Norntates PUBLISHED by the AMERICAN MUSEUM of NATURAL HISTORY CENTRAL PARK WEST at 79TH STREET, NEW YORK, NY 10024 Number 3235, 13 Pp., 3 Figures August 27, 1998

Home , Didelphodon, Herpetotherium, Mayulestes, Peradectes

AMERICANt MUSEUM Norntates PUBLISHED BY THE AMERICAN MUSEUM OF NATURAL HISTORY CENTRAL PARK WEST AT 79TH STREET, NEW YORK, NY 10024 Number 3235, 13 pp., 3 figures August 27, 1998

Basicranial Anatomy of Herpetotherium (Marsupialia: Didelphimorphia) from the Eocene of Wyoming

SHERRI L. GABBERT'

ABSTRACT Cranial anatomy of early Tertiary Herpetother- panic floor, and it is not known how much was ium is described, with an emphasis on the audi- contributed by the ectotympanic. The exit for tory region. The preserved anatomy presents a mandibular nerve (V3) through a rostral incisure suite of characters indicating a conservative mor- of the piriform fenestra is most similar to the con- phological trend in the ear region. The tympanic dition observed in Marmosa and Monodelphis. In cavity floor has a large alisphenoid contribution general, the anatomy of the ear region observed that resembles the condition seen in Caluromys. in Herpetotherium resembles that observed in ex- No other bony structures from the petrosal or sur- tant didelphids; however, changes in spatial mor- rounding bones contribute to closure of the tym- phology appear to distinguish taxonomic groups.

INTRODUCTION The cranial morphology of three very from an ash layer in a section 2.5 mi south- well-preserved, nearly complete marsupial east of the Jim Christian Hills (Flat Top skulls, referred to the genus Herpetotherium Mountain) in Niobrara County (AMNHP (Fox, 1983; Korth, 1994), are described here- 22304). The latter is roughly assigned to in. All three fossils were collected by Frick what is now considered the late Eocene and expeditions in Wyoming; two come from the the White River Formation is Chadronian, White River Formation in Converse County which is also late Eocene (Prothero and (AMNHP 127684, 127685) and one comes Swisher, 1992). Because most of the North

' Research Fellow, Department of Vertebrate Paleontology, American Museum of Natural History; author mailing address: 4505 E. Harvard Ave., Denver, CO 80222-6527.

American marsupial fossils from the Creta- [1996] for an alternative interpretation). ceous and early Tertiary consist of isolated Thus, taxa from the order Didelphimorphia teeth and maxillary, dentary, and rostral frag- sensu Kirsch et al. (1997) that were available ments, these fossils are especially significant. to the author were used for the comparisons, Their completeness adds significantly to our regardless of the variable subordinal relation- knowledge, not only of the genus Herpeto- ships. Two taxa from two different orders of therium, but also of the evolution of the au- marsupials were arbitrarily selected to permit ditory region in marsupials. The goals of this study of character variation: Caenolestes (or- study are (1) to describe the anatomy of the der Paucituberculata) and Dasyurus (order basicranium of Herpetotherium, (2) to com- Dasyuromorphia). The following specimens pare that anatomy with extant members of of extant marsupial genera were used to an- the Didelphidae as defined by Marshall et al. alyze the comparative anatomy: Caenolestes, (1990) and Kirsch et al. (1997), and (3) to AMNHM 62914, 62902, 64381, 64414, discuss the implications for character evolu- 64389; Caluromys, AMNHM 133200, tion of the alisphenoid tympanic process and 78101, 96641, 230001, 139783; Caluromy- other features of the auditory region with ref- siops, AMNHM 208101; Dasyurus, erence to extinct and extant marsupials. AMNHM 154518, 196845; Didelphis, AMNHM 128992, 28961, 176700, 172168; METHODS AND MATERIALS Marmosa, AMNHM 96698, 184846, 206765, 254508; Metachirus, AMNHM Descriptions of three skulls of Herpeto- 130597, 133110, 267362; Micoureus, therium sp. (AMNHP 127684, represented AMNHM 210397, 267818; Monodelphis, by a nearly complete cranium and two den- AMNHM 261243, 130516, 16125, 130565, taries; AMNHP 22304, a complete cranium; 77291; and Philander, AMNHM 260033, and AMNHP 127685 represented by a nearly 96593, 96736, 133064, 67295. Abbrevia- complete cranium and a right dentary) em- tions: AMNHP American Museum of Natu- ploy the terminology of MacPhee (1981), ral History, Department of Vertebrate Pale- Novacek (1986), and Wible (1990). All three ontology; AMNHM American Museum of skulls were utilized to generate the most Natural History, Department of Manmmalogy. complete and accurate descriptions. The cra- nia were allocated to the genus Herpetother- DESCRIPTION ium on the basis of their dental morphology, which did not present any new features to MID-CRANIAL REGION: Among all of the report. Traits present in the skulls described Herpetotherium skulls described here, crush- here that are diagnostic of Herpetotherium ing and fragmentation in the region of the Cope, 1873 (Korth, 1994) include the follow- mid-cranium and cranial vault has resulted in ing: dominant stylar cusp D on M1-M2, the loss of much relational information. In dominant stylar cusp C on M3-M4, and en- spite of this, nasal and frontal bones appear larged and procumbent 11 and 12. Herpe- to have a positional relationship to each other totherium was established as a North Amer- that is found in most marsupials: i.e., with ican genus distinct from Peratherium by the nasals expanded posteriorly along their Korth (1994). contact with the frontals. The frontals extend The marsupial phylogeny of Marshall et posteriorly to a position dorsal to the most al. (1990) was employed intially to identify posterior limit of the orbitotemporal fossa. appropriate comparative taxa. Marshall et al. Parietals continue from the coronal suture (1990) designated caluromyines a subfamily posteriorly until they contact the supraoccip- of didelphids, contra Kirsch (1977) who rec- ital, and form transversely oriented, slightly ognized the family Caluromyidae as distinct curved crests. These crests are fragmentary, from Didelphidae. More recent work by but it appears they would have eventually Kirsch and Palma (1995) and Kirsch et al. met along the sagittal suture. (1997) upheld the assignment of the subfam- The jugal articulates anteriorly with the ily Caluromyinae, along with Glironiinae, to maxilla and lacrimal, curving outward later- the family Caluromyidae (see Patton et al. ally and horizontally from the rostrum. A 1998 GABBERT: BASICRANIAL ANATOMY OF HERPETOTHERIUM 3

PGF \ // BG_a :M ER I PSCC

HF Fig. 1. Herpetotherium sp., AMNH 22304. Right auditory region, ventral view. Bulla is not preserved. Abbreviations: AC aqueductus cochleae; ACF anterior carotid foramen; ATP alisphenoid tympanic process has been broken along its rostromedial-caudolateral axis and folded up into the tympanic roof; BCG basicapsular groove; ER epitympanic recess; FC fenestra cochleae; FO foramen ovale; HF hypoglossal foramina; IPSC canal for the inferior petrosal sinus; P promontorium; PGF postglenoid foramen; PLF posterior lacerate foramen. small postorbital process can be seen on the crushing in this region. It conveys the spheno- dorsal aspect of the jugal. The distal portion palatine artery, vein, and accompanying nerve of the jugal curves gently downward and be- (Archer, 1976). Posterior to the sphenopala- neath the zygomatic process of the squamo- tine foramen, there is a vertically oriented sal, extending posteriorly to form the anterior sliver of bone with a well-defined border. This lip of the glenoid fossa. The glenoid fossa of is probably the posterior margin of the sphen- the squamosal is shallow and slightly con- orbital fissure. It is not possible to determine vex. It is bounded posteriorly by the post- which bones border the fissure; however, in glenoid process and posterodorsally by a extant didelphids, this fissure is at the junction large postglenoid foramen, which conveys of the alisphenoid and orbitosphenoid (Maier, the postglenoid artery and the sphenoparietal 1987). The sphenorbital fissure transmits the emissary vein in extant marsupials (Wible, ophthalmic branch of cranial nerve V and the 1990: fig. 1). supraorbital ramus of the ophthalmic branch In the region of the orbits, there are frag- of the internal carotid artery, in addition to ments of the lacrimal, orbitosphenoid, pala- cranial nerves H through IV and VI in mar- tine, and alisphenoid bones. However, their supials (Archer, 1976). The foramen rotun- topological relationships cannot be deter- dum, which transmits the maxillary branch of mined with any certainty. Within the orbito- cranial nerve V, is located in the alisphenoid temporal region, there are several well-pre- lateral and slightly posterior to the sphenor- served features: from anterior to posterior, bital fissure. the sphenopalatine foramen, the posterior BASICRANIUM: From the basal aspect, the edge of the sphenorbital fissure, and the fo- most conspicuous parts of the alisphenoid are ramen rotundum. (The optic canal is conflu- the pterygoid processes (not figured) and the ent with the sphenorbital fissure in marsupi- tympanic processes (figs. 1, 2, 3). The pter- als [Maier, 1987].) The sphenopalatine fora- ygoid processes are crushed against the basi- men is in the anteroventral corner of the or- sphenoid, but the tympanic processes are, for bit, presumably in the palatine bone; the most part, intact. Sutures separating the however, its position is equivocal because of basisphenoid and alisphenoid have fused 4 AMERICAN MUSEUM NOVITATES NO. 3235

ATP TCF

Fig. 2. Herpetotherium sp., AMNH 127685. Right auditory region, ventral posterolateral view. Ab- breviations: ACF anterior carotid foramen; ATP alisphenoid tympanic process; BF basal fissure containing openings for posterior lacerate foramen and inferior petrosal sinus canal; CTP caudal tympanic process (broken edge); HF hypoglossal foramen; M mastoid region; PFF primary facial foramen; RTP rostral tympanic process of petrosal; TCF transverse canal foramen; TH tympanohyal (broken edge). forming a "sphenoid complex" that is tainty. The basioccipital contribution to the bounded posteriorly by the auditory bulla occpital complex is subrectangular and un- and the basioccipital contribution to the oc- remarkable. The exoccipital portion of the cipital. The anterior carotid foramen is lo- occipital complex is pierced by two small hy- cated in the posterolateral edge of the basi- poglossal foramina that are situated antero- sphenoid component of the sphenoid com- ventral to each occipital condyle (figs. 1, 2). plex, immediately rostral to the sphenoid-ba- Condylar foramina, one on each side, are sioccipital suture. Based on this position, it dorsolateral to the condyles. There is no jug- is inferred that Herpetotherium had an inter- ular process (= "paroccipital process of ex- nal carotid artery that ran medial to the au- occipital" sensu Wible, 1990) associated ditory bulla, ventral to the basicapsular fis- with the exoccipital. sure and entered the braincase directly The auditory bulla consists of two ele- through the anterior carotid foramen. The ments: the tympanic process of the putative exit for the mandibular nerve (= foramen alisphenoid component of the sphenoid and ovale) lies just lateral to the anterior carotid the ectotympanic (figs. 2, 3). A small pos- foramen in a common depression anterome- terolateral part of the rostral tympanic pro- dial to the apex of the promontory (fig. 1). cess of the petrosal may have contacted the The foramen ovale as a distinct feature of the posterior crus of the ectotympanic, thus par- alisphenoid is not present; the exit is demar- ticipating in posterior closure of the tympan- cated in the anteromedial part of the piriform ic floor. The tympanic process of the alisphe- fenestra and more accurately should be noid is the most significant element, account- called an incisure. The foramen for the trans- ing for greater than 50% of the bullar floor. verse canal (fig. 2) is anterior to the depres- It is moderately inflated dorsally, ventrally, sion containing the foramen ovale and ante- and ventromedially. Crushing has displaced rior carotid foramen and it appears to be in the bullae medially in AMNH 127684 and the ventral part of the alisphenoid; however, 127685, so the medial extent and subsequent because the sutures have fused in this area, associations most likely do not represent life its position cannot be determined with cer- relationships. Medially, the alisphenoid tym- 1998 GABBERT: BASICRANIAL ANATOMY OF HERPETOTHERIUM 5

ATP GF

PGP E

CTP Fig. 3. Herpetotherium sp., AMNH 127684. Left auditory region, ventral view. Abbreviations: AC aqueductus cochleae; ATP alisphenoid tympanic process; CTP caudal tympanic process (broken); E ectotympanic (proximal end); ER epitympanic recess; GF glenoid fossa; IPSC inferior petrosal sinus canal; PGP postglenoid process. panic process approximates a groove that is mentioned, this contact may be the result of aligned rostrocaudally. This groove is asso- preservation conditions. ciated with closure of the basicapsular fis- Inside the tympanic cavity, the epitympan- sure, which provides a canal for the inferior ic recess and the tympanic surface of the pe- petrosal sinus that runs immediately dorsal to trosal (figs. 1, 3) are visible. The bone in the the groove (fig. 1) (Presley, 1979; Wible, area of the recess is fragmented, but the epi- 1983, 1986). The alisphenoid tympanic pro- tympanic recess appears to be situated in the cess contacts the medial petrosal border, al- extreme lateral edge of the petrosal; the fossa though postmortem displacement may be re- incudis is recognizable at the posterior bor- sponsible for this contact. The posterior crus der of the recess. Medial to the fossa incudis and part of the anterior crus of the ectotym- is the crista parotica, which is continuous panic have been preserved in AMNH 127684 with the caudal tympanic process posteriorly. and have been displaced up and under the The canal for the lateral head vein (= prootic posterior edge of the alisphenoid tympanic canal), the tympanic aperture of which is at process (fig. 3). It is probable that the ecto- the posterior margin of the epitympanic re- tympanic would have contributed some por- cess in marsupials, is not identifiable in any tion to the posterolateral wall of the bulla. of the specimens (Wible and Hopson, 1995). The anterior crus of the ectotympanic is nar- The promontorium is small relative to the row and columnar, and gradually expands bulla and fairly simple in structure; there are transversely toward the posterior end. It ap- two consecutive, anteriorly facing convex pears that a small sliver of the incudal artic- ridges on the promontorium (fig. 1). These ular surface of the malleus is preserved along represent the turns of the cochlear duct. In with the anterior crus of the ectotympanic. It AMNH 127684, the bone over this duct has cannot be determined whether the ectotym- not been preserved, so the first two turns of panic was attached to the alisphenoid tym- the fossilized duct are visible (fig. 3). There panic process or suspended as a "free ring" are no features of the promontorium that as in some Recent marsupials. The posterior could be interpreted as vascular grooves or tip of the ectotympanic is in contact with the canals; therefore, it is inferred that Herpe- posterolateral portion of the rostral tympanic totherium, like other extinct marsupials, process of the petrosal, though as previously lacks a stapedial artery as do extant marsu- 6 AMERICAN MUSEUM NOVITATES NO. 3235 pials (Wible, 1990; Meng and Fox, 1995a, phatic duct and its vein, also opens into this 1995b). A small bump on the posterior sec- fissure from the medial aspect of the petrosal. tion of the promontorium, ventral to the fenestra cochleae, represents the minimally de- COMPARATIVE MORPHOLOGY veloped rostral tympanic process of the petrosal. The stapedial ratio (Segall, 1970), as As discussed in the Introduction, most de- measured on AMNHP 22304, is 1.75. scriptions of marsupial fossils from North Though within the limits as defined by Segall America concern teeth-bearing fragments. (1970), there may have been some dorsoven- However, research on the cranial anatomy of tral distortion due to crushing, which affects extinct marsupials has been accumulating. this measurement. The fenestra vestibuli fac- Matthew (1916) assigned a fragment of tem- es laterally and the fenestra cochleae faces poral bone to the stagodontid genus Eodel- posterolaterally. The facial sulcus arcs phis from the Oldman Formation of Alberta. around the lateral aspect of the promonto- Gazin (1935) described a partial skeleton of rium of the petrosal, curving laterally and Herpetotherium that had been preserved on posteriorly along the roof of the middle ear. a slab of shale, mostly as impressions. The The primary facial foramen opens into the skull and right mandible were present, but sulcus anteromedial to the fenestra vestibuli crushed flat. Clemens (1966: figs. 55, 56, pp. (fig. 2). The fossa for the stapedius muscle 75 and 76) described a squamosal fragment is a shallow, but well-circumscribed depres- and a petrosal-squamosal fragment referred sion medial to the posterior continuation of to the genus Didelphodon, a stagodontid the crista parotica. The tympanohyal and marsupial from the Late Cretaceous. Archi- caudal tympanic process of the petrosal are bald (1979) identified a marsupial petrosal not well-preserved in any of the specimens from the Hell Creek Formation as possibly (fig. 2). Their anatomy cannot be described belonging to Pediomys or Alphadon. Wible with any certainty; however, based on (1990) presented a comprehensive study on AMNH 22304, which preserves the most of fossil marsupial ear anatomy, describing iso- both these features, it is possible that one or lated petrosals from the Late Cretaceous. both of them could have approached the pro- Meng and Fox (1995a, 1995b) also describe montorium at or near the rostral tympanic the anatomy of the Late Cretaceous marsu- process to form a ledge under the postero- pial petrosals from Montana and Alberta, re- lateral segment of the facial sulcus. A sty- spectively. Wible's (1990) descriptions serve lomastoid notch is not apparent in any of the as the anatomical atlas for this analysis of specimens, presumably because of breakage. Herpetotherium. An indistinct mastoid prominence can be ob- Herpetotherium was considered to be a served on the ventral aspect of the mastoid member of the subfamily Herpetotheriine by region. At the posteromedial apex of the au- Marshall et al. (1990). Kirsch et al. (1997) ditory region, at the junction of the basioc- elevated the herpetotheriines to family level cipital and petrosal, just opposite the hypo- within Didelphimorphia but designated them glossal foramina, is a large ventral fissure in incertae sedis. To make discussion of char- the shape of an hourglass (fig. 2). There are acter evolution of the ear region appropriate, two openings that converge at this fissure, comparative anatomical analysis focused on one anterior and the other posterior; they are didelphimorphs of the family Didelphidae positioned at right angles to each other. The and subfamily Caluromyiinae. Herpetother- anterior opening is the canal for the inferior ium specimens AMNH 127684, 127685, and petrosal sinus, directed anteroposteriorly. 22304 were compared to the living genera The posterior opening is the posterior lacer- Didelphis, Metachirus, Monodelphis, Micou- ate foramen, which is directed dorsoventral- reus, Marmosa, Philander (subfamily Didel- ly. In living marsupials, the posterior lacerate phinae), and Caluromys and Caluromysiops foramen conducts cranial nerves IX, X, and (subfamily Caluromyinae). (Caenolestes, a XI, and, occasionally, a small vein from the caenolestid and Dasyurus, a dasyurid, were transverse sinus (Archer, 1976). The aque- also used for outgroup comparison.) ductus cochleae, which carries the perilym- The anatomy of the bullar floor of the mar- 1998 GABBERT: BASICRANIAL ANATOMY OF HERPETOTHERIUM 7 supials studied herein is associated with sev- variation among taxa. One of the features eral variables: number of contributory ele- that displays observable variation is the ros- ments, inflation of contributory elements, tral tympanic process of the petrosal. Unlike and relative closure of the bullar floor, all of any of the didelphids in this study, Herpe- which have been discussed by other authors totherium possesses a rudimentary rostral (e.g., Archer, 1976; Reig et al., 1987; Maier, tympanic process that does not appear to play 1989). The minimal condition of two bullar a role in forming the bullar floor. In didel- floor elements, uninflated or expanded and phids, the rostral tympanic process attains its ventrally open, can be observed in Philander greatest development at its most posterolat- and Metachirus. In these taxa, the alisphen- eral extent. There, it is tabular and projects oid tympanic process and the rostral tympan- laterally. Metachirus and Didelphis are the ic process of the petrosal define the anterior least developed of the didelphids with re- and posterior limits of the bullar floor, respect to this trait, but both attain a condition spectively. The ectotympanic in both taxa is of greater development than does Herpeto- narrow, contacts the rostral tympanic pro- therium. The rostral tympanic process of the cess, and contributes little to the lateral wall petrosal in caluromyines is greatly expanded of the bullar floor. Didelphis differs only in rostrally and rostrolaterally to form a kind of having a relatively more inflated alisphenoid petrosal plate. This condition was also ob- tympanic process and an ectotympanic that served in Dasyurus. Caenolestes demon- does not contact the rostral tympanic process strates a condition similar to that obtained in of the petrosal. Monodelphis, Micoureus, and Herpetotherium, but with slightly greater lat- Marmosa all expand the posterior crus of the eral development. ectotympanic toward the midline of the bul- The tympanohyal and the caudal tympanic lar floor and thereby effectively raise the process of the petrosal also showed variable number of bullar floor elements to three. The relationship to each other and to the stylo- bullar floor is open ventromedially, as is true mastoid notch/foramen. In Didelphis, Mar- to some degree in all these taxa. Of these mosa, and Philander, the tympanohyal and three, only Marmosa shows any significant the caudal tympanic process of the petrosal inflation of the alisphenoid tympanic process. are both small. They are situated lateral and Among the didelphid taxa included in this medial, respectively, to the stylomastoid study, Caluromys and Caluromysiops exhibit notch, which gives egress for the facial nerve the greatest extent of bullar floor closure; in (Wible, 1990). The morphology is similar in both, the inflated alisphenoid tympanic pro- Micoureus, where the stylomastoid notch is cess contributes 50% or greater to the bullar bordered on each side by the caudal tympan- floor. The alisphenoid tympanic process and ic process and tympanohyal. In some taxa, the rostral tympanic process meet (or nearly such as Metachirus, Dasyurus, and the cal- meet) over the promontorium, leaving only a uromyines, the squamosal has a platelike small ventromedial slit in the bullar floor. process that overlays the lateral wall of the Unlike Marmosa, Monodelphis, and Micou- facial canal near the tympanohyal. The most reus, Caluromys and Caluromysiops have extreme condition was observed in the cal- only two bullar floor elements, the ectotym- uromyines, where the caudal tympanic pro- panic being aphaneric. In addition, the latter cess of the petrosal and the posttympanic taxa are easily distinguished from the other process of the squamosal converge to close taxa bearing two floor elements (Philander, the posterior space near the fenestra cochle- Metachirus, and Didelphis), which do not ae. This has the effect of creating a stylo- elaborate either of the contributory elements. mastoid "foramen" in Caluromys and Cal- The alisphenoid tympanic process in Her- uromysiops. It is of interest to note the con- petotherium resembles that of the caluromy- ditions in the outgroup taxa: in Caenolestes, ine oppossums: it is greatly inflated and the tympanohyal and caudal tympanic pro- makes up approximately 50% of the bullar cesses are strongly developed ventrally, cre- floor. ating a deep stylomastoid notch. The most Most of the anatomical characters from the elaborate condition was observed in Dasyu- tympanic aspect of the petrosal showed little rus: the tympanohyal and caudal tympanic 8 AMERICAN MUSEUM NOVITATES NO. 3235 process of the petrosal meet laterally, poste- observed to be variable. In all taxa, the an- rolaterally, and ventrally to form a stylomas- terior carotid foramen was situated in the ros- toid canal that opens in the usual posterolat- tral end of a groove in the posterolateral cor- eral corner of the auditory region. The extent ner of the basisphenoid, but the groove to which the tympanohyal and caudal tym- length varied. In Didelphis, Philander, and panic process are developed in Herpetother- Herpetotherium, for example, the anterior ca- ium is not known, but judging from the anat- rotid foramen was closer to the basisphenoid- omy of the broken surfaces, it is not likely basioccipital juncture than in Caluromysiops. that these features were extensive. Nevertheless, the anterior carotid foramen With the exception of Philander, the jug- was constrained anatomically as a feature of ular process contributes a rostral process that the putative basisphenoid portion of the varies in degree of participation of posterior sphenoid complex. The positional anatomy wall closure in extant didelphids (Note: of the foramen ovale, on the other hand, was Archer [1976] discussed a feature called the more variable. A rostral incisure in the piri- tympanic wing of the paroccipital process, form fenestra serves as the exit for the man- which is undoubtedly the same as the rostral dibular branch of the trigeminal nerve (V3) process of the jugular process. Archer's ter- in Marmosa, Monodelphis, and Micoureus. minology is not used here for two reasons. An intermediate condition is observed in First, the feature in question on the jugular some specimens of Didelphis, most speci- process does not always contribute to tym- mens of caluromyines, and Metachirus. In panic anatomy. Second, the so-called paroc- these taxa, the foramen ovale is mostly con- cipital process of the petrosal in cynodonts tained in the alisphenoid; however, the fora- and early mammals on the ventral surface of men is partitioned posteriorly from the piri- the petrosal [Rougier et al., 1992] is nonhom- form fenestra by small struts from the ali- ologous with the "paroccipital process of ex- sphenoid that are incomplete, thus leaving occipital." Therefore, the term jugular pro- the foramen largely continuous with the fe- cess [World Assoc. of Vet. Anat., N.A.V., nestra. Finally, there is the condition ob- 1994] is preferred to avoid confusion.) The served in Philander, Caluromysiops, and jugular process in Didelphis sends a rostral most specimens of Didelphis where the fo- process toward the petrosal without meeting ramen ovale is definitively in the alisphenoid, it, constraining the space ventral to the fe- separated from the pirifirm fenestra, and dis- nestra cochleae. In Monodelphis, the rostral plays a complex anatomy of primary and process is incipient, but noticeably present secondary foramina (for a detailed analysis and similar in morphology to that of Didel- of this anatomy, see Gaudin et al., 1996). The phis. The rostral process of the jugular pro- condition in Herpetotherium is the same as cess is developed ventrally, medially, and lat- observed in Marmosa, Monodelphis, and Mi- erally to contact the caudal tympanic process coureus, where an incisure in the piriform of the petrosal and the processus recessus in fenestra is the marker for the exit of V3. The Micoureus and Metachirus, thereby limiting outgroups proved to have variable anatomy the posterior extent of the tympanic cavity. of the foramen ovale-with Dasyurus ap- In Caenolestes, the jugular process curves proximating the condition obtained in Her- anteroventrally to meet the caudal tympanic petotherium, Marmosa, Monodelphis, and process of the petrosal and close the postero- Caluromys; and Caenolestes showing com- medial aspect of the tympanic cavity; thus, plete bony separation of the foramen ovale there is no obvious rostral process. Dasyurus from the piriform fenestra. presents extensive development of the jugu- As a final note, it is of interest to mention lar process, which sends out a massive, fan- that among the didelphimorphs included in shaped wing covering the caudal tympanic this study, Caluromys does not have a trans- process and contacting the rostral tympanic verse canal in the basisphenoid in the usual process. As stated above, Herpetotherium sense of that term (Archer, 1976). The ab- has no jugular process. sence of a transverse canal has been reported The relative positions of the anterior ca- for Pucadelphys andinus as well (Marshall rotid foramen and the foramen ovale were and Muizon, 1995). A structure present in 1998 GABBERT: BASICRANIAL ANATOMY OF HERPETOTHERIUM 9

Caenolestes has been called by the same resolution of this requires a phylogenetic name, but there is now some doubt as to analysis, but it is conceivable that bullar evo- whether it is, in fact, homologous across lution in North American didelphids pro- marsupial taxa (Sanchez-Villagra, in prep.). ceeded by the successive acquisition or elab- oration of bony elements in the auditory re- DISCUSSION AND CONCLUSIONS gion beginning with the alisphenoid tympanic process. The recovery of new structural data con- Recent papers by Muizon (1991, 1994), cerning the basicranial anatomy of Herpe- Muizon et al. (1997), Marshall and Muizon totherium brings up several questions con- (1995), Trofimov and Szalay (1994), and cerning character evolution among late Cre- Szalay and Trofimov (1996) presented im- taceous/early Tertiary marsupials. The most portant information about bullar composition controversial of these is the anatomy of the in Paleocene and Cretaceous marsupials that bulla. The morphology of the piriform fe- is pertinent to future phylogenetic studies. nestra/foramen ovale and the observable The two Paleocene marsupials described by variation of the caudal tympanic process, Muizon are from South America. One is a tympanohyal, and rostral process of the jug- didelphid, Pucadelphys (Muizon, 1991; Mar- ular process also merit discussion. shall and Muizon, 1995) and the other a bor- Bullar floor composition in Herpetother- hyaenoid, Mayulestes (Muizon, 1994), both ium is most like the condition obtained in the from the early Paleocene of Bolivia. Muizon caluromyine didelphids, Caluromysiops, and (1991: 589-590 and fig. 6, p. 589) reported Caluromys. The alisphenoid complement of the absence of an alisphenoid tympanic pro- the bulla in all three taxa is quite inflated and cess in Pucadelphys; in fact, no evidence of makes up at least 50% of the bullar floor. a bulla of any kind; the same was true for Unlike Herpetotherium, the caluromyines Mayulestes (Muizon, 1994). Muizon (1994) elaborate the rostral tympanic process of the argued that the primitive condition of the petrosal to wall off the posterior aspect of the tympanic cavity floor for marsupials is the tympanic cavity. However, it is certain that lack of any contribution by bones surround- the rostral tympanic process of the petrosal ing the auditory region. This contrasts with did not contribute to the closure of the pos- the consensus opinion (e.g., Marshall, 1979; terior part of the bulla in Herpetotherium be- Clemens, 1979; Reig et al., 1987; Marshall cause of its small size. The contribution of et al., 1990) that the presence of an alisphen- the ectotympanic to the bullar floor in Her- oid tympanic process is diagnostic for Mar- petotherium cannot be assessed; therefore, it supialia (this process is present in all living is not possible to designate the bullar floor marsupials except vombatids [Springer and as "bipartite" or "tripartite" as discussed by Woodburne, 1989] and is absent in Creta- Reig et al. (1987). Given their definition of ceous eutherians and the prototribosphenidan "bipartite" bullae as consisting of alisphen- Vincelestes [Kielan-Jaworowska, 1981 and oid anteriorly, rostral tympanic process pos- Rougier et al., 1992, respectively]). teriorly, and a thin, noncontributory ectotym- Based on the cladogram included in Mui- panic, the morphology observed in Herpe- zon (1994), derived borhyaenoid marsupials totherium, with its very small rostral tym- acquired tympanic processes from adjacent panic process, would represent a departure bones, but not the same as those in didel- from this arrangement. However, if the ec- phids; thus, tympanic floor contributions totympanic did contribute to the lateral as- arise at least twice in marsupials from com- pect of the bullar floor, then the bulla in Her- pletely different precursor bones. Muizon's petotherium would be "bipartite," but from (1994) interpretation is based on a second- different constituents. A minimal reconstruc- order inference concerning the putative re- tion hypothesis would leave the entire pos- lationship between borhyaenids and deltath- terior part of the bullar floor in Herpetother- eroidans suggested by Marshall and Kielan- ium open, with little or no contribution from Jaworowska (1992). In the latter paper, del- either the ectotympanic or the rostral tym- tatheroidans were determined to be the sister panic process of the petrosal. Obviously, the taxa to the borhyaenoids + didelphoids 10 AMERICAN MUSEUM NOVITATES NO. 3235 based on the presence of an alisphenoid tym- that Pucadelphys might be a very primitive panic process (Kielan-Jaworowska and Nes- borhyaenoid based on the secondary loss of sov, 1990) and other dental and skeletal char- the alisphenoid contribution to the bulla. Un- acters (some of which are of uncertain po- til a phylogenetic analysis of all relevant taxa larity, e.g., jugal participation in the glenoid is conducted, this issue will remain unresol- fossa), making the alisphenoid complement ved. of the bulla a primitive trait given the Cre- The relationship between the piriform fe- taceous age of deltatheriodans. Muizon nestra and the foramen ovale is variable (1994), however, argued, on the basis of a across Mammalia and particularly complex cladogram that included only borhyaenids within Marsupialia (Gaudin et al., 1996). and Pucadelphys, that the alisphenoid tym- Nevertheless, Gaudin et al. (1996) were able panic process is not primitive for marsupials. to partition the observable variation of the More recently, Muizon et al. (1997) have fur- true foramen ovale among extinct and extant thered this hypothesis, arguing that their in- marsupials into three character states: (1) terpretation is supported by the condition in wholly within the alisphenoid, (2) between the stagodontids Eodelphis and Didelphodon the alisphenoid and petrosal, and (3) between (Matthew, 1916; Clemens, 1966, respective- the alisphenoid and the squamosal. The third ly). Neither of these taxa are represented by condition is the least common taxonomically. material from the pertinent region of the al- Gaudin et al. (1996) were unable to deter- isphenoid. There is a very slim fragment of mine the primitive state for the foramen the alisphenoid preserved at its sutural ovale with their data set and accordingly boundary with the squamosal in Didelpho- scored Marsupialia as a polymorphic taxon. don, and Clemens (1966: 80) remarked that The morphology of the piriform fenestra/ it was possible that the alisphenoid contrib- foramen ovale obtained for Herpetotherium, uted a bullar process. and also observed for Marmosa, Monodel- Trofimov and Szalay (1994) and Szalay phis, and Caluromys, is described by char- and Trofimov (1996) described and named acter state 2 above. The condition of the fo- Asiatherium, from the Late Cretaceous of ramen ovale in the early Paleocene didel- Mongolia and placed it in a new order, the phoid Pucadelphys was reported by Muizon Asiadelphia, which they speculated was a (1991) to be between the alisphenoid and lineage of marsupials distinct from delta- squamosal, but this interpretation was theroidans and North American marsupials. changed in Marshall and Muizon (1995) to They reported that Asiatherium possesses an be between the alisphenoid and petrosal. The alisphenoid tympanic process, in addition to condition in the early Paleocene borhyaenoid contributions by the rostral and caudal tym- Mayulestes is the same as that of Herpeto- panic processes of the petrosal. This would therium. This would imply that the primitive ostensibly support the idea that an alisphen- condition for marsupials is between the al- oid tympanic process is primitive for Mar- isphenoid and petrosal, making the condition supialia, given the purported basal position in Herpetotherium a retention. Wroe (1997) of asiadelphians and deltatheroidans. With arrived at a similar conclusion for the prim- this in mind, it would be more parsimonious itive condition of the foramen ovale in mar- to assume the loss of the alisphenoid tym- supials in an analysis of dasyurid marsupial panic process in Pucadelphys and basal bor- synapomorphies. hyaenids, than it would be to infer that it was Finally, the morphological variations ob- independently acquired in asiadelphians, del- served in the rostral process of the jugular tatheroidans, and living marsupials. More- process, as well as those of the tympanohyal over, the condition of the alisphenoid is not and caudal tympanic process of the petrosal, known in certain relevant extinct taxa, such suggest that there may be some systematic as stagodontids (the putative sister group to significance to be gleaned with respect to sty- didelphids according to Marshall and Kielan- lomastoid foramen morphology and closure Jaworowska [1992]) and the Late Creta- of the posterior bullar floor. For example, the ceous/Early Paleocene didelphids Peradectes outline of a morphocline for the rostral pro- and Albertotherium. Finally, it is conceivable cess of the jugular process can be seen, rang- 1998 GABBERT: BASICRANIAL ANATOMY OF HERPETOTHERIUM I1I ing from Herpetotherium, where there is no ic boundaries. This would be pertinent to jugular process, to Philander, where there is work being done by many researchers (e.g., a jugular process but no rostral process, to Marshall et al., 1990; Muizon, 1991, 1994; Caluromys, where the rostral process of the Szalay, 1994; Trofimov and Szalay, 1994; jugular process contributes to the closure of Luckett, 1994; Gaudin et al., 1996; Wroe, the posterior tympanic cavity. The stylomas- 1997; Kirsch et al., 1997, Springer et al., toid notch/foramen presents similar transfor- 1997) that has been directed toward better mations related to the variable development understanding of the higher-level relation- and relationship of the tympanohyal and the ships among the Marsupialia. caudal tympanic process of the petrosal. The egress for the facial nerve varies in mor- ACKNOWLEDGMENTS phology from a simple, relatively shallow notch, as observed in Didelphis, to a circum- My thanks to Ross MacPhee, John Wible, scribed foramen in caluromyines. Marcelo Sanchez-Villagra, Judd Case, Pat- It is obvious that the recovery of anatom- rick Luckett, and Jennifer Rodman Snyder ical information about Herpetotherium and for their constructive criticisms and impor- the subsequent comparative survey has en- tant perspectives. Malcolm McKenna and gendered the need for a reevaluation of ba- Michael Novacek provided valuable com- sicranial characters among didelphids. The ments on earlier drafts of this manuscript. Ed documented variation in many important an- Hecht skillfully produced the drawings and atomical traits makes simple evolutionary photographs and patiently endured many re- transformation assessments impossible. The quested modifications. This work was sup- reassessment of basicranial anatomy would ported by a Frick Predoctoral Fellowship contribute to a more robust hypothesis of from the American Museum of Natural His- marsupial evolution across several taxonom- tory.

REFERENCES Archer, M. Gaudin, T. J., J. R. Wible, J. A. Hopson, and W. 1976. The basicranial region of marsupicar- D. Turnbull nivores (Marsupialia), interrelation- 1996. Cohort Epitheria (Mammalia, Euther- ships of carnivorous marsupials, and ia): the morphological evidence reex- affinities of the insectivorous marsupial amined. J. Mammal. Evol. 3: 31-79. peramelids. Zoo. J. Linn. Soc. 59: 217- Gazin, C. L. 322. 1935. A marsupial from the Florissant Beds Archibald, J. D. (Tertiary) of Colorado. J. Paleontol. 9: 1979. Oldest known eutherian stapes and a 57-62. marsupial petrosal bone from the Late Kielan-Jaworowska, Z. Cretaceous of North America. Nature 1981. Evolution of the therian mammals in 281: 669-670. the Late Cretaceous of Asia. Part IV. Clemens, W. A., Jr. Skull structure in Kennalestes and 1966. Fossil mammals of the type Lance For- Asioryctes. Paleontologia Polonica 42: mation Wyoming. Part II. Marsupialia. 25-78. Univ. Calif. Publ. Geol. Sci. 62: 122 Kielan-Jaworowska, Z., and L. A. Nessov PP. 1990. On the metatherian nature of the Del- 1979. Marsupialia. In J. A. Lillegraven, Z. tatheroida, a sister group of Marsupi- Kielan-Jaworowska, and W. A. Clem- alia. Lethaia 23: 1-10. ens (eds.), Mesozoic mammals, the first Kirsch, J. A. W. two-thirds of mammalian history: 192- 1977. The comparative serology of Marsupi- 220. Berkeley: Univ. California Press. alia, and a classification of marsupials. Fox, R. C. Australian J. Zool. (supp. ser.) 52: 1- 1983. Notes on the North American marsu- 152. pials Herpetotherium and Peradectes. Kirsch, J. A. W., and R. E. Palma Can. J. Sci. 20: 1565-1578. 1995. DNA/DNA hybridization studies of 12 AMERICAN MUSEUM NOVITATES NO. 3235

carnivorous marsupials. V. A further upon the affinities of the Cretaceous estimate of relationships among op Mammalia. Bull. Am. Mus. Nat. Hist. possums (Marsupialia: Didelphidae). 35: 477-500. Mammalia 59: 403-425. Meng, J., and R. C. Fox Kirsch, J. A. W., F-J. Lapointe, and M. S. Springer 1995a. Osseous inner ear structures and hear- 1997. DNA-hybridization studies of marsu- ing in early marsupials and placentals. pials and their implications for meta- Zool. J. Linn. Soc. 115: 47-71. therian classification. Australian J. 1995b. Therian petrosals from the Oldman and Zool. 45: 211-280. Milk River Formations (Late Creta- Korth, W. W. ceous), Alberta, Canada. J. Vertebr. Pa- 1994. Later Tertiary marsupials (Mammalia: leontol. 15: 122-130. Marsupialia) from North America. J. Muizon, C. de Paleontol. 68: 376-396. 1991. La Fauna de mamfferos de Tiupampa Luckett, W. P (Paleoceno inferior, Formacion Santa 1994. Suprafamilial relationships within Mar- Lucia), Bolivia. In R. Suarez-Sorucco supialia: resolution and discordance (ed.), Fosiles y Facies de Bolivia, Vol. from multidisciplinary data. J. Mam- I. Vertebrados: 575-624, Revista Tec- mal. Evol. 2: 255-283. nica de YPFB, Santa Cruz, Bolivia. MacPhee, R. D. E. 1994. A new carnivorous marsupial from the 1981. Auditory regions of primates and eu- Palaeocene of Bolivia and the problem therian insectivores. Morphology, on- of marsupial monophyly. Nature 370: togeny and character analysis. Contrib. 208-211. Primatol. 18: 1-282. Muizon, C. de, R. L. Cifelli, and R. Cespedes Paz Maier, W. 1997. The origin of the dog-like borhyaenoid 1987. The ontogenetic development of the or- marsupials of South America. Nature bitotemporal region in the skull of 389: 486-489. Monodelphis domestica (Didelphidae, Novacek, M. J. Marsupialia) and the problem of the 1986. The skull of leptictid insectivorans and mammalian alisphenoid. In H.-J. Kuhn the higher-level classification of euther- and U. Zeller (eds.), Morphogenesis of ian mammals. Bull. Am. Mus. Nat. the mammalian skull: 71-90. Hamburg: Hist. 183: 1-112. Verlag Paul Parey. Patton, J. L., S. F dos Reis, and M. N. F da Silva 1989. Morphologische Untersuchungen am 1996. Relationships among didelphid marsu- Mittelohr der Marsupialia. Z. zool. pials based on sequence variation in the Syst. Evolut.-forsch. 27: 149-168. mitochondrial cytochrome B gene. J. Marshall, L. G. Mammal. Evol. 3: 3-29. 1979. Evolution of metatherian and eutherian Presley, R. (mammalian) characters: a review 1979. The primitive course of the internal ca- based on cladistic methodology. Zool. rotid artery in mammals. Acta Anat. J. Linn. Soc. 66: 369-410. 103: 238-244. Marshall, L. G., and Z. Kielan-Jaworowska Prothero, D. R., and C. C. Swisher 1992. Relationships of the dog-like marsupi- 1992. Magnetostratigraphy and geochronolo- als, deltatheroidans and early tribo- gy of the terrestrial Eocene-Oligocene sphenic mammals. Lethaia 25: 361- transition in North America. In D. R. 374. Prothero and W. A. Berggren (eds.), Marshall, L. G., and C. de Muizon Eocene-Oligocene climatic and biotic 1995. Pucadelphys andinus (Marsupialia, evolution: 46-73. Princeton: Princeton Mammalia) from the early Paleocene of Univ. Press. Bolivia. Part II. The skull. Mem. Mus. Reig, 0. A., J. A. W. Kirsch, and L. G. Marshall Natl. Hist. Nat. 165: 21-90. 1987. Systematic relationships of the living Marshall, L. G., J. A. Case, and M. 0. Woodburne and Neocenozoic American "opossum- 1990. Phylogenetic relationships of the fami- like" marsupials (suborder: Didelphi- lies of marsupials. In Hugh H. Geno- morphia), with comments on the clas- ways (ed.), Current mammalogy 2: sification of these and of the Creta- 433-505. New York: Plenum Press. ceous and Paleogene New World and Matthew, W. D. European metatherians. In M. Archer 1916. A marsupial from the Belly River Cre- (ed.), Possums and opossums: studies taceous. With critical observations in evolution: 1-89. Sydney: Surrey 1998 GABBERT: BASICRANIAL ANATOMY OF HERPETOTHERIUM 13

Beatty & Sons and the Royal Zoologi- Trofimov, B. A., and F S. Szalay cal Society of New South Wales. 1994. New Cretaceous marsupial from Mon- Rougier, G. W., J. R. Wible, and J. A. Hopson golia and the early radiation of Meta- 1992. Reconstruction of the cranial vessels in theria. Proc. Natl. Acad. Sci. 91: the Early Cretaceous mammal Vince- 12569-12573. lestes neuquenianus: implications for Wible, J. R. the evolution of the mammalian cranial 1983. The internal carotid artery in early vascular system. J. Vertebr. Paleontol. eutherians. Acta Palaeontol. Polonica 12: 188-216. 28: 281-293. Segall, W. 1986. Transformation in the extracranial 1970. Morphological parallelism of the bulla course of the internal carotid artery in and auditory ossicles in some insecti- mammalian phylogeny. J. Vertebr. Pa- vores and marsupials. Fieldiana Zool. leontol. 6: 313-325. 51: 169-205. 1990. Petrosals of Late Cretaceous marsupials Springer, M. S., and M. 0. Woodburne from North America, and a cladistic 1989. The distribution of some basicranial analysis of the petrosal in therian mam- characters within the Marsupialia and a mals. Ibid. 10: 183-205. phylogeny of the Phalangeriformes. J. Wible, J. R., and J. A. Hopson Vertebr. Paleontol. 9: 210-221. 1995. Homologies of the Springer, M. S., Kirsch, J. A. W., and J. A. Case prootic canal in 1997. The chronicle of marsupial evolution. mammals and non-mammalian cyno- In T. Givnish and K. Sytsma (eds.), donts. J. Vertebr. Paleontol. 15: 331- Molecular evolution and adaptive ra- 336. diation: 129-161. New York: Cam- World Association of Veterinary Anatomists, In- bridge Univ. Press. ternational Committee on Veterinary Szalay, F S. Gross Anatomical Nomenclature 1994. Evolutionary history of the marsupials 1994. Nomina Anatomica Veterinaria, 4th and an analysis of osteological charac- ed., distrib. by Cornell University, Ith- ters. New York: Cambridge Univ. aca, New York, 198 pp. Press, 481 pp. Wroe, S. Szalay, F S., and B. A. Trofimov 1997. A reexamination of proposed morphol- 1996. The mongolian Late Cretaceous Asia- ogy-based synapomorphies for the fam- therium, and the early phylogeny and ilies of Dasyuromorphia (Marsupialia). paleobiogeography of Metatheria. J. I. Dasyuridae. J. Mammal. Evol. 4: 19- Vertebr. Paleontol. 16: 474-509. 51.

Recent issues of the Novitates may be purchased from the Museum. Lists of back issues of the Novitates and Bulletin published during the last five years are available at World Wide Web site http://nimidi.amnh.org. Or address mail orders to: American Museum of Natural History Library, Central Park West at 79th St., New York, NY 10024. TEL: (212) 769-5545. FAX: (212) 769- 5009. E-MAIL: [email protected]

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