A tamerican museum Novitates PUBLISHED BY THE AMERICAN MUSEUM OF NATURAL HISTORY CENTRAL PARK WEST AT 79TH STREET, NEW YORK, NY 10024 Number 3557, 27 pp., 17 figures March 8, 2007 A Small Derived Theropod from Oosh, Early Cretaceous, Baykhangor Mongolia ALAN H. TURNER,1’2 SUNNY H. HWANG,3 AND MARK A. NORELL4 ABSTRACT A new theropod dinosaur, Shanag ashile, from the Early Cretaceous Oosh deposits of Mongolia is described here. The new specimen (IGM 100/1119) comprises a well-preserved right maxilla, dentary, and partial splenial. This specimen exhibits a number of derived theropod features, including a triangular anteriorly tapering maxilla, a large antorbital fossa, and maxillary participation in the caudally elongate external nares. These features resemble the Early Cretaceous dromaeosaurids Sinornithosaurus millenii and Microraptor zhaoianus, as well as the basal avialan Archaeopteryx lithographica. A comprehensive phylogenetic analysis including 58 theropod taxa unambiguously depicts the new Oosh theropod as a member of Dromaeosauridae. Relative to other dromaeosaurids, Shanag ashile is autapomorphic in its lack of a promaxillary fenestra and in the presence of interalveolar pneumatic cavities. The discovery of IGM 100/1119 expands our knowledge of Early Cretaceous dromaeosaurids and the faunal similarity between the Oosh and the Jehol biotas. more rare than the plentiful fossils of INTRODUCTION Psittacosaurus mongoliensis and sauropods at Theropods were some of the first dinosaur the locality (Osborn, 1923, 1924b). In some fossils discovered from the Early Cretaceous places, however, isolated theropod material of locality of Oosh (= Ohshih or Oshih), but a range of sizes is encountered frequently— their described remains are limited to undiag¬ especially unguals, phalanges, and teeth. nostic teeth of the large theropod “Pro- The deposits at Oosh are considered to be deinodon” (Osborn, 1924a; but see Currie, Early Cretaceous (Berriasian-Barremian). 2000). Theropod remains are generally much This is roughly coeval with beds forming the 1 Division of Paleontology, American Museum of Natural History ([email protected]). 2 Corresponding author. 3 Division of Paleontology, American Museum of Natural History ([email protected]). Current address: New York College of Osteopathic Medicine, Old Westbury, NY 11568. 4 Division of Paleontology, American Museum of Natural History ([email protected]). Copyright © American Museum of Natural History 2007 ISSN 0003-0082 2 AMERICAN MUSEUM NOVITATES NO. 3557 Fig. 1. Map of Mongolia showing geographical relationship of the Oosh locality to other Gobi fossil localities. Jehol group of northeastern China. The Jehol and the American Museum of Natural History beds of the Yixian and Juifotang formations include a wide array of fossil materials, have produced abundant vertebrate fossil including gobiconodontids (Rougier et al., remains (reviewed in Zhou et al., 2003). 2001), a pterosaur (Andres and Norell, These discoveries include the ubiquitous 2005), lizards, large and small sauropods, Psittacosaurus (also present in the Oosh beds) and psittacosaurus ranging in size from as well as avialan and nonavialan dinosaurs, hatchlings to senescent adults. including small feathered theropods (reviewed The Jehol biota and Oosh locality contain in Norell and Xu, 2005). Additionally, the some elements in common such as the Jehol beds contain a diverse mammalian fauna primitive ceratopsian Psittacosaurus. Un¬ including one of the largest Mesozoic mam¬ fortunately, dates of the Jehol and Oosh mals, Repenomamus (Li et al., 2001). The faunas are poorly constrained. Despite an Jehol flora is abundant, containing the puta¬ abundance of andesitic layers that allow for tive earliest angiosperm Archaeofructus (Sun radiometric dating of rocks that produce Jehol et al., 1998). fossils, several different ages have been pub¬ The Oosh locality lies in the Altai region of lished. Recent dating efforts agree with pre¬ Central Mongolia (see Watabe and Suzuki, vious results (Swisher et al., 1999; Wang et al., 2000; this paper, fig. 1). Fossils were first 2001) in proposing a Barremian date for the collected here by American Museum field feathered dinosaur beds in Sihetun and parties in the 1922 (Andrews, 1932). New bracket the lowermost Yixian beds (one of collections from the Oosh locality by field the most fossiliferous formations in the Jehol parties of the Mongolian Academy of Sciences group) as being between 128 m.y. and 2007 TURNER ET AL: OOSH THEROPOD 3 139 m.y. (Swisher et al., 2002). Older dates process projecting into the orbit, like that in have been published that would put much of derived tyrannosaurids. In overall appearance the Yixian Formation in the Late Jurassic (Lo it is similar to the postorbital of Sinraptor et al., 1999). These older dates, however, have dongi (Currie and Zhao, 1993) and tentatively been criticized on a number of grounds (Zhou considered a noncoelurosaur (allosauroid?) or et al., 2003). basal tyrannosauroid postorbital. Similarly, the stratigraphic and temporal A partial pes along with other identified divisions for the deposits at Oosh are poorly material pertaining to a small theropod was constrained (Jerzykiewicz and Russell, 1991). collected and is currently being prepared. According to Shuvalov (2000), rocks that Slightly weathered postcranial remains col¬ overlie the strata of the Oosh locality exam¬ lected as float further indicated the presence of ined here have yielded dates of 141 ± 8 m.y. at least one small, derived theropod taxon. and 126 ± 9 m.y. This indicates a Berriasian- This material includes a proximal femur and Hauterivian stage for these strata. Correlated tibia. The femoral head is weathered but exposures have been dated at around 130 m.y. preserves a prominent, moundlike lateral ridge (Samilov et al., 1988; Rougier et al., 2001), and a well-developed posterior trochanter. and are consistent with the results of Shuvalov The proximal tibia lacks a medial cnemial crest, unlike that seen in Mononykus olecranus (2000). A broader overview of the geology of (Perle et al., 1994) and other alvarezsaurids, Oosh is given by Andres and Norell (2005). and bears a well-developed fibular crest. Unfortunately, the dorsal surface of the femur COMPONENTS OF THE is poorly preserved, so it is unclear if the lesser OOSH LOCALITY and greater trochanters were confluent as in derived avialans. Nonetheless, the moundlike Consistent with previous collecting, lateral ridge and posterior trochanter indicate Mongolian Academy of Sciences-American this material likely is referable to a small Museum of Natural History Mongolian ex¬ paravian theropod. peditions in 1997, 1999, 2001, and 2005 In 1999, the Mongolian Academy of collected ubiquitous Psittacosaurus material Sciences-American Museum of Natural and numerous undiagnostic large theropod History Mongolian expedition recovered teeth. In addition to this material, partial a partial skull of a small theropod dinosaur vertebrate material was recovered that is from Oosh in a coarse, well-sorted sand referable to varying levels of clade specificity. channel. Although incomplete, this skull Isolated squamate vertebrae lacking zygo- (IGM 100/1119; fig. 2) is notable for its phenes/zygantra, exhibiting an unconstricted possession of several extremely derived coe¬ precondylar region, and nonoblique condyles lurosaur characters, clearly distinguishing it as are insufficient for referral to a specific clade, a small dromaeosaurid. but these characters do rule out varanoid, cordyliform, lacertiform, or snake affinities. Additional squamate material includes a prob¬ SYSTEMATIC PALEONTOLOGY able nonvaranoid, nonsnake autarchoglossan (based on the presence of small dorsolateral THEROPODA MARSH, 1884 zygosphenes) and a nonautarchoglossan cau¬ COELUROSAURIA VON HUENE, 1914 dal vertebra with a fracture plane that is on the transverse process. A nearly complete MANIRAPTORA GAUTHIER, 1986 gekkonomorph skull is also known from Oosh (Conrad and Norell, 2006). DROMAEOSAURIDAE MATTHEW AND A large (~6 cm), heavily rugose postorbital BROWN, 1922 was collected. Its straight frontal process Shanag ashile, new taxon precludes its referral to most coelurosaur clades except ornithomimosaurs or tyranno- Holotype: IGM 100/1119, nearly com¬ sauroids. Its size and rugose texture excludes it plete right maxilla, dentary, and partial from ornithomimosaurs. It lacks an anterior splenial. 4 AMERICAN MUSEUM NOVITATES NO. 3557 Fig. 2. IGM 100/1119 in right lateral view. Anatomical labels in appendix 3. Etymology: Shanag, Black-hatted dan¬ a medial lamina, or the portion of the maxilla cers in the Buddhist Tsam festival, and ashile, that resides within the antorbital fossa, and in reference to the old Oosh locality and a lateral lamina that forms the lateral surface formation name used by Dr. Henry F. of the maxilla outside the antorbital fossa. Osborn. These laminae correspond to the lamina Diagnosis: Small dromaeosaurid thero- lateralis and lamina medialis of Witmer pod diagnosed by the following combination (1997). Nomenclature for the structures of of characters: triangular, anteriorly tapering the antorbital cavity follows that used by maxilla; lateral lamina of nasal process of Witmer (1997). maxilla reduced to small triangular exposure; The maxilla is triangular, tapering anter¬ absence of a promaxillary fenestra; presence of iorly as in Archaeopteryx lithographica interalveolar pneumatic cavities; incipient (Wellnhofer, 1974: fig. 5; Mayr et al., 2005: dentary groove on posterolateral surface of fig. 2) and Sinornithosaurus millenii (Xu and dentary. Wu, 2001). The maxilla contributes to the narial border, as suggested by the slight depression on its dorsal margin near its DESCRIPTION anterior tip. This is unlike the condition in IGM 100/1119 consists of an almost com¬ derived dromaeosaurids, but similar to the plete right maxilla and dentary in articulation basal dromaeosaurids Microraptor zhaoianus (fig. 2). The maxilla is eroded laterally, re¬ (Xu et al., 2000) and Sinornithosaurus millenii vealing details of its alveoli, pneumatic cham¬ (Xu et al., 1999) and derived troodontids such bers, and fenestrae. Like most theropods, the as Troodon formosus (Makovicky and Norell, maxilla of Shanag ashile can be divided into 2004), Saurornithoides junior (IGM 100/1) and three parts: a posteriorly projecting subantor- Byronosaurus jaffei (IGM 100/983). The artic¬ bital process that contacts the jugal, a dorsally ulation with the premaxilla is complex. projecting ascending ramus of the nasal pro¬ Anteromedially, a thin projection of bone, cess, and an anteriorly projecting ventral the surface of which is slightly depressed, ramus of the nasal process (Witmer, 1990, extends rostrally past the limit of the dentiger¬ 1997). The maxilla can be further divided into ous portion of the maxilla. A complementary 2007 TURNER ET AL: OOSH THEROPOD 5 groove separates this process from the lateral the internal antorbital fenestra. Furthermore, lamina of the maxilla. In articulation, a por¬ there is no indication that the maxilla was tion of the posterior ramus of the premaxilla recessed medially from the nasal and thus it is would have fit into this groove. A similar unlikely that the nasal formed the dorsal construction is present in a disarticulated boundary of the antorbital fossa. maxilla of Velociraptor mongoliensis (IGM The antorbital cavity contains two fenes- 100/976). Neurovascular foramina are present trae, the antorbital and maxillary fenestrae. on the labial surface with each foramen There is no sign of a promaxillary (tertiary generally corresponds to a single tooth antorbital) fenestra. Archaeopteryx lithogra¬ position. phica, which possesses the most plesiomorphic The antorbital fossa is large and occupies antorbital cavity of any avialan (Witmer, much of the lateral surface of the maxilla. 1997), has both a maxillary and promaxillary Anterior to the fossa, a small triangular lateral fenestra in its recessed antorbital cavity. lamina is present below the external nares. Confuciusornis sanctus, a basal avialan phylo- This is similar to the condition in Si- genetically intermediate between Archae¬ nornithosaurus millenii (Xu and Wu, 2001), opteryx lithographica and Ornithothoraces Microraptor zhaoianus (Xu et al., 1999), and (sensu Chiappe, 1995), has only a small, round Archaeopteryx lithographica (Wellnhofer, accessory antorbital fenestra (Chiappe et al., 1974: fig. 5; Mayr et al., 2005: fig. 2). The 1999). Chiappe et al. (1999) tentatively ho- rostral boundary of the antorbital fossa is mologized this fenestra with the maxillary situated approximately 7 mm caudal to the fenestra. All other known avialans have premaxillary contact, above the 3rd maxillary complex diverticula extending medially, ros- tooth position. The caudal margin of the naris trally, and caudally from one large antorbital overlaps the rostral border of the antorbital cavity (Witmer, 1990), but laterally, the fossa as in Archaeopteryx lithographica antorbital opening is not subdivided into (Wellnhofer, 1974: fig. 5; Mayr et al., 2005: separate accessory fenestrae. Witmer (1997) fig. 2) but unlike dromaeosaurids (Norell suggested that the promaxillary fenestra char¬ and Makovicky, 2004) and troodontids acterizes Neotheropoda. The lack of a pro¬ (Makovicky and Norell, 2004). The antorbital maxillary fenestra is not, however, entirely fossa is defined shallowly anteriorly, but the unusual among nonavialan theropods. The dorsal and posteroventral posterior margins troodontids Saurornithoides junior (Barsbold, are sharply bounded by thin, laterally project¬ 1974; Norell et al., in prep.) and Byronosaurus ing shelves of bone. Although the proportions jaffei (IGM 100/983) lack a promaxillary of the antorbital fossa and general shape of fenestra, while the more basal troodontid the maxilla are similar to that of Sinovenator changii (Xu et al., 2002) retains Archaeopteryx, the raised dorsal rim of the a promaxillary fenestra. Additionally, the antorbital fossa differs from the recessed oviraptorosaur Incisivosaurus gauthieri (Xu et antorbital fossa seen in Archaeopteryx al., 2002) and the therizinosauroid Erliko- (Witmer, 1990). In almost all avialans, the saurus andrewsi (Clark et al., 1994) lack ascending ramus of the maxilla is lost, and as a promaxillary fenestra. a result, the nasal forms the majority of the The internal antorbital fenestra is large and dorsal border of the internal antorbital fenes¬ is defined anteriorly (and probably dorsally) tra and antorbital fossa (Cracraft, 1986). by the ascending process of the maxilla. The Archaeopteryx lithographica is the only avia- antorbital fossa does not extend ventrally Ian that retains an extensive ascending ramus beyond the border of the internal antorbital of the maxilla, but the ramus is recessed fenestra, as in troodontids {Saurornithoides medially so that the nasal forms the dorsal junior, IGM 100/1; Saurornithoides mongolien¬ border of the antorbital fossa as in other basal sis, AMNH FR 6515; Troodon formosus avialans (Witmer, 1990), but not the internal [Witmer, 1997]). The surface bone is missing antorbital fenestra unlike other avialans. IGM from the anterior rim of the internal antorbital 100/1119 retains the typical theropod condi¬ fenestra exposing a substantial rostral excava¬ tion in which maxilla excludes the nasal from tion. Anteriorly, the excavation extends to the 6 AMERICAN MUSEUM NOVITATES NO. 3557 root of the 6th maxillary tooth. The thin caudodorsally open fossa, both of which are lamina of bone forming the anterior wall of located within the antorbital fossa. The fossa the excavation wraps around the entire length containing the internal aperture of the maxil¬ of the root. This excavation may correspond lary fenestra of Shanag ashile, however, is to the deep fossa invading the pila postantralis much deeper and more sharply defined than (the strut of bone reinforcing the poster¬ those in the previously mentioned taxa. It is, omedial wall of the maxillary antrum) noted therefore, more similar to Bambiraptor fein¬ by Witmer (1997: 43) in Ornitholestes herman- bergorum (AMNH FR 30554), which has ni, Allosaurus fragilis, Marshosaurus bicentesi- a maxillary fenestra that is nearly continuous mus, and Deinonychus antirrhopus. The la¬ with the excavatio pneumatica. In Bam¬ teral surface of the medial wall of the biraptor feinborgorum (AMNH FR 30554) excavation is not smooth, but crossed by both structures are contained within a caudo¬ ridges and sculptured with rounded depres¬ dorsally opening fossa. Thus, it is possible that sions. Ventrally, the dorsal surfaces of the the oval structure in IGM 100/1119, consid¬ posterior alveoli protrude slighting into the ered here to be the maxillary fenestra, may in fossa. fact represent a maxillary fenestra confluent The maxillary fenestra in Shanag ashile is with an excavatio pneumatica. positioned high on the ascending ramus of the Some unusual accessory cavities—interal¬ maxilla. This location more generally corre¬ veolar pneumatic cavities—are present be¬ sponds to the excavatio pneumatica of other tween the pairs of adjacent alveoli separated theropods (Witmer, 1997: 43). The plesio- by a diastema. The interalveolar chambers of morphic position of the maxillary fenestra is IGM 100/1119 would have been closed off more ventral, such that the center of the from the rest of the craniofacial air sac system maxillary fenestra is aligned with the center of in life, surrounded on all sides by a thin layer the internal antorbital fenestra. Our interpre¬ of bone, and are visible only because of the tation of the structure as the maxillary fenestra eroded labial surface of the maxilla. The roots is supported, however, by the presence of of the teeth that bracket them define the a maxillary antrum medial to the opening, boundaries of the chambers. Thin laminae of a similarly positioned maxillary fenestra in bone envelop the roots of the flanking teeth, Deinonychus antirrhopus (Ostrom, 1969; Wit¬ defining the anterior and posterior limits of mer, 1997), and a generally dorsally displaced the chambers. In addition, the chambers maxillary fenestra in dromaeosaurids. extend dorsally no farther than the dorsal The maxillary fenestra is a relatively large, margin of the shorter bounding tooth, no anteroventrally inclined oval structure with farther ventrally than the crown-root junction both an internal and external aperture, similar of the teeth, and no farther medially than the in structure to the well-developed antorbital medial limit of the teeth. The medial walls of cavities of Erlikosaurus andrewsi (Clark et al., the chambers have the same irregular surface 1994), Bambiraptor feinbergorum (AMNH FR as that of the internal antorbital fenestra. 30554), and Saurornithoides mongoliensis These are not the “recessi pneumatici inter- (AMNH FR 6516). The external aperture is alveolares” described by Witmer (1997), which confluent with the lateral surface of the are ventrolateral invaginations of the maxil¬ maxilla and defines a fairly deep fossa that is lary antrum into the alveolar process. continuous with the maxillary antrum. The Structures possibly homologous to the pneu¬ internal aperture is positioned at the rostro- matic interalveolar recesses are present within ventral end of this fossa and perforates the the antorbital fenestra in IGM 100/1119 where maxilla. A somewhat similar arrangement is the alveoli protrude into the fenestra. seen in taxa such as Marshosaurus bicentesi- At least nine widely and unevenly spaced mus (see Witmer, 1997), Deinonychus anti¬ alveoli, six of which contain teeth, are present rrhopus (Ostrom, 1969), and Velociraptor in the maxilla. The number of maxillary teeth mongoliensis (AMNH FR 6515, IGM 100/ is similar to the eight in Archaeopteryx 982). In these taxa, the maxillary fenestra is lithographica (Wellnhofer, 1988), the nine in contained within a shallow, caudally or Dromaeosaurus albertensis (AMNH FR 5356; 2007 TURNER ET AL: OOSH THEROPOD 7 becoming more inset posteriorly. The ventral and dorsal margins of the dentary are sub¬ parallel, as in dromaeosaurids (Currie, 1995) and Archaeopteryx lithographica (Wellnhofer, 1992: fig. 6). Two rows of mental foramina are present on the labial surface, although a series of smaller anteriorly directed foramina cluster rostrally. The dorsal row contains a greater number of foramina that are anteroposteriorly elliptical and elongate. The dorsal row of mental foramina does not lie in a deep subalveolar groove as in troodontids (Makovicky et al., 2003) and Buitreraptor gonzalezorum (Makovicky et al., 2005). Posteriorly, the dorsal foramina lengthen to Fig. 3. Detail of posterior tooth morphology in form a shallow groove similar to that present IGM 100/1119. Note serrations on posterior in Archaeopteryx lithographica (Wellnhofer, carinae and unconstricted root-crown transition. 1974: plate 22), the holotype of Velociraptor mongoliensis (AMNH FR 6515), and Tsaagan Currie, 1995), and the 10 in Velociraptor mangas (IGM 100/1015; Norell et al., 2006). mongoliensis (AMNH FR 6515). The tooth There are fewer foramina in the lower row, count in IGM 100/1119 is, however, different although they are of the same general shape. from the numerous maxillary teeth present in A relatively deep Meckelian groove runs Jinfengopteryx elegans (Ji et al., 2005) and along the medial surface of the dentary. troodontids—even basal taxa such as Anteriorly, the groove begins just caudal to Sinovenator changii (Xu et al., 2002) and Mei the rugose symphyseal tip of the dentary. A long (Xu and Norell, 2004). The lateral surface small fragment of the splenial covers the of the maxilla is missing from the central posterior end of the Meckelian groove, so portion of the maxilla so that the roots of two the posterior extent of the groove cannot be teeth are exposed. Although there are several determined. The posterior margin of the empty alveoli in the maxillary tooth row, some dentary is broken, so the posterior articula¬ of the gaps are solid stretches of interdental tions are unknown. bone separating adjacent alveoli. The maxil¬ There are at least 15 teeth in the dentary. lary teeth are laterally compressed, with the Due to the overhanging maxilla and its teeth, carinae positioned lingually. Posterior teeth it is difficult to determine the exact number of have serrations on their posterior carinae alveoli present. Like the maxillary tooth (fig. 3). The three central maxillary teeth are count, the dentary tooth count of Shanag unusual in that they have long roots consti¬ ashile is comparable to that of some dromaeo¬ tuting approximately 70% of the entire tooth saurids, such as the 16 in Deinonychus length. Tooth root bulges of these teeth are antirrhopus (Ostrom, 1969), 15-16 in visible on the medial face of the ascending Saurornitholestes langstoni (Sues, 1978), and ramus of the maxilla. 14-15 in Velociraptor mongoliensis (AMNH The dentary is long and relatively dorso- FR 6515; Currie, 1995). The dentary teeth, ventrally shallow, its height comprising only like the maxillary teeth, are interspersed with about 12% of its length, although it is broken diastemata. The anterior teeth are quite small posteriorly. The dentary is straight and does and especially closely packed like in troodon¬ not curve medially at the symphysis as in tids (Makovicky and Norell, 2004) and derived troodontids like Troodon formosus Microraptor zhaoianus (Xu et al., 2000; this (Currie, 1985) or Saurornithoides (Barsbold, paper, fig. 4). Overall, the dentary teeth are 1974; Norell et al., in prep). It is relatively smaller and more closely packed than the labiolingually thick, with the tooth row maxillary teeth (1.0 versus 1.4 mm basal slightly inset from the labial margin and crown widths and 0.37 versus 0.70 mm gaps AMERICAN MUSEUM NOVITATES NO. 3557 Fig. 4. Detail of the tooth morphology and size variation seen in the dentary of IGM 100/1119. Anatomical labels in appendix 3. between adjacent teeth not separated by a di¬ Makovicky, 2004), Archaeopteryx lithographi- astema). The anterior dentary teeth are devoid ca (Wellnhofer, 1993), and the troodontid of serrations, while the posterior teeth have Saurornithoides junior (Barsbold, 1974; Norell small serrations only on their posterior carinae et al., in prep). A long, narrow notch at the (12.4 serrations per millimeter). This is the posterior of the splenial fragment, correspond¬ same pattern seen in Microraptor zhaoianus ing perhaps to the mylohyoid foramen (see (Hwang et al., 2002). Unlike Microraptor Currie, 1995: fig. 7E), is confluent with the zhaoianus teeth and the teeth of troodontids Meckelian groove. and early avialans, the teeth of IGM 100/1119 do not have constrictions between their roots DISCUSSION and crowns. This unconstricted root/crown transition is present in Buitreraptor gonzale- The taxon preserves an interesting mosaic of zorum (Makovicky et al., 2005), Sin- avialan and dromaeosaurid characteristics. ornithosaurus millenii (Xu and Wu, 2001), Like most avialans, Shanag ashile possesses and other dromaeosaurids. a tapering, triangular shaped maxilla, which Only a small sliver of the splenial protrudes participates in the narial border and has a large past the posterior edge of the dentary, but the antorbital fossa. Additionally, the caudal anterior portion is nearly intact. It cannot be margin of the external nares overlaps the determined if the splenial was laterally ex¬ rostral border of the antorbital fossa, as in posed at the contact of the dentary and avialans. A number of dromaeosaurid syna- postdentary bones as in deinonychosaurs, as pomorphies, however, are preserved in IGM this region of the mandible is not preserved. It 100/1119. These include a straight, parallel¬ overlaps the posterior third of the dentary, sided dentary and a dorsally displaced maxil¬ reaching to about the level of the 13th dentary lary fenestra that is itself recessed in a caudo- alveolus. The preserved portion of the splenial dorsally directed depression. is thin, long, and sharply triangular, much like To test the phylogenetic placement of the splenial of the dromaeosaurids (Norell and Shanag ashile we have expanded the Norell 2007 TURNER ET AL: OOSH THEROPOD 9 et al. (2006) study by including the new taxon Saurornitholestes langstoni. The Adams con¬ and 15 additional characters, several of which sensus demonstrates that consistent structure are novel and pertain to the maxilla morphol¬ exists within the large non-unenlagiine dro¬ ogy of dromaeosaurids and avialans. A total maeosaurid clade. This phylogenetic structure of 56 coelurosaurian taxa and 251 characters is nearly identical to that recovered in the (19 ordered) were used in the analysis, with analysis of Norell et al. (2006). In the present Allosaurus fragilis and Sinraptor dongi used to analysis, an Adams consensus shows the root the most parsimonious trees. The dataset consistent presence of a Deinonychus + was treated with equally weighted parsimony Velociraptor sister group relationship when analysis implemented in TNT v. 1.0 (Goloboff the position of Saurornitholestes is not consid¬ et al., 2003). A heuristic tree search strategy ered as well as the consistent presence of an was conducted performing 1000 replicates of Achillobator + Utahraptor clade. However, in Wagner trees (using random addition se¬ the latter case, both Adasaurus and Dro- quences) followed by TBR branch swapping maeosaurus are depicted closer to Achillo¬ (holding 10 trees per replicate). The best trees bator than Utahraptor or closer to Utah¬ obtained at the end of the replicates were raptor than Achillobator in some of the funda¬ subjected to a final round of TBR branch mental trees. The largely incomplete nature of swapping. Zero length branches were col¬ Dromaeosarus albertensis and Adasaurus mon- lapsed if they lack support under any of the goliensis reduced the resolution within this most parsimonious reconstructions (i.e., rule 1 subclade of large-bodied dromaeosaurids. of Coddington and Scharff, 1994). This Although the exact sister taxon of Shanag analysis resulted in 552 most parsimonious ashile is currently ambiguous, it is most simi¬ trees of 764 steps (Cl = 0.381, RI = 0.709) lar in general shape and morphology to found in 625 out of the 1000 replicates. Sinornithosaurus millenii (Xu et al., 1999). Shanag ashile was found to be a member of Both taxa possess a triangle-shaped, anteriorly a dromaeosaurid clade less inclusive than tapering maxilla as well as maxillary partici¬ Unenlagiinae (i.e., Buitreraptor gonzalezorum, pation in the external nares. In the most Unenlagia comahuensis, and Rahonavis ostromi) parsimonious trees that recover Shanag ashile in all of the most parsimonious reconstructions closely related to Sinornithosaurus millenii, the (fig. 5). This placement is supported by the topology is supported by a single synapomor- presence in Shanag ashile of three of the clade’s phy—the lateral lamina of the ventral ramus six unambiguous synapomorphies: a solid of the maxilla reduced to a small triangular quadrate (char. 52.0—unknown in Shanag exposure (char. 244.1). Shanag ashile remains ashile)', dentary with subparallel dorsal and distinguishable from Sinornithosaurus millenii ventral edges (char. 70.1); large maxillary and due to its lack of a promaxillary fenestra. dentary teeth, less than 25 in dentary (char. Prior to the discovery of IGM 100/1119, the 84.0); anterior cervical centra level with or only theropod material from the Early shorter than posterior extent of neural arch Cretaceous Oosh deposits were undiagnostic. (char. 96.0—unknown in Shanag ashile); ex¬ Shanag ashile marks the first theropod taxon tremely long extension of the prezygapophyses from the deposits referable to a specific thero¬ of the distal caudal vertebrae (char. 120.1— pod clade—the Dromaeosauridae. The discov¬ unknown in Shanag ashile)', and dorsal dis¬ ery of IGM 100/1119 is important as it extends placement of maxillary fenestra (char. 237.1). our knowledge of Early Cretaceous dromaeo¬ Given the large percentage of missing data for saurids. The best known Early Cretaceous IGM 100/1119, Shanag ashile was recovered in dromaeosaurids—Sinornithosaurus and Micro¬ numerous placements within the large clade raptor—are known from the Jehol beds of more derived than Unenlagiinae clade (i.e., China. The presence Shanag ashile in the Sinornithosaurus + Velociraptor), resulting in an Jehol-contemporaneous Oosh deposits of unresolved strict consensus. The Adams con¬ Mongolia suggests a wider diversity of Early sensus depicted in figure 5 shows this lack of Cretaceous dromaeosaurids that show a resolution is due to the labile position of broader distribution of character states and IGM 100/1119 and to a lesser extent, avianlike cranial features. 10 AMERICAN MUSEUM NOVITATES NO. 3557 A - Alloseurus fragffis - Sinraptor dongi ■ Tyrannosaurus rex - Aibertosaurus iibratus ■ Petecanimimus polydon - Harpymimus okiadnikovi - Shenzhousaurus orientalis ■ Archaeomithomimus asiaticus Garudimimus brevipes -Struthiomimus aitus -GaUimimus bullatus ■ Omithomimus edmonticus - Anserimimus planinychus Omitholestes hermanni ■ Huaxiagnathus orientalis -IGM 100/1119 ■ Sinosauropteryx prima ■ Compsognatbus hngipes - Microraptor zhaoianus - Alvarezsaurus calvoi ■ Sinomithosaurus millenii ■ Patagonykus puertai - Sauromitholestes iangstoni ■ Mononykus olecranus - Tsaaganmangas ■ Shuuvia deserii - Dainonychus antirrhopus - Velociraptor mmgoiiensis - Segnosaurus galbirmmis -Adasaums mongoiiensis ■ Erlikosaurus andrewsi -Dromaeosaurus albertensis - Alxasaurus efesitaiensis ■ tncisivosaurus gauthieri - Achiiiobator giganticus ■ Utahraptor ostrommaysamm ■ Caudipteryx zoui — Microvenator celer — Chirostenot&s pergracitis — Avimimm portentosus — Ingenia yanshani — Citipati osmolskaa — Oviraptor mongoliensis — Oviraptor philoceratops — Conchoraptar gracilis Archaeopteryx iithographica Confudusornis sanctus Mei long Sinovenatar changii EKtroodontid -Byronosaurus iaffei -Sinomithoidas youngi Troodon formosus Sauromithoides mongoiiensis Sauromithoides junior Buitreraptor gonzalezorum Rahonavis ostromi Unenla^ia^omahuens^ - Sinomithosaurus millenii ■ Microraptor zhaoianus Dromaeosauridae ■ Deinonychus antirrhopus ■ Velociraptor mongoiiensis ■ Tsaaganmangas Utahraptor ostrommaysorum - Dromaeosaurus albertensis ' Achiiiobator giganticus ■ Adasaums mongoiiensis ■ Sauromitholestes Iangstoni IGM 100/1119 Fig. 5. Phylogenetic placement of IGM 100/1119. A. Strict consensus of 552 most parsimonious reconstructions of coelurosaurian interrelationships found in our phylogenetic analysis of 251 characters and 56 coelurosaurian taxa. The new taxon is indicated in bold. B. Adams consensus topology of non- unenlagiine dromaeosaurids recovered in from this analysis. ACKNOWLEDGMENTS field crew of the 1999 field season for their hard work, Chris Brochu for providing images Collection and study of this specimen was of FMNH PR2081, and Peter Makovicky, supported by NSF grants DEB-9300700, DEB Jack Conrad, and Sterling Nesbitt for helpful 0608003, and ATOL 0228693. We thank the discussions. Amy Davidson prepared the