteristically xenosaurid type, with the postorbital branch widened and sculptured. The sculpture pattern 64 BULLETIN CARNEGIE MUSEUM OF NATURAL HISTORY NO. 33 pjg 27. Exostinus lancensis, Lance and Hell Creek formations, Wyoming and Montana: A, UALVP 29849, incomplete right maxillary, lateral view; B, UALVP 29850, fragment of left maxillary, medial view; C, UALVP 29919, incomplete frontal, dorsal view; D, UALVP 29847. incomplete parietal, dorsal view; E, UALVP 29920, left jugal, lateral view; F, UALVP 29838, nearly complete left dentary, medial view. 1 Exostinus sp„ Oldman Formation, Alberta: G, UALVP 944, fragment of fused frontal, dorsal view. Xenosauridae, genus and species undetermined. Milk River Formation, Alberta: H, UALVP 29914, incomplete right jugal, lateral view; I, UALVP 29915, left maxillary fragment, lateral view; J, UALVP 29917, right maxillary fragment, medial view. Scale = 2 mm. 1996 GAO AND FOX— LATE CRETACEOUS LIZARDS 65 matches that on the frontal and parietals; therefore, these disar- ticulated specimens can be referred to the same species in spite of their size differences. The jugal is large, wide, and heavily ornamented with tuberculate osteodermal mounds on the lateral surface. Similar, but much smaller, osteodermal mounds coalesce along the posterior and ventral border of the orbit. The antero- ventral process is elongated and thickened, forming most of the ventral border of the orbit. Although incomplete, the anterior end of this process bears a ventral crest, which fits into the posterior groove of the maxillary above the supradental shelf. The dorsal process of the jugal is largely broken off, but the base shows that it was widened and extensively ornamented (a synapomor- phy of the Xenosauridae, see Estes et al., 1988). Dentary. The dentary structure of Exostinus Umcensis is not well known, except for Estes’ (1964) brief description. UALVP 29838 (Fig. 26, 27F) from the BTB locality represents by far the most nearly complete dentary for this species, and provides the basis for a more extensive description of the dentary than was possible before. The dentary is elongate, lightly built, and rela- tively straight in dorsal view. The lateral surface is smooth and only weakly convex at the posterior part of the bone. Nine un- equally spaced inferior alveolar foramina open laterally, with the posteriormost one located lateral to the 15th tooth position. Me- dially, it bears a horizontally straight, slender, but clearly defined subdental shelf, which has a subdental ridge bordering a shallow sulcus dentalis. The shelf gradually increases in depth anteriorly, and reaches a maximum depth about one-third of the tooth height. The Meckelian canal is anteriorly narrow and open ven- trally below the first eight tooth positions; it increases in depth and width posteriorly, indicating that a splenial covered the most ventromedial aspect of the dentary. The maximum depth of the Meckelian canal is about equal to the tooth height. Both dorsal and ventral spleniodentary articulation surfaces terminate ante- riorly under the ninth tooth position. The large posterior interior alveolar foramen opens into the Meckelian canal under the 18th and 19th teeth. As Gauthier (1982) has noted, this location is more anterior than that in the Oligocene Exostinus serratus. The intramandibular septum (sensu Estes et al., 1988) retains a prim- itive configuration: the septum is obliquely directed and lacks a free ventral border. This condition is similar to that in extant Xenosaurus grandis (see Gauthier, 1982: fig. 44) and some skinks (e.g., Mabuya macularia, personal observation), but is different from that in advanced anguids, in which the septum is vertically directed and has a free ventral border (Estes, 1964). The dentary contains a nearly complete tooth row, consisting of 22 tooth positions (12 well-preserved teeth and tooth bases as well as spaces for ten others). We estimate that only the last one or two teeth are missing. The dentary teeth are closely spaced and pleurodont, with slightly less than one-third of the tooth height projecting above the lateral parapet. The tooth shafts are subcylindrical, lacking either compression or expansion. The tooth crowns are slightly compressed laterally and recurved, with the leading cutting edge much longer than the posterior one, giving the teeth a dagger-like appearance. The crowns are not serrated along the cutting edge, but they are faintly striated on the medial surfaces of at least some teeth (in agreement with Estes [1964], who described the teeth as having “faint lingual striations”). Although several teeth have their anterior cutting edge chipped, as if they were “bicuspid,” well-preserved teeth clearly show no indication of a second cusp or cuspule on the crowns. This observation is contrary to Estes (1964) and Gau- thier (1982), who described the teeth as “slightly bicuspid” or “incipiently bicuspid” (Estes [1965:104] suggested that “the condition is better described as bilobate, as in many Anguidae’ ). The five anteriormost teeth are not preserved, but the remainder show that the dentary teeth of this lizard are homodont (isodont), without differentiation of tooth form along the tooth row, except that the anterior ones are slightly more slender than those more posterior. Several teeth on the specimen clearly show that deep replacement pits are developed lingually in the tooth bases, im- plying a direct tooth replacement mode (Edmund, 1969). Discussion. — The family Xenosauridae is a prob- lematic taxon, and the evolutionary history of the family is poorly known because of its scant fossil record. The North American record of the family includes only three species in two genera: the Oli- gocene Exostinus serratus, the late Paleocene and early Eocene Restes rugosus, and the Late Creta- ceous Exostinus lancensis. These fossil taxa are based mostly on disarticulated materials that are an- atomically not comparable between the species. Gauthier (1982) separated Restes rugosus from Ex- ostinus based on “its unique cephalic osteoderms” and pointed out that Exostinus lancensis could well be generically distinguishable from the type species Exostinus serratus. On the basis of new evidence revealed from the newly discovered specimens, we offer the following comments on this important but problematic taxon. The genus Exostinus Cope, 1873 (see also Cope, 1884, 1886) includes two species: the type species E. serratus (see Cope, 1873) and E. lancensis (see Gilmore, 1928). Gauthier (1982) suggested that E. lancensis should be removed from Exostinus be- cause of the retention of a sulcus dentalis and less extensive intramandibular septum, while Estes (1983a) explained the difficulty in taking such a course, owing to the lack of materials that are an- atomically comparable between the two species: the type series of E. serratus consists of a frontal, a jugal, and a dentary, while a frontal and jugal were unknown for E. lancensis, the holotype of which is a maxillary. Now, with the discovery of the frontal (UALVP 29919) and the jugal (UALVP 29920) that are of “ E . lancensis- type,” Gauthier’s suggestion seems to be more plausible. The newly recovered frontal shows that the prefrontal incision does not extend to the midpoint of the orbit, and, hence, is significantly different from that of the Oligocene E. serratus or that of the Paleocene-Eocene Restes ru- gosus. The jugal and dentary specimens of the Lan- cian “E. lancensis ” are also significantly different from the same elements of E. serratus, as described above. These differences are apparently great enough to separate the Lancian species from the Oligocene E. serratus at a generic level; therefore, 66 BULLETIN CARNEGIE MUSEUM OF NATURAL HISTORY NO. 33 a new generic name should be proposed for the Lancian form. However, it is premature to make such a revision in this paper, before the holotype of the type species Exostinus serratus is carefully ex- amined. Therefore, Exostinus lancensis is retained in this paper as a valid name. Estes et al. (1969) referred an unfused frontal (MCZ 3662b) from the Hell Creek Formation to Exostinus lancensis. The specimen was not figured, but the description of it raises some uncertainty about this identification: Estes et al. (1969:20) first claimed that “the frontal (MCZ 3662b) may be re- ferable to E. lancensis on the basis of dermal sculp- ture pattern,” then stated that the frontal “does not display a sculpture pattern” because it is eroded. With the discovery of new frontal material from the same horizon of definitely Exostinus lancensis mor- phology, the identification of MCZ 3662b is uncer- tain. In extant xenosaurids, the frontals are fused in the embryo (Estes et al., 1988). This, together with new fossil evidence, indicates that the fused frontal is a primitive character state within the Xenosauri- dae. Estes (1964) misspelled Harpagosaurus Gilmore, 1928 as “ Marpagosaurus ” in the caption for his plate 5, and later (1983a: 130) listed “ Marpagosau- rus parvus ” as a synonym of Gilmore’s Harpago- saurus parvus (= Exostinus lancensis Gilmore, 1928). In this particular case, Estes’ “ Marpagosau- rus” is an “incorrect subsequent spelling” (not an available name), that has no standing in zoological nomenclature, and hence, should not be recorded in synonymy (ICZN, 1985:article 33c). lExostinus sp. (Fig. 27G) Referred Specimen. — UALVP 944, a small but completely co-ossified frontal. Locality and Horizon. — Outcrop of the Oldman Formation, in Lsd. 10, Sec. 29, Tp. 21, R 12, W 4 (near the Railway Grade locality), southern Alberta; Upper Cretaceous. Description. — UALVP 944 (Fig. 27G) is the posterior part of a fused frontal, and is the only specimen of this morphology known from the Oldman Formation. Although small, the frontal is so well co-ossified as a single unit that the ventral surface shows no midline suture. The subolfactory processes are broken on both left and right sides, but the broken bases suggest that the processes were ventrally directed and probably not in contact at the midline. In dorsal view, the specimen shows important characters on which its taxonomic assignment is based: it is strongly constricted between the orbits, giving an hourglass shape to the bone; and small tuberculate osteoderms with ver- miculate sculpture are fused to its external surface. These osteo- dermal mounds are all about equal in size and are separated from one another by deep and wide grooves. The specimen is too incomplete to show the prefrontal or postfrontal incision. Discussion. — UALVP 944 is the first xenosaurid specimen known from the Oldman Formation, which is about 1 1 Myr older than the Lance and Hell Creek formations (Thomas et al., 1990), from which Exostinus lancensis is known. The different osteodermal sculpture clearly distinguishes the Ju- dithian morph from the Lancian E. lancensis, and in keeping with the wide geologic gap between the two, is probably taxonomically significant. The osteodermal sculpture of the Judithian form is structurally more similar to that of the Oligocene Exostinus serratus (see Gauthier, 1982:fig. 7) than to either Lancian E. lancensis or the Paleocene Res- tes rugosus (see Gauthier, 1982:fig. 5). This condi- tion may suggest that two different lineages are in- volved in the evolutionary history of the Xenosaur- idae. However, the fossil record for these taxa is so poor that many taxonomically important characters are unknown as yet. Clarification of their phyloge- netic relationships must await discovery of more nearly complete material. Sahni (1972) reported two xenosaurids from the Judith River Formation by referring AMNH 8497 to “ Exostinus lancensis ” and AMNH 8498 to “cf. Exostinus sp.” The former specimen, AMNH 8497, is too fragmentary to provide a valid basis for its referral to Exostinus lancensis, and seems to be in- distinguishable in tooth morphology from AMNH 8491 (Sahni, 1972:fig. 8K), a maxillary that Sahni referred to Leptochamops denticulatus (see relevant discussion above). The latter specimen (AMNH 8498) is a left maxillary with the tooth crowns bro- ken off. However, the specimen has cylindrical tooth shafts without a basal expansion (Sahni, 1972: fig. 8F), and is similar in that respect to Exostinus. The lateral surface has vermiculate rugosities, but the particular sculpture pattern is somewhat differ- ent from that of E. lancensis (compare Sahni, 1972: fig. 8E with Estes, 1964:pl. 5). Therefore, it is pos- sible when more nearly complete specimens are found that the Judithian form may prove to be ge- nerically different from Exostinus (see discussion below). Genus and species undetermined (Fig. 27H-J) Specimens. — UALVP 29914, jugal; UALVP 29915-29918, incomplete maxillaries; and several unnumbered specimens (fragmentary premaxillary, maxillaries, and jugals). 1996 GAO AND FOX— LATE CRETACEOUS LIZARDS 67 Locality and Horizon. — MR-6 locality. Verdigris Coulee, approximately 30 km east of the village of Milk River, southern Alberta; upper member of the Milk River Formation (Aquilan), Upper Cretaceous. Description. — UALVP 29914 (Fig. 27H) is an incomplete right jugal with both of its posterodorsal and anteroventral pro- cesses broken. The jugal is substantially smaller than UALVP 29920 assigned to Exostinus lancensis (see above), but is pro- portionally compatible with the maxillaries described below from the same locality and horizon. Unlike the Lancian Exostinus lan- censis and extant xenosaurs, the posterodorsal process of this jugal is narrow, probably the primitive condition. However, the lateral surface is of xenosaurid-type, with vermiculate rugosities on the postorbital branch of the bone. These rugosities are less prominent and arranged in a more random pattern than those in Exostinus lancensis, which are concentrated to form tuberculate mounds. Four maxillaries (UALVP 29915-29918) are tentatively iden- tified as "Xenosauridae, genus and species undetermined,” ba- sedon their characteristic osteodermal ornamentation on the lat- eral surface. UALVP 29915 (Fig. 271) is the anterior part of a left maxillary, with three broken teeth and vacant spaces for two others. Part of the dorsal process is preserved; the lateral surface is extensively ornamented with vermiculate sculpture. The sculp- ture is as extensive as, and is similar in pattern to, that of Ex- ostinus lancensis (compare with UALVP 29849; Fig. 27A), but the tubercles are less prominent and less concentrated than in that species. Another specimen (UALVP 29918), the anterior part of a right maxillary, shows the same sculpture pattern. As in the Lancian Exostinus lancensis, the superior alveolar foramina are small, circular, and equally spaced. UALVP 29916, 29917 are the posterior parts of maxillaries, both showing a small palatine process associated with a small posterior interior alveolar foramen. The tooth form of this Milk River xenosaurid is best shown in UALVP 29917 (Fig. 27J), which has two complete teeth preserved. The two teeth are pos- terior to the posterior interior alveolar foramen and probably rep- resent the sixth and fifth positions from the back. The teeth are pleurodont and slightly recurved. The crowns are trenchant, with sharp anterior and posterior blades, but are not serrated. As in other anguimorphans, the tooth bases bear a small medial fora- men. The bases are not expanded and also have no infoldings, differing in these respects from those in the Varanoidea. Discussion. — The specimens described above from the upper member of the Milk River Forma- tion represent an early fossil record for the family Xenosauridae, as the age of the formation has been determined as Aquilan (i.e., early Campanian; see section on geological setting above). The lower tax- onomic assignment of this early xenosaurid is dif- ficult, because all the specimens known are frag- mentary and other fossil taxa of the same group are still poorly understood (see above discussion). Mor- phologically, the Milk River jugal (UALVP 29914) is clearly distinguishable from those of the Lancian Exostinus lancensis and Oligocene E. serratus by its narrow, less heavily sculptured postorbital branch. The sculpture pattern on the lateral surface of the maxillaries (UALVP 29915-29918) closely resembles that on the maxillary that Sahni (1972) described from the Judith River Formation, eastern Montana, but is different from that in the Lancian E. lancensis. The resemblance may suggest that the Milk River and Judith River materials are conge- neric; however, a reliable taxonomic assignment of these fossils must await the discovery of better pre- served specimens. Finally, it must be pointed out that there is no direct evidence for the association of the Milk River jugal with the maxillaries discussed above. These materials are tentatively placed together, leaving only one xenosaurid lizard recognized from the Milk River Formation. However, further study on more extensive materials could well change this as- sociation. Family Anguidae Gray, 1825 The family Anguidae includes the so-called glass lizards, alligator lizards, and related species. Most members of the family are heavily scaled or armour- plated, and have a lateral fold of body squamation (McDowell and Bogert, 1954) “which allows for distention of their otherwise inflexible skin when they are carrying eggs or have eaten a large meal” (Mattison, 1989:167). A strong tendency within the family is towards limb reduction or loss, and some are snake-like in the complete loss of limbs (for other recognized synapomorphies, see Estes et al., 1988). The Anguidae contain some 80 extant spe- cies, and have a geographical distribution in North and South America and the West Indies, and in Eu- rope, central, south, and southeast Asia, and north- westernmost Africa. All species are carnivorous, and their prey includes insects, small mammals, oth- er lizards, and molluscs. The family has its best fos- sil record from North America and Europe (see Es- tes, 1983a:fig. 65). Genus Odaxosaurus Gilmore, 1928 Type Species. — Peltosaurusl piger Gilmore, 1928. Range. — Upper Cretaceous, North America. Sev- eral Paleocene and Oligocene specimens were pre- viously referred to Odaxosaurus , but lately have been placed in Proxestops Gauthier, 1982, or Ma- chaerosaurus Gilmore, 1928 (see discussion be- low). Accordingly, the geologic range of Odaxo- saurus may not extend to the Paleocene; for pre- Lancian Odaxosaurus , see later discussion. Diagnosis ( after Estes, 1983^:147). — “Differs from other glyptosaurine lizards in having paired 68 BULLETIN CARNEGIE MUSEUM OF NATURAL HISTORY NO. 33 Fig. 28. — Odaxosaurus piger, Scollard (Alberta), Frenchman (Saskatchewan) and Lance (Wyoming) formations: A, UALVP 33381, incomplete left dentary, medial view; B. UALVP 33382, incomplete left maxillary, medial view; C, D, UALVP 29851, incomplete right maxillary, lateral and medial views; E, UALVP 29852, nearly complete left dentary, medial view. Scale = 2 mm. frontals (which may fuse in Paleocene individuals if correctly referred); vermiculate (nontubercular) osteodermal sculpture; rectangular osteoderms with poorly developed lateral bevels; teeth obtuse, shafts expanded transversely, crowns squared off with an anteroposteriorly oriented but slightly oblique cut- ting edge striated lingually and labially” (but see discussion below). Odaxosaurus piger (Gilmore, 1928) (Fig. 28) Peltosaurus ? piger Gilmore, 1928:136, pi. 26, fig. 4, 6. Odaxosaurus obliquus Gilmore, 1928:158, pi. 26, fig. 3, 5. Peltosaurus piger Estes, 1964:1 19. Pancelosaurus piger Meszoely, 1970:105. Holotype. — USNM 10687, incomplete right den- tary with six teeth. Collected by O. A. Peterson in 1889, from “Peterson’s quarry,” Lance Creek, Ni- obrara County, Wyoming (Gilmore, 1928). Referred Specimens. — Scollard Formation. KUA-3 locality: UALVP 33381, dentary; Frenchman Formation, Wounded Knee locality: UALVP 33382, maxillary. Known Distribution. — Upper Cretaceous Lance Formation, eastern Wyoming; Hell Creek Forma- tion, eastern Montana; Laramie Formation, Colo- rado (Carpenter, 1979); Scollard Formation, central Alberta; and Frenchman Formation, southwestern Saskatchewan (all Lancian). Some specimens of Ju- dithian and Paleocene age have been referred to this species, but none of the referrals are convincing (see later discussion). Diagnosis. — A Lancian Odaxosaurus differing from the new species described below in the fol- lowing combination of character states: dentary teeth crowded along tooth row; tooth crowns strongly truncate, heavily striated both medially and laterally; tooth shafts strongly expanded transverse- ly, with well-developed medial shoulder-like expan- sion, rectangular in cross section; intramandibular septum short, deep, not extending anteriorly to mid- point of tooth row. Description. — UALVP 33381 (Fig. 28 A) from the KUA-3 lo- cality (Scollard Formation) is an incomplete left dentary with seven tooth positions preserved (three teeth and broken bases as well as vacant spaces for four others). This specimen is referred to Odaxosaurus piger as it shows the same tooth form and den- tary structure as the type Lance specimens (see Gilmore, 1928; Estes, 1964). As in the Lance specimens, this dentary bears pleu- rodont teeth, which have subrectangular bases, anteroposteriorly compressed and medially expanded shafts, and chisel-shaped crowns with a horizontal cutting edge and vertical striations. These are the diagnostic characters for Odaxosaurus piger (Gil- more, 1928; Estes, 1964). The dentary is broken posteriorly at the anterior inferior alveolar foramen, and hence, the notch for the foramen on the tooth-bearing border is not preserved, nor is the intramandibular septum posteriorly within the Meckelian ca- nal. UALVP 33382 (Fig. 28B) from the Wounded Knee locality (Frenchman Formation) is a left maxillary with nine tooth po- sitions (seven teeth and broken bases for two others). The referral of this maxillary to Odaxosaurus piger is principally on a large 1996 GAO AND FOX— LATE CRETACEOUS LIZARDS 69 posterior interior alveolar foramen that is associated with a large palatine process. These characters are distinct for Odaxosaurus, and are different from gerrhonotines, which have a much smaller or no palatine process and have two posterior interior alveolar foramina (Meszoely, 1970; personal observation). As a result of erosion, the tooth crowns show no horizontal cutting edge and vertical striations, but enough remains to show that the teeth are squared off and laterally compressed. The tooth bases are like those of specimens from the Lance, subrectangular in shape and bearing a small basal foramen. In addition to the specimens described above, some 13 max- illaries and 27 dentaries from the Lance and Hell Creek forma- tions are included in this study because these specimens provide new information for a better understanding of Odaxosaurus pig- er, and provide comparative materials for the study of the Scol- lard and Frenchman specimens. The Lance specimens were col- lected from the BTB locality, and include one maxillary (UALVP 29851) and five dentaries (UALVP 29852-29856). The Hell Creek specimens were from the BCA locality, and include 12 maxillaries (UALVP 29857-29861, 29879-29885) and 22 well- preserved dentaries (UALVP 29862-29878, 29886-29890). UALVP 29851 (Fig. 28C, D) is an incomplete right maxillary having four teeth preserved. Although fragmentary, the specimen is of importance as it shows unworn tooth crowns and osteo- derms on its lateral surface. The marginal teeth of this lizard have the characteristic combination of a chisel-shaped lateral cut- ting edge (previously known) and a remnant medial cusp (un- known before). A well-preserved dentary (UALVP 29852, see below) shows the same crown pattern, while most of the other specimens have no medial cusp as the result of wear. The tooth crowns are striated both lingually and labially, but the lingual striations are obviously stronger than those on the labial side. An osteoderm is fused to the lateral surface of the maxillary (a sim- ilar structure is seen on UALVP 29887, a dentary from the BCA locality). The osteoderm is thick and tuberculate, and is different from those on the previously referred parietals and frontals (see Estes, 1964, 1983a), which are thin plates with a gerrhonotine- diploglossine type of sculpture (Estes, 1964; Meszoely, 1970). The crown pattern of the dentary teeth is best shown on UALVP 29852 (Fig. 28E), an incomplete left dentary having nine teeth and the spaces for four others. As in the maxillary described above, the teeth of this specimen show a combination of a chisel-like lateral cutting edge with a remnant medial cusp. However, this crown pattern varies along the tooth row: the slightly procumbent anterior teeth have a less well-developed cutting edge but a stronger medial cusp, while the posterior teeth, which are shorter than the anterior ones, have a stronger cutting edge and a weaker medial cusp. The Meckelian canal is anteri- orly restricted to form a shallow and ventrally facing groove, but gradually turns more ventromedially in the posterior half of the dentary. The ventral edge of the strongly reduced subdental shelf bears a lateral extension in front of the anterior inferior alveolar foramen, forming a flat surface for attachment of the splenial. Discussion. — Odaxosaurus piger has a complex taxonomic history, one in which its taxonomic name has been changed several times. Gilmore (1928:136, 158) named Peltosaurus ? piger and Odaxosaurus obliquus based on several specimens from the Lance Formation, eastern Wyoming. Estes (1964) synonymized Odaxosaurus obliquus with Peltosaurusl piger, and firmly referred the species to Peltosaurus Cope, 1872; meanwhile, he also pro- vided solid evidence for the familial assignment of Peltosaurus piger to the Anguidae from the con- struction of the dentary. More recently, Meszoely (1970) removed the species from Peltosaurus based on differences in jaw and osteodermal sculpture, and named a new genus, Pancelosaurus, for Pel- tosaurus piger. Meszoely et al. (1978) published the most recent revision of the genus, and finally clar- ified the taxonomic status of this species. They re- viewed all of the nomenclatural controversy, restu- died the holotype of Odaxosaurus obliquus, and came to the following conclusions: 1 ) Gilmore’s Odaxosaurus obliquus is indistinguishable from Peltosaurus ? piger, 2) Gilmore’s Peltosaurus ? piger is generically different from Peltosaurus, and should be removed from the genus; 3) Pancelosau- rus Meszoely, 1970, is a synonym of Odaxosaurus Gilmore, 1928; and 4) according to ICZN rules, both the generic name “ Odaxosaurus ” and specific epithet “ piger ” have priority; thus, the proper name for this lizard is Odaxosaurus piger. Gilmore (1928) classified Odaxosaurus piger (his “ Peltosaurus ? piger') in the Anguidae, and first noticed its similarity in tooth morphology to Glyp- tosaurus Marsh, 1871. This familial placement was questioned by McDowell and Bogert (1954), but has been solidly confirmed by Estes (1964) on the basis of dentary structure. However, the taxonomic status of Odaxosaurus piger at the subfamilial level is still in dispute. The family Anguidae is common- ly subdivided into four subfamilies: Anguinae, Glyptosaurinae, Gerrhonotinae, and Diploglossinae (see, e.g., McDowell and Bogert, 1954; Meszoely, 1970; see also Estes, 1983a, for a different classi- fication). Meszoely (1970) recognized the basal po- sition of Odaxosaurus in anguid phylogeny, and re- ferred it to the subfamily Anguinae on the basis of its Ophisaurus-Wke osteoderms. Sullivan (1982) proposed a subfamilial name Odaxosaurinae for Odaxosaurus piger, separating this species from Glyptosaurinae on the basis of its nontubercular os- teoderms (but see description above). This idea is further developed in his later publications (see Sul- livan, 1986a, 1986b). After careful evaluation of the character states of this lizard, Estes (1983a) put greater weight on tooth morphology than sculpture pattern of the osteoderms, and classified O. piger in the Glyptosaurinae, as its most primitive member (Estes, 1983a: 148). The evidence that Sullivan (1982, 1986a, 1986b) used to separate Odaxosaurus from the Glyptosaur- inae as a subfamily is quite problematic: first, the 70 BULLETIN CARNEGIE MUSEUM OF NATURAL HISTORY NO. 33 holotype of Odaxosciurus piger is a dentary with teeth, not a skull element with osteoderms; thus, characters of tooth morphology and dentary struc- tures must be relied on for interpreting the relation- ships of this lizard. Second, the referral of those parietals and frontals having nontuberculated sculp- ture to Odaxosaurus piger is questionable, as two jaw specimens (UALVP 29851, 29887) show osteo- dermal sculpture different from those on the re- ferred parietals and frontals (thick tuberculate mounds vs. thin plates with a vermiculate sculp- ture). The referral of the Lance disarticulated pari- etals and frontals to O. piger is based mainly on size and relative abundance of individuals (see Meszoely, 1970:105). Estes (1964:122) pointed out that "the pitted and ridged sculpture pattern of these osteoderms resembles most Recent Gerrhonotus .” Meszoely (1970:108) correctly noted that: “the frontals of Odaxosaurus piger resemble closely those of Recent Diploglossus\ they agree in general outline. . .and in having nearly identical sculpture on the osteoscutes covering these bones.” However, Estes (1983a) saw that the species has a derived, glyptosaurine-like tooth form. Here, we have a Cre- taceous anguid lizard with glyptosaurine tooth mor- phology and jaw structure but diploglossine skull elements and osteoderms, with association with the skull elements based on size and relative abundance of the fossils. Therefore, the question is whether this referral is correct or the skull elements actually rep- resent an anguid lizard other than Odaxosaurus. In fact, Armstrong-Ziegler (1980:21) had already questioned this referral by stating that “the dentar- ies of both Gerrhonotus and Pancelosaurus occur in the Lance Formation; Meszoely (1970, p. 105) assigned the above defined osteoscutes to Pance- losaurus instead of Gerrhonotus on the basis of the greater abundance of skull remains of Pancelosau- rus relative to Gerrhonotus. This criterion of rela- tive abundance used by Meszoely does not seem sufficient reason to assign these osteoscutes to Pan- celosaurus or Gerrhonotus." The discovery of the jaw materials with fused osteoderms (see descrip- tion above) implies more strongly than before that the association of these disarticulated materials is incorrect. It appears that the early history of Anguidae in- volved two major evolutionary lines: the diploglos- sine (gerrhonotine) line and the glyptosaurine line. The former includes those anguids that are essen- tially predatory, with sharp teeth and relatively thin, nontuberculated osteoderms; and the latter line in- cludes those nonpredatory anguids having blunt teeth with a thick enamel cap for a durophagous diet, and a heavy armor of thick, tuberculated os- teoderms. From jaw structure and tooth form, Odax- osaurus piger is obviously on the glyptosaurine line; and the skull elements previously referred to O. piger are possibly associated with some kind of predatory anguid, as the osteodermal sculpture on these bones is closely similar to the Diploglossus- Ophisaurus type (Estes, 1964; Meszoely, 1970). This type of osteoderm is also known from the Oli- gocene Parophisaurus Sullivan, 1987, the anguid lizard that is ideal as the structural ancestor of Ophi- saurus (Sullivan, 1987). Odaxosaurus piger is known only from Lancian horizons, including the Lance, Hell Creek, Laramie, Scollard, and Frenchman formations. Several Paleo- cene and Oligocene specimens were previously re- ferred to this species (Meszoely, 1970; Estes, 1975, 1976; Sullivan, 1979, 1981, 1982), but have re- cently been reassigned to Proxestops silberlingi or Machaerosaurus (Gauthier, 1982; Estes, 1983a; Sullivan, 1986b, 1991); therefore, no definite Odax- osaurus piger specimen is known from post-Lan- cian horizons. Pre-Lancian Odaxosaurus lizards were previously reported from the Judith River For- mation (Sahni, 1972), Mesaverde Formation (see Estes, 1983a), and Oldman Formation (Fox, 1976). Our study of the specimens from the Oldman For- mation reveals that the Judithian Odaxosaurus rep- resents another species different from Lancian O. piger. Odaxosaurus priscus, new species (Fig. 29, 30A-F) Etymology.— Priscus (Latin), meaning “former, aboriginal.” Holotype. — UALVP 29896, incomplete right dentary with eight teeth and spaces for six others. Type Locality and Horizon. — Outcrop of the Old- man Formation near Irvine, in Sec. 31, Tp. 11, R 2, W 4, about 40 km east of Medicine Hat, south- eastern Alberta; Upper Cretaceous Oldman Forma- tion (Judithian). Referred Specimens. — Railway Grade locality: UALVP 29891, maxillary. Irvine locality: UALVP 29892, 29893-29895, maxil- laries; UALVP 29897-29902, 33380, dentaries. Known Distribution. — Type Irvine and Railway Grade localities, Oldman Formation (Judithian), southeastern Alberta. Diagnosis. — A Judithian Odaxosaurus differing from the Lancian O. piger in the following combi- nation of character states: dentary teeth fewer, fur- ther apart from each other; tooth crowns less trun- 1996 GAO AND FOX— LATE CRETACEOUS LIZARDS 71 l i i I i i Fig. 29. — Odaxosaurus priscus, new species, Oldman Formation, Alberta: UALVP 29896 (holotype), incomplete right dentary, lat- eral (above) and medial (below) views. Scale = 5 mm. Abbre- viations as in Fig. 2. cate, less flaring anteriorly and posteriorly in side view, and less heavily striated medially; intraman- dibular septum not so strongly developed, with free ventral border extending further anteriorly, to mid- point of tooth row. Description. — The holotype, UALVP 29896 (Fig. 29, 30A), is a right dentary showing that the jaw configuration of this lizard is similar to that of the Lancian Odaxosaurus piger. This spec- imen retains a complete tooth row, with eight teeth and the spac- es for six others, giving a total of 14 positions. As in O. piger, the tooth shafts are strongly compressed anteroposteriorly and have a strong, shoulder-like, medial expansion. The tooth crowns on this specimen are partially dissolved, but are apparently less truncated than those in Odaxosaurus piger, especially for the anterior and middle teeth. These teeth are also more widely spaced than in Odaxosaurus piger. Medially, most of the subdental shelf has been broken away; thus, the notch for the anterior inferior alveolar foramen is not shown on this specimen. However, four other dentaries (UALVP 29897-29898, 29900, 29902) clearly show that the notch forms the dorsal and anterior border of the anterior inferior alveolar foramen. The Meckelian canal is as ventrally restricted as in O. piger, but the intramandibular septum is significantly shallower and less vertically directed compared to the latter species, and has a free ventral border extending anteriorly to the midpoint of the tooth row. Another dentary, UALVP 29897 (Fig. 30B), has seven teeth and the bases for four others preserved. The crowns of most of the teeth are badly worn, but one posterior tooth (probably the fourth from the back) clearly shows that the crown is chisel- shaped. The sulcus dentalis is entirely lacking, and the subdental shelf is greatly reduced to form a smooth, sloping border with a small notch for the anterior inferior alveolar foramen. Like the holotype, this dentary also shows that the free ventral border of the intramandibular septum extends anteriorly beyond the ante- rior inferior alveolar foramen to the midpoint of the tooth row, differing in that respect from Odaxosaurus piger, in which the septum extends only to the point below the foramen. The dentary teeth are best shown on UALVP 33380 (Fig. 30C), a specimen from the Irvine locality. On this specimen, two well-preserved posterior teeth above the intramandibular septum show that the tooth shafts, like those of Odaxosaurus piger, are anteroposteriorly compressed and have a shoulder-like medial ex- pansion. However, unlike the posterior teeth in O. piger, the two teeth have slightly recurved crowns with faint medial striations (but lack lateral striations). Well-preserved posterior teeth on UALVP 29894, a maxillary, consistently show the same type of crown pattern (see later description). Among the five maxillaries referred to this species, two are obviously better preserved than the others. UALVP 29891 (Fig. 30D) from the Railway Grade locality is from the right side and bears four teeth and the bases as well as spaces for seven others. The specimen shows essentially the same configuration as that of Lancian Odaxosaurus piger, but has a much more prominent premaxillary process. The posterior interior alveolar foramen is single, large, and located above the ninth or tenth tooth position from the front. Unlike the maxillary in the Lancian species, how- ever, this specimen shows that the crowns of the anterior and middle teeth are bluntly unicuspid, slightly recurved, but clearly show a tendency towards development of a lateral cutting edge. In occlusal view, the tooth crowns have a central cusp that curves lingually and posteriorly, forming an “inwardly pointing V” (see Estes, 1964). The medial striations on these teeth are much weak- er than in O. piger and the lateral surfaces show no striations at all. The posterior teeth of UALVP 29891 are not preserved, but must have had chisel-shaped crowns, as clearly shown on the dentary specimens. Another maxillary, UALVP 29892, is from the left side and has five teeth and the bases for four others. Like UALVP 29891 , this specimen also shows a prominent premaxillary process (see Fig. 30E). The lateral parapet of the maxillary is broken imme- diately behind the ninth tooth position, but it still clearly shows that a single and large posterior interior alveolar foramen is lo- cated above the tenth tooth position. The tooth crowns on this specimen are not well preserved, but are similar in shape to those of UALVP 29891. The other three specimens (UALVP 29893-29895) are less well preserved than the two described above, in terms of showing the general shape of the maxillary. However, these are signifi- cant, because the teeth on these specimens show slightly re- curved crowns with faint medial, but no lateral striations, con- sistent with the posterior dentary teeth described above. UALVP 29894 (Fig. 30F) has badly worn tooth crowns, indicating it is from an old individual, while UALVP 29895 shows a prominent palatine process closely associated with a single large posterior interior alveolar foramen, as in Lancian O. piger. Discussion. — The holotype described above from the Oldman Formation represents a new, Judithian species of Odaxosaurus, which differs principally from the Lancian O. piger in having fewer, less truncate dentary teeth, and a shallower intramandi- bular septum extending anteriorly to the midpoint 72 BULLETIN CARNEGIE MUSEUM OF NATURAL HISTORY NO. 33 Pig 30. — Odaxosaurus priscus, new species, Oldman Formation, Alberta: A, UALVP 29896 (holotype), incomplete right dentary, medial view; B, UALVP 29897, incomplete right dentary, medial view; C, UALVP 33380, right dentary fragment, medial view; D, UALVP 29891, right maxillary, medial view; E, UALVP 29892, left maxillary, medial view; F, UALVP 29894, right maxillary, medial view. Anguidae, genus and species undetermined. Milk River Formation, Alberta: G, UALVP 33378, incomplete left dentary, medial view; H, UALVP 33379, fragmentary left dentary. medial view. Scale = 2 mm. of the tooth row. Referred specimens from the same formation show other taxonomically significant character states for the new species, including the possession of a more prominent premaxillary pro- cess of the maxillary and tooth crowns having much weaker medial striations (while no lateral striations are shown on these specimens). Odaxosaurus pris- cus is clearly on the glyptosaurine line of evolution. as evidenced by its close resemblance in dentary structure and tooth morphology to the Lancian O. piger, which is currently regarded as a precusor of the Tertiary true glyptosaurines (Estes, 1983a; Sul- livan, 1986«, 1986 b). The dentary of this Judithian form has a shallow but anteriorly extending intra- mandibular septum, and obtuse teeth with faint me- dial striations, from which the Lancian O. piger can 1996 GAO AND FOX— LATE CRETACEOUS LIZARDS 73 easily be derived. Therefore, the Judithian species represents the earliest known stage in glyptosaurine evolution. There are no osteoderms directly asso- ciated with O. priscus (e.g., fused to the maxillary or dentary specimens), but its characteristic tooth morphology and dentary structure indicate that the splitting of glyptosaurines from their ancestral group can be traced at least back to Judithian time. Several authors have referred specimens of Ju- dithian age to the Lancian species Odaxosaurus pig- er, but this study shows that these referrals are ei- ther erroneous or unconvincing: Sahni (1972:354) referred two specimens (AMNH 8494, 8495) from the Judith River Formation to “ Peltosciurus piger ” (= Odaxosaurus piger) based on the characters that are symplesiomorphic for the family Anguidae (e.g., “anteriorly the Meckelian fossa is ventral in posi- tion”). Regarding Sahni’s referral of these two spec- imens to Odaxosaurus piger, we here offer the fol- lowing comments: 1) AMNFI 8494 (Sahni, 1972: fig. 8H), an incomplete right dentary with all of the teeth broken off, is certainly an anguid as it shows a notch on a greatly reduced subdental shelf in com- bination with the loss of the sulcus dentalis. How- ever, it differs from the dentary of Odaxosaurus in having a straight tooth-bearing border, and more im- portantly, having cylindrical but not transversely ex- panded tooth shafts (personal observation). Similar materials are also known from the Oldman and Milk River formations, which are identified as “Angui- dae, genus and species undetermined” in this paper. 2) AMNH 8495 (Sahni, 1972:fig. 8G) as described appears to be referrable to Odaxosaurus, as Sahni (1972:354) stated that “the teeth are pleurodont, lin- guolabially expanded, and the transverse width is only slightly smaller than the height. The crown is expanded relative to the shaft and consists of a dis- tinct anteroposteriorly directed crest rather than sep- arate cuspules.” However, the clear figure of the specimen (Sahni, 1972:fig. 8G) contradicts this de- scription, and indicates that it is largely artificial. In fact, all but one tooth on the maxillary are broken off, and the one preserved is slender, definitely nei- ther chisel-shaped nor striated (personal observa- tion). A large resorption pit of this tooth is also substantially different from the small, rounded basal foramen in Odaxosaurus. Therefore, this specimen should be excluded from Odaxosaurus. The speci- men is more or less Contogenys- like, but its frag- mentary nature does not allow a familial referral. Although AMNH 8494, 8495 may not pertain to Odaxosaurus, some of the maxillary and dentary fragments in a specimen box (labelled as AMNH 8496) are indeed of the Odaxosaurus kind, indicat- ing occurrence of this genus in the Judith River For- mation. Estes (1983 a) referred several specimens from the “Mesaverde” Formation in Wyoming to the species Odaxosaurus piger in his monographic re- view of fossil lizards. Without a description or fig- ure, it is difficult to evaluate the reliability of this identification. Genus and species undetermined (Fig. 30G, H) Referred Specimens. — Milk River Formation, MR-4 locality: UALVP 33378, incomplete left dentary. MR-6 locality: UALVP 33379, fragmentary left dentary. Many unnumbered maxillary and dentary fragments from MR-4, MR-6, MR-8, and MR-20 localities. Description. — UALVP 33378 (Fig. 30G) consists of the mid- dle and posterior part of a small left dentary, bearing six teeth and vacant positions for two others. The lateral surface of the dentary is smooth, lacking any sculpture. Three inferior alveolar foramina preserved on the specimen are large, open anterolat- erally, and aligned horizontally about midway in the height of the dentary. These foramina are more or less equally spaced from one another. Medially, the subdental shelf is greatly reduced as a smoothly sloped tooth-bearing border, which is posteriorly notched for the anterior inferior alveolar foramen. The Meckelian canal anterior to the notch cannot be seen in medial view; how- ever, a ventral view shows that the canal is greatly restricted, opening ventrally throughout most of its length as a shallow and narrow gutter. A slightly flattened ventral surface of the dentary anterior to the notch indicates that a greatly reduced splenial terminated anteriorly close to the midpoint of the tooth row. Pos- teriorly within the Meckelian canal, a clearly defined intraman- dibular septum is vertically directed: it has a free ventral border and is posteriorly notched for the Meckelian cartilage. The dentary teeth are high-crowned, having a cylindrical shaft and a small basal foramen opening medially at the tooth base. Most of the teeth have the crowns broken off, but two are well preserved and show that the crowns have faint medial striations. These teeth are pleurodont, having two-thirds of their height at- tached to the well-developed lateral parapet of the dentary. A more fragmentary dentary of this kind (UALVP 33379; Fig. 30H) is known from the MR-6 locality. It has the same jaw construction and proportions as the above-described specimen: the size, notched tooth-bearing border, and well-developed intra- mandibular septum match UALVP 33378 and allow both to be referred to the same species. On the basis of similar jaw con- struction and proportions, several unnumbered maxillary and dentary fragments from the Milk River Formation are also re- ferrable to Anguidae, genus and species undetermined. These specimens were collected from MR-4, MR-6, MR-8, and MR-20 localities. Discussion. — Fox (1972a, in schedula) made the original identification of the above-described spec- imens as anguids. As a result of this study, the placement of these specimens in the Anguidae is supported by a well-recognized synapomorphy: the strongly reduced subdental shelf is notched for the 74 BULLETIN CARNEGIE MUSEUM OF NATURAL HISTORY NO. 33 anterior inferior alveolar foramen, so that the den- tary forms the dorsal and anterior border of the fo- ramen (Estes, 1964; Estes et al., 1988). Two other character states shown on these specimens are syn- apomorphic for a higher taxon, Anguimorpha; these include: a Meckelian canal that opens ventrally an- terior to the anterior inferior alveolar foramen, and a posterior part of the Meckelian canal that is sub- divided by a ventrally directed intramandibular sep- tum (Estes, 1964; Estes et al., 1988). These speci- mens also show a great reduction of the subdental shelf corresponding with the total loss of the sulcus dentalis, as is commonly seen in other anguimor- phans. Although undetermined at generic and specific levels, the identification of these specimens as An- guidae (Fox, 1972a, in schedula) is significant, as these early Campanian materials document the ear- liest undoubtedly fossil record for this family. Pre- vious to the present study, Winkler et al. (1990) reported “Anguidae?” from the Comanchean Pa- luxy Formation, central Texas. However, the spec- imen figured (SMU 72297; see Winkler et al., 1990: fig. 8E) has a robust subdental shelf and a sulcus dentalis, suggesting a scincomorphan affiliation, rather than anguid. Therefore, the Comanchean specimen should be excluded from the Anguidae, making the Milk River specimens described above represent the earliest known record of the family. Family Necrosauridae* Hoffstetter, 1943 The Necrosauridae* are primitive varanoids char- acterized by having an unretracted naris, and lack- ing a ventral contact of the subolfactory processes of the frontal (Estes, 1983a). The extinct family in- cludes fossil taxa such as: Necrosaurus Filhol, 1876, and Eosaniwa Haubold, 1977, from the Ter- tiary of Europe; and Parasaniwa Gilmore, 1928, Colpodontosaurus Estes, 1964, (Late Cretaceous), and Provciranosaurus Gilmore, 1942, (Paleocene) from North America. The family may also include several Cretaceous forms known from the Gobi Desert (Borsuk-Bialynicka, 1984), but the phylo- genetic relationships of the Gobi taxa need further study. The family Necrosauridae* is cladistically treated as a metataxon (Estes et al., 1988), as neither its monophyly nor positive paraphyly can be rec- ognized at present. Genus Colpodontosaurus Estes, 1964 Type Species. — Colpodontosaurus cracens Estes, 1964. Range. — Upper Cretaceous, North America. Diagnosis. — As for the type and only known spe- cies. Colpodontosaurus cracens Estes, 1964 (Fig. 31A-C) Holotype. — UCMP 46608, an incomplete left dentary having broken bases for three, and vacant spaces for 12, dentary teeth. The holotype was fig- ured by Estes (1964:fig. 60) with several teeth re- stored from other specimens. Referred Specimens. — Scollard Formation, KUA-3 locality: UALVP 33388, fragmentary left dentary. In addition, some spec- imens from outside the study area are also included in this paper, because they show significant individual variation within the spe- cies. These specimens include: Lance Formation, BTB locality: UALVP 29781-29784, incomplete dentaries. Hell Creek For- mation, BCA locality: UALVP 29785, 29786, fragmentary den- taries. Known Distribution. — Upper Cretaceous Lance Formation, eastern Wyoming; Hell Creek Forma- tion, eastern Montana; and Scollard Formation, cen- tral Alberta, (all Lancian). Diagnosis (Revised from Estes, 1983a). -—A Late Cretaceous necrosaurid lizard differing from Pro- varanosaurus Gilmore, 1942a, its closest known relative, in having: dentary more slenderly built; marginal teeth lower crowned; tooth bases expand- ed, lacking striations; and having a distinct small ventral process on the weakly developed intraman- dibular septum, which lacks a free ventral border. Description. — UALVP 33388 (Fig. 31 A) is the only specimen of this species collected from the Scollard Formation. The frag- mentary left dentary bears a broken tooth and the vacant spaces for three others. Like the specimens from the Lance and Hell Creek formations, this dentary shows taxonomically significant features: the subdental shelf (without sulcus dentalis) is in a very low position; the tooth base is expanded, lacks striations, and has a basal foramen opening posteromedially; the teeth are hollow and have strongly constricted and possibly recurved crowns (comparing UALVP 33388 with UALVP 29782; see Fig. 31). The size and robustness of this dentary compared with those from the Lance and Hell Creek formations indicate that the Scol- lard specimen is probably from a relatively old individual. Among the several specimens of Colpodontosaurus cracens from the BTB locality in the UALVP collections, UALVP 29781 (Fig. 3 1 B) is relatively complete, with only the anterior end miss- ing. The dentary shows the same general configuration as the holotype of the species, but differs from the latter in having four (rather than two) inferior alveolar foramina on its lateral surface, although it is smaller than the holotype. There are no teeth pre- served on the specimen, but spaces for 13 tooth positions can be clearly recognized. The total number of dentary teeth for this lizard is estimated at close to 20, as in the holotype. Another specimen (UALVP 29782; Fig. 3 1C) is the posterior part of a right dentary, including the posteriormost eight tooth positions. This dentary has three incomplete teeth preserved, showing that the tooth bases are expanded, nonstriated, and each 1996 GAO AND FOX— LATE CRETACEOUS LIZARDS 75 Fig. 31. — Colpodontosaurus cracens, Scollard and Lance formations, Alberta and Wyoming: A, UALVP 33388, left dentary fragment, medial view; B, UALVP 29781, left dentary, medial view; C, UALVP 29782, right dentary fragment, medial view. Necrosauridae*, new genus and species (unnamed). Milk River Formation, Alberta: D, UALVP 29787, incomplete right dentary, medial view; E, UALVP 33383, incomplete right dentary, medial view. Scale = 2 mm. bears a small foramen. The tooth crowns are strongly constricted, and are obviously less recurved than those Estes (1964) figured for the holotype. The intramandibular septum is similar to that of the holotype and the other specimens included in this collec- tion, in lacking a free ventral border but having a tiny ventral process. Discussion. — The Lancian Colpodontosaurus cracens was originally diagnosed by Estes (1964) as a lizard having thin-walled, pointed, recurved, and short teeth; and having a tiny free ventral pro- cess of the intramandibular septum. The specimens in the UALVP collections from the Lance and Hell Creek formations provide no new information rel- evant to the diagnosis of this lizard, but show some variation in the curvature of the dentary teeth and the number of inferior alveolar foramina. Estes (1964:fig. 60) figured the holotype with strongly re- curved teeth (restored from other specimens, ac- cording to Estes), while UALVP 29782 shows that the teeth have a relatively straight shaft with only the tip of the crown slightly recurved. The holotype, as Estes (1964) described, has only two inferior al- veolar foramina on its lateral surface, while UALVP 29781 (also UALVP 29785) shows four foramina. Recognition of these variations is significant for a better understanding of this poorly known species. Estes (1964) originally placed Colpodontosaurus cracens in the Diploglossa (Anguioidea) incertae sedis, on the basis of the tiny ventral process of the intramandibular septum, absence of basal infoldings of the teeth, and the presumed lack of an intraman- dibular hinge. More recently, Estes (1983a), re- ferred this species to the Necrosauridae* on the ba- sis of resemblances in jaw and tooth structure to some necrosaurids (e.g., lack a free ventral border of the intramandibular septum as in necrosaurids generally; thin-walled teeth as in Provaranosaurus). This familial placement is followed in this paper. However, Colpodontosaurus cracens is based on dentary materials (in most cases on specimens with the teeth broken off), and maxillary and other skull elements are unknown or poorly known as yet. A better understanding of the relationships of this liz- ard at and below the familial level depends on fur- ther study of better preserved materials. Genus and species new (unnamed) (Fig. 31D, E) Referred Specimens. — UALVP 29787, 33383; both right den- taries with several broken teeth. Locality and Horizon. — MR-6 locality, Verdigris Coulee, approximately 30 km east of the village of Milk River, southern Alberta; upper member of the Milk River Formation (Aquilan), Upper Cretaceous. Known Distribution. — Upper Cretaceous Milk River Formation (Aquilan), southern Alberta. Description. — UALVP 29787 (Fig. 3 ID) is a short, slender, and delicate dentary (7 mm long as preserved) that abruptly ta- pers anteriorly and ends at a weak mandibular symphysis. The lateral surface of the dentary is smooth and much less convex than that of Colpodontosaurus cracens. As in the latter species, the anteroventral edge of the dentary turns sharply medially, forming part of the floor of Meckelian canal. Four inferior al- veolar foramina (mental foramina) are small and unequally spaced from one another. Medially, the dentary bears a delicately built but prominent 76 BULLETIN CARNEGIE MUSEUM OF NATURAL HISTORY NO. 33 subdental shelf, which is nearly equal in thickness along its ex- tent; in Colpodontosciurus cracens, the shelf is much less prom- inent and thins posteriorly. The sulcus dentalis is absent in this specimen, as in C. cracens. The Meckelian canal narrows sharply anteriorly, and turns ventrally. The ventral border of the dentary bears an articular surface for the splenial, which indicates that the splenial extended anteriorly to about the midpoint of the tooth row, leaving a long and ventral opening of the Meckelian canal anteriorly. As in Colpodontosaurus cracens, the intramandibular septum lacks a free ventral border, but bears a small ventral pro- cess; however, this process is much less prominent than that of Colpodontosaurus cracens. The preserved portion of the dentary bears broken bases and spaces for 13 teeth. The four tooth bases that are preserved show that the teeth are short and thin-walled, and have swollen and nonstriated bases. The dentary tooth row is incomplete, but can be estimated as having about 16 positions, significantly fewer than in Colpodontosaurus cracens. Every tooth has a small, cir- cular basal foramen, which is medioposterior in position. Another specimen (UALVP 33383; Fig. 3 IE), also a right den- tary from the MR-6 locality, has one tooth and the broken bases for seven others preserved. This specimen is slightly larger and more robust than UALVP 29787, but it can be referred to the same species on the basis of thin-walled teeth and a horizontally wide but dorsoventrally slender subdental shelf. Discussion. — The two specimens (UALVP 29787, 33383) are referrable to the Necrosauridae* on the basis of their general tooth morphology and dentary structure (e.g., thin-walled teeth and poorly developed intramandibular septum). Although the two dentaries are not well preserved, the signifi- cance of this discovery is obvious, as the specimens document the earliest North American fossil record of the Necrosauridae*, and represent an Aquilan species that was closely related to the Lancian Col- podontosaurus cracens (Fox [1972a, in schedula] recognized this relationship by identifying UALVP 29787 as “ Colpodontosaurus ? sp.”). The Milk Riv- er specimens share with the Lancian Colpodonto- saurus cracens character states such as having thin- walled, low-crowned, and nonstriated teeth, and a small ventral process on the weakly developed in- tramandibular septum; but differ from the latter (at the generic level) in having a significantly shorter dentary with a much more prominent subdental shelf. These morphological differences, combined with the 1 8 Myr chronologic gap between Milk Riv- er and Lance horizons, suggest that the Milk River form represents a new genus and species; however, the new taxon is not named in this paper, because the two specimens described above are too poorly preserved for an adequate diagnosis. Genus Parasaniwa Gilmore, 1928 Type Species. — Parasaniwa wyomingensis Gil- more, 1928. Range. — Upper Cretaceous, North America. Diagnosis. — As for the type species. Parasaniwa wyomingensis Gilmore, 1928 (Fig. 32A-H) Parasaniwa obtusa Gilmore, 1928:26, fig. 10, 12. Holotype. — USNM 10797, fragmentary left den- tary having one complete and two broken teeth, col- lected by J. B. Hatcher in 1892 from the Lance Formation, eastern Wyoming (see Gilmore, 1928; Estes, 1964, 1983a). Referred Specimens. — Frenchman Formation, Wounded Knee locality: UALVP 33347, premaxillary with two teeth; UALVP 33348, right maxillary fragment with two teeth; UALVP 33349, incomplete left dentary having ten tooth positions; UALVP 33350, dorsal vertebra. Scollard Formation, KUA-1 locality: UALVP 33384, fragmentary maxillary with two teeth. Lance Formation, BTB locality: UALVP 33351-33353, in- complete tooth-bearing maxillaries; UALVP 33354, frontal; UALVP 1612, 1616, 33355, 33356, incomplete tooth-bearing dentaries. Hell Creek Formation, BCA locality: UALVP 33357- 33360, tooth-bearing maxillary fragments; UALVP 33361, and several unnumbered dentaries; UALVP 33362, 33363, toothed pterygoids; UALVP 33364, 33365, parietals. Known Distribution. — Upper Cretaceous Lance, Hell Creek, Frenchman, and Scollard formations (all Lancian). Congeneric specimens are also known from the Oldman Formation (Judithian), southern Alberta (see later description and discussion). Diagnosis ( Revised from Estes, 1983a). — Late Cretaceous necrosaurid lizard from North America, differing from Necrosaurus Filhol, 1876, and other necrosaurids in having skull osteoderms fused, flat- tened, and separated by shallow and wide grooves; maxillary having higher and vertically-directed dor- sal process, which rises abruptly behind narial open- ing; marginal teeth strongly compressed laterally, trenchant, but lacking serrations; tooth bases dilated and having pronounced basal striations. Description. — Gilmore (1928) gave an excellent description of the holotype dentary from the Lance Formation. Estes (1964) revised the description on more nearly complete specimens, and for the first time described and figured maxillary, frontal, and parietal materials of this species from the same formation. The following description focuses on the specimens in UALVP col- lections, especially on those from the Scollard and Frenchman formations, central Alberta and southwestern Saskatchewan, where the newly discovered specimens indicate a northern ex- tension of the geographical distribution of this species. Other specimens from the BTB and BCA localities are included in this study as comparative materials from the same horizon but dif- ferent geological formations. Premaxillary. UALVP 33347 (Fig. 32A) from the Wounded Knee locality. Frenchman Formation, is the first premaxillary specimen to be described for Parasaniwa wyomingensis. The premaxillary is a single unit (unpaired), having five tooth posi- 1996 GAO AND FOX— LATE CRETACEOUS LIZARDS 77 Fig. 32. — Parascmiwa wyomingensis, Lancian formations in Alberta, Saskatchewan, and Wyoming: A, UALVP 33347, incomplete premaxillary, lingual view; B, UALVP 33384, right maxillary fragment, medial view; C, UALVP 33349, left dentary, medial view; D, UALVP 33361, left dentary, medial view; E, UALVP 33354, right frontal, dorsal view; F, UALVP 33362, toothed left pterygoid, ventral view; G, H, UALVP 33350, dorsal vertebra, dorsal and ventral views. IParasaniwa sp.. Milk River Formation, Alberta: I, UALVP 33923, fragmentary left dentary, medial view; J, UALVP 33924, left maxillary fragment, medial view; K, UALVP 33925, right dentary fragment, medial view. Scale = 2 mm. 78 BULLETIN CARNEGIE MUSEUM OF NATURAL HISTORY NO. 33 tions. Two poorly preserved teeth show basal infoldings, the same tooth form as seen in the maxillary and dentary specimens. Dorsally, the premaxillary spine is tall, splint-like, and slightly recurved in correspondence with the high postnarial process of the maxillary. On each side of the base of this spine are two well-defined premaxillary foramina, through which passed nerves and blood vessels to the bases of the premaxillary teeth. Although no osteoderms are shown on the premaxillary, this specimen is different from the premaxillary of Paraderma in having a narrow and tall splint-like spine, and lacking a plate- like palatal process; and different from Colpodontosciurus and Provarcinosaurus in the presence of basal striations on the teeth. Therefore, this specimen is here referred to Parasaniwa. Maxillary. UALVP 33384 (Fig. 32B) consists of a fragmen- tary tooth-bearing right maxillary from the Scollard Formation, central Alberta. The two teeth preserved in front of the posterior interior alveolar foramen are short and strongly trenchant, but have no serrations on their anterior and posterior blades. The tooth bases are strongly expanded and striated with basal infold- ings (Estes, 1964, noted that the degree of these striations is individually variable within the species). The supradental shelf is flat and horizontal, lacking a thickened dorsal fold, and the lateral wall of the maxillary is thin and shows no fused osteo- derms, in these aspects differing from the maxillary of Parader- ma (see later description). Other maxillaries from the Frenchman, Lance, and Hell Creek formations numbered above are referred to Parasaniwa wyomingensis on the basis of their tooth form and/or sculpture pattern on the lateral surface of the maxillaries. Dentary. Among the dentary specimens referred to this spe- cies, two (UALVP 33349, 33361) are more extensively preserved than others. UALVP 33349 (Fig. 32C) is a left dentary from the Wounded Knee locality. Frenchman Formation. The specimen as preserved has ten tooth positions, including four teeth and the bases as well as spaces for six others. These teeth are pleurodont (contra Estes, 1964: “subpleurodont’"), as only half or less of the tooth height projects above the lateral parapet of the jaw. The dentary shows the same configuration and structure as does UCMP 46041, a nearly complete dentary from the Lance For- mation (see Estes, 1964:fig. 61): it is slender, lightly built, has a delicate symphyseal process, and a crescentic tooth-bearing bor- der (see Estes, 1964:129). However, the Frenchman dentary is much smaller and more slender than UCMP 46041, indicating that it is from a young individual. The other dentary (UALVP 33361; Fig. 32D), also from the left side, is from the BCA locality. Hell Creek Formation. This specimen reinforces Estes’ statement that “the intramandibular septum is fused to the floor of the Meckel ian groove throughout its length, and has a small notch posteriorly for Meckel’s carti- lage” (Estes, 1964:129); and, together with other specimens (UCMP 46041, UALVP 1612), shows that the groove for the anteromedial process of the coronoid bone is developed below the last four tooth positions, where the tooth-bearing border sud- denly becomes narrow. Frontal. UALVP 33354 (Fig. 32E) and another unnumbered specimen in the UALVP collections are clearly referrable to this species, as they show the same osteodermal pattern as the pari- etals of P. wyomingensis (see Estes, 1964). UALVP 33354 is a right frontal broken away from the left side at the midline. The unnumbered frontal from the same locality is similarly preserved. Estes (1964) described a fused frontal (AMNH 8103) with the midline suture clearly visible ventrally. Discovery of the new specimens from the BTB locality (Lance Formation) indicates that fusion of the frontals is an ontogenetic character for this species. Parietal. Two specimens (UALVP 33364, 33365) from the BCA locality are identified as parietals of Parasaniwa wyomin- gensis: they are closely similar to UCMP 54200 (Estes, 1964: fig. 63), a well-preserved parietal from the type Lance Formation. Although the two parietals are poorly preserved both in terms of completeness and sculpture pattern, their general shape, size, and especially the lateral flange for the temporal musculature indicate that they belong to the same species as UCMP 54200. Pterygoid. Two small pterygoid fragments, UALVP 33362, 33363 from the BCA locality, are referrable to Parasaniwa wyomingensis. Both of the specimens are too slender to be re- ferrable to Paraderma or Palaeosaniwa, but are consistent with the maxillaries and dentaries referred to Parasaniwa wyomingen- sis from the same locality. Both specimens show that the pter- ygoid of this lizard is toothed, a primitive character state. UALVP 33362 (Fig. 32F) clearly shows that the pterygoid of Parasaniwa differs from that of helodermatids (e.g., UALVP 33338; see later description) not only in size and robustness, but more importantly, in the manner that the pterygoid teeth are ar- ranged: the pterygoid teeth of Parasaniwa are arranged along the medial border of the palatine process, while those of heloder- matids are inset from the border. Vertebra. UALVP 33350 (Fig. 32G. H) and several unnum- bered vertebrae from the Wounded Knee locality are referrable to Parasaniwa wyomingensis. UALVP 33350 is a well-preserved dorsal vertebra, which is clearly different from that of Palaeo- saniwa in having a much weaker diapophysis below the prezy- gapophysis, and in having a much less constricted condylar base. In addition, UALVP 33350 and several unnumbered specimens are obviously too small to be referrable to Palaeosaniwa or Par- aderma', they are consistent in size with the jaw materials for Parasaniwa. Discussion.— Based on fragmentary dentary ma- terial from the Lance Formation, Gilmore (1928) founded Parasaniwa, and designated P. wyomin- gensis as the type species. His Parasaniwa included another species, P. obtusa, from the same locality. He separated the latter species from the type species on the basis that “it differs from the single complete tooth known of P. wyomingensis from the same for- mation and locality in having an erect crown (not recurved) with an obtusely pointed apex, and with striations restricted to the extreme basal surfaces” (Gilmore, 1928:86). Estes (1964) correctly placed P. obtusa in synonymy with P. wyomingensis, and recognized that the obtuse crown of P. obtusa is the result of erosion, and the degree of basal striations is variable for this lizard. A comparison of a large number of specimens in the UALVP collections shows that larger individuals have stronger basal striations than the smaller ones, indicating that this variation is an ontogenetic character. Here, we add another point in support of Estes’ revision: the max- illary that Gilmore placed in Parasaniwa obtusa (USNM 10800; see Gilmore, 1928:pl. XXVI, fig. 10) shows obviously stronger basal striations than 1996 GAO AND FOX— LATE CRETACEOUS LIZARDS 79 his holotype dentary (USNM 10798), and is indis- tinguishable in structure from those later referred to P. wyomingensis (see Estes, 1964, and this paper). The familial assignment of Parcisaniwa has been changed since Gilmore’s (1928) original descrip- tion. Gilmore (1928) first placed Parcisaniwa in the family Varanidae, but noted its striking similarity in tooth form to Palaeovaranus Filhol, 1877, which is a junior synonym of Necrosaurus Filhol, 1876 (see Hoffstetter, 1943; Estes, 1983a). Later, McDowell and Bogert (1954:32) accepted Gilmore’s familial designation of Parasaniwa as a saniwine varanid, but noted that “there is a suggestive similarity in dental form between Exostinus lancensis and its contemporaries and compatriots Parasaniwa obtusa and P. wyomingensis . ” In fact, a trenchant crown pattern is a character state shared by many carniv- orous anguimorphans. Estes (1964) erected a new family, Parasaniwidae, which includes both Para- saniwa and Paraderma, and diagnosed the family as including genera that are “characterized by teeth with simple basal fluting extending into the med- ullary cavity; development of an intramandibular jaw hinge; unretracted nares; skull roof with well- defined scute areas; scutes tending toward multipli- cation” (Estes, 1964:128). Later, Hoffstetter (1969) recognized the close relationships between Parasan- iwidae and Necrosauridae* on the evidence from maxillary structure and skull osteoderms, and sug- gested that the differences between the two groups are not great enough to separate them at a familial level. Accepting Hoffstetter’s interpretation and on more extensive discussion with other colleagues, Estes (1983a) synonymized his Parasaniwidae with Necrosauridae* Hoffstetter, 1943, and tentatively al- located Paraderma to the Helodermatidae Gray, 1 837 (see later discussion on Paraderma). The new specimens from several localities at Lancian hori- zons add no new information concerning the famil- ial affiliation of this lizard; thus, in this paper Par- asaniwa wyomingensis remains a Cretaceous nec- rosaurid as Estes (1983a) classified it. The geologic distribution of Parasaniwa wyom- ingensis, in current understanding, is limited to Lan- cian horizons. The specimens of this particular spe- cies were previously known from the Lance and Hell Creek formations (Gilmore, 1928; Estes, 1964; Estes et al., 1969), and now from the Frenchman and Scollard formations (Fox, 1989; and this paper). Post-Lancian Parasaniwa sp. has been reported by Hecht (1959) from Eocene deposits in Wyoming, and by McKenna (1960) from the Eocene of Col- orado. However, these materials likely belong to an- other genus (Estes, 1983a); accordingly, the Eocene record of Parasaniwa is rejected. Pre-Lancian con- generic materials have been reported in faunal lists from the Oldman Formation of Alberta (Fox, 1976; Currie, 1986; Eberth, 1987; Koster et ah, 1987), but these Judithian materials likely represent another species (see below). Three fragmentary specimens (UALVP 33923- 33925) from the Milk River Formation are Para- saniwa- like, as they have laterally compressed teeth with a trenchant cutting blade and weak basal in- foldings. However, these specimens are poorly pre- served, and cannot be definitely referred to Para- saniwa', they are identified as “? Parasaniwa sp.” (see Fig. 32I-K). Parasaniwa, new species, cf. P. wyomingensis (Fig. 33) Specimens. — UALVP 33346, 33366-33369; RTMP 65.24.6, 81.38.4; and 27 unnumbered spec- imens, all incomplete maxillaries. UALVP 33370- 33373, RTMP 79.8.631, and more than 30 unnum- bered fragment specimens, all dentaries; UALVP 33374, frontal; UALVP 33375-33377, RTMP 82.20.36, 84.168.2, parietals. Locality and Horizon. — The specimens listed above are collected from various localities of the Oldman Formation, southeastern Alberta; the UALVP specimens are from the Irvine locality; and the RTMP specimens are from the Railway Grade, White Rock Coulee, Hope Johnson's microfossil site (DPP), Sandy Point, and Manyberries localities (coordinates on file in UALVP and RTMP). Description. — All of the specimens listed above can be distin- guished from those referred to Palaeosaniwa or Paraderma (see later description), and can be recognized as Parasaniwa- like on the basis of general jaw proportions, tooth form, or texture of osteodermal sculptures. Several relatively well-preserved speci- mens are described as follows. Maxillary. UALVP 33346 (Fig. 33A. B) is a nearly complete right maxillary from the Irvine locality. Oldman Formation. In proportions and sculpture pattern generally this specimen is Par- asaniwa- like, but differs from that of the type species ( P . wyom- ingensis) in having a roughly triangular dorsal process and a greater number of tooth positions anterior to the posterior interior alveolar foramen (seven vs. five for P. wyomingensis in UCMP 49935, personal observation). Anteriorly, the two premaxillary processes are short and not clearly separated from each other, because the notch between them is poorly developed. The ante- rior interior alveolar foramen is small, rounded, and opens en- tirely on the internal side of the nasal process, so that it is clearly visible in medial view. The nasal process (= the dorsal or facial process of other authors) rises behind the narial opening at about a 45-degree angle, then curves along a gentle slope associated with the inward bending of the dorsalmost part of the process. There is no such step in UCMP 49935, but a vertical anterior 80 BULLETIN CARNEGIE MUSEUM OF NATURAL HISTORY NO. 33 Fig. 33. — Parasaniwa, new species, cf. P. wyomingensis, Oldman Formation, Alberta: A, B, UALVP 33346, nearly complete right maxillary, lateral and medial views; C, RTMP 81.38.4. right maxillary fragment, medial view; D, UALVP 33370, posterior part of left dentary, medial view; E, UALVP 33371, right dentary, medial view; F, UALVP 33374, right frontal, dorsal view; G, H, UALVP 33375, incomplete parietal, dorsal and ventral views. Scale = 3 mm. border of the nasal process (see Estes, 1964:fig. 62). The poste- rior border of the nasal process descends at a similar angle, giv- ing the entire process a roughly triangular shape. UALVP 33346 is so far the most nearly complete maxillary known for Para- saniwa and is the best preserved with respect to the texture of the osteodermai sculpture. The lateral surface of the nasal process is vertically oriented, except for the medial curvature dorsally. The surface above the superior alveolar foramina is extensively covered with vermiculate osteoderms that are fused to the bone beneath (see Fig. 33A). The superior alveolar foramina, as many as eight, are small and open anteroventrally. Internally, the maxillary bears a well-developed supradental shelf, which is a flat, shelf-like structure lacking the fold seen in Paraderma. As in Parasaniwa wyomingensis (UCMP 49935), a long trough-ike depression for the nasal cavity above the shelf extends posteriorly and terminates in front of the posterior inte- rior alveolar foramen (see Fig. 33B). Flowever, the foramen opens above the eighth tooth position, while that in P. wyomin- 1996 GAO AND FOX— LATE CRETACEOUS LIZARDS 81 gensis opens above the sixth position (personal observation). The eight tooth positions preserved include the bases of five and the spaces for three others. The tooth bases have simple and short basal infoldings, as in the Lancian Parasaniwa wyomingensis. The posterior end of the maxillary is broken, so that the total number of maxillary teeth is unknown, but it can be estimated as 1 1-, since four or five positions are often seen along the tooth row posterior to the posterior interior alveolar foramen. RTMP 81.38.4 (Fig. 33C) is a fragmentary right maxillary from White Rock Coulee (Sec. 16, Tp. 17, R 3, W 4), about 50 km north of Medicine Hat. The fragment shows one complete tooth, which is trenchant and slightly recurved, as in Parasaniwa wyomingensis (UALVP 33368 from the Irvine locality shows the same type of tooth form). This maxillary fragment also shows that there are at least four positions for teeth behind the posterior interior alveolar foramen. Dentary. Among the dentaries referred to this species, two specimens from the Irvine locality (UALVP 33370, 33371) are better preserved than the others. UALVP 33370 (Fig. 33D) con- sists of the posterior half of a left dentary, which is broken di- rectly in front of the sixth tooth position from the rear. This specimen shows that the intramandibular septum of this Judithian species remains fused to the floor of the Meckelian canal through most of its length, with only the posteriormost end having a weak separation. This is the same condition as seen in Lancian spec- imens, indicating that there has been little change in this aspect of dentary structure for the approximately 1 1 Myr known for the evolution of Parasaniwa. However, the tooth-bearing border of the Judithian species is much less crescentic than that in the Lancian P. wyomingensis. The tooth bases in UALVP 33370 have simple basal infoldings, as in the Lancian specimens. The lateral surface of the dentary is smooth and more nearly flat than convex. The last inferior alveolar foramen opens lateral to the third posteriormost tooth position. Another specimen, UALVP 33371 (Fig. 33E), is an even better preserved right dentary, but the intramandibular septum is bro- ken. The tooth bases have the same kind of infoldings as UALVP 33370, but these are stronger, as this is a larger specimen from a presumably older individual. On five of the teeth the crowns are broken, but one (second to last of the complete tooth row) is nearly complete. This tooth has strongly trenchant, blade-like cutting edges, but no serrations can be seen. These dentary teeth are pleurodont, having the lower half of the tooth height attached to the high lateral parapet of the dentary. Frontal. UALVP 33374 (Fig. 33F) from the Irvine locality is an incomplete right frontal referred to Parasaniwa on the basis of its thin, large patches of osteoderms. Once again, this speci- men shows that the frontals of Parasaniwa are paired, and the fused condition (Estes, 1964) is likely an ontogenetic character. On the ventral side of the frontal, the subolfactory process (flange) is well developed, but seems not to meet at the ventral midline. Parietal. Five parietals from the Oldman Formation are re- ferrable to this Judithian species on the basis of the general shape of the parietal table and the texture of the osteoderms; these include UALVP 33375 (Fig. 33G, H), 33376, 33377 from the Irvine, RTMP 82.20.36 from the Railway Grade, and RTMP 84.168.2 from the Sandy Point localities. All four specimens have the parietal table well preserved, but the temporal process is broken. These specimens together indicate that the parietal of this Judithian species is similar to that of Lancian Parasaniwa wyomingensis in general proportions and osteodermal pattern, but differs from the latter in having both the parietal foramen and the parietal fossa in a more posterior location and the fused osteoderms covering only the anterior two-thirds or less of the parietal table. Discussion. — The specimens described above represent a Judithian lizard that is referrable to Par- asaniwa, as they show close resemblances to the type species, Parasaniwa wyomingensis, in general jaw proportions, tooth morphology, and osteoder- mal pattern. However, this Judithian form differs from the type species in having a roughly triangular dorsal process of maxillary, less crescentic tooth- bearing border of dentary, a more posterior location for the parietal foramen, and a less extensive cov- ering of fused osteoderms on the parietal. The pa- rietal foramen in the Judithian specimens is close to the center of the parietal table, while that on the Lancian specimens (e.g., UCMP 54200; see Estes, 1964:fig. 63) is close to the frontoparietal suture; the fused osteoderms on the Judithian specimens cover only the anterior two-thirds or less of the pa- rietal table, while those on the Lancian specimens are more extensive, covering two-thirds. These dif- ferences are so consistent on all the four specimens in the collection that they cannot be regarded as individual variations (or sexual dimorphism). In- stead, the morphological differences plus the 1 1 Myr-chronologic gap from P. wyomingensis suggest that the Judithian specimens represent another spe- cies. However, we are unable to name this new spe- cies at present, because none of the known dentaries is complete enough to be compared to that of Lan- cian P. wyomingensis ; similarly, the new species is represented by well-preserved maxillary material, while in Lancian P. wyomingensis, the maxillary is poorly known. Discovery of a better preserved den- tary may reveal more significant characters for this new species. Family Helodermatidae Gray, 1837 The Helodermatidae have the only two living species of venomous lizards: Heloderma horridum Wiegmann. 1829 (the Mexican beaded lizard) and Heloderma suspectum Cope. 1869 (the Gila mon- ster), which have a restricted present-day distribu- tion in southwestern United States and Mexico. The monophyly of the family is well supported by some 23 synapomorphies (Estes et al., 1988), among which the well-recognized osteologic characters are: teeth with venom grooves (Odermatt, 1940); and osteoderms thick, subconical. polygonal, and cov- ering the entire body (McDowell and Bogert, 1954). Contrary to their poor diversity and restricted dis- tribution today, the fossil record of Helodermatidae 82 BULLETIN CARNEGIE MUSEUM OF NATURAL HISTORY NO. 33 or helodermatid-like forms in North America, Eu- rope, and Asia "suggests] that they inhabited a once broader spectrum of environments than would be inferred from their present distribution” (Pregill et al., 1986:199). Among the fossil forms, Parad- erma Estes, 1964, has been recognized as the ear- liest definite record for the family (Pregill et al., 1986); later records include Eurheloderma Hoffs- tetter, 1957, from the upper Eocene of France, and indeterminate helodermatids are known from the lower Eocene of Belgium and the lower Oligocene of France (Rage and Auge, 1993); and Lowesaurus Pregill et al., 1986 and Helodermci taxana Stevens, 1977, from the Oligocene and Miocene of Colora- do, Nebraska, and Texas. In addition, Gobiderma Borsuk-Bialynicka, 1984, and Estesia Norell et al., 1992, are known from the Gobi; they are in some degree related to the Helodermatidae, but their clas- sification is still an open question (Borsuk-Bialyn- icka, 1984; Norell et al., 1992). Genus Parciderma Estes, 1964 Type Species. — Parciderma bogerti Estes, 1964. Range. — Upper Cretaceous, North America. Diagnosis. — As for the type species. Paraderma bogerti Estes, 1964 (Fig. 34, 35) Holotype. — UCMP 54261, incomplete left max- illary with four broken teeth and the spaces for two others. Type Locality and Horizon. — UCMP loc. V-5817 (False Disciple Point), Niobrara County, Wyoming (Clemens, 1963; Estes, 1964); Upper Cretaceous Lance Formation (Lancian). Referred Specimens. — Scollard Formation, KUA-1 locality: UALVP 29845 and several unnumbered specimens, tooth-bear- ing dentaries; UALVP 33385, incomplete parietal; UALVP 33902-33910, disarticulated cervical and dorsal vertebrae. Frenchman Formation, Wounded Knee locality: UALVP 29846, incomplete right dentary; and UALVP 33387, dorsal vertebra. Known Distribution. — Upper Cretaceous Lance Formation, eastern Wyoming; Hell Creek Forma- tion, eastern Montana; Scollard Formation, central Alberta; and Frenchman Formation, southwestern Saskatchewan (all Lancian). Pre-Lancian distribu- tion of the species is reported from the Judith River Formation (Sahni, 1972), but see later discussion. Diagnosis (Revised from Estes , 1983a). — A Late Cretaceous helodermatid from North America, dif- fering from other fossil and extant forms of the same family by a combination of the following character states: marginal teeth robust, trenchant crown having shallow anterior but no posterior ven- om groove; parietal foramen present, but small; ce- phalic osteoderms polygonal, with pitted sculpture and separated by wide groove; neural spines on trunk vertebrae tall and vertically directed. Description. — Several specimens in the UALVP collection from the KUA-1 locality provide significant material for a better understanding of Paraderma bogerti. The specimens described below include a parietal, several vertebrae, and limb bones that are the first known for this species, and a well-preserved dentary bearing teeth with an anterior venom groove. Dentary. Among the dentaries referred to Paraderma bogerti in this paper, two are well preserved. The first, UALVP 29845 (Fig. 34A; 35A, B), is a left dentary from the KUA-1 locality, Scollard Formation. The specimen is anteriorly complete, having the symphyseal part well preserved, but is broken posteriorly at the tenth tooth position. Nevertheless, it is the best dentary spec- imen known for this species. Anteromedially, the dentary bears a prominent symphyseal process, corresponding with the strong medial curvature of the jaw and making the mandibular arch widely U-shaped. The subdental shelf and the sulcus dentalis are entirely absent; instead, a robust tooth-bearing border forms most of the medial side of the Meckelian canal. The anterior half of the canal opens ventrally, but the posterior half gradually turns ventromedially in correspondence with the posterior deepening of the jaw. Because of breakage at the tenth tooth position, the nature of the intramandibular septum cannot be determined from this specimen (but see description of another specimen below). UALVP 29845 shows the first nine tooth positions of the jaw, including the first, third, fifth, seventh, and eighth dentary teeth, and the spaces for the second, fourth, sixth, and ninth. This tooth- and-space pattern is obviously the result of an alternating pattern of tooth replacement. The first tooth on the jaw is mostly broken off, leaving only the basal part preserved. The eighth tooth is completely preserved, showing that the posterior teeth of this lizard are laterally compressed, having a trenchant crown without serrations. The remaining three teeth have the tips of the crowns broken off, but still show that the anterior teeth are more colum- nar than compressed. These teeth are pleurodont, having the low- er half of the tooth attached to the lateral parapet of the jaw. In agreement with the holotype maxillary (UCMP 54261; see Pre- gill et al., 1986:fig. 9), the teeth in this dentary also show a taxonomically important character for this species: a clearly de- fined venom groove is developed anteriorly on the tooth shaft (see Fig. 34A), which is, however, less conspicuous and shallow- er than that in the extant Heloderma', and it has no posterior groove, differing from the latter genus. The tooth base is weakly infolded, and each tooth has a basal foramen developed medially. The lateral surface of the dentary is smooth and is anteriorly strongly turned towards the mandibular symphysis in correspon- dence with the anterior curvature of the jaw. Four inferior al- veolar foramina can be seen on the surface; and the fifth is not present, although the approximate location for this foramen is preserved on the jaw. These foramina are small in relation to the size of the jaw, and are slightly elongated and widely spaced from one another. Another specimen, UALVP 29846 (Fig. 35C), is the posterior part of a right dentary from the Wounded Knee locality. Al- though fragmentary, this specimen has one tooth and the spaces for three others preserved; more importantly, the specimen clear- ly shows that the ventral border of the intramandibular septum 1996 GAO AND FOX— LATE CRETACEOUS LIZARDS 83 Pig. 34. — Paraderma bogerti, Scollard Formation, Alberta: A, UALVP 29845, nearly complete left dentary, medial view (arrow indicates venom groove); B, C, UALVP 33385, incomplete parietal, dorsal and ventral views (arrow indicates incomplete parietal foramen); D- F, UALVP 33903, dorsal vertebra, lateral, ventral, and anterior views. Scale = 5 mm. is largely free from the floor of the Meckelian canal (contra Es- tes, 1983a: 177), in contrast to Parasaniwa from the same horizon in which the septum is mostly fused to the floor of the canal. Parietal. UALVP 33385 (Fig. 35D, E) is the first parietal known for Paraderma bogerti, since previous referral of YPM 1063 to this species (Estes, 1964) was erroneous (see later dis- cussion). Although incomplete with only the left side preserved, the specimen clearly shows a helodermatid synapomorphy: origin of the adductor muscles from the ventral surface of the parietal table (Pregill et al., 1986), as indicated by a strong roof-like lateral flange. The lateral border of the bone is comparatively straight (not strongly concave medially), indicating that the pa- rietal must have been roughly trapezoidal in shape. The dorsal surface of the bone is covered with heavy osteoderms that in pattern and surface texture are similar to those on the holotype maxillary (Estes, 1964:fig. 64; Pregill et al., 1986:fig. 9). Another 84 BULLETIN CARNEGIE MUSEUM OF NATURAL HISTORY NO. 33 Fig. 35. — Paraderma bogerti, Scollard and Frenchman formations, Alberta and Saskatchewan: A, B. UALVP 29845, nearly complete left dentary, lateral and medial views; C, UALVP 29846, fragment of right dentary, medial view; D, E, UALVP 33385, incomplete parietal, dorsal and ventral views; F, UALVP 29902, cervical vertebra, lateral view; G-I, UALVP 29903, complete dorsal vertebra, anterior, lateral, and ventral views; J, K, UALVP 33387, dorsal vertebra, dorsal and lateral views. Scale = 5 mm. taxonomically and evolutionarily significant feature that this pa- rietal shows is the presence of a small but clearly defined parietal foramen (Fig. 34C), which is close to the frontoparietal suture in position. In all extant and Tertiary forms in which the parietal is known, the foramen is absent. Vertebra. Nine disarticulated vertebrae from the KUA-1 lo- cality and one from the Wounded Knee locality are referrable to Paraderma bogerti on the basis of their general helodermatid morphology (see below). One of these (UALVP 33902; Fig. 35F) can be identified as a cervical vertebra, as it has a relatively low and wide neural spine compared to other vertebrae. This cervical shows no hypapophysis on the ventral surface of the centrum, and absence of the hypapophysis is considered as a synapomor- phy for the Helodermatidae (Hoffstetter and Gasc, 1967, 1969; Pregill et al., 1986; Estes et al., 1988). The remainder are well- preserved dorsal vertebrae, having a more or less tapered cen- trum, and a condyle with a poorly developed flange. They are definitely of helodermatid type, as they have a slightly concave lateral border of the centrum, and in the posterior half of the centrum, the lateral borders are subparallel (Hoffstatter and Gasc, 1996 GAO AND FOX— LATE CRETACEOUS LIZARDS 85 1969; Pregill et al., 1986). The neural spine is narrow and tall, but is vertically directed on the dorsal vertebrae, rather than ris- ing at an acute angle as in other helodermatids (Pregill et al., 1986). In addition, the dorsal vertebrae of Paraderma bogerti differ from those of Palaeosaniwa canadensis and other varanids in having a less pronounced precondylar constriction, an anteri- orly oriented condylar cup, and well-developed zygantra and zygosphenes. These character states are clearly shown on UALVP 33903 (Fig. 34D-F, 35G-I) from the KUA-1 locality, and UALVP 33387 (Fig. 35J, K) from the Wounded Knee lo- cality. Another important feature that these vertebrae show is that the diapophysis and parapophysis are separated; the condition in extant Heloderma in this aspect has not been clearly described in previous publications to our knowledge, but Boulenger’s ( 1 89 1 :fig. 5) figure seems to show that the two processes in Hel- odernia are jointed to form a synapophysis (sensu Romer, 1956), as in varanids. Limb Bone. An unnumbered right humerus from the KUA-1 locality is so far the only limb bone known for Paraderma. Al- though not articulated, the humerus was preserved close to the vertebrae described above: its proximal end is about 10 mm away from a dorsal verteba in the claystone matrix. The following characters are used to distinguish the humerus of Paraderma from that of other large varanoids; the humerus is robustly built, with a short shaft and weakly expanded proximal and distal ends. The proximal articular surface is wide and suboval in outline, and the deltoid crest is short and nearly perpendicular to the long axis of the proximal articular surface, differing from the condi- tion in Saniwa and Varanus (see Gilmore, 1922). The distal end of this humerus lacks well-ossified condyles and an ectepicon- dylar foramen, which are present in the two varanids mentioned above (see Gilmore, 1922). Discussion. — Estes (1964) founded Paraderma bogerti on several specimens from the Lance For- mation, Wyoming. The holotype (UCMP 54261; see Estes, 1964:fig. 64) is a nearly complete left maxillary with thick osteoderms fused to its lateral surface. As Estes (1964, 1983a) stressed, the obvi- ous characters for this form are the Heloderma- like osteoderms on the maxillary and the strong medial curvature of the nasal process. However, in contrast to Estes’ (1964:133) statement that “internally the maxillary is identical with that of Parasaniwa," the holotype of Paraderma bogerti differs from the maxillary of Parasaniwa in having a great dorsal thickening of the supradental shelf, which forms a strong fold of the shelf, and having double openings for the superior alveolar canal (personal observa- tion). The dentary morphology of Paraderma bogerti is poorly known, as previous collections include only one dentary fragment (UCMP 49939), which was briefly described and figured (Estes, 1964; Pre- gill et al., 1986:fig. 8). The newly discovered spec- imens from the Scollard and Frenchman formations reveal, contrary to what Estes (1964, 1983a) pre- sumed, that the teeth of this lizard are much more high-crowned (UALVP 29845) than those of Par- asaniwa, and the intramandibular septum is free from the floor of the Meckelian canal (UALVP 29846). The dentary of Paraderma is clearly distin- guishable from that of Parasaniwa from the same horizon: in addition to its much larger size and mas- siveness, the dentary of Paraderma has a more ro- bust but less crescentic tooth-bearing border, much stronger symphyseal process, and a free ventral bor- der of the intramandibular septum, while the septum in Parasaniwa is mostly fused to the floor of the Meckelian canal (Estes, 1964). Estes (1964) originally placed Paraderma togeth- er with Parasaniwa in his Parasaniwidae (= Nec- rosauridae* Hoffstetter, 1943) “on the basis of gen- eral tooth form [which was largely unknown then] and simple basal infolding of tooth bases” (Estes, 1964:134); he also noted close resemblances to Hel- oderma in premaxillary morphology, snout shape, and osteodermal pattern. Accepting Hoffstetter’s (1969) interpretation that Parasaniwidae and Nec- rosauridae* are closely related groups, and recog- nizing the incipient venom groove on the teeth of Paraderma, Estes (1983a) synonymized Parasani- widae with Necrosauridae*, and allocated Parader- ma to the Helodermatidae, albeit with some uncer- tainty. That Paraderma is a helodermatid has been confirmed by a more recent revision of fossil hel- odermatids (Pregill et al., 1986). The discovery of the parietal of Paraderma bog- erti (UALVP 33385) is of special importance, as it provides solid evidence to clarify some taxonomic and evolutionary controversies concerning the Hel- odermatidae. Gilmore (1928:158) placed an incom- plete parietal (YPM 1063) from the Lance Forma- tion that Marsh (1892:450) had referred to Chamops segnis as “the upper part of the cranium” in his “ Megasaurus robustus ” (this name is a nomen du- bium, although Estes, 1983a listed it as the syn- onym of Palaeosaniwa canadensis ). Estes (1964: 135; 1983a:95, 177) referred YPM 1063 to Par- aderma bogerti, but ambiguously commented that it “appears to belong to Palaeosaniwa as shown by the UCMP specimen of the latter.” With the dis- covery of UALVP 33385, the characters of which demonstrate a taxonomic association with the ho- lotype maxillary of Paraderma bogerti, the referral of YPM 1063 to P. bogerti is probably in error. In fact, Gilmore’s (1928:pl. 27, fig. 3) figure shows that YPM 1063 is certainly not helodermatid-like, since its strongly developed lateral flange indicates that the temporal musculature originated from the dorsolateral surface, while in helodermatids the 86 BULLETIN CARNEGIE MUSEUM OF NATURAL HISTORY NO. 33 temporal muscles are attached to the ventrolateral surface of the parietal table (Pregill et al„ 1986). Alternatively, an association of YPM 1063 with Pa- laeosaniwa, as Estes (1983a) suggested, is possible but quite uncertain, because skull elements of Pa- laeosaniwa are poorly known. The vertebrae of Paraderma bogerti were previ- ously unknown (Pregill et ah, 1986). With the dis- covery of well-preserved vertebrae of this species from the Scollard and Frenchman formations, now it is possible to draw the following conclusions: The vertebrae of Paraderma share with those of other helodermatids character states such as a constricted neural canal and a laterally compressed centrum; however, they are distinguishable from those of oth- er forms (e.g., Lowesaurus', see Pregill et ah, 1986: fig. 5) in having much stronger zygantra and zyg- osphenes, and the narrow and tall neural spine, at least in the dorsal vertebrae, is vertically directed (contra Pregill et ah, 1986: neural spines rising at an acute angle as a diagnostic feature for the Hel- odermatidae). A Tertiary (upper Eocene or lower Oligocene) European helodermatid, Eurheloderma Hoffstetter, 1957, is characterized by having an elongate pari- etal that is strongly constricted in its middle portion (Hoffstetter, 1957; Estes, 1983a; Pregill et ah, 1986). Yatkola (1976) regarded the constricted pa- rietal as a derived condition, but lacked support from fossil evidence. Pregill et ah (1986:190) point- ed out that Yatkola’s consideration “may be correct, but the parietal of Paraderma must be found to ver- ify the polarity.” Now, with the discovery of UALVP 33385, it is clear that a wide and trapezoi- dal parietal is a primitive condition in the Helod- ermatidae. This statement is also supported by the discovery of another helodermatid parietal from the Oldman Formation (see later description on “Hel- odermatidae, genus and species undetermined”). Genus Labrodioctes, new genus Etymology. — Labros + dioktes (Greek, masculine), meaning “greedy hunter,” in reference to the presumed predatory habit (feeding style) of this helodermatid. Type and Only Known Species. — Labrodioctes montanensis, new species. Range. — Upper Cretaceous, North America. Diagnosis. — As for the type and only known spe- cies. Labrodioctes montanensis, new species (Fig. 36A-D) Holotype. — ANSP 18664, incomplete left den- tary having five teeth and spaces and broken bases of six others. Type Locality and Horizon. — Blackbird Ridge site, in NW 1/4, Sec. 6, Tp. 7, R 20, Golden Valley County, central Montana; Upper Cretaceous Judith River Formation (Judithian). Referred Specimen. — RTMP 78.18.1, a fragmentary frontal with thick, fused osteoderms; this referral is tentative, since the specimen is anatomically noncomparable to the holotype dentary and is from a different locality. Known Distribution. — Upper Cretaceous Judith River Formation, eastern Montana; Oldman For- mation, southeastern Alberta (both Judithian). Diagnosis. — A large helodermatid differing from Paraderma as follows: teeth nontrenchant, strongly recurved, having posterior venom groove; dentary tooth-bearing border thickened, strongly curved ventrally. Differing from all other helodermatids by having the following character states: dentary great- ly elongate but robustly built; marginal teeth lack- ing anterior venom groove; posterior groove shal- low, only extending to half of crown height. Description. — The holotype, ANSP 18664 (Fig. 36A), is a large, robust dentary of about three times the size of the dentary of Paraderma. The specimen is anteriorly complete, but has a broken surface lateral to the seventh through 1 1th tooth positions. The dentary has a robust, crescentic tooth-bearing border, which is more strongly dorsoventrally concave than in Paraderma. but the curvature is not as strong as in extant Heloderma. The an- terior end of the bone strongly curves to the mandibular sym- physis and bears a prominent medioposterior projection (the symphyseal process). The anterior half of the Meckelian canal is an extremely shallow sulcus that opens ventrally, curving to the strong mandibular symphysis. The posterior half of the canal is greatly deepened and triangular in medial view, widely open me- dially with the splenial missing. A well-developed intramandi- bular septum posteriorly separates the Meckelian canal from the lateral inferior alveolar canal. The septum is posteriorly broken, but the preserved part has a free ventral border, as in L.ancian Paraderma (see above description). The posterior notch for the Meckelian cartilage is not preserved on this specimen. As preserved, the dentary has 1 1 tooth positions, including five teeth (four complete and one with broken tip), three broken bases, and three vacant spaces as a result of tooth replacement. The dentary teeth are robust, having greatly thickened and fluted bases, and pointed and strongly recurved crowns. Although the tooth crowns have weak anterior and posterior cutting edges (without serrations), they are not laterally compressed and are nontrenchant compared to those in Paraderma bogerti. The fifth and the seventh have a shallow groove on the posteromedial surface of the teeth. These are probably venom grooves, which are different from the grooves of other helodermatids: other fossil and extant helodermatids have a clearly defined anterior venom groove extending to or near the tooth apex, whereas the groove on ANSP 18664 is shallow and posteromedial, and only extends to half of the crown height. These teeth are widely spaced from each other and are pleurodont, having the upper half of the tooth height projecting above the lateral parapet of the jaw. The last two tooth positions on the preserved part of the dentary are as- sociated ventrally with a sudden narrowing of the tooth-bearing 1996 GAO AND FOX— LATE CRETACEOUS LIZARDS 87 Fig. 36. — Labrodioctes montanensis, new genus and species, Judith River and Oldman formations, Montana and Alberta: A, ANSP 18664 (holotype), incomplete left dentary, medial view; B-D, RTMP 78.18.1, incomplete left frontal, lateral, dorsal, and ventral views. Helodermatidae, genus and species undetermined, Oldman Formation, Alberta: E, F, RTMP 88.36.212, incomplete frontals, dorsal and ventral views; G, H, RTMP 85.58.58, incomplete parietal, dorsal and ventral views; I, UALVP 33338, toothed left pterygoid, ventral view. Scale = 6 mm. border, where the anterior inferior alveolar foramen is located. This condition, plus the free ventral border of the intramandibular septum, indicates that probably only the posteriormost one or two teeth are missing from this specimen, yielding a total count of about 12-13 for the dentary tooth row. The lateral surface of the dentary is mostly preserved, with the posterodorsal part broken. The surface is smooth, more flat than convex, and anteriorly turns medially to the mandibular symphysis. The four lateral mental foramina (lateral inferior al- veolar foramina) that open to the surface are about equal in size and are evenly spaced from one another. These foramina are more or less almond-shaped, and are horizontally aligned along the middle depth of the jaw. RTMP 78.18.1 (Fig. 36B-D), a robust frontal from White 88 BULLETIN CARNEGIE MUSEUM OF NATURAL HISTORY NO. 33 Rock Coulee, is compatible in size with ANSP 18664, and is definitely helodermatid-like in having thick, polygonal osteo- derms fused to the skull elements (McDowell and Bogert, 1954), and the prefrontal and postfrontal incisions approaching each other above the orbits (Estes et al., 1988; see Fig. 36B). The Oldman Formation in Alberta is geologically equivalent to the upper part of the Judith River Formation in Montana (e.g., L. S. Russell. 1964; Eberth and Hamblin, 1993). Therefore, ANSP 18664 and RTMP 78.18.1 may pertain to the same species, un- less another helodermatid of that size within the same geologic and geographic range is involved (for which there is no evi- dence). Discussion. — The dentary described above (ANSP 18664) represents a Judithian varanoid, as it bears teeth with basal infolding, which is a var- anoid synapomorphy (see Estes et al., 1988). More- over, ANSP 18664 is Paraderma- like, as it shows a robust tooth-bearing border that is strongly curved dorsomedially at the anterior end. However, the specimen is very large and elongate, and is different from Lancian Paraderma in having a much stronger mandibular symphysis, and strongly recurved, much less trenchant teeth that lack an anterior ven- om groove. These differences are great enough to separate it from Paraderma bogerti at the generic level. The lack of an anterior venom groove on the dentary teeth may weaken its referral to the Hel- odermatidae, as the presence of such a groove is regarded as one of the few unambiguous synapo- morphies for the family (Bogert and del Campo, 1956; Estes et al., 1988; but see also Norell et al., 1992); however, several other character states strongly support a placement of it within the family. These include: a robust and medially curved tooth- bearing border, strong mandibular symphysis with a prominent process, and robust teeth that are broad- ly based and weakly infolded. On the basis of these character states, Labrodioctes montanensis is ten- tatively placed in the family Helodermatidae, while the lack of an anterior venom groove should be re- garded as a primitive condition in the family (and it now is clear that the Lancian Paraderma has a weaker groove than Tertiary and Recent heloder- matid species). Mesozoic helodermatid or helodermatid-like fos- sils are uncommon. Besides the Lancian Paraderma bogerti and Judithian Labrodioctes montanensis from North America, Gobiderma pulchra Borsuk- Bialynicka, 1984, and Estesia mongoliensis Norell et al., 1992, are known from the Barun Goyot For- mation (late Campanian in age. Fox, 1978; Jerzyk- iewicz and Russell, 1991), East Asian Gobi Desert. The holotype of Gobiderma pulchra (ZPAL MgR- II 1/64) is a nearly complete skull with a partial man- dible from the Khermeen Tsav II locality. Although she noted the helodermatid similarities of this Gobi species in general morphology of the skull, osteo- dermal pattern, and construction of the jaw, Borsuk- Bialynicka (1984:39) placed Gobiderma in “nec- rosaurian grade, family uncertain.” In a review of helodermatid evolution, Pregill et al. ( 1986) empha- sized that Gobiderma is more generalized in skull morphology than Heloderma, and stated that “the overall appearance of the skull recalls that of Xe- nosaurus” (Pregill et al., 1986:195). From personal examination of the holotype and referred specimens of Gobiderma pulchra, we here offer the following comments: 1) Gobiderma is definitely a varanoid, but not a xenosaurid, because it has plicidentine in- foldings of the tooth bases (Borsuk-Bialynicka, 1984; personal observation). 2) Gobiderma shares several characters that are in combination unique to helodermatids: the jaw adductors have a ventral or- igin on the parietal; the supradental shelf curves medially anteriorly; the osteoderms are tuberculate and fused to the skull elements; and the tooth num- ber is reduced in correspondence with the shorten- ing of the jaws. 3) However, Gobiderma differs from all known helodermatids in having a more pointed and longer snout, and in retaining a supra- temporal opening. From the above comparisons, Gobiderma appears to be more closely related to the Helodermatidae than to any other varanoids. We agree with Borsuk- Bialynicka (1984), who regarded Gobiderma as “an Asian substitute of the American helodermatids,” and not ancestral to the extant American heloder- matids. North American Mesozoic helodermatids are mostly based on disarticulated jaws and skull elements while the teeth of Gobiderma are poorly known; therefore, the precise relationships and tax- onomic status of Gobiderma will remain obscure until further study of better preserved materials for both North American and Asian forms. No matter whether or not the East Asian Gobi- derma and Estesia are included in the Heloderma- tidae, the North American Judithian Labrodioctes represents the earliest fossil record for the family, because the age of the Barun Goyot Formation in the Gobi Desert is probably younger than the Judith River and Oldman formations (e.g.. Fox, 1978; Jer- zykiewicz and Russell, 1991). Genus and species undetermined (Fig. 36E-I) Specimens. — RTMP 88.36.212, incomplete par- tially fused frontals; RTMP 85.58.58, incomplete parietal; UALVP 33338, a left pterygoid. 1996 GAO AND FOX— LATE CRETACEOUS LIZARDS 89 Locality and Horizon. — See description of each specimen below. Description. — The incomplete partially fused frontals, RTMP 88.36.212 (Fig. 36E, F), from the Railway Grade locality, are roughly triangular in outline as preserved, and are covered with patches of pitted osteoderms. In structure and thickness, these osteoderms are different from those on RTMP 78.18.1 (described above), but are compatible with those on a parietal described below. Although anteriorly broken, symmetrical articular sur- faces on both sides indicate that the nasal extended posteriorly to overlap the frontals. Pre- and postfrontal incisions are not pre- served, and it cannot be determined whether or not the pre- and postfrontals were in contact above the orbit. The frontals are posteriorly broken, and thus the nature of the frontoparietal su- ture is unknown. In ventral view, the specimen shows a midline suture and breakage at the anterior end of the left subolfactory process such that the morphology of the process is unknown. RTMP 85.58.58 (Fig. 36G, H), also from the Railway Grade locality, consists of the main body of a parietal table, with the posterior part and the supratemporal processes broken off. The parietal is broad and trapezoidal, lacking a strong constriction of the middle part of the table. Besides its large size and different texture of osteoderms, this parietal differs from that of Parasan- ixva in having the parietal fossa (ventral view) close to the pa- rietal foramen, and more importantly, in having a roof-like dor- solateral extension for the attachment of temporal musculature to the ventral surface of the parietal table (see Fig. 36H). The latter feature is one of the characters that distinguish heloder- matids from varanids (Pregill et al., 1986). In ventral view, the specimen shows a small but deep parietal fossa, which is pro- portionally much smaller than that in Paraderma bogerti and is more closely located to the parietal foramen. The specimen is distinguishable from the parietal of Paraderma by its large size, large parietal foramen at the center of the parietal table, and thin, rounded tuberculate osteoderms. The parietal of Paraderma has a smaller parietal foramen close to the frontoparietal suture, and thick, polygonal osteoderms divided by deep grooves. Although collected from roughly the same horizon, this parietal is not com- patible in size and robustness with the dentary and referred fron- tal of Labrodioctes (see above); therefore, it probably represents a helodermatid other than Labrodioctes. UALVP 33338 (Fig. 361) was collected by D. Brinkman from the bank of the South Saskatchewan River, about 8 km upstream from Sandy Point. The specimen is the anterior part of a toothed left pterygoid that can be tentatively referred as to “Heloder- matidae, genus and species undetermined” based on its signifi- cantly large size and Heloderma-Y\ke. morphology. On this spec- imen, the basal plate is well preserved, and the anteromedial (or palatine) process is nearly complete, with only the anterior tip broken. The process is short and broad, having an anteromedial depression for the pterygoid— palatine articulation. The anterolat- eral (or ectopterygoid) process is broken off from the basal plate; however, the specimen clearly shows a triangular ventral depres- sion between the diverging palatine and ectopterygoid processes; such a triangular depression indicates a short and wide skull, as it is developed in Heloderma. but not in Lanthanotus and Var- anus, in which the skull tends to be longer and more slender. The posterior (or quadrate) process, arising in front of the pter- ygoid—basisphenoid contact, is also missing. Posteriorly from the anterior part of the palatine process, the pterygoid bears a heavily toothed area parallel but inset from the medial border of the basal plate. The teeth are fused to the plate in an apparently random manner such that no particular pattern can be recognized. Discussion. — The general morphology of RTMP 85.58.58 clearly indicates that the parietal is hel- odermatid-like, as it shows a roof-like lateral flange for a ventral origin of the temporal musculature, and osteoderms are fused to the dorsal surface of the bone. However, the specimen is incompatible with RTMP 78.18.1 (frontal referred to Labrodioctes) in robustness and surface texture of the osteoderms, but in these aspects is compatible with the frontals (RTMP 88.36.212) from the same locality (compare Fig. 36C, E, G). Therefore, the parietal and the fron- tals probably represent another helodermatid of Ju- dithian age. The parietal of Labrodioctes montanen- sis is unknown, but is likely to have been much more robust and to have had thicker and more dense osteoderms than in this specimen. In spite of its relatively large size, RTMP 85.58.58 is more lightly built than the parietal of Paraderma bogerti, which is only half the size of the former; and this Judithian form has a proportionally larger parietal foramen at the center of the parietal table, which represents a more primitive condition than in Lancian Parader- ma. As described above, the pterygoid UALVP 33338 is large and likely belongs to a helodermatid. The size and robustness of this specimen are compatible with both Palaeosanixva and Labrodioctes (in both genera, the pterygoid is unknown), and the possi- bility of its association with the former genus can- not be ruled out at present. Here we tentatively refer this specimen to “Helodermatidae, genus and spe- cies undetermined” based on its helodermatid-like morphology, leaving its referral indefinitive pending study of more and better preserved materials when these become available. Family Varanidae Gray, 1827 The extant varanids include approximately 34 species in the single genus Varanus Merrem. 1820, and are confined to the Old World, with a distri- bution in Australasia, southeast Asia, and Africa. Although absent from the New World today, fossil varanids have long been known from North Amer- ica since Leidy reported Saniwa ensidens from the Eocene of Wyoming in 1870. In western Canada, Palaeosanixva canadensis Gilmore, 1928, from the Oldman Formation in Alberta marks the earliest def- inite appearance of the Varanidae in North America, and may provide the earliest record of the family, pending on the age and relationships of Telmasau- 90 BULLETIN CARNEGIE MUSEUM OF NATURAL HISTORY NO. 33 i i i i i i i i i i i Fig. 37. — Palaeosaniwa canadensis, Oldman Formation, Alber- ta: RTMP 89.121.1, incomplete right dentary, lateral (above) and medial (below) views. Scale = 10 mm. rus Gilmore, 1943, from the Djadochta Formation, Gobi Desert (see also Borsuk-Bialynicka, 1984). Genus Palaeosaniwa Gilmore, 1928 Type Species. — Palaeosaniwa canadensis Gil- more, 1928. Range. — Upper Cretaceous, North America. Diagnosis. — As for the type and only known spe- cies. Palaeosaniwa canadensis Gilmore, 1928 (Fig. 37, 38) Megasaurus robustus Gilmore, 1928:pl. 27, fig. 1, la. Holotype. — USNM 10864, a nearly complete dorsal vertebra collected by W. E. Cutler in 1913 from the Oldman (then Belly River) Formation “near Steveville” (Gilmore, 1928:84), southeastern Alberta. Referred Specimens. — RTMP 80.29.120, 80.20.37, incomplete maxillaries; UALVP 33337, RTMP 85.36.184, 89.36.101, 89.121.1, tooth-bearing dentaries; RTMP 73.23.1, 80.16.439, 81.22.55, 81.26.22, 82.14.42, 82.19.106, 82.31.30, 85.43.24, 85.65.13, 85.67.24 (total: ten), and UALVP 33339-33345 (total: seven), all isolated vertebrae; RTMP 73.7.1, 81.23.47, tibiae. Locality and Horizon. — The specimens listed above were collected from various localities of the Oldman Formation in southeastern Alberta; UALVP specimens are mostly from Sandy Point and nearby localities, and RTMP specimens are mostly from the Bonebed microfossil sites in DPP and nearby lo- calities (coordinates on file in UALVP and RTMP). Known Distribution. — Upper Cretaceous Oldman Formation, southeastern Alberta; Judith River For- mation, Montana (Sahni, 1972; see discussion be- low). Post-Judithian occurrences of this species have been reported from the Lance Formation (Es- tes, 1964), Hell Creek Formation (Estes et al., 1969), and Fort Union Formation (Sullivan, 1982), but see discussion below. Diagnosis ( Revised from Gilmore, 1928; Estes, 1983a). — A Late Cretaceous large varanid from North America, closely resembling Saniwa Leidy, 1870, in vertebral morphology, but differing from the latter and Recent Varanus in having the follow- ing combination of character states: narial retraction short, restricted to three or four anteriormost tooth positions; marginal teeth relatively short, robust, and strongly recurved; tooth bases greatly expand- ed, crowns narrower, slightly trenchant, having mi- croserrations on both anterior and posterior cutting edges; vertebral neural spines short and robust, rod- like, with strong anterior crest. Description. — Among the specimens available for this study, several maxillaries and dentaries as well as other skull elements can be confidently referred to Palaeosaniwa canadensis on the basis of their large size (compatible with the size of the holotype vertebra) and varanid tooth form. These specimens are important for providing, for the first time, information about characters oth- er than vertebrae, on which the species was primarily based and from which the species has been long known, except for isolated teeth reported by Estes (1964) from the Lance Formation. Maxillary. Included in the collections for this study are two maxillaries that are the first known for Palaeosaniwa canadensis: RTMP 80.29.120 (Fig. 38A, B), a left maxillary fragment from Sandy Point, South Saskatchewan River, about 20 km south of Empress; and RTMP 80.20.37 (Fig. 38C, D), a right maxillary fragment from BB 118 microfossil site (DPP), southeastern Al- berta. The specimen RTMP 80.29.120 is especially important for it is the only known maxillary that bears a complete tooth of this species. The maxillary tooth is short, robust, and strongly re- curved. The tooth base is greatly expanded and infolded, forming the so-called “honeycomb structure” (term used by Estes, 1964). A small, round foramen is developed at the medial side of the tooth base. From the expanded tooth base, the tooth is greatly reduced in its thickness towards the apex, but has a medially swollen shaft. The tooth crown is more or less laterally com- pressed, strongly recurved, and has slightly trenchant anterior and posterior edges that bear microserrations. The maxillary shows a well-developed posterior interior alveolar foramen, above and slightly posterior to the tooth that remains. This fo- ramen indicates that the single tooth on the specimen is one of the posterior maxillary teeth, which may be slightly shorter and more robust than the teeth adjacent anteriorly, which are in the middle part of the tooth row. Although fragmentary, RTMP 80.29.120 shows that the lateral surface of the maxillary is or- 1996 GAO AND FOX— LATE CRETACEOUS LIZARDS 91 Fig. 38. — Palaeosaniwa canadensis, Oldman Formation. Alberta: A, B, RTMP 80.29.120, left maxillary fragment, lateral and medial views; C, D. RTMP 80.20.37, right maxillary fragment, lateral and medial views; E. F, RTMP 89.121.1, incomplete right dentary, lateral and medial views; G, H, RTMP 89.36.101, incomplete left dentary, lateral and medial views; I, RTMP 85.36.184. right dentary fragment, medial view; J, UALVP 33342, dorsal vertebra, lateral view; K, UALVP 33344. dorsal vertebra, dorsal view. Scale = 6 mm. 92 BULLETIN CARNEGIE MUSEUM OF NATURAL HISTORY NO. 33 namented at least partially with bony rugosities that are in texture and thickness different from the fused osteoderms on the frontal and parietal of helodermatids from the same horizon (see fore- going description). Most varanids have osteoderms reduced or absent, and when present, they are not fused to the skull ele- ments; however, Varanus kornodoensis has heavy osteoderms bonded to skull elements (McDowell and Bogert, 1954:pl. 12). RTMP 80.20.37 (Fig. 38C, D) is the anterior fragment of a right maxillary, which bears a broken tooth and the base of an- other. The maxillary has a strong base for the premaxillary pro- cess, and the facial process rises as a gentle slope behind this base. The length of the premaxillary process and the gentle slope of the nasal process are sharply different from Parasaniwa, but indicate a narial retraction to the point of the third or the fourth tooth position. Medial and anterior to the facial process is the anterior interior alveolar foramen, which is almost as large as the posterior interior alveolar foramen (shown on RTMP 80.29.120). This maxillary fragment also shows surface sculpture similar to that on RTMP 80.29.120 described above. Dentary. In this study, four dentary specimens (RTMP 89.36.101, 89.121.1, 85.36.184, and UALVP 33337) are referred to Palaeosaniwa canadensis. The large size and robustness of these dentaries are consistent with the size and proportions of the holotype vertebra. The dentary teeth, having microserrations, are easily distinguishable from those of other North American Cretaceous varanoid taxa, such as Paradenna and Parasaniwa. Among the four dentaries mentioned above, RTMP 89.121.1 (Fig. 37; 38E, F) from a locality (RTMP LI 700) near Onefour, Alberta, is the best preserved in terms of showing tooth form and the dentary structure of this early varanid. This specimen has four broken teeth and the spaces for four others, representing the last eight positions of the dentary tooth row. Medially, as commonly seen in other varanoids, RTMP 89.121.1 shows that the dental gutter on the medial side of the tooth row is entirely lost, and the subdental shelf is replaced by a ventrally growing ridge that forms a crescentic tooth-bearing border; however, this border is much less crescentic than that of Parasaniwa and is different from that in helodermatids in having a sharp ventral edge. The Meckelian canal on this specimen is mostly open ven- trally, with only the posterior portion below the last three teeth facing more medially than ventrally. A short but well-defined intramandibular septum is developed below the last three and one-half tooth positions. It has a free ventral border that is clearly separated from the floor of the Meckelian canal and is developed as a prominent ridge-like structure, differing in those respects from the septum in Parasaniwa. The posterior edge of the sep- tum is deeply notched for the Meckelian cartilage. Dorsal to the intramandibular septum is the coronoid facet, a deep groove run- ning under the posterior tooth-bearing border for an interlocking articulation with the internal anterior process of the coronoid bone. The facet extends anteriorly beyond the anterior end of the intramandibular septum by half of a tooth position; that is, the facet is located below the last four tooth positions. Four teeth are preserved on RTMP 89.121.1, but all have the crown apex missing by breakage. The teeth are pleurodont, with half of their height attached to the lateral parapet of the jaw. As seen on the maxillary (RTMP 80.29.120), the tooth bases are greatly expanded and infolded with complex plicidentine. Each tooth has a basal foramen opening medially at a strong basal cement ring. The tooth shaft above the basal infoldings has a flattened lateral surface, but is medially swollen and anteriorly depressed, giving a roughly semicircular but anteriorly concave cross section (see Fig. 37). Although the tooth apices are all broken on this specimen, the parts that remain indicate that the crowns are laterally compressed and slightly recurved. Two teeth on the jaw (the six and seventh from the back) clearly show microserrations on the anterior cutting edge of the crown. The last three teeth are obviously smaller than those anterior to them. Anteriorly, the tooth-bearing border of this specimen curves dor- sally, indicating that the anterior breakage is close to the man- dibular symphysis. Judging from tooth size and the curvature of the tooth-bearing border, we estimate that the complete dentary tooth row of this lizard contained 11-12 positions. The lateral surface of RTMP 89.121.1 is smooth with no os- teodermal rugosities, and bears inferior alveolar foramina that are widely spaced from each other; the last foramen is located close to the last dentary tooth, as in extant Varanus. The lateral surface of this specimen is shallowly concave posterodorsally. The con- cavity in Heloderma houses the venom gland, while in other varanoids it houses a large seromucous gland (Bogert and del Campo, 1956; Kochva, 1974; Saint-Girons, 1976). The depres- sion of Palaeosaniwa is probably for a seromucous gland, as there are no venom grooves developed on the dentary teeth of this lizard. In spite of this depression, the dentary is generally straight in dorsal view, lacking the distinct sigmoid curvature of helodermatids (McDowell and Bogert, 1954). Of the other three dentaries referred to the same species, RTMP 89.36.101 (Fig. 38G, H) is an incomplete left dentary having six tooth positions (three broken teeth and the spaces for three others). The specimen shows that the jaw structure and the tooth form agree in every aspect with those of RTMP 89.121.1. except for its slightly larger size (perhaps indicating an older individual). The other two specimens (UALVP 33337 and RTMP 85.36.184) are both fragmentary right dentaries with broken teeth. UALVP 33337, from a site about 8 km upstream from Sandy Point, represents the largest individual among the four specimens. RTMP 85.36.184 (Fig. 381) from BB 25 (DPP) clear- ly shows the basal infoldings and the cement around the tooth base. Both specimens show that the inner surface of the Meck- elian canal is more deeply grooved than in the jaw of Paradenna. Vertebra. As listed above, some 17 disarticulated vertebrae are referred to Palaeosaniwa canadensis on the basis of their resemblances in size and morphology to the holotype (USNM 10864) and other vertebrae that Gilmore (1928) described. There are no cervical vertebrae, as none of these show a hypapophysis, a structure that is diagnostic for the cervicals. Most of these are dorsal vertebrae, showing a morphology that is basically the same as Gilmore ( 1928) described; however, UALVP 33342 (Fig. 38J) has the neural spine completely preserved, which has pre- viously been unknown in this species. The discovery of this ver- tebra supports Estes’ (1964) interpretation that the spine is not expanded or plate-like, and in this aspect differs from Saniwa and extant Varanus. The zygantrum and zygosphene of these dorsal vertebrae are far less well developed than those in Par- aderma. and are sharply distinguishable from the latter in having an arched neural canal in anterior view (best shown on UALVP 33344; Fig. 38K), while in Paradenna, the canal is compressed owing to the development of strong articular surfaces of the zyg- antrum. RTMP 82.3 1 .30 from BB 23 (DPP) is probably the only caudal vertebra in the available collections for this study. It differs from presacral vertebrae in having much weaker zygapophyses and basically no zygosphene. Ventrally, the centrum differs from that of the dorsal vertebrae in lacking a flattened surface but having a longitudinal groove between a pair of prominent ridges. The transverse processes are mostly broken off, but the preserved 1996 GAO AND FOX— LATE CRETACEOUS LIZARDS 93 bases indicate that they are posterolaterally directed and probably slender and short. The posterior half of the centrum is broken, and thus, neither the condyle nor the articulation for the chevron is preserved. Tibia. Two specimens (RTMP 73.7. 1 , 81 .23.47) from DPP are both left tibiae, and are referred to Palaeosaniwa canadensis on their compatibility in size with the vertebrae described above and resemblance to the tibia known for Saniwa (Gilmore, 1928). The two tibiae differ from that of Varanus in having a well-developed sigmoid crest laterally on the proximal end; Saniwa has a similar but longer and stronger crest at the same location (Gilmore, 1928:pl. 6). The distal end of RTMP 73.7.1 has a strong lateral condyle that extends well beyond the medial condyle and in this aspect differs from Varanus. Discussion. — Gilmore (1928) founded Palaeo- saniwa canadensis on four isolated dorsal vertebrae (UA 112; USNM 10864, 11045; GSC 8510). The type specimen (USNM 10864) that he selected for the species was collected by W. E. Cutler in 1913 from a site near Steveville, Red Deer River, south- eastern Alberta. Gilmore’s naming of Palaeosaniwa canadensis is significant, as it documents the first North American Mesozoic varanid recognized and the first discovery of a terrestrial lizard from the Oldman Formation. On the other hand, his naming of this lizard has created a long-term problem, be- cause the species is founded solely on vertebral specimens, and vertebrae of lizards are much less diagnostic than jaw materials especially at lower taxon (generic and specific) levels. Accordingly, this lizard has long been poorly understood. Gilmore (1928:83) placed Palaeosaniwa cana- densis in the family Varanidae on the basis of “the tapering form of the centrum with constricted con- dylar end’’ and the general similarity compared to the vertebrae of Saniwa Leidy, 1870, a North Amer- ican Eocene varanid. This familial assignment was questioned by McDowell and Bogert (1954:53), who stated that Palaeosaniwa “although definitely platynotan, [is] not positively varanid.” However, Estes (1964) added tooth form (trenchant crown with microserrations plus complex basal infoldings) as new evidence to support the varanid affinity of Palaeosaniwa. Estes (1983a) cited a Palaeosaniwa skull associated with vertebrae in the UCMP col- lections; unfortunately, this important discovery has never been described or otherwise mentioned since. In the present study, maxillary and dentary ma- terials are described for the first time for this early varanid. These materials are obviously compatible in size with the vertebrae on which the species is founded, and are morphologically clearly distin- guishable from homologous parts of Parasaniwa, Paraderma, and Labrodioctes, the three other known North American varanoids of the same age. Therefore, these specimens are referred to Palaeo- saniwa. Estes (1964) referred several vertebrae and iso- lated teeth from the Lance Formation to Palaeosan- iwa canadensis. If the referral is correct, it will ex- tend the geologic range of this varanid species from Judithian to Lancian. However, there are some un- certainties regarding the identification of the speci- mens: first, the isolated teeth without referrence to jaw structure are hard to demonstrate as different from those of Saniwa or other varanoids; and the “microserrations” that Estes (1964) figured seem too small and dense, compared to those on Oldman specimens described in this paper, to pertain to the same species. Second, the Lancian dorsal vertebrae (Estes, 1964:fig. 66) show much weaker diapophy- ses and a more convex ventral surface of the cen- trum than those from the Oldman Formation. These, plus a well-developed zygantrum and zygosphene, suggest that the Lancian vertebrae may represent a varanoid other than Palaeosaniwa canadensis ; com- parison should be made with helodermatids. Estes et al. (1969) reported “ Palaeosaniwa , cf. P. canadensis ” from the Hell Creek Formation, eastern Montana. The specimen for this identifica- tion was a single large vertebra (MCZ 3665) that was neither figured nor described in detail. The specimen was characterized as having “convex lat- eral borders of the centrum, as do Eocene Saniwa and both Oldman Formation and Lance Formation Palaeosaniwa" (Estes et al., 1969:22); however, lack of information on the morphologies of the con- dylar cup, precondylar constriction, and neural spine prevents a definite referral of this specimen to Palaeosaniwa. Sullivan (1982:1007) recorded “cf. Palaeosani- wa canadensis ” from a fragmentary dentary (AMNH 15957) from the Fort Union Formation. Wyoming, and stated that “the labyrinthine mor- phology of the tooth bases is the major criterion for referring this specimen to Palaeosaniwa canaden- sis. Assuming the assignment is correct, Palaeosan- iwa’s range is extended from the Late Cretaceous (Gilmore, 1928; Estes, 1964) to the middle Paleo- cene (Torrejonian).” It is clear now that presence of basal dentine infolding is broadly synapomorphic for Varanoidea (Estes et al., 1988), not a unique feature for Palaeosaniwa canadensis ; and thus, both Sullivan’s identification and his claim of a Paleo- cene record of cf. Palaeosaniwa are unconvincing. Estes (1983a: 184) made the following comments on this matter: “The fragmentary nature of the speci- 94 BULLETIN CARNEGIE MUSEUM OF NATURAL HISTORY NO. 33 men, as well as the fact that Sullivan gave no mea- surements or figure, makes any definite statement impossible. In fact, Sullivan (1982) reported a Saniwa vertebra from the same locality. From his description, however, we see no reason for assign- ing the vertebra to “cf. Saniwa sp.” or the dentary fragment to “cf. Palaeosaniwa canadensis A more likely possibility is that the two specimens pertain to the same species (which would be “cf. Saniwa sp.”). GEOLOGICAL DISTRIBUTION AND EVOLUTION OF LATE CRETACEOUS LIZARDS IN WESTERN CANADA Late Cretaceous lizards from western Canada were previously documented in scattered publica- tions (see Introduction), but no synthetic studies have ever been attempted, largely owing to lack of fossil specimens. However, over the past 25 years collections from Alberta and Saskatchewan have provided important materials for undertaking the re- search reported in this paper. The purpose of this section is to summarize the fossil record of lizards from the Upper Cretaceous of western Canada, and to discuss the evolutionary changes through time of the relevant groups in the study area. The geological distribution of these lizards is summarized in the form of two charts (Fig. 39, 40), in which the ra- diometric ages are from Harland et al. (1990) and Thomas et al. (1990). In the following discussion, we recognize three lizard assemblages on the basis of differences in geological age and taxonomic composition. Aquilan ( Early Campanian) Assemblage Aquilan, or early Campanian, lizards are known in western Canada from the upper member of the Milk River Formation, southern Alberta. The Milk River assemblage documents the earliest North American records for the families Iguanidae*, Scin- cidae, Xenosauridae, Necrosauridae*, and the ear- liest certain record of the Anguidae; and it also in- cludes several early teiids ( Chamops , Sphenosi- agon, and, possibly, Glyptogenys). Iguania. — The Milk River iguanians include Cnephasaurus and an unnamed new genus and spe- cies, which can be referred to the Iguanidae* (sensu lato, = nonacrodontan iguanians). In tooth mor- >S 5 Stage NALM Age Iguanidae Teiidae Seine idae ?Xantusiidae ?Cordylldae Family Incertae sedis 65 76 c JZ o k. 0) e a 5 -?- Lane lan Genus & species new (B) Chamops Leptochamops Meniscognathus Haptosphenus Stypodontosaurus Genus & species new (unnamed) Contogenys Genus & species undetermined (B) Genus & species new (A) Genus & species new (B) , Edmontonian ? Chamops sp. 7 S £ ■a 3 ~s Chamops sp. Socognathus Sphenosiagon Glyptogenys Gerontoseps Leptochamops Orthrioscincus Genus & species undetermined (A) Genus & species undetermined 1 J Cnephasaurus Chamops sp. Penemabuya Genus & species c . Osteology of Polvglyphanodon, an Upper Cre- taceous lizard from Utah. Proceedings of the United States National Museum, 92:229-265. . 1943a. Fossil lizards of Mongolia. Bulletin of the American Museum of Natural History, 81:361-384. . 1943 b. Osteology of Upper Cretaceous lizards from Utah, with a description of a new species. Proceedings of the United States National Museum, 93:209-214. Golz, D., and J. A. Lillegraven. 1977. Summary of known occurrences of terrestrial vertebrates from Eocene strata of southern California. The University of Wyoming, Contri- butions to Geology, 15:43-65. Goodrich, E. S. 1930. Studies on the Structure and Develop- ment of Vertebrates. Constable and Company Ltd., London, United Kingdom. Gorman, G. C. 1970. Chromosomes and the systematics of the family Teiidae (Sauria, Reptilia). Copeia, 1970:230-245. Greene, H. 1982. Dietary and phenotypic diversity in lizards: Why are some organisms specialized? Pp. 107-128, in En- vironmental Adaptation and Evolution (D. Mossakowski and G. Roth, eds.). Gustav Fischer Verlag, Stuttgart, Germany, and New York, New York. Greer, A. E. 1970. A subfamilial classification of scincid liz- ards. Bulletin of the Museum of Comparative Zoology, Har- vard University, 139:151-184. . 1991. Tooth number in the scincid lizard genus Cten- otus. Journal of Herpetology, 25:473-477. Harland, W. B., R. L. Armstrong, A. V. Cox, L. E. Craig, A. G. Smith, and D. G. Smith. 1990. A Geologic Time Scale 1989. Cambridge University Press, Cambridge, United Kingdom. Harris, V. A. 1963. The Anatomy of the Rainbow Lizard. Hutchinson and Co., Ltd., London, United Kingdom. . 1964. The Life of the Rainbow Lizard. Hutchinson and Co., Ltd., London, United Kingdom. Hatcher, J. B. 1896. Some localities for Laramie mammals and horned dinosaurs. American Naturalist, 30:112-120. Hay, O. P. 1902. Bibliography and catalogue of the fossil Ver- tebrata of North America. Bulletin of the United States Geo- logical Survey, 179:1-868. Hecht, M. 1951. Fossil lizards of the West Indian genus Aris- telliger (Gekkonidae). American Museum Novitates, 1538: 1-33. . 1956. A new xantusiid lizard from the Eocene of Wy- oming. American Museum Novitates, 1774:1—8. . 1959. Amphibians and reptiles. Pp. 130-146, in The Geology and Paleontology of the Elk Mountain and Taber- nacle Butte Area, Wyoming (P. O. McGrew, ed.). Bulletin of the American Museum of Natural History, 1 17:1 17-176. Hoffstetter, R. 1943. Varanidae et Necrosauridae tossiles. Bul- letin du Museum National d'histoire Naturelle, Paris, 15: 134-141. . 1957. Un Saurien helodermatide (Eurheloderma galli- cum nov. gen. et sp.) dans la faune fossile des phosphorites du Quercy. Bulletin de la Societe Geologique de France, 7: 775-786. . 1969. Presence de Varanidae (Reptilia, Sauria) dans le Miocene de Catalogne — Considerations sur l’histoire de la famille. Bulletin du Museum National d’histoire Naturelle, Paris, 40:1051-1064. Hoffstetter, R., and J. P. Gasc. 1967. Observations sur le squelette cervical et specialement sur les hypapophyses des sauriens varanoides actuels et fossiles. Bulletin du Museum National d'histoire Naturelle, 39:1028-1043. . 1969. Vertebrae and ribs of modern reptiles. Pp. 201- 310, in Biology of the Reptilia, Volume 1 (C. Gans, A. d’A. Bellairs, and T. S. Parsons, eds.). Academic Press, London, United Kingdom, and New York, New York. Hotton, N. 1955. A survey of adaptive relationships of denti- tion to diet in the North American Iguanidae. American Midland Naturalist, 53:88-114. ICZN. 1985. International Code of Zoological Nomenclature Adopted by the XX General Assembly of the International Union of Biological Sciences. International Trust for Zoo- logical Nomenclature and University of California Press, London, United Kingdom, and Berkeley, California. Irish, E. J. W. 1970. The Edmonton Group of south-central Al- berta. Bulletin of Canadian Petroleum Geology, 18:125-155. Jerzykiewicz, T., and D. A. Russell. 1991. Late Mesozoic stra- tigraphy and vertebrates of the Gobi Basin. Cretaceous Re- search, 12:345-377. Johnston, P. A., and R. C. Fox. 1984. Paleocene and Late Cre- taceous mammals from Saskatchewan, Canada. Palaeonto- graphica, Abteilung A, 186:163-222. Kochva, E. 1974. Glandes specialises de la machoire inferieure chez les Anguimorphes. Pp. 281-286, in Recherches Biol- ogiques Contemporaines (L. Arvy, ed.). Imprimerie Vagner, Nancy, France. Koster, E. H., P. J. Currie, D. Eberth, D. Brinkman, P. John- ston, and D. Braman. 1987. Sedimentology and palaeon- tology of the Upper Cretaceous Judith River/Bearpaw for- mations at Dinosaur Provincial Park, Alberta. Tyrrell Mu- seum of Palaeontology, Drumheller, Alberta, Canada. Langston, W., Jr. 1965. Pre-Cenozoic vertebrate paleontology in Alberta: Its past and future. Pp. 9-13, in Vertebrate Pa- leontology in Alberta (C. R. Stelck, ed.). Department of Ge- ology Bulletin 2, University of Alberta, Edmonton, Alberta, Canada. Laurin, M. 1991. The osteology of a Lower Permian eosuchian from Texas and a review of diapsid phylogeny. Zoological Journal of the Linnean Society (1991), 101:59-95. Lecuru, S. 1969. Etude morphologique de l'humerus des lac- ertiliens. Annales des Sciences Naturelles, Zoologie, 11: 515-558. Leidy, J. 1870. (Descriptions of Einys jecinsei, E. haydeni, Bae- na arenosa, and Saniwa ensidens). Proceedings of the Acad- emy of Natural Sciences of Philadelphia, 1870:123-124. Lerbekmo, J. F. 1987. Magnetostratigraphic restrictions on the age of the Frenchman Formation and the magnitude of the sub-Frenchman disconformity in southwest Saskatchewan. Bulletin of Canadian Petroleum Geology, 35:454-459. . 1989. The stratigraphic positon of the 33-33R (Cam- panian) polarity chron boundary in southeastern Alberta. Bulletin of Canadian Petroleum Geology, 37:43-47. Lerbekmo, J. F., and K. C. Coulter. 1985. Late Cretaceous to early Tertiary magnetostratigraphy of a continental se- 1996 GAO AND FOX— LATE CRETACEOUS LIZARDS 105 quence: Red Deer Valley, Alberta, Canada. Canadian Journal of Earth Sciences, 22:567-583. Li, J. 1985. A new lizard from Late Jurassic of Subei, Gansu. Vertebrata PalAsiatica, 23:13-18. Lillegraven, J. A. 1969. Latest Cretaceous mammals of upper part of Edmonton Formation of Alberta, Canada, and review of marsupial-placental dichotomy in mammalian evolution. The University of Kansas Paleontological Contributions, Ar- ticle 50 (Vertebrata 12): 1-122. . 1991. Stratigraphic placement of the Santonian-Cam- panian boundary (Upper Cretaceous) in the North American Gulf Coastal Plain and Western Interior, with implications to global geochronology. Cretaceous Research, 12:115-136. Lillegraven, J. A., and M. C. McKenna. 1986. Fossil mam- mals from the “Mesaverde” Formation (Late Cretaceous, Judithian) of the Big Horn and Wind River basins, with def- initions of the Late Cretaceous Land Mammal “Ages.” American Museum Novitates, 2840:1-68. Lillegraven, J. A., and L. M. Ostresh, Jr. 1990. Late Creta- ceous (earliest Campanian/Maastrichtian) evolution of west- ern shorelines of the North American Western Interior Sea- way in relation to known mammalian faunas. Pp. 1-30, in Dawn of the Age of Mammals in the Northern Part of the Rocky Mountain Interior, North America (T. M. Bown and K. D. Rose, eds.). Geological Society of America, Special Paper 243. Lonnberg, E. 1903. On the adaptation to a molluscivorous diet in Varanus niloticus. Arkiv for Zoologi, 1 :65-82. Lucas, S. G. 1986. Proper syntax when using aff. and cf. in taxonomic statements. Journal of Vertebrate Paleontology, 6:202. Lucas, S. G., N. J. Mateer, A. P. Hunt, and F. M. O’Neill. 1987. Dinosaurs, the age of the Fruitland and Kirtland for- mations, and the Cretaceous-Tertiary boundary in the San Juan Basin, New Mexico and Colorado. Pp. 35-50, in The Cretaceous-Tertiary Boundary in San Juan and Raton Ba- sins, New Mexico and Colorado (J. E. Fassett and J. K. Rigby, Jr., eds.). Geological Society of America, Special Pa- per 209, Boulder, Colorado. MacLean, W. P. 1974. Feeding and locomotion mechanisms of teiid lizards; Functional morphology and evolution. Papeis Avulsos Zoologia, 27:179-213. Malan, M. E. 1963. The dentitions of the South American Rhynchocephalia and their bearing on the origin of the rhyn- chosaurs. South African Journal of Science, 59:214-220. Marsh, O. C. 1872. Preliminary description of new Tertiary reptiles. American Journal of Science, Series 3, 4:298-309. . 1889. Discovery of Cretaceous Mammalia. American Journal of Science, Series 3, 38:81-92. . 1892. Notice of new reptiles from the Laramie For- mation. American Journal of Science, Series 3, 43:449-453. Mattison, C. 1989. Lizards of the World. Facts on File, Inc., New York, New York. Mayr, E., and P. D. Ashlock. 1991. Principles of Systematic Zoology. McGraw-Hill, Inc., New York, New York. McDowell, S., and C. M. Bogert. 1954. The systematic po- sition of Lanthanotus and the affinities of the anguinomor- phan lizards. Bulletin of the American Museum of Natural History, 105:1-142. McGowan, G., and S. E. Evans. 1995. Albanerpetontid am- phibians from the Cretaceous of Spain. Nature, 373: MS- MS. McKenna, M. C. 1960. Fossil Mammalia from the early Was- atchian Four Mile fauna. Eocene of northwest Colorado. University of California Publications in Geological Sciences, 37:1-130. McLean, J. R. 1971. Stratigraphy of the Upper Cretaceous Ju- dith River Formation in the Canadian Great Plains. Sas- katchewan Research Council, Geological Division Report, 1 1:1-96. . 1977. Lithostratigraphic nomenclature of the Upper Cretaceous Judith River Formation in southern Alberta: Phi- losophy and practice. Bulletin of Canadian Petroleum Ge- ology, 25:1 105-1 1 14. Meszoely, C. 1970. North American fossil anguid lizards. Bul- letin of the Museum of Comparative Zoology, Harvard Uni- versity, 139:87-149. Meszoely, C., R. Estes, and H. Haubold. 1978. Eocene anguid lizards from Europe and a revision of the genus Xestops. Herpetologica, 34:156-166. Moffat, L. 1973. The concept of primitiveness and its bearing on the phylogenetic classification of the Gekkota. Proceed- ings of the Linnean Society of New South Wales, 97:275— 301. Mones, A. 1989. Nomen dubium vs. nomen vanum. Journal of Vertebrate Paleontology, 9:232—234. Murry, P. A., D. A. Winkler, and L. L. Jacobs. 1989. Small tetrapods from the Comanchean (Early Cretaceous) of cen- tral Texas. Journal of Vertebrate Paleontology, 9:33A-34A. Supplement to No. 3: Abstracts of Papers, 49th Annual Meeting, Society of Vertebrate Paleontology, University of Texas, Austin, Texas. Nessov, L. A. 1985. Rare bony fishes, terrestrial lizards and mammals from the lagoonal zone of the littoral lowlands of the Cretaceous of the Kyzylkumy. Year-Book of All-Union Palaeontological Society, Leningrad. 28:199-219 (in Rus- sian). . 1988. Late Mesozoic amphibians and lizards of Soviet Middle Asia. Acta Zoologica Cracoviensia. 3 1 :475-486. Nicholls, E. L., and A. P. Russell. 1990. Paleobiogeography of the Cretaceous Western Interior Seaway of North Amer- ica: The vertebrate evidence. Palaeogeography, Palaeocli- matology, Palaeoecology, 79:149-169. Nopcsa, F. 1908. Zur Kenntniss der Fossilen Eidechsen. Beitra- ge zur Palaontologie und Geologie osterreich-Ungarns und des Orients, 21:33-62. Norell, M. A., M. C. McKenna, and M. J. Novacek. 1992. Estesia mongoliensis, a new fossil varanoid from the Late Cretaceous Barun Goyot Formation of Mongolia. American Museum Novitates, 3045:1—24. Northcutt, R. G. 1978. Forebrain and midbrain organization in lizards and its phylogenetic significance. Pp. 1 1-64, in Behavior and Neurology of Lizards (N. Greenberg and P. D. MacLean, eds.). United States Department of Health. Edu- cation and Welfare, Publication No. 77-49 1 . . 1979. The comparative anatomy of the nervous system and the sense organs. Pp. 615—769, in Hyman's Comparative Vertebrate Anatomy (M. H. Wake, ed.). The University of Chicago Press, Chicago, Illinois. Odermatt, C. 1940. Beitrage zur Kenntnis des Gebisses von Heloderma. Vierteljahrsschrift der Naturforschenden Ge- sellschaft, Zurich, 85:98-141. Olmo, E., and G. Odierna. 1980. Chromosomal evolution and DNA of cordylid lizards. Herpetologica, 36:311-316. OsmOlska, H., and R. Barsbold. 1990. Troodontidae. Pp. 259- 268, in The Dinosauria (D. B. Weishampel, P. Dodson, and 106 BULLETIN CARNEGIE MUSEUM OF NATURAL HISTORY NO. 33 H. Osmolska, eds.). University of California Press, Berkeley and Los Angeles, California. Parris, D. C., and B. S. Grandstaff. 1989. Nonmarine mi- crovertebrates of the Ellisdale local fauna: Campanian of New Jersey. Journal of Vertebrate Paleontology, 9:35A, Sup- plement to No. 3: Abstracts of Papers, 49th Annual Meeting, Society of Vertebrate Paleontology, The University of Texas, Austin, Texas. Peters, J. A., R. Donoso-Barros, B. Orejas-Miranda, and P. E. Vanzolini. 1986. Catalogue of the Neotropical Squa- mata. Smithsonian Institution Press, Washington, D. C., and London, United Kingdom. Pregill, G. K., J. A. Gauthier, and H. W. Greene. 1986. The evolution of helodermatid squamates, with description of a new taxon and an overview of Varanoidea. Transactions of the San Diego Society of Natural History, 21:167-202. Presch, W. 1970. The evolution of macroteiid lizards: An os- teological interpretation. Unpublished Ph.D. Dissert., Uni- versity of Southern California, Los Angeles, California. . 1974a. Evolutionary relationships and biogeography of the macroteiid lizards (family Teiidae, subfamily Teiinae). Bulletin of the Southern California Academy of Science, 73: 23-32. . 1974£>. A survey of the dentition of the macroteiid liz- ards (Teiidae: Lacertilia). Herpetologica, 30:344-349. . 1975. The evolution of limb reduction in the teiid lizard genus Bachia. Bulletin of the Southern California Academy of Science, 74: 1 13-121 . . 1983. The lizard family Teiidae: Is it a monophyletic group? Zoological Journal of the Linnean Society, 77:189- 197. Price, R. A., and E. W. Mountjoy. 1970. Geologic structure of the Canadian Rocky Mountains, between Bow and Atha- basca Rivers: A progress report. Pp. 7-25, in Structure of the Southern Canadian Cordillera (J. O. Wheeler, ed.). The Geological Association of Canada, Special Paper 6. Rage, J. C. 1991. Squamate reptiles from the early Paleocene of the Tiupampa area (Santa Lucia Formation), Bolivia. Re- vista Tecnica de Yacimientos Petroliferos Fiscales Bolivi- anos, 12:503-508. Rage, J. C., and M. Aug£. 1993. Squamates from the Cainozoic of the western part of Europe: A review. Revue de Paleo- biologie, volume special 7:199—216. Rahmani, R. A., and J. F. Lerbekmo. 1975. Heavy-mineral anal- ysis of Upper Cretaceous and Paleocene sandstones in Al- berta and adjacent area of Saskatchewan. Pp. 607-632, in The Cretaceous System in the Western Interior of North America (W. G. E. Caldwell, ed.). The Geological Associ- ation of Canada, Special Paper 13. Ray, C. E. 1965. Variation in the number of marginal tooth positions in three species of iguanid lizards. Breviora, 236: 1-15. Richter, A. 1994. Lacertilia aus der Unteren Kreide urn Una und Galva (Spanien) und Anoual (Marokko). Berliner Geowissenschaftliche Abhandlungen, Reihe E, Band 14:1- 138. Rjeppel, O. 1978. Tooth replacement in anguimorph lizards. Zoomorphologie, 91:77-90. . 1988. The classification of the Squamata. Pp. 261-293, in The Phylogeny and Classification of the Tetrapods, Vol- ume I : Amphibians, Reptiles, Birds (M. J. Benton, ed.). The Systematics Association, Special Volume No. 35A, Claren- don Press, Oxford, United Kingdom. Rigby, J. K., Jr., and D. L. Wolberg. 1987. The therian mam- malian fauna (Campanian) of Quarry 1, Fossil Forest study area, San Juan Basin, New Mexico. Pp. 51-79, in The Cre- taceous/Tertiary Boundary in the San Juan and Raton Ba- sins, New Mexico and Colorado (J. E. Fassett and J. K. Rigby, Jr., eds.). Geological Society of America, Special Pa- per 209, Boulder, Colorado. Robinson, P. L. 1967. The evolution of the Lacertilia. Colloques Internationaux du Centre National de la Recherche Scienti- fique, 163:395-407. Romer, A. S. 1956. Osteology of the Reptiles. The University of Chicago Press, Chicago, Illinois. . 1966. Vertebrate Paleontology. The University of Chi- cago Press, Chicago, Illinois. Rowe, T., R. L. Cifelli, T. M. Lehman, and A. Weil. 1992. The Campanian Terlingua local fauna, with a summary of other vertebrates from the Aguja Formation, Trans-Pecos Texas. Journal of Vertebrate Paleontology, 12:472-493. Ruibal, R. 1952. Revisionary studies of some South American Teiidae. Bulletin of the Museum of Comparative Zoology, Harvard University, 106:477-529. Russell, D. A. 1969. A new specimen of Stenonychosaurus from the Oldman Formation (Cretaceous) of Alberta. Ca- nadian Journal of Earth Sciences, 6:595-612. . 1989. An Odyssey in Time: The Dinosaurs of North America. University of Toronto Press, Toronto, Ontario, Canada. Russell, L. S. 1935. Fauna of the upper Milk River beds, south- ern Alberta. Transactions of the Royal Society of Canada, Series 3, 29:115-128 . 1964. Cretaceous non-marine faunas of northwestern North America. Contributions Royal Ontario Museum, Life Sciences, 6:1-24. . 1975. Mammalian faunal succession in the Cretaceous system of western North America. Pp. 137-161, in The Cre- taceous System in the Western Interior of North America (W. G. E. Caldwell, ed.). The Geological Association of Canada, Special Paper 13. Russell, L. S., and R. W. Landes. 1940. Geology of the south- ern Alberta Plains. Geological Survey of Canada, Memoir, 221:1-223. Sahni, A. 1972. The vertebrate fauna of the Judith River For- mation, Montana. Bulletin of the American Museum of Nat- ural History, 147:321-412. Saint-Girons, H. 1968. La morphologie comparee des glandes endocrines et la phylogenie des reptiles. Bijdragen tot de Dierkunde, 37:61-79. . 1976. Comparative histology of the endocrine glands, nasal cavities and digestive tract in anguimorph lizards. Pp. 205-278, in Morphology and Biology of Reptiles (A. Bel- lairs and C. Cox, eds.). Linnean Society Symposium Series 3. Savage, J. M. 1963. Studies on the lizard family Xantusiidae. Los Angeles County Museum, Contributions in Science, 71: 1-38. Schatzinger, R. 1975. Later Eocene (Uintan) lizards from the Greater San Diego area, California. Unpublished M.Sc. The- sis, San Diego State University, San Diego, California. . 1980. New species of Palaeoxantusia (Reptilia: Sauria) from the Uintan (Eocene) of San Diego County, California. Journal of Paleontology, 54:460-471. Schmidt, K. P. 1919. Herpetology of the Belgian Congo. Bui- 1996 GAO AND FOX— LATE CRETACEOUS LIZARDS 107 letin of the American Museum of Natural History, 39:385- 624. Schwenk, K. 1988. Comparative morphology of the lepidosaur tongue and its relevance to squamate phylogeny. Pp. 569- 598, in Phylogenetic Relationships of the Lizard Families (R. Estes and G. Pregill, eds.). Stanford University Press, Stanford, California. Seiffert, J. 1973. Upper Jurassic lizards from central Portugal. Servicos Geologicos de Portugal, Separata da Memoria 22: 1-85. Sheehan, P. M., and D. E. Fastovsky. 1992. Major extinctions of land-dwelling vertebrates at the Cretaceous-Tertiary boundary, eastern Montana. Geology, 20:556-560. Simpson, G. G. 1948. The beginning of the Age of Mammals in South America. Bulletin of the American Museum of Nat- ural History, 91:1-232. Sloan, R. E. 1987. Paleocene and latest Cretaceous mammal ages, biozones, magnetozones, rates of sedimentation, and evolution. Pp. 165-200, in The Cretaceous/Tertiary Bound- ary in the San Juan and Raton Basins, New Mexico and Colorado (J. E. Fassett and J. K. Rigby, Jr., eds.). Geological Society of America, Special Paper 209, Boulder, Colorado. Sloan, R. E., and L. Van Valen. 1965. Cretaceous mammals from Montana. Science, 148:220-227. Smith, H. M. 1946. Handbook of Lizards. Comstock Publication Co., Inc., New York, New York. Stanton, T. W. 1910. Fox Hills sandstone and Lance Formation (“Ceratops beds”) in South Dakota, North Dakota, and east- ern Wyoming. American Journal of Science, Series 4, 30: 172-188. Sternberg, C. M. 1951. The lizard Chamops from the Wapiti Formation of northern Alberta; Polyodontosaurus grandis not a lizard. Bulletin of the National Museum of Canada, 123:256-258. Storer, J. E. 1991. The mammals of the Gryde local fauna, Frenchman Formation (Maastrichtian: Lancian), Saskatche- wan. Journal of Vertebrate Paleontology, 11:350-369. Stott, D. F. 1975. The Cretaceous System in northeastern Brit- ish Columbia. Pp. 441-467, in The Cretaceous System in the Western Interior of North America (W. G. E. Caldwell, ed.). The Geological Association of Canada, Special Paper 13. Sulimskj, A. 1972. Adamisaurus magnidentatus n. gen., n. sp. (Sauria) from the Upper Cretaceous of Mongolia. Palaeon- tologia Polonica, 27:33-40. . 1975. Macrocephalosauridae and Polyglyphanodonti- dae (Sauria) from the Late Cretaceous of Mongolia. Palaeon- tologia Polonica, 33:25-102. . 1978. New data on the genus Adamisaurus Sulimski 1972 (Sauria) from the Upper Cretaceous of Mongolia. Pa- laeontologia Polonica, 38:43-56. Sullivan, R. M. 1979. Revision of the Paleogene genus Glyp- tosaurus (Reptilia, Anguidae). Bulletin of the American Mu- seum of Natural History, 163:1-72. . 1981. Fossil lizards from the San Juan Basin, New Mexico. Pp. 76-88, in Advances in San Juan Basin Pale- ontology (S. G. Lucas, J. K. Rigby, Jr., and B. Kues, eds.). University of New Mexico Press, Albuquerque, New Mex- ico. . 1982. Fossil lizards from Swain Quarry, “Fort Union Formation,” middle Paleocene (Torrejonian), Carbon Coun- ty, Wyoming. Journal of Paleontology, 56:996-1010. . 1986a. The skull of Glyptosaurus sylvestris Marsh, 1871 (Lacertilia: Anguidae). Journal of Vertebrate Paleon- tology, 6:28-37. . 19866. A new fossil anguid lizard Parodoxosaurus san- juanensis, new genus, new species from the middle Paleo- cene (Torrejonian) Nacimiento Formation, San Juan Basin, New Mexico. Journal of Herpetology, 20:109-1 1 1. . 1987. Parophisaurus pawneensis (Gilmore, 1928) new genus of anguid lizard from the middle Oligocene of North America. Journal of Herpetology, 21:115—133. . 1991. Paleocene Caudata and Squamata from Gidley and Silberling Quarries, Montana. Journal of Vertebrate Pa- leontology, 11:293-301. Tanke, D. 1988. Ontogeny and dimorphism in Pachyrhinosau- rus (Reptilia, Ceratopsidae), Pipestone Creek, N. W. Alberta, Canada. Journal of Vertebrate Paleontology, 8:27A, Supple- ment to No. 3: Abstracts of Papers, 48th Annual Meeting, Society of Vertebrate Paleontology, Tyrrell Museum of Pa- laeontology, Drumheller, Alberta, Canada. Thomas, R. G., D. A. Eberth, A. L. Deino, and D. Robinson. 1990. Composition, radioisotopic ages, and potential signif- icance of an altered volcanic ash (bentonite) from the Upper Cretaceous Judith River Formation, Dinosaur Provincial Park, southern Alberta, Canada. Cretaceous Research, 1 1 : 125-162. Tokaryk, T. T., and P. C. James. 1989. Cimolopteryx sp. (Aves, Charadriiformes) from the Frenchman Formation (Maas- trichtian), Saskatchewan. Canadian Journal of Earth Sci- ences, 26:2729—2730. Twente, J. W. 1952. Pliocene lizards from Kansas. Copeia, 1952:70-73. Vorobyeva, E. I., and T. J. Chugunova. 1986. The dental sys- tem in lizards: An intergrated approach. Pp. 315-318, in Studies in Herpetology (Z. Rocek, ed.). Prague, Czechoslo- vakia. Waldman, M. 1970. A teiid lizard jaw from the Cretaceous of Alberta, Canada. Canadian Journal of Earth Sciences, 7: 542-547. Wall, J. H. 1975. Diatoms and radiolarians from the Cretaceous System of Alberta — A preliminary report. Pp. 391-410, in The Cretaceous System in the Western Interior of North America (W. G. E. Caldwell, ed.). The Geological Associ- ation of Canada, Special Paper 13. Williams, G. D., and C. F. Burke. 1964. Upper Cretaceous. Pp. 169-189, in Geological History of Western Canada (R. G. McCrossan and R. P. Glaister, eds.). Alberta Society of Pe- troleum Geology, Calgary, Alberta, Canada. Williams, G. D., and C. R. Stelck. 1975. Speculations on the Cretaceous palaeogeography of North America. Pp. 1-20, in The Cretaceous System in the Western Interior of North America (W. G. E. Caldwell, ed.). The Geological Associ- ation of Canada, Special Paper 13. Wilson, M. V. H. 1988. Taphonomic process: Information loss and information gain. Geoscience Canada, 15:131—148. Winkler, D. A., P. A. Murry, and L. L. Jacobs. 1990. Early Cretaceous (Comanchean) vertebrates of central Texas. Jour- nal of Vertebrate Paleontology, 10:95-116. Yatkola, D. A. 1976. Fossil Heloderma (Reptilia, Heloder- matidae). The University of Kansas, Occasional Papers of the Museum of Natural History, 51:1-14. Zangerl, R. 1944. Contributions to the osteology of the skull of the Amphisbaenidae. American Midland Naturalist, 31: 417-454. ISSN 0145-9058 BULLETIN OF CARNEGIE MUSEUM OF NATURAL HISTORY DAWN OF THE AGE OF MAMMALS IN ASIA 4 N . ; Edited by K. CHRISTOPHER BEARD and MARY R. DAWSON Section of Vertebrate Paleontology NUMBER 34 PITTSBURGH, 1998