sign in sign up
<<
Home , Adelobasileus, Gobiconodontidae, Holotheria, Theria

AMERICAN MUSEUM Norntates PUBLISHED BY THE AMERICAN MUSEUM OF NATURAL HISTORY CENTRAL PARK WEST AT 79TH STREET, NEW YORK, N.Y. 10024 Number 3149, 19 pp., 3 figures November 16, 1995

A Mammalian Petrosal from the Early of Mongolia: Implications for the Evolution of the Ear Region and Mammaliamorph Interrelationships*

JOHN R. WIBLE,1 GUILLERMO W. ROUGIER,2,3 MICHAEL J. NOVACEK,2 MALCOLM C. MCKENNA,2 AND DEMBERLYIN DASHZEVEG4

ABSTRACT A mammalian petrosal from the Early Creta- conodontidae and Prototribosphenida (Vince- ceous Khoobur locality, Guchin Us Somon, Mon- lestes + ). In the Khoobur fauna are gobi- golia, is described and the major vessels and nerves conodontids and symmetrodonts, which could ac- associated with it are restored. A cladistic analysis count for such a systematic allocation. Previous of 44 characters of the petrosal and associated hypotheses ofmammaliamorph interrelationships structures across 15 ingroup taxa, including all are compared with the results of our analysis. A mammaliamorphs for which petrosal morphology close relationship ofmonotremes with therians and is known, identified nine equally parsimonious a position ofdocodonts outside trees. The consensus tree placed the Khoobur pe- are not supported here. trosal within Mammalia at a trichotomy with Tri-

INTRODUCTION Since 1990, the Mongolian Academy of research in and Cenozoic outcrops Sciences and the American Museum of Nat- of Mongolia. In the summer of 1991, screen- ural History have conducted paleontological washing at the locality of Khoobur (also var- * This is contribution number 20 of the Mongolian-American Museum Paleontological Project. 1 Research Associate, Department of , American Museum of Natural History; Department of Ana- tomical Sciences and Neurobiology, School of Medicine, University of Louisville, Louisville, KY 40292. 2 Department of Paleontology, American Museum of Natural History. 3Consejo Nacional de Investigaciones Cientificas y Tecnicas (CONICET), Argentina. 4Geological Institute, Mongolian Academy of Sciences, Ulaan Bataar, Mongolia.

Copyright © American Museum of Natural History 1995 ISSN 0003-0082 / Price $2. 10 AMERICAN MUSEUM NOVITATES NO. 3149 iously spelled Khobur, Khoboor, and The definition of taxonomic units em- Khovboor), Guchin Us Somon, yielded a col- ployed here follows Gauthier (1986) and lection ofsmall . Included were iso- Rowe (1987). We use the terms Eucynodon- lated teeth, fragmentary upper and lowerjaws, tia (Kemp, 1988; see comments by Hopson, postcranial remains, and two petrosals, the 1991), Mammaliamorpha, Mammalia- more complete of which is reported here. formes, Mammalia, Theriimorpha, and Khoobur beds are believed to be of ?Aptian- Theria (Rowe, 1988), (Hopson, Albian Age () based on the 1994), Cladotheria and (Mc- known fauna (Clemens et al., 1979; Kielan- Kenna, 1975), (Prothero, Jaworowska and Dashzeveg, 1989). 1981), and Prototribosphenida (Rougier, The petrosal is a topologically complex 1993). and - bone of the mammalian skull that is associ- tidae are treated here as monophyletic taxa, ated with a variety of soft-tissue structures, even though relationships among "Tricono- including the organs of hearing and balance, donta" are not fully resolved. At least in one the central and peripheral nervous systems, instance (Rowe, 1993), and major cranial arteries and veins, and muscles Triconodon have been identified as sister taxa derived from the mandibular and hyoid arch- and we refer to this group informally as tri- es (Goodrich, 1930; Novacek, 1993). Many conodonts. of these soft tissues leave impressions on the To evaluate the affinities of the Khoobur petrosal in extant mammals and, therefore, petrosal, we scored the distribution of 44 can be reconstructed in fossil material (Wi- characters across 15 ingroup taxa. We had ble, 1990; Rougier et al., 1992). Additionally, the opportunity to study all the taxa included the impressions left by various soft tissues on here, with the exception of . the petrosal have proven to be of systematic Most of the characters employed were taken value for determining higher-level phyloge- from the literature and corrected or supple- netic relationships among mammals (Wible, mented according to our observations (see 1986, 1987, 1990; Wible and Hopson, 1993). Appendix 1). The literature sources are: Ker- Petrosals, and cranial remains in general, mack (1963), Kuhn (1971), Kermack et al. of Mesozoic mammals are rare. In fact, pe- (1981), Sun (1984), Crompton and Sun trosals ofEarly Cretaceous age have been re- (1985), Gow (1986), Kielan-Jaworowska et covered for only two identifiable taxa: Vin- al. (1986), Novacek (1986), Sues (1986), Hahn celestes from the Hauterivian La Amarga (1988), Rowe (1988), Zeller (1989), Wible Formation of Argentina (Bonaparte and (1990, 1991), Lillegraven and Krusat (1991), Rougier, 1987) and partially described, un- Rougier et al. (1992), Crompton and Luo named triconodontids from the Aptian Clov- (1993), Lillegraven and Hahn (1993), Wible erly Formation of Montana (Crompton and and Hopson (1993, 1995), Lucas and Luo Jenkins, 1979). Another fragmentary petro- (1993), Rougier (1993), Luo (1994), Luo and sal ofindeterminate mammalian affinities was Crompton (1994), and Luo et al. (1995). To reported from the Aptian-Albian Kelvin For- polarize characters, we chose as the outgroup mation of Utah by Prothero (1983). Discov- the following non-mammalian eucynodonts: ery of a petrosal in the Khoobur beds thus traversodontids, probainognathids, and chi- affords another opportunity to investigate the niquodontids. The outgroup was polymor- anatomy of this bone in an Early Cretaceous phic for five characters, which were scored as and to explore the evolution ofthis missing data. The resulting taxon-character complex part of the skull in early mammals matrix (see Appendix 2) was subjected to the and their relatives. branch and bound algorithm ofPAUP (Swof- ford, 1993) and the ie command ofHennig86 (Farris, 1988). In both programs, all multi- MATERIALS AND METHODS state characters were treated as unordered. The Khoobur petrosal is cataloged as PSS- The scope of our analysis is limited by the MAE- 104 in the collections ofthe Geological available material, that is, an isolated petro- Institute of the Mongolian Academy of Sci- sal. However, we do not believe that char- ences, Ulaan Bataar, Mongolia. acters from this bone have more relevance 1 995 WIBLE ET AL.: MAMMALIAN PETROSAL FROM MONGOLIA 3 for the study of eucynodont phylogeny than ken medial surface of the petrosal exposes a do those from other anatomical systems. deep, smooth-walled channel, interpreted The terminology for anatomical terms used here as housing the inferior petrosal sinus. here follows Wible (1987, 1990) and Rougier The posterior opening ofthis channel lies an- et al. (1992). teroventral to the jugular notch where a par- tially preserved foramen is visible. The an- DESCRIPTIONS terior end of this channel curves laterally at The petrosal bone houses the organs of the anterior pole of the promontorium (fig. hearing and equilibration. It can be divided 2) where it communicates with the lateral into two parts in extant mammals: the pars trough via a minute foramen. cochlearis, which is anteroventrally placed On the posterolateral aspect of the pro- and includes the promontorium (the cochlear montorium is the elliptical fenestra vestibuli, housing), and the pars canalicularis, which is which faces ventrolaterally and has its long posterodorsally placed and includes the semi- axis oriented anteroposteriorly (fig. 1A, B). circular canals. For descriptive purposes, we On the posterior aspect ofthe promontorium will describe the petrosal in four views-ven- is the perilymphatic foramen. It is subequal tral, dorsal, lateral, and posterior-with the in size to the fenestra vestibuli, nearly vertical orientation based on the presumed position in its orientation, and has a roughly qua- in the skull. drangular outline. Although a conspicuous The specimen is a nearly complete, isolated groove for the perilymphatic duct is lacking, right petrosal (figs. 1, 2). Articular surfaces we identify this opening as a perilymphatic are preserved for the exoccipital, supraoccip- foramen because there is no certain cochlear ital, and squamosal bones. Missing is the an- aqueduct. A tiny foramen in the posterior terodorsal part ofthe lateral flange, where the part of the petrosal's contribution to the jug- anterior lamina of the petrosal is expected. ular foramen (not visible in fig. 1A) is too The ventral surface ofthe lateral flange is also small and improperly positioned to be a broken in two places: a small break at its cochlear aqueduct. It leads into the substance anteriormost end and a more extensive break of the petrosal and probably transmitted a posteriorly. On the promontorium, the an- vein. The jugular foramen is similar in di- teriormost tip is missing. Where the articular ameter to the perilymphatic foramen, as de- surface of the basioccipital is predicted on duced from what is preserved on the petrosal. the medial border of the petrosal, breakage Separating the perilymphatic foramen from exposes a deep channel. Small chips of bone the fenestra vestibuli is a low, narrow bridge are also missing from the tip of the paroc- of bone, the crista interfenestralis, which cipital process, and there is slight erosion pos- gradually decreases in height posteriorly and terodorsal to the does not reach the paroccipital process. The subarcuate fossa along the term crista interfenestralis has been used pre- articular surface for the supraoccipital. viously in sauropsids (Oelrich, 1956) for a structure that we interpret to be homologous VENTRAL VIEW (fig. IA, B, C) with that in . The pars cochlearis is represented by a nar- Running parallel to the promontorium is row, elongate promontorium about 4 mm in a deep, wide trough, the lateral trough. The length. Exposed at the broken anteriormost deepest point of the lateral trough is located tip of the promontorium is a round, matrix- anteriorly where it is bounded medially by filled space. We are uncertain about the true the nearly vertical lateral wall of the pro- nature of this space, but it might represent montorium (figs. IA, B, 2). Posterior to this the rostral end of a straight cochlear duct, striking depression are three foramina. The which is the shape predicted from the exter- two more anteriorly placed foramina are in- nal morphology of the promontorium. Me- terpreted as exits for branches of the facial dial to the ventralmost prominence of the nerve: the one in front for the palatine ramus promontorium is a slightly concave surface and the other for the hyomandibular ramus that extends medially toward the presumed (fig. 1 C). The precise location ofthe aperture basioccipital suture. As stated above, the bro- for the palatine ramus (the hiatus Fallopii) 4 AMERICAN MUSEUM NOVITATES NO. 3149 cannot be determined because there is some bone missing along the preserved margin. However, we believe that the missing bone would only slightly increase the distance be- tween the two preserved exits for the facial nerve branches. Posterolateral to the aperture for the hyomandibular ramus (the secondary facial foramen) is a slightly larger foramen (fig. 1A, B) that leads into a canal that opens endocranially on the dorsal surface (fig. 1D, E). This foramen is the tympanic aperture of the prootic canal, which in life transmitted the prootic vein. The lateral extent of the lateral trough is bounded by a thickened crest of bone, the lateral flange (figs. 1A, B, 2). The preserved anterior portion of the lateral flange shows no contact facet for a quadrate ramus of the epipterygoid (alisphenoid). The preserved posterior portion is inflected medially to a

Fig. 1. Three views of the right petrosal from Khoobur, PSS-MAE-104. A, B, C, Ventral view. D, E, F, Dorsal view. G, H, I, Lateral view. Cross- hatching indicates damaged surfaces. Facial nerve and vessels are reconstructed in C, F, and I (see text). Abbreviations: adm, arteria diploetica mag- na; ar, anterolateral recess; av, aqueductus vesti- buli; ce, cavum epiptericum; ci, crista interfenes- tralis; cp, crista parotica; cr, crista petrosa; ctpp, caudal tympanic process of petrosal; dag, dorsal ascending groove; fcn, foramen for cochlear nerve; fs, fenestra semilunaris; fv, fenestra vestibuli; gadm, groove for arteria diploetica magna; goa, groove for occipital artery; gri, groove for ramus inferior; grs, groove for ramus superior; grt, groove for ramus temporalis; hf, hiatus Fallopii; gg, genicu- late ganglion; iam, internal acoustic meatus; ijv. internal jugular vein; ips, inferior petrosal sinus; jn, jugular notch; If, lateral flange; Ihv, lateral head vein; It, lateral trough; me, mastoid exposure; oa, occipital artery; pc, prootic canal; pf, perilym- phatic foramen; pp, paroccipital process; pr, pro- montorium; ps, prootic sinus; pts, post-promon- torial tympanic sinus; ri, ramus inferior; rs, ramus superior; rso, ramus supraorbitalis; rt, ramus tem- poralis; sa, stapedial artery; saf, subarcuate fossa; sff, secondary facial foramen; sips, sulcus for in- ferior petrosal sinus; sps, sulcus for prootic sinus; ss, sigmoid sinus; sss, sulcus for sigmoid sinus; ts, transverse sinus; vag, ventral ascending groove; vii-h, hyomandibular branch of facial nerve; vii- p, palatine branch of facial nerve. 1995 WIBLE ET AL.: MAMMALIAN PETROSAL FROM MONGOLIA 5

ri 99

-IPs

sa

jv

Rostral L Medial F ri

-PC ,'Ps .sps - rs

- rso

Rostral I sss L Lateral

g9

ri

Dorsal L Rostra 6 AMERICAN MUSEUM NOVITATES NO. 3149 small extent. Presumably, when complete, the deep and contains three large foramina. The lateral flange would have contributed to a medial foramen, which is separated from the partial floor below the lateral trough. The ex- others by a low crista transversa, is oval and tent of this floor is uncertain, but it did not transmits the cochlear nerve. The anterolat- reach medially to the promontorium because eral foramen, also oval, is the primary facial contact scars are lacking. Lateral to the pos- foramen through which the facial nerve leaves terior margin ofthe lateral flange is a smooth, the cranial cavity to enter the cavum supra- concave surface, which is broken anteriorly. cochleare. The circular posterolateral fora- The nature of this concavity, which we call men is for the vestibular nerve. the anterolateral recess, will be treated sep- Posterolateral to the internal acoustic me- arately below. Posterior to this recess is a atus is the deeply recessed subarcuate fossa, moderate eminence, which is continuous lat- which in life housed the paraflocculus of the erally with a low crest and medially with the cerebellum. The gyrus of the anterior semi- broken base ofthe lateral flange. Immediately circular canal forms the posterior border of posterior to this region is a broad, transverse- the subarcuate fossa. Between the subarcuate ly oriented sulcus. This sulcus continues onto fossa and the internal acoustic meatus is a the lateral surface of the petrosal and is in- moderate-sized aperture for the endolym- terpreted as a ventral ascending groove for phatic duct, the aqueductus vestibuli. the ramus superior of the stapedial artery. Anterolateral to the subarcuate fossa is the At the posterolateral corner ofthe petrosal cranial aperture ofthe prootic canal. Leading is a stout, ventrally projecting paroccipital posteriorly from the aperture is a sulcus that process (figs. 1G, H, 2). It is compressed la- bends medially posterior to and follows the teromedially to form a crest whose long axis outer contour of the subarcuate fossa. This is oriented in the same direction as that of sulcus cannot be traced beyond the level of the promontorium. Two ridges extend from the aqueductus vestibuli. In the prootic canal the paroccipital process: the crista parotica and in the sulcus running posteriorly from it, and the caudal tympanic process of the pe- we reconstruct as the major occupant the trosal. The crista parotica runs anterodorsally prootic sinus, a distributary ofthe transverse and near its anterior end is a medially pro- sinus. In the part of the sulcus posterior and jecting broken prong of bone, which is iden- medial to the subarcuate fossa, we restore the tified as a tympanohyal, an ossified segment sigmoid sinus, the other distributary of the of Reichert's cartilage (fig. 2). The caudal transverse sinus. We believe that the major tympanic process runs medially, almost exit for the blood drained in the sigmoid sinus reaching the jugular notch. Between the back was the foramen magnum and not thejugular of the promontorium and the paroccipital foramen, given the latter aperture's small size. process, crista parotica, and caudal tympanic A shallow and incompletely preserved ca- process of the petrosal is a recess, the post- vum epiptericum for the trigeminal (semi- promontorial tympanic sinus (figs. IA, B, 2). lunar) ganglion lies lateral to and is separated In this recess anteromedial to the tympano- by a thick crista petrosa from the internal hyal is an oval depression, the fossa for the acoustic meatus. The cavum is connected to stapedius muscle (fig. 2). Visible in ventral the cavum supracochleare within the petrosal view along the entire lateral surface of the via a bony gap, the fenestra semilunaris, which petrosal is an extensive articular surface, pre- also occurs in Vincelestes (Rougier et al., 1992) sumably for the squamosal (fig. 1A). and some extant where it trans- mits no structures (Wible, 1990). DoRsAL VIEW (fig. ID, E, F) LATERAL VIEW (fig. 1G, H, I) This view shows three major structures: the internal acoustic meatus, subarcuate fos- The lateral view is dominated by extensive sa, and cavum epiptericum (most of which articular surfaces for the squamosal and su- is missing). The internal acoustic meatus lies praoccipital, both of which are largely com- in the anteromedial part of the petrosal. It is plete. Also shown in this view are four well- 1995 WIBLE ET AL.: MAMMALIAN PETROSAL FROM MONGOLIA

pp

PC -sf f *ar

hf

Fig. 2. Anteroventral views of the right petrosal from Khoobur, PSS-MAE- 104. Cross-hatching indicates damaged surfaces. Abbreviations: ar, anterolateral recess; cp, crista parotica; ctpp, caudal tympanic process of petrosal; hf, hiatus Fallopii; If, lateral flange; It, lateral trough; pc, prootic canal; pp, paroccipital process; pr, promontorium; pts, post-promontorial tympanic sinus; rlt, recess of lateral trough; sips, sulcus for inferior petrosal sinus; sf, stapedius fossa; th, tympanohyal; vag, ventral ascending groove. developed vascular grooves: from back to VAScuLAR RECONSTRUCTION (fig. 1C, F, I) front, these are grooves for the occipital ar- tery, arteria diploetica magna, ramus su- Selected soft tissues are restored in grooves, praorbitalis (dorsal ascending groove), and canals, and foramina on the Khoobur petro- ramus superior (ventral ascending groove). sal based on the anatomy ofliving mammals There are also two smaller grooves for rami as discussed by Rougier et al. (1992) and Wi- temporali, one continuous with the groove ble and Hopson (1995). Though not illus- for the arteria diploetica magna and the other trated in the figures, most ofthe arteries were with the ventral ascending groove. The likely accompanied by small companion veins grooves on the lateral surface ofthe petrosal, (Wible and Hopson, 1995). with the exception of that for the occipital The major components of the arterial pat- artery, would have been enclosed in bony tern interpreted to be present in non-mam- canals in life the malian eucynodonts (Rougier et al., 1992) by overlying squamosal. can be restored on the Khoobur petrosal. VIEW However, our arterial reconstruction differs POSTERIOR from the primitive pattern in having an ex- The dominant feature in this view (not fig- tremely short channel for the arteria diplo& ured) is the broad mastoid exposure of the tica magna, a very long, horizontal, intra- petrosal on the occiput (partially seen in fig. mural ventral ascending channel for the ra- IA, B, G, H). Along the dorsolateral edge of mus superior ofthe stapedial artery, and two the mastoid exposure is a notch that con- narrow channels for temporal rami (fig. I). As tained the arteria diploetica magna. Recon- in non-mammalian eucynodonts, there is no structing the squamosal in place would create direct indication on the petrosal for the in- a composite posttemporal foramen and canal ternal carotid and stapedial arteries. Never- for the arteria diploetica magna, as is the gen- theless, given the distribution ofboth vessels eralized condition in . among extant taxa, they were likely present 8 AMERICAN MUSEUM NOVITATES NO. 3149 on the Khoobur petrosal. In living mammals, in interpreting the morphology of the Khoo- the internal carotid artery follows one ofthree bur petrosal: the position ofthe fossa incudis alternative pathways, which may or may not and the nature ofthe anterolateral recess (figs. be indicated by a sulcus on the petrosal: either IA, B, 2). In most ofthe ingroup taxa studied transpromontorial, perbullar, or extrabullar here, there is a depression either lateral to or (Wible, 1986). We do not reconstruct the in- on the crista parotica for the dorsal process ternal carotid here, because there is no evi- of the quadrate (the crus breve of the incus). dence supporting any one alternative. The In contrast, there is no fossa incudis on the large size of the sulci for the ramus superior petrosal ofthe outgroup taxa tritheledontids, and ramus inferior suggests the presence of tritylodontids, Adelobasileus, and Sinocono- the stapedial artery and, consequently, we re- don. The only surface on the Khoobur petro- construct it. sal that resembles a fossa incudis is the an- The groove for the occipital artery in the terolateral recess. However, this recess is too Khoobur petrosal indicates that vessel was large and anteriorly placed to be a suitable very large and probably the major contrib- candidate. Moreover, the bone underlying the utor to the arteries on the sidewall of the recess is very thick, suggesting a mechanical braincase (fig. 1G, H, I). This was also prob- robustness beyond that required to support ably the case in non-mammalian eucyno- the quadrate (incus). The only structure in donts, given the relative sizes of their vas- this region that could require such a stout cular channels (Rougier et al., 1992). This construction is the glenoid fossa or an area supposedly primitive pattern is altered in abutting the glenoid. Vincelestes has a similar marsupials and placentals (except xenar- thickened, laterally directed process on its pe- thrans) by the reduction of the arteria di- trosal that is continuous laterally with the ploetica magna and the enlargement ofeither glenoid fossa (Rougier, 1993). Therefore, we the external carotid or stapedial system (Wi- interpret the anterolateral recess of the ble, 1987). Among , the arteria Khoobur petrosal as either a part of the gle- diploetica magna is large in the noid itself or, more likely, a bone providing (Kuhn, 1971), but is comparatively small in support for it. We have seen a comparable the (Rougier et al., 1992; Wible and arrangement in Ornithorhynchus, where the Hopson, 1995). glenoid fossa continues rostrally onto the an- Several major veins can also be recon- terior lamina. However, in contrast to the structed on the Khoobur petrosal, including condition in the Khoobur petrosal and Vin- the sigmoid, prootic, and inferior petrosal si- celestes, the anterior lamina in the platypus nuses and the lateral head vein. The most is not especially thickened near the glenoid striking difference from the pattern thought and is not a conspicuous laterally directed to be primitive for non-mammalian eucy- process. Given our interpretation for the an- nodonts (Rougier et al., 1992; Wible and terolateral recess, there is no suitable place Hopson, 1995) is the enclosure ofthe inferior for the fossa incudis on the Khoobur petrosal. petrosal sinus in a canal largely between the The articular facet for the squamosal indi- petrosal and basioccipital. The primitive cates that the squamosal bone contacted the course for this vessel is likely an endocranial lateral surface of the crista parotica. Conse- one, as occurs in monotremes (Hochstetter, quently, we believe that the fossa incudis must 1896) and is reconstructed for Morganuco- have been on the squamosal lateral to the don by Kermack et al. (1981). crista parotica, which as a primitive feature is peculiar in light of other, more advanced DISCUSSION attributes of the Khoobur petrosal. Appendix 2 shows the taxon-character ma- AFFINITIES OF THE KHOOBUR PETROSAL trix employed in our phylogenetic analysis. As stated above, a cladistic analysis of 44 The branch and bound algorithm of PAUP characters across 15 ingroup taxa was per- and the ie command of Hennig86 identified formed to evaluate the affinities of the nine equally most-parsimonious trees. The Khoobur petrosal. In the course of our char- strict consensus tree is shown in figure 3. In acter analysis, we encountered two problems an additional run on both programs, four 1 995 WIBLE ET AL.: MAMMALIAN PETROSAL FROM MONGOLIA 9 multistate characters (2, 9, 34, 41) for which Sigogneau-Russell et al. (1992). The material hypotheses of transformation could be sup- described to date consists of isolated teeth ported were coded as ordered, and the same and upper and lowerjaws. An additional tax- nine trees were obtained. on, Prodelttheridium kalandadzei [sic] Tro- As shown in the strict consensus tree, the fimov (1984) cited in Reshetov and Trofimov Khoobur petrosal falls at a trichotomy with (1984), was said to be from Khoobur. How- and a clade comprising ever, it was never figured or described and, Theria and Vincelestes, the Prototribosphen- consequently, was considered to be nomen ida ofRougier (1993). In six ofthe nine trees nudum by Kielan-Jaworowska and Nessov obtained, the Khoobur petrosal forms a (1990). monophyletic group with Triconodontidae. To associate the isolated petrosal with a This assemblage is supported by the presence taxon known from dentition and given the of a deep recess in the anterior part of the possible relationships suggested by our anal- lateral trough (character 26) and an intra- ysis, we used nearly complete skulls and low- mural ventral ascending canal between the er jaws of a Cloverly triconodontid (MCZ petrosal and squamosal (41). However, we 19974) and Vincelestes (Rougier, 1993) to are confident that the Khoobur petrosal does predict skull length for the Khoobur petrosal. not belong to the family Triconodontidae be- The estimated skull length was about 4.0 cm cause of the striking differences, including, based on the triconodontid and 3.5 cm based for example, the anterolateral recess and cau- on Vincelestes. There are two taxa reported dal tympanic process in the Khoobur speci- thus far from Khoobur with predicted skull men. Yet it may belong to a triconodont or lengths (based on lower jaw length) approx- to its sister group as we use the term here (see imating that calculated for the Khoobur pe- Materials and Methods). In the remaining trosal: the gobiconodontid Gobiconodon bor- three trees, the Khoobur petrosal is linked to issiaki (Trofimov, 1978) at about 5.0 cm and the prototribosphenidan clade by a caudal the symmetrodont Gobiodon infinitus (Tro- tympanic process ofthe petrosal (12), a post- fimov, 1980) at between 4.1 and 4.6 cm. If promontorial tympanic sinus (13), a thick- gobiconodontids form a monophyletic group ened laterally directed surface ofthe petrosal with triconodontids, as tentatively suggested related to the glenoid fossa (32), and two re- by Rowe (1993), either assignment is possi- versals, i.e., presence ofcrista parotica-squa- ble: triconodontid affinities on the one hand mosal contact (17) and absence of crests de- (with Gobiconodon borissiaki) and prototri- marcating a tensor tympani fossa (27). Al- bosphenidan affinities on the other (with Go- though more characters support linkage of biodon infinitus). the Khoobur petrosal with the prototribos- Gobiconodon is also known from the Clov- phenidan clade than with Triconodontidae, erly Formation of North America (Jenkins both hypotheses are equally parsimonious and Schaff, 1988). A separate species, G. os- considering the tree as a whole. tromi, was recognized for these specimens, Regarding the affinities ofthe Khoobur pe- with a skull length estimated to be more than trosal, the results of our computer analysis 10.0 cm by Jenkins and Schaff(1988). Among are not conclusive. Considering the fossil rec- other cranial material, an isolated left squa- ord from Khoobur, a preliminary faunal list mosal was recovered. This squamosal pre- was published by Clemens et al. (1979), com- serves a nearly complete glenoid fossa that is piled from Russian bibliographies, mainly separated from the squama by a distinct neck. those ofBeliajeva et al. (1974) and Dashzeveg The presence of a neck would preclude any (1975). Additional taeniolabidoid multitu- contact between the glenoid fossa and the berculate material was described by Trofi- petrosal, which we have interpreted above to mov (1980) and Kielan-Jaworowska et al. be present in the Khoobur petrosal, given the (1987); amphidontid symmetrodonts by Tro- morphology ofthe anterolateral recess. Ifthe fimov (1980); cladotherians by Dashzeveg Khoobur petrosal is affiliated with Mongo- (1979, 1994); and tribosphenidans by Dash- lian triconodonts, either Gobiconodon from zeveg and Kielan-Jaworowska (1984), Kie- the Gobi and Gobiconodon from North lan-Jaworowska and Dashzeveg (1989), and America are not closely related or the glenoid AMERICAN MUSEUM NOVITATES NO. 3149

C-l 0 i

Fig. 3. Strict consensus tree ofnine equally most parsimonious trees produced by PAUP and Hennig86. The strict consensus (nelsen) on Hennig86 has the following parameters: branch length = 90; consistency index = 0.58; and retention index = 0.73; these values on the nine individual trees in Hennig86 are 85, 0.62, and 0.77, respectively. ofGobiconodon varies remarkably at the spe- some ofthe taxa included here and have em- cies level (i.e., continuity with the petrosal ployed some petrosal characters: Rowe (1988, on the one hand versus separation via a dis- 1993), Wible (1991), Lillegraven and Krusat tinct neck on the other). Given that the lower (1991), Crompton and Luo (1993), Wible and jaws and dentitions of the North American Hopson (1993), Lucas and Luo (1993), and and Mongolian forms are remarkably simi- Luo (1994). Our analysis includes all Meso- lar, we agree with Jenkins and Schaff (1988) zoic mammaliamorphs for which the petro- that they belong to the same genus. Further- sal is currently known. Comparisons of our more, we would not expect such an extreme results with those of the authors mentioned difference in glenoid morphology between two above highlight the following problematic is- species of the same genus. Consequently, we sues in mammaliamorph phylogeny. think it unlikely that the Khoobur petrosal 1) and belongs to Gobiconodon sp. We conclude that Relationships. There is currently debate as to it belongs to either an as yet unknown tri- whether tritheledontids or tritylodontids are conodont or to a primitive holotherian, which more closely related to mammaliaforms. Our in the context of the currently known Khoo- analysis does not provide a rigorous test of bur fauna would be the symmetrodont Go- this problem, which requires that more non- biodon infinitus. Holotheria (Hopson, 1994) mammalian be considered as in- includes the common ancestor of Kuehneo- group taxa. Nevertheless, our analysis of pe- therium and therians plus all its descendants. trosal characters supports tritylodontids in that position, in agreement with Rowe (1988, 1993), Wible (1991), Wible and Hopson IMPLICATIONS FOR (1993), Lucas and Luo (1993: fig. 14b), and MAMMALIAMORPH PHYLOGENY Luo (1994: fig. 6.8). Two unequivocal and Recently, there have been a number of three equivocal synapomorphies support this studies that have discussed the affinities of relationship (appendix 3, node 10). Other i995 WIBLE ET AL.: MAMMALIAN PETROSAL FROM MONGOLIA I I analyses using some petrosal characters to- is necessary to add four steps to the shortest gether with characters from other systems trees of our study. A monophyletic grouping reach the reverse conclusion (Lucas and Luo, ofHaldanodon and Megazostrodon is weakly 1993: fig. 14a; Luo, 1994; Luo and Cromp- supported here by two unequivocal synapo- ton, 1994). Given these conflicting results, it morphies, one ofwhich is a reversal, plus one appears that this phylogenetic problem can- equivocal trait (appendix 3, node 5). Simi- not be resolved with currently available in- larly, Butler (1988) suggested a closer rela- formation. Of crucial importance is a better tionship ofdocodonts to Megazostrodon than understanding of the cranial and postcranial to . skeleton of tritheledontids. 4) Interrelationships of Mammals. Mam- 2) Mammaliaform Relationships. As malia, as defined by Rowe (1988), comprises defined by Rowe (1988), Mammaliaformes the latest common ancestor of monotremes, comprises the common ancestor of Morgan- marsupials, and placentals plus all its de- ucodon and Mammalia plus all its descen- scendants. In our analysis, Mammalia is sup- dants. In our analysis of petrosal characters, ported by four unequivocal and two equiv- and Adelobasileus are the suc- ocal synapomorphies (see appendix 3, node cessive sister taxa to Mammaliaformes. In 4). As follows from Rowe's definition, the their description ofAdelobasileus, Lucas and scope of Mammalia is dependent on the po- Luo (1993: fig. 14) presented two phyloge- sition of monotremes among Mesozoic taxa, netic hypotheses: a strict consensus tree, which is a controversial topic in recent lit- which includes Adelobasileus within Mam- erature. Prior to the discovery of the Early maliaformes and Sinoconodon as the first Cretaceous Steropodon galmani outgroup, and their preferred tree, which (Archer et al., 1985), there was a general con- places Adelobasileus and Sinoconodon at a sensus that monotremes were "non-therian trichotomy with other mammaliaforms. Cri- mammals" (Kermack, 1967; Hopson, 1970; teria for their preference oftrees are not spec- Kielan-Jaworowska, 1970; Crompton and ified. Whereas their strict consensus tree con- Jenkins, 1979). However, after that discov- tradicts ours, their preferred tree does not. If ery, monotremes were allied with different our hypothesis is correct, there is congruence groups of holotherians: tribosphenidans between the branching pattern and age ofthe (Archer et al., 1985), peramurids (Kielan-Ja- taxa, because Adelobasileus is known from worowska et al., 1987; Jenkins, 1990), dry- the and the others from the Liassic. olestoids (Bonaparte, 1990; Archer et al., The basal position of Adelobasileus is sup- 1993), and symmetrodonts (Hopson, 1994). ported here by two unequivocal and three Our analysis does not support holotherian equivocal synapomorphies (appendix 3, node affinities for monotremes (see also Rougier, 9). 1993), unless multituberculates and tricon- 3) Relationships. The only odonts are also considered to be holotherians. basicranial material known for docodonts is We see no compelling evidence for including the or Kimmeridgian Haldanodon multituberculates and triconodonts within exspectatus described by Krusat (1980) and Holotheria, because they do not share the Lillegraven and Krusat (1991). In the latter reversed-triangles pattern used to di- paper, 149 craniomandibular characters were agnose that grouping. The same conclusion scored among four ingroup taxa: Haldano- is implicit in the cladograms published in don, Sinoconodon, Morganucodon, and a col- Rowe (1988, 1993) and Wible and Hopson lective group called other mammals. The re- (1993), but these authors did not explicitly sulting phylogenetic hypothesis places Hal- address the problem ofthe hypothesis ofhol- danodon as the outgroup to the other ingroup otherian affinities for monotremes. Removal taxa. This contrasts with the more wide- of monotremes from Holotheria has impor- spread view that docodonts are mammali- tant implications for predicting the biology aforms-that is, related to morganucodon- of basal holotherians (e.g., reproduction, tids or even more derived (Crompton and thermoregulation) as attempted by Jenkins Jenkins, 1979; Crompton and Sun, 1985; (1990). McKenna, 1987). To obtain Lillegraven and Another controversial topic in recent lit- Krusat's (1991) position for Haldanodon, it erature is the phylogenetic position of mul- 12 AMERICAN MUSEUM NOVITATES NO. 3149 tituberculates. As was the case for mono- borissiaki questionable. We conclude that the tremes, multituberculates have been tradi- Khoobur petrosal is either that of an as yet tionally allied with "non-therian mammals," unknown triconodont or a holotherian, which and several recent authors (e.g., McKenna, is in the size range of Gobiodon infinitus. 1987; Lillegraven and Krusat, 1991; Kielan- Given the nature ofthe Khoobur fossil de- Jaworowska, 1992) have considered multi- scribed here, we were compelled to restrict tuberculates as an ancient lineage possibly our analysis to characters ofthe petrosal bone outside mammaliaforms. In contrast, Kemp and associated structures. Wible and Hopson (1983, 1988) tentatively grouped multituber- (1993) presented an analysis that was similar culates with cladotherians, and Rowe (1988) in scope, both morphologically and taxo- and Wible (1991) placed multituberculates at nomically. Subsequent to their publication, an unspecified rank among holotherians. In new relevant taxa and additional information our analysis of petrosal characters, three of from previously known forms became avail- the nine trees as well as reweighting of the able. The inclusion ofthese data in our anal- data set support a monophyletic grouping of ysis resulted in trees with topologies different multituberculates and monotremes (see also from those in Wible and Hopson (1993). We Wible and Hopson, 1993); in the remaining attribute these differences to an increase in six trees, multituberculates are more closely the number of taxa studied from 11 to 15, related to triconodontids, the Khoobur pe- an increase in the number of characters (30 trosal, and prototribosphenidans than to to 44 from the petrosal), and fewer missing monotremes. data. In spite of a more complete data set, Triconodontids, also traditionally consid- we obtained poorer resolution than did Wible ered "non-therians," are linked in our anal- and Hopson (1993). Some nonconflicting ysis to the Khoobur petrosal and prototri- characters used by those authors were ren- bosphenidans by two unequivocal and six dered homoplastic in our analysis with the equivocal synapomorphies (appendix 3, node increased number oftaxa studied and the de- 3). Following Rowe (1993), we apply the term creased amount of missing data. Theriimorpha for this grouping, but our anal- The topological changes resulting from the ysis renders it a less inclusive category be- addition of a few taxa and characters are sig- cause multituberculates are excluded. No nificant. Our results highlight the impact on other analyses have postulated such affinities phylogeny of a dramatically increasing mor- for triconodontids. Several complete skulls phological database for Mesozoic mammal- from the Cloverly Formation have been par- iamorphs. tially described (Crompton and Jenkins, 1979; Crompton and Sun, 1985; Crompton and Luo, 1993), and their full description will ACKNOWLEDGMENTS greatly increase the database for resolving the affinities of this group. For permission to study material under his care we are grateful to Dr. Farish A. Jenkins Jr. of the Museum of Comparative Zoology, Harvard University. We also thank Ed Heck CONCLUSIONS for the illustrations and photographs. For Our phylogenetic analysis of the Khoobur comments and improvements on an earlier petrosal restricts its affiliation to two possible version of the manuscript, we acknowledge groups: either with triconodontids or with Jim Hopson, Zofia Kielan-Jaworowska, Zhexi holotherians. In the fauna, the triconodont Luo, and Jin Meng. This research was sup- Gobiconodon borissiaki and the symmetro- ported by NSF Grant DEB 930070 to M. J. dont Gobiodon infinitus match the predicted Novacek and BSR 91-19212 to J. R. Wible. size. However, complementary morpholog- G. W. Rougier's research was supported by ical information from Gobiconodon ostromi a Frick Research Fellowship from the AMNH makes the assignment of the petrosal to G. and by CONICET, Argentina. 1995 WIBLE ET AL.: MAMMALIAN PETROSAL FROM MONGOLIA

REFERENCES Archer, M., T. F. Flannery, A. Ritchie, and R. E. first two-thirds of mammalian history, Molnar pp. 59-73. Berkeley: Univ. ofCalifornia 1985. First Mesozoic mammals from Austra- Press. lia-an Early Cretaceous monotreme. Crompton, A. W., and Z. Luo Nature 318: 363-366. 1993. Relationships of the Liassic mammals Archer, M., P. Murray, S. Hand, and H. Godthelp Sinoconodon, Morganucodon oehleri, 1993. Reconsideration of monotreme rela- and . In F. S. Szalay, M. tionships based on the skull and den- J. Novacek, and M. C. McKenna (eds.), tition of the Miocene Obduradon dick- Mammal phylogeny: Mesozoic differ- soni. In F. S. Szalay, M. J. Novacek, entiation, multituberculates, mono- and M. C. McKenna (eds.), Mammal tremes, early therians, and marsupials, phylogeny: Mesozoic differentiation, pp. 30-44. New York: Springer-Verlag. multituberculates, monotremes, early Crompton, A. W., and A. L. Sun therians, and marsupials, pp. 75-94. 1985. Cranial structure and relationships of New York: Springer-Verlag. the Liassic mammal Sinoconodon. Zool. Beliajeva, E. I., B. A. Trofimov, and V. Yu. Resh- J. Linn. Soc. 85: 99-119. etov Dashzeveg, D. 1974. General stages ofevolution ofLate Me- 1975. Kielantherium gobiensis, a primitive sozoic and Early Tertiary mammalian therian from the Early Cretaceous of fauna in central Asia. In N. N. Kra- Mongolia. Nature 227: 402-403. marenko, B. Luvsandansan, Yu. I. Vo- 1979. Arguimus khosbajari gen. n., sp. n., ronin, R. Barsbold, A. K. Rozhdestven- (Peramuridae, Eupantotheria) from the sky, B. A. Trofimov, and V. Yu. Resh- lower Cretaceous ofMongolia. Acta Pa- etov (eds.), Joint Soviet-Mongolian Pa- laeontol. Polonica 24: 199-204. leontological Expedition Transactions 1994. Two previously unknown eupanto- 1, pp. 19-45. [In Russian] theres (Mammalia, Eupantotheria). Am. Bonaparte, J. F. Mus. 1990. New mammals from Novitates 3107: 11 pp. Los Alamitos Formation, northern Pa- Dashzeveg, D., and Z. Kielan-Jaworowska tagonia. Natl. Geogr. Res. 6: 63-95. 1984. The lower jaw of an aegialodontid Bonaparte, J. F., and G. W. Rougier mammal from the Early Cretaceous of 1987. Mamiferos del Cretacico Inferior de Pa- Mongolia. Zool. J. Linn. Soc. 82: 217- tagonia. Actas IV Congr. Latinoam. Pa- 227. leontol., Bolivia 1: 343-359. Farris, J. S. Butler, P. M. 1988. Hennig86 Reference, vers. 1.5. Private- 1988. Docodont molars as tribosphenic ana- ly published computer software. logues (Mammalia, ). In D. E. Gauthier, J. S. Russell, J.-P. Santoro, and D. Sigog- 1986. Saurischian monophyly and the origin neau-Russell (eds.), Teeth revisited, of birds. In K. Padian (ed.), The origin Mem. Mus. Natl. Hist. Nat., Paris, ser. ofbirds and the evolution offlight. Cal- C, 53: 329-340. ifornia Acad. Sci. Mem. 8: 1-55. Clemens, W. A., J. A. Lillegraven, E. H. Lindsay, Goodrich, E. S. and G. G. Simpson 1930. Studies on the structure and develop- 1979. Where, when, and what-a survey of ment of . London: Macmil- known Mesozoic mammal distribution. lan. In J. A. Lillegraven, Z. Kielan-Jawo- Gow, C. E. rowska, and W. A. Clemens (eds.), Me- 1986. A new skull of Megazostrodon (Mam- sozoic mammals, the first two-thirds of malia: Triconodonta) from the Elliot mammalian history, pp. 7-58. Berke- Formation (Lower Jurassic) ofsouthern ley: Univ. of California Press. Africa. Palaentol. Afrricana 26: 13-23. Crompton, A. W., and F. A. Jenkins, Jr. Hahn, G. 1979. Origin of mammals. In J. A. Lillegrav- 1988. Die Ohr-region der Paulchoffatiidae en, Z. Kielan-Jaworowska, and W. A. (, Ober-Jura). Pa- Clemens (eds.), Mesozoic mammals, the laeovertebr. Montpellier 18: 155-185. 14 AMERICAN MUSEUM NOVITATES NO. 3149

Hochstetter, F. Kielan-Jaworowska, Z., R. Presley, and C. Poplin 1896. Beitraige zur Anatomie und Entwick- 1986. The cranial vascular system in taeniola- lungsgeschichte des Blutgefassystems der bidoid multituberculate mammals. Phi- Monotremen. Semon Zool. For- los. Trans. R. Soc. London B313: 525- schungsreisen in Australien 5: 189-243. 602. Hopson, J. A. Kielan-Jaworowska, Z., A. W. Crompton, and F. 1970. The classification of non-therian mam- A. Jenkins, Jr. mals. J. Mammal. 51: 1-9. 1987. The origin of -laying mammals. Na- 1991. Systematics of the non-mammalian ture 326: 871-873. Synapsida and implications for patterns Kielan-Jaworowska, Z., D. Dashzeveg, and B. A. of evolution in synapsids. In H.-P. Trofimov Schultze and L. Trueb (eds.), Origins of 1987. Early Cretaceous multituberculates from the higher groups of : contro- Mongolia and a comparison with Late versy and consensus, pp. 635-693. Ith- Jurassic forms. Acta Palaeontol. Polon- aca, NY: Comstock. ica 32: 3-47. 1994. evolution and the radiation of Krusat, G. non-eutherian mammals. In R. S. Spen- 1980. Contribuigco pora o conhecimento da cer (ed.), Major features of vertebrate Fauna do Kimeridgiano da Mina de evolution, Paleontol. Soc. Short Cours- Lignito Guimarota (Leira, Portugal). IV es in Paleontol. 7: 190-219. Parte-Haldanodon exspectatus Kiihne Jenkins, F. A., Jr. and Krusat 1972 (Mammalia, Doco- 1990. Monotremes and the biology of Meso- donta). Mem. Serv. Geol. Portugal 27: zoic mammals. Netherlands J. Zool. 40: 1-79. 5-31. Kuhn, H.-J. Jenkins, F. A., Jr., and C. R. Schaff 1971. Die Entwicklung und Morphologie des 1988. The Early Cretaceous mammal Gobi- Schaidels von Tachyglossus aculeatus. conodon (Mammalia, Triconodonta) Abh. Senckenb. Naturforsch. Ges. 528: from the Cloverly Formation in Mon- 1-224. tana. J. Vertebr. Paleontol. 8: 1-24. Lillegraven, J. A., and G. Hahn Kemp, T. S. 1993. Evolutionary analysis ofthe middle and 1983. The interrelationships of mammals. inner ear of multituber- Zool. J. Linn. Soc. 77: 353-384. culates. J. Mammal. Evol. 1: 47-74. 1988. Interrelationships of Synapsida. In M. Lillegraven, J. A., and G. Krusat J. Benton (ed.), The phylogeny and clas- 1991. Cranio-mandibular anatomy of Hal- sification of tetrapods, vol. 2: mam- danodon exspectatus (; mals, pp. 1-22. Oxford: Clarendon. Mammalia) from the Late Jurassic of Kermack, K. A. Portugal and its implications to the evo- 1963. The cranial structure of triconodonts. lution of mammalian characters. Con- Philos. Trans. R. Soc. London B246: trib. Geol., Univ. Wyo. 28: 39-138. 83-102. Lucas, S. G., and Z. Luo 1967. The interrelationships of early mam- 1993. Adelobasileus from the Upper mals. Zool. J. Linn. Soc. 47: 241-249. of west : the oldest mammal. J. Kermack, K. A., F. Mussett, and H. W. Rigney Vertebr. Paleontol. 13: 309-334. 1981. The skull of Morganucodon. Zool. J. Luo, Z. Linn. Soc. 71: 1-158. 1994. Sister taxon relationships of mammals Kielan-Jaworowska, Z. and transformations of diagnostic 1970. Unknown structures in multitubercu- mammalian characters. In N. C. Fraser late skull. Nature 226: 974-976. and H.-D. Sues (eds.), Life in the shad- 1992. Interrelationships of Mesozoic mam- ow of dinosaurs: Early Mesozoic tetra- mals. Hist. Biol. 6: 185-202. pods, pp. 98-128. Cambridge: Cam- Kielan-Jaworowska, Z., and D. Dashzeveg bridge Univ. Press. 1989. Eutherian mammals from the Early Luo, Z., and A. W. Crompton Cretaceous of Mongolia. Zool. Scr. 18: 1994. Transformation of the quadrate (incus) 347-355. through the transition from non-mam- Kielan-Jaworowska, Z., and L. A. Nessov malian cynodonts to mammals. J. Ver- 1990. On the metatherian nature of the Del- tebr. Paleontol. 14: 341-374. tatheroida, a sister group of the Mar- Luo, Z., A. W. Crompton, and S. G. Lucas supialia. Lethaia 23: 1-10. 1995. Evolutionary origins ofthe mammalian 1995 WIBLE ET AL.: MAMMALIAN PETROSAL FROM MONGOLIA

promontorium and cochlea. J. Vertebr. 1988. Definition, diagnosis and origin of Paleontol. 15: 113-121. Mammalia. J. Vertebr. Paleontol. 8: McKenna, M. C. 241-264. 1975. Towards a phylogenetic classification of 1993. Phylogenetic systematics and the early the Mammalia. In W. P. Luckett and F. history ofmammals. In F. S. Szalay, M. S. Szalay (eds.), Phylogeny of the pri- J. Novacek, and M. C. McKenna (eds.), mates: a multidisciplinary approach, pp. Mammal phylogeny: Mesozoic differ- 21-46. New York: Plenum Press. entiation, multituberculates, mono- 1987. Molecular and morphological analysis tremes, early therians, and marsupials, of higher-level mammalian interrela- pp. 129-145. New York: Springer-Ver- tionships. In C. Patterson (ed.), - lag. cules and morphology in evolution: Sigogneau-Russell, D., D. Dashzeveg, and D. R. conflict or compromise, pp. 55-93. Russell Cambridge: Cambridge Univ. Press. 1992. Further data on Prokennalestes (Mam- Novacek, M. J. malia, inc. sed.) from the Early 1986. The skull of leptictid insectivorans and Cretaceous of Mongolia. Zool. Scr. 21: the higher-level classification ofeuther- 205-209. ian mammals. Bull. Am. Mus. Nat. Hist. Sues, H. D. 183: 1-112. 1986. The skull and dentition of two tritylo- 1993. Patterns of diversity in the mammalian dontid synapsids from the Lower Ju- skull. In J. Hanken and B. K. Hall (eds.), rassic of western North America. Bull. The skull, vol. 2, patterns of structural Mus. Comp. Zool. 151: 217-268. and systematic diversity, pp. 438-545. Sun, A. L. Chicago: Univ. of Chicago Press. 1984. Skull morphology of the tritylodont ge- Oelrich, T. M. nus of Sichuan. Scientia 1956. The anatomy ofthe head of Ctenosaura Sinica B27: 205-217. pectinata (Iguanidae). Misc. Publ., Mus. Swofford, D. L. Zool., Univ. Mich. 94: 1-122. 1993. PAUP: phylogenetic analysis using par- Prothero, D. R. simony, version 3.1.1. Washington, D. 1981. New Jurassic mammals from Como C.: Smithsonian Institution. Bluff, Wyoming, and the interrelation- Trofimov, B. A. ships ofNon-Tribosphenic Theria. Bull. 1978. The first triconodonts (Mammalia, Tri- Am. Mus. Nat. Hist. 167: 277-326. conodonta) from Mongolia. Dokl. Akad. 1983. The oldest mammalian petrosals from Nauk SSR 243: 213-216. [In Russian] North America. J. Paleontol. 57: 1040- 1980. Multituberculata and Symmetrodonta 1046. from the Lower Cretaceous from Mon- Reshetov, V. Yu., and B. A. Trofimov golia. Dokl. Akad. Nauk SSR 251: 209- 1984. Review of the study of fossil mammals 212. [In Russian] from the USSR. In W. E. Sokolov and Wible, J. R. W. W. Kucheryk (eds.), Theriology in 1986. Transformation in the extracranial the USSR, pp. 6-29. Moscow. [In Rus- course of the internal carotid artery in sian] mammalian phylogeny. J. Vertebr. Pa- Rougier, G. W. leontol. 6: 313-325. 1993. Vincelestes neuquenianus Bonaparte 1987. The eutherian stapedial artery: charac- (Mammalia, Theria) un primitivo ter analysis and implications for super- mamifero del Cret'acico Inferior de la ordinal relationships. Zool. J. Linn. Soc. Cuenca Neuquina. Diss., Univ. of Bue- 91: 107-135. nos Aires, 720 pp. 1990. Petrosals ofLate Cretaceous marsupials Rougier, G. W., J. R. Wible, and J. A. Hopson from North America and a cladistic 1992. Reconstruction of the cranial vessels in analysis ofthe petrosal in therian mam- the Early Cretaceous mammal Vince- mals. J. Vertebr. Paleontol. 10: 183-205. lestes neuquenianus: implications for the 1991. Origin of Mammalia: the craniodental evolution of the mammalian cranial evidence reexamined. Ibid. 11: 1-28. vascular system. J. Vertebr. Paleontol. Wible, J. R., and J. A. Hopson 12: 188-216. 1993. Basicranial evidence for early mammal Rowe, T. phylogeny. In F. S. Szalay, M. J. No- 1987. Definition and diagnosis in the phylo- vacek, and M. C. McKenna (eds.), genetic system. Syst. Zool. 36: 208-211. Mammal phylogeny: Mesozoic differ- 16 AMERICAN MUSEUM NOVITATES NO. 3149

entiation, multituberculates, mono- 18. Quadrate (incus) articulation in fossa incudis tremes, early therians, and marsupials, on petrosal - absent (0) or present (1). pp. 45-62. New York: Springer-Verlag. 19. Pterygoparoccipital foramen position relative 1995. Homologies of the prootic canal in to fenestra vestibuli - at same level or pos- mammals and non-mammalian cyno- terior to (0) or anterior to (1). donts. J. Vertebr. Paleontol. 15. 20. Prootic canal - absent (0), present within pe- Zeller, U. trosal (1), or present between petrosal and 1989. Die Entwicklung und Morphologie des squamosal (2). Schiidels von Ornithorhynchus anatinus 21. Cranial aperture of prootic canal - in rear of (Mammalia: : Monotrema- cavum epiptericum (0) or at anterodorsal ta). Abh. Senckenb. Naturforsch. Ges. margin of subarcuate fossa (1). 545: 1-88. 22. Tympanic aperture ofprootic canal - separate from (0) or confluent with pterygoparoccipital foramen (1). APPENDIX 1 23. Cavum epiptericum floor- open ventrally (0), partially floored by petrosal (1), floored, pri- Petrosal Characters marily by petrosal (2), or floored, primarily 1. Sutural relation of prootic and opisthotic - by alisphenoid (3). unfused (0) or fused to form petrosal (1). 24. Facial ganglion floor - open ventrally (0) or 2. Petrosal promontorium - absent (0), incipient floored by petrosal (1). but distinctive with anterior part covered by 25. Lateral trough ofpetrosal - absent (0) or pres- basisphenoid wing (1), or inflated without ba- ent (1). sisphenoid wing (2). 26. Deep recess in anterior part of lateral trough 3. Promontorium shape - triangular in outline, - absent (0) or present (1). wider posteriorly (0) or fingerlike outline (1). 27. Crests demarcating distinct fossa for tensor 4. Ventral crest on promontorium - absent (0) tympani on petrosal - absent (0) or present or present (1). (1). 5. Fenestra vestibuli margin - with thickened 28. Craniomandibular joint position relative to osseous ring (0) or smooth, without osseous fenestra vestibuli - level with (0) or anterior ring (1). to (1). 6. Fenestra vestibuli shape - round, stapedial 29. Petrosal (prootic) contact with quadrate ra- ratio less than 1.5 (0) or elongate, stapedial mus of alisphenoid (epipterygoid) - elongate ratio greater than 1.5 (1). (0) or greatly reduced or absent (1). 7. Wall separating perilymphatic foramen from 30. Lateral flange of petrosal (prootic) - laterally jugular foramen - absent (0) or present (1). directed shelf (0), ventrally directed crest (1), 8. Channel for perilymphatic duct - open (0) or or absent (2). enclosed in osseous canal, a cochlear aqueduct 31. Vascular foramen in lateral flange of petrosal (1). (prootic) - absent (0) or present (1). 9. Cochlear duct - straight (0), bent (1), or coiled, 32. Thick, laterally directed surface of petrosal 3600 (2). related to glenoid fossa - absent (0) or present 10. Secondary spiral lamina - absent (0) or pres- (1). ent (1). 33. Sulcus along basioccipital surface of petrosal 11. Semicircular canals - enclosed by opisthotic, (for inferior petrosal sinus) - absent (0) or supraoccipital, and exoccipital (0) or enclosed present (1). by opisthotic (petrosal) (1). 34. Cavum supracochleare - absent (0), present, 12. Caudal tympanic process of petrosal - absent with incomplete wall separating it from ca- (0) or present (1). vum epiptericum (fenestra semilunaris pres- 13. Post-promontorial tympanic sinus - absent ent) (1), or present, completely enclosed by (0) or present (1). bone (2). 14. Fossa for stapedius muscle - absent (0) or 35. Separate foramina for cochlear and vestibular present (1). nerves - absent (0) or present (1). 15. Paroccipital process with distinct, ventrally 36. Common depression housing foramina for fa- directed projection for muscle attachment - cial and vestibulocochlear nerves - absent (0) absent (0) or present (1). or present (1). 16. Tympanohyal - separate (0) or attached to 37. Pila antotica - ossified base present (0) or ab- petrosal (1). sent (1). 17. Crista parotica-squamosal contact - present 38. Anterior lamina ofpetrosal (prootic) - present (0) or absent (1). (0) or greatly reduced or absent (1). 1995 WIBLE ET AL.: MAMMALIAN PETROSAL FROM MONGOLIA 17

39. Contribution of anterior lamina of petrosal Most ofthe above characters are taken directly or (prootic) to cranial wall - extends forward to are modified from the following references, to posterior border of exits for mandibular and which the reader is referred for additional infor- maxillary nerves (0), expanded forward an- mation. terior and dorsal to nerve exits, no contact Characters: with orbitosphenoid (1), or expanded forward 1, 7, 9, 16, 28, 35, 37 - Rowe (1988) and contacting orbitosphenoid (2). 2, 4, 5, 44 - Lucas and Luo (1993) 40. Number of exits for mandibular and maxil- 6, 12, 13-Wible (1990) lary nerves - single (0) or multiple (1). 8, 38 - Kermack et al. (1981) 41. Ventral ascending groove or canal - open lat- 11 - Luo et al. (1995) erally (0), intramural, within petrosal (1), or intramural, between petrosal and squamosal 25, 33, 40 - Lillegraven and Krusat (1991) 14, 15, 20, 21, 22, 23, 27, 34, 36, 39, 41, 43 - (2). Wible and Hopson (1993) 42. Orientation ofventral ascending groove or ca- 24 - Crompton and Sun (1985) nal - vertical (0) or horizontal (1). 29, 31 - Wible (1991) 43. Dorsal ascending groove or canal - open lat- 30 - Luo (1994) erally (0) or intramural or endocranial (1). 44. Posterior opening of posttemporal canal - separated from margin of occiput by tabular (0) or positioned close to margin of occiput, bordered dorsolaterally by squamosal (1). 18 AMERICAN MUSEUM NOVITATES NO. 3149

APPENDIX 2 Taxon-Character Matrix Character* Taxon 5 10 15 20 25 30 35 40 44 Outgroup OONNO 00000 00000 0000? ??000 NOOOO ?0000 ?0000 0000 Tritheledontidae ?ONNO 0100? 00000 00000 NNOOO NOOOO 00001 00000 0000 Tritylodontidae lONNO 0100? 00010 11000 NNOOO NOOOO 10001 10001 0010 Adelobasileus 11001 0100? ?0000 ?0001 ?0101 000?1 1000? ?0011 0011 Sinoconodon 12011 0100? 10011 00001 00111 00000 10001 ?0011 0011 Morganucodon 12001 01000 10011 01101 10111 00001 10001 00011 0011 Megazostrodon 12011 0100? 10011 ?1101 101?1 00001 100?1 ??010 00?1 Haldanodon 12011 01000 10011 1?111 10?11 00001 10001 11010 00?? Dinnetherium 12001 0100? 10011 11101 10111 ?1001 100?1 ??0?? 00?1 Triconodontidae 12101 11000 10011 11111 10211 11101 00021 1?011 2111 Multituberculata 12101 01000 10011 11111 11211 01011 00101 10021 1111 Vincelestes 12001 11111 11111 11111 10211 00111 11011 11011 1111 Monotremata 12001 01010 10001 11101 11211 00011 00001 10021 1011 Marsupialia 12001 01121 11110 101N2 1N310 N0112 00121 lllNl NNN1 12001 11121 11110 10110 NN310 N0112 00121 lllNl NN11 Khoobur Petrosal 12101 11000 11111 10311 10211 10111 01111 1?0?? 211? * ? = missing data or multistate condition for outgroup; N = not applicable; 0-3 = character states (see appendix 1). 1995 WIBLE ET AL.: MAMMALIAN PETROSAL FROM MONGOLIA 19

APPENDIX 3 Node 5 - Megazostrodon + Haldanodon 4 - ventral crest on promontorium Diagnoses of Nodes 40(0) - single exit for mandibular and maxillary nerves Nodes are diagnosed by listing the derived con- Node 6 ditions for characters in appendix 1. Multistate 16 - characters or reversals are depicted within paren- tympanohyal on cranium theses () following the character number. Aster- 36* - facial and vestibulocochear nerves in com- isks (*) denote equivocal characters. These are list- mon depression [node 10] ed at the less inclusive node, and the more inclu- Node 7 - Mammaliaformes sive distribution is indicated within brackets []. 17 - crista parotica/squamosal contact absent 18 - fossa incudis present on petrosal Node 1 - Theria 21 - cranial aperture ofprootic canal near subar- 15(0) - paroccipital process absent cuate fossa 17(0) - crista parotica/squamosal contact pres- 30(1)* - lateral flange ventrally directed [node ent 23(3) -cavum epiptericum with alisphenoid floor 9] 25(0) - lateral trough absent Node 8 30(2) - lateral flange absent 15 - paroccipital process with ventral projection 33 - inferior petrosal sulcus present 24 - facial ganglion floored by petrosal 38 - anterior lamina greatly reduced or absent 2(2)* - promontorium inflated without basi- 9(2)* - cochlear duct coiled [node 2] sphenoid wing [node 9] 34(2)* - cavum supracochleare enclosed [node 3] 11 * - semicircular canals enclosed by petrosal [node 9] Node 2 - Prototribosphenida 14* - stapedius fossa present [node 10] 8 - cochlear aqueduct present 10 - secondary spiral lamina present Node 9 12* - caudal tympanic process of petrosal pres- 5 - margin of fenestra vestibuli without ring ent [node 3] 20(1) - prootic canal present in petrosal 13* - post-promontorial sinus present [node 3] 23(1) - cavum epiptericum with partial petrosal 37* - pila antotica absent [node 3, 5] floor 25 - lateral trough present Node 3 - Theriimorpha 39(l) - anterior lamina expanded dorsal to man- 6 - fenestra vestibuli elongated dibular and maxillary nerve exits 28 - craniomandibularjoint anterior to fenestra 44 - posttemporal canal with squamosal border vestibuli Node 10 - Mammaliamorpha 19* - pterygoparoccipital foramen anterior to 31 - lateral flange vascular foramen present fenestra vestibuli [node 4] 40 - multiple foramina for mandibular and 42* - ventral ascending groove or canal hori- maxillary nerves zontal [node 4] 1* - prootic and opisthotic fused [node 11] Node 4 - Mammalia Node 1 1 23(2) - cavum epiptericum with petrosal floor 7 - distinct jugular and perilymphatic foramina 29 - quadrate ramus of alisphenoid absent 35 - cochlear and vestibular nerves in separate 31(0) - lateral flange vascular foramen absent foramina 41(1) - ventral ascending groove or canal intra- 43 - dorsal ascending groove or canal intramural mural in petrosal or endocranial Recent issues of the Novitates may be purchased from the Museum. Lists of back issues of the Novitates, Bulletin, and Anthropological Papers published during the last five years are available free of charge. Address orders to: American Museum of Natural History Library, Department D, Central Park West at 79th St., New York, N.Y. 10024. TEL: (212) 769-5545. FAX: (212) 769- 5009. E-MAIL: [email protected]

This paper meets the requirements of ANSI/NISO Z39.48-1992 (Permanence of Paper).