INTRODUCTION

The lesser Sulawesi cuscus Strigocuscus celebensis and the scaly-tailed possum Wyulda squamicaudata are small to medium-size members of the marsupial family Phalangeridae. Both species bear a superficial resemblance to members of the most diverse and abundant group of phalangerids, the true cuscuses of the genus Phalanger. In the case of S. celebensis this similarity led to inclusion of the species in Phalanger (e.g., Tate, 1945), a situation which prevailed as the accepted taxonomy until comparatively recently. By contrast, Wyulda has always been recognized as a distinct taxon, with craniodental characters that indicate a close affinity to the brushtailed possums of the genus Trichosurus.

The division of the Phalangeridae into two groups, the monogeneric cuscuses and the two genera of possums, remained unchallenged until the phylogenetic revision of Flannery et al. (1987), who proposed that cuscuses were paraphyletic with respect to the possums. The Phalangeridae were divided into two subfamilies, the Ailuropinae, containing a single species Ailurops ursinus, and the Phalangerinae, containing all the remaining species of phalangerid. The Phalangerinae were divided into a further two groups: the tribes Phalangerini and Trichosurini. Flannery et al. (1987) resurrected the generic name Strigocuscus from synonymy within Phalanger and placed the genus within the Trichosurini, along with Wyulda and Trichosurus. As originally defined by Flannery et al., Strigocuscus contained four species formerly included within Phalanger: S. celebensis, S. gymnotis, S. ornatus, and S. rothschildi. A fifth taxon, S. mimicus, was raised from subspecific status within Phalanger orientalis. Subsequent studies have removed all species but S. celebensis from within the group and added a new species, S. pelengensis (George, 1987; Springer et al., 1990; Norris, 1992; Flannery, 1994; Hamilton and Springer, 1999), previously a subspecies of S. celebensis (Tate, 1945). The generic placement of this taxon is still under debate.

A study by Norris (1994) described the periotic bones of various phalangerid species and categorized three periotic morphologies which more or less corresponded to the divisions originally proposed by Flannery et al. (1987), together with the narrower definition of Strigocuscus made by later authors. Norris only studied one trichosurin periotic in detail, that of Trichosurus vulpecula. No loose periotic of either S. celebensis or W. squamicaudata was available for examination.

The fact that periotic morphology in phalangerids corresponded so closely to the phylogenetic divisions created by analysis of a broad character set of morphological and molecular characters suggested that a study of this region in specimens of fossil phalangerids from Tertiary deposits at Riversleigh, Australia, might provide insights into the early evolution and diversification of the group. In superficial morphology, the Riversleigh species appear to be trichosurin. However, they also possess a number of cranial characters that are plesiomorphic for phalangerids as a whole. Flannery et al. (1987) hinted at this in their list of synapomorphies for the trichosurins, where they noted that the Riversleigh species Trichosurus dicksoni exhibited the plesiomorphic state for some characters. A comparison of the periotic of this species and of Strigocuscus reidi, also from Riversleigh, showed a very different morphology than that of T. vulpecula, as discussed by Norris (1994). In fact, examination of these, and other, extinct taxa from Riversleigh suggest that a much broader range of periotic morphologies may be present within the group than is apparent from its extant members. Since other craniodental characters of these fossil species suggest that the majority of them have trichosurin affinities (K.C., in prep.), a broader knowledge of the periotic anatomy of the living trichosurins is required. To this end, periotics of Strigocuscus celebensis and Wyulda squamicaudata were dissected from skulls housed in the Department of Mammalogy at the American Museum of Natural History (AMNH) and studied in detail.

MATERIALS AND METHODS

The right periotic was dissected from the skull of AMNH 196502, an adult female specimen of Strigocuscus celebensis. The skull of AMNH 196919, an adult female specimen of Wyulda squamicaudata, was prepared from the alcohol-preserved carcass of this animal by dermestid beetle action, and the left periotic was dissected. Detailed descriptions of the two bones were made, and the morphology was compared with specimens of a variety of extant and extinct phalangerid taxa (table 1).

For the most part, the terminology employed in this paper follows MacIntyre (1972), with modifications from Wible (1990) and Norris (1994). There are certain structures on the periotic, however, whose homology needed to be further evaluated before they could give solid phylogenetic information. These were the rostral tympanic process, the ectotympanic process, and the crista promontorii medioventralis. Problems with these structures include whether they are homologous with those seen in other groups, as well as their relative positions.

MacPhee (1981) defined the rostral tympanic process (RTPP) as being a “tympanic process that arises (apparently always periosteally) from all or part of the ventral surface of the pars cochlearis.” Wible (1990), in a study of periotics of Cretaceous marsupials, described a crest that “arises from the promontorial surface in front of the fenestra cochleae and runs towards the midline along the posterior edge of the medial expansion.” He noted that this crest is in the same position as the RTPP in modern marsupials and that the morphology of this crest in his Petrosal Type A (probably a species of Pediomys) resembles the RTPP of phalangerids. Norris (1994) described the RTPP in Trichosurus vulpecula as the medial border of the pars cochlearis, a “sharp ridge running antero-ventrally to fuse with the margin of the trigeminal fossa.” In direct contrast to this, however, Marshall et al. (1995: p. 75) contended that phalangerids have a smooth promontorium with no RTPP.

The confusion arises because the RTTP identified by Wible in Didelphis is only the apex of a much larger structure; this extends both rostrally and medially from the summit of the mound formed by the underlying cochlea, which we term the cupula cochleae. In phalangerids, the cupula intrudes into this structure, and the ventrally extended process seen in Didelphis is absent. As a result, there appear to be two structures on the ventral surface of the pars cochlearis: a crest running anteriorly from the cupula cochleae onto the promontorium (the RTTP of Norris, 1994), and a variably developed, ventrally oriented process. The latter structure was labeled by Norris (1994) as the ectotympanic process (ECTP). In fact, this process is merely the caudo-lateral extension of the RTTP (fig. 1). Thus, in this paper, we have subsumed the ECTP sensu Norris (1994) into the RTPP.

The final structure that is in question is the crista promontorii medioventralis (CPM). MacIntyre (1972) defined this as a ridge running anteriorly from the margin of the posterior lacerate foramen that forms the boundary between the cerebellar and tympanic faces. In MacIntyre's figure this structure is shown as lying lateral to the sulcus of the inferior petrosal sinus (SIPS). In the periotics of Wyulda and phalangerins, however, the ventral edge of the cerebellar face can extend down medial to the SIPS, forming a ventrally directed flange. In these cases, the lowest point on the pars cochlearis, and the apparent margin between the cerebellar and tympanic faces, tends to be the ventral edge of this flange. Despite this, however, we continue to apply the term CPM to the crest that runs lateral to the SIPS, as this is the structure that most closely resembles the original definition of MacIntyre.

Terminology of the nerves and blood vessels surrounding the periotic follows Archer (1976). In this paper, the term “trichosurin” refers only to the modern species. Riversleigh specimens are discussed separately, as their exact phylogenetic position has not yet been determined (this includes the two described Riversleigh species, Trichosurus dicksoni and Strigocuscus reidi).

DESCRIPTIONS OF PERIOTIC MORPHOLOGY

DISCUSSION

In a number of respects, the periotics of Strigocuscus celebensis and Wyulda squamicaudata are very similar in morphology to those seen in species of Trichosurus. These form a tight morphotype when compared to the periotics of species of Ailurops, Phalanger, and Spilocuscus (fig. 4). The morphotype is distinguished primarily by considerable elongation of the promontorium (around 45% of the total length of the pars cochlearis, compared with 30% in Ailurops and the phalangerins). Further features, most notably the extensive expansion of the crista petrosa and the crista promontorii medioventralis, support a closer relationship between Trichosurus and Wyulda than between either of these taxa and Strigocuscus. All three show a more simply built promontorium than that of the phalangerins. In the latter group, a deeply grooved sulcus of the inferior petrosal sinus separates the rostral tympanic process and crista promontorii medioventralis, while the sulcus of the facial nerve cuts deeply into the pars cochlearis anteroventral to the hiatus Fallopii. These two features greatly increase the structural complexity of the promontorium in phalangerins and cause a marked discontinuity between the promontorium and the rest of the pars cochlearis from both cerebellar and tympanic aspects, which is absent in the trichosurins. Finally, while the anteroposterior proportions of both the subarcuate fossa and the internal auditory meatus are fairly consistent across all phalangerid taxa, the dorsoventral axis for the two openings shows consistent differences when the trichosurins are compared to the phalangerins. The subarcuate fossa accounts for around 70% of the dorsoventral axis of the periotic, compared with 55% in the phalangerins, and appears much larger. By contrast, the equivalent axis for the internal auditory meatus in trichosurins accounts for only 50% of the height of the pars cochlearis, compared to 75% in phalangerins, and appears much smaller. These results appear to solidify the relationship between Trichosurus, Wyulda, and S. celebensis, showing that the latter species is definitely a trichosurin, rather than a phalangerin or the sister group of a trichosurin + phalangerin clade.

One aim of this study was to see what light, if any, the periotics of the modern trichosurin genera could throw on the radiation of phalangerid marsupials in the Tertiary. Two phalangerid species were described by Flannery and Archer (1987) from the Miocene of Riversleigh and assigned to the modern genera Strigocuscus and Trichosurus. A preliminary look at their associated periotics suggests that these species are more closely related to each other than to either of the modern genera. The distinctive morphology of Miocene phalangerids will be looked at in a separate paper (K.C. in prep.), but a general comparison of the various specimens of periotic from Tertiary phalangerid species with those of modern trichosurins follows.

Periotic morphology of Riversleigh specimens shows more similarities to trichosurins than to phalangerins. Like trichosurins, Riversleigh specimens have an elongate promontorium, small internal auditory meatus, and a large epitympanic recess (fig. 5). However, there are also some striking differences. The Riversleigh specimens lack the expanded crista petrosa that is seen in the trichosurins, and although the crista promontorii medioventralis is expanded ventrally, the space bounded by the edge of the crista, the rostral tympanic process, and the cupula cochleae is concave, rather than smoothly sloping as in the trichosurins. This morphology of the pars cochlearis is also seen in a number of other diprotodontian groups, including pseudocheirids and potoroids and, although absent from the ailuropine petrosal (Norris, 1994), it may be plesiomorphic within phalangerids. All Riversleigh phalangerid taxa show a distinctive discontinuity across the cerebellar face of the periotic formed by the sulcus of the inferior petrosal sinus. This is also seen in Wyulda. Unlike Wyulda, the Riversleigh periotics are all delicate rather than robust. The promontorium is also quite curved rostrally when compared to the modern trichosurins.

Other features of the periotic in the trichosurins are harder to apportion to individual clades and seem likely to be autapomorphic for the taxa concerned. Most notable here is the shape of the pars cochlearis: slender and elongate in Strigocuscus, broad and stocky in Wyulda, or with a prominent, ventrally directed “blade” in Trichosurus.

Although a broader examination of craniodental morphology in the Trichosurini is outside the scope of this particular study (but is the subject of ongoing work by K.C.), it is worth making a brief overview of the situation in order to put the anatomy of the periotic into a wider context. Flannery et al. (1987) listed a set of five synapomorphies defining the tribe Trichosurini and a further eight defining the clade Trichosurus + Wyulda (table 3). Subsequent investigation by a number of authors (Springer et al., 1990; Norris, 1992, 1994; Hamilton and Springer, 1999), drawing on both molecular and morphological evidence, has led to a number of these characters being discarded as cases of homoplasy between the trichosurins and certain members of the phalangerin radiation, but have also produced new characters, most notably in the otic region. A revised list of diagnostic characters for the Trichosurini is given in table 4. When combined with the results of our study of periotic anatomy, we think that this character set provides a sufficient basis for support of a monophyletic Trichosurini including S. celebensis.

One outstanding issue concerning the Trichosurini is the phylogenetic position of Phalanger pelengensis, which is still uncertain. This species was originally considered a subspecies of Strigocuscus celebensis (Tate, 1945). Flannery et al.. (1987) suggested that the species might be a plesiomorphic sister taxon to both Phalanger and Spilocuscus, but left the species as “Phalanger”. George (1987) also kept P. pelengensis in Phalanger rather than following S. celebensis to Strigocuscus. Groves (1987) maintained the close relationship with S. celebensis, but kept both in Phalanger. Flannery (1994, 1995) also maintained this relationship, but used the genus Strigocuscus. However, preliminary study of skull morphology by the authors suggests that, despite the outward similarity of these species, the skull of Phalanger pelengensis is much more similar to phalangerins than to trichosurins (including S. celebensis): the periotic of P. pelengensis lacks the elongate promontorium and expanded crista petrosa seen in the trichosurin possums; there is no I3-C1 diastema; the ventral surface of the orbits shows only a limited overhang of the cheek tooth row; and the molars are slightly crenulate. It is possible that the taxonomic affinities of this species may lie with the endemic cuscuses of the neighboring Moluccan archipelago, P. alexandrae, P. ornatus, and P. rothschildi: this hypothesis is still under study, and will be explored in detail in a future paper (Crosby and Norris, in prep.).