Cranial anatomy of Oligo-Miocene koalas (Diprotodontia: Phascolarctidae): stages in the evolution of an extreme leaf-eating specialization more

Journal of Vertebrate Paleontology 29(4):981–992, December 2009 # 2009 by the Society of Vertebrate Paleontology FEATURED ARTICLE CRANIAL ANATOMY OF OLIGO-MIOCENE KOALAS (DIPROTODONTIA: PHASCOLARCTIDAE): STAGES IN THE EVOLUTION OF AN EXTREME LEAF-EATING SPECIALIZATION 1 JULIEN LOUYS,*,1 KEN APLIN,2 ROBIN M.D. BECK1 and MICHAEL ARCHER1,3 School of Biological, Earth and Environmental Sciences, University of New South Wales, Sydney, New South Wales 2052 Australia; 2 Australian National Wildlife Collection, CSIRO Sustainable Ecosystems, GPO Box 284, Canberra, ACT, 2601, Australia, ken.aplin@csiro.au; 3 Faculty of Science, University of New South Wales, Sydney, New South Wales 2052 Australia, m.archer@unsw.edu.au. ABSTRACT—Partial crania of two fossil species of koala (family Phascolarctidae) from Oligo-Miocene deposits in the Riversleigh World Heritage Area, one referable to Litokoala kutjamarpensis and another to Nimiokoala greystanesi, are described. Comparison with the extant koala Phascolarctos cinereus and other diprotodontian marsupials reveals a high degree of similarity in basicranial morphology between fossil and modern phascolarctids but substantial differences in the architecture of the masticatory system. Key specialisations present in Phascolarctos but absent in both Litokoala and Nimiokoala include forward displacement of the palate, enlargement of the occlusal surface of the molar teeth, thickening of the maxillae above the toothrow with resultant lowering of the occlusal plane of the cheekteeth relative to the glenoid fossa, and a decrease in the size of the pterygoid fossae. These extreme aspects of the cranial morphology of Phacolarctos probably reflect its dependence on eucalypt leaves, a nutrient-poor food resource that became increasingly abundant in the Australian environment through the Neogene. Derived similarities in basicranial structure, notably the large size of the auditory bulla, between the fossil and modern phascolarctids raises the possibility that two distinctive behavioural characteristics of the modern koala, sedentism and vociferousness, may have developed relatively early during phascolarctid evolution. INTRODUCTION Eucalypts dominate the tree layer of most Australian forests and woodlands but their foliage is underutilised by mammals on account of its low nutritional content and toxicity caused by high levels of tannins and other phenolics (reviewed by Cork and Sanson, 1991). One mammal that has overcome these limitations in spectacular fashion is the koala (Phascolarctos cinereus), the sole living member of the diprotodontian marsupial family Phascolarctidae. Koalas are the only mammals that consume eucalypt foliage as a dominant component of their diet. Moreover, with body weights sometimes exceeding 20 kg in males, they also are among the largest of all arboreal folivores (Cork and Sanson, 1991). To attain this remarkable condition, koalas have evolved a suite of anatomical and physiological specializations that combine extractive efficiency (e.g. specialized masticatory and digestive anatomies; Lanyon and Sanson, 1986; Davison and Young, 1990; Hume 1999) with reduced energy demands based on low basal and field metabolic rates (Degabriele and Dawson, 1979; Nagy and Martin, 1985) and a highly sedentary lifestyle. Koalas are further unusual among marsupials for their loud and complex vocalisations; these are thought to compensate for infrequent faceto-face social contacts within low density, sedentary populations (Smith, 1980; Mitchell, 1991a). Koalas (family Phascolarctidae) are most closely related among living marsupials to wombats (family Vombatidae; Aplin and Corresponding author. Current address: School of Natural Sciences and Psychology, Liverpool John Moores University, Liverpool, United Kingdom; j.louys@ljmu.ac.uk, louys_julien@hotmail.com * Archer, 1987; Munson 1992; Kirsch et al., 1997; Horovitz and ´nchez-Villagra, 2003; Asher et al., 2004; Beck, 2008). HowSa ever, the divergence between koalas and wombats is quite ancient (30-40 MYA based on molecular data; Drummond et al., 2006; Beck, 2008) and it is generally agreed that wombats are more closely related to a number of extinct diprotodontian families (some or all of Diprotodontidae, Palorchestidae, Wynyardiidae, Maradidae, Ilariidae and Thylacoleonidae; Aplin and Archer, 1987; Archer et al. 1999). The koala thus has no living close relatives that might be consulted to reveal aspects of its evolutionary history. Koalas are generally scarce in the fossil record and most taxa are known from a few isolated teeth or jaw fragments. Nevertheless, the diversity of fossil koalas is surprisingly high; with a total of 18 named species representing five genera and spanning the period from the late Oligocene to the Quaternary (see Long et al., 2002 for summary). Comparisons of dental morphology have shed some light on the derivation of the modern koala dentition (Woodburne et al., 1987; Black & Archer, 1997) but this reveals little or nothing regarding the mode or time of origin of its unique ecophysiological adaptations. Over the last two decades, numerous exquisitely preserved remains of fossil marsupials have been recovered from fluviatile to lacustrine limestones of Oligo-Miocene age in the Riversleigh World Heritage Area, northwestern Queensland (Archer et al. 1991). Although koala fossils are generally uncommon throughout the Riversleigh sedimentary sequence, partial crania are now available for two taxa: Nimiokoala greystanesi and Litokoala kutjamarpuensis. The excellent state of preservation of these specimens provides the impetus for a detailed comparison of cranial morphology of fossil and modern koalas, and for a consi- 981 982 JOURNAL OF VERTEBRATE PALEONTOLOGY, VOL. 29, NO. 4, 2009 2003), Norris (1994) and Crosby and Norris (2003). A detailed osteological description of the cranium of Phascolarctos, with excellent accompanying figures, can be found in Wegner (1964). Throughout this descriptive section, we compare the fossil crania with examples of the modern koala (henceforth Phascolarctos; Fig. 1A) and the brush-tailed possum Trichosurus vulpecula (henceforth, Trichosurus; Fig. 1B). The latter taxon, a member of the extant marsupial family Phalangeridae, is comparable in size to each of the two fossil koalas. More importantly, it also possesses a relatively plesiomorphic cranial morphology that may be close to the ancestral ‘bauplan’ for all diprotodontian marsupials (Ride, 1964; Aplin, 1987; Aplin and Archer, 1987; Weisbecker and Archer, 2008). Interpretation of muscular and neurovascular features of the cranium of Phascolarctos was facilitated by dissection of the masticatory, pharyngeal and hyoid muscles and assorted structures in several formalin preserved specimens (including both adults and juveniles) and detailed study of a serially sectioned head of a pouch young individual (see Aplin, 1990 for details). Comparative specimens of modern koalas and a range of other marsupial taxa were examined from the collections of the Vertebrate Palaeontology Laboratory, University of New South Wales, and the Australian National Wildlife Collection, Canberra. Abbreviations—QMF, specimen prefix of the Queensland Museum, Australia; SAM, South Australian Museum, AR, J, specimen prefixes of the Vertebrate Palaeontology Lab, University of New South Wales, Sydney, Australia. deration of the evolutionary pathways that resulted in the highly specialised adaptations of the modern koala. MATERIALS AND METHODS The fossil crania were freed from a limestone matrix by acid dissolution (this process is described in detail in Archer et al. 1991). Fossil specimens are registered in the Vertebrate Palaeontology collection of the Queensland Museum (QMF). The partial cranium of Nimiokoala greystanei (QMF 30483) was recovered from Boid’s Site East site in Riversleigh Faunal Zone B, interpreted to be early to middle Miocene in age (biostratigraphic nomenclature follows Arena, 2004). Black and Archer (1997) figured and gave a brief description of this specimen; their account is emended and extended here. The partial cranium of Litokoala kutjamarpensis (QMF 51382) was recovered from Jim’s Carousel Site, also in Faunal Zone B. Louys et al. (2007) described the dentition of this and other specimens of Litokoala from Riversleigh and gave reasons for placing L. kanukaensis in synonymy with L. kutjamarpensis. Species-level systematic nomenclature for fossil koalas otherwise follows Black & Archer (1997). Higher level systematic nomenclature of marsupials follows Aplin & Archer (1987). Dental terminology follows Archer (1978) as modified by Tedford & Woodburne (1987) and Luckett (1993). Cranial terminology follows Aplin (1990); basicranial terminology follows Archer (1976), with revisions from Aplin (1987, 1990). Petrosal terminology follows MacIntyre (1972), as modified by Wible (1990, FIGURE 1. Crania of modern species discussed in the text. A, Phascolarctos cinereus, ventral view; B, Trichosurus vulpecula, ventral view. Scale bar equals 2 cm. LOUYS ET AL.—FOSSIL KOALA CRANIA 983 Subsequently, a second species, L. kanunkaensis, was described from two isolated teeth and dental fragments from Kanunka North Local Fauna, Etadunna Formation, South Australia (Springer, 1987). Black and Archer (1997) referred Riversleigh specimens to L. kanunkaensis but noted some minor differences in dental morphology. The recently recovered specimen from Jim’s Carousel Site, Riversleigh, Queensland (QMF 51382) is a partial skull of a young adult with a complete and lightly worn cheektooth series. Its dentition was described by Louys et al. (2007), and this specimen was referred to L. kutjamarpensis. Louys et al. revised the systematics of Litokoala and concluded that the distinguishing features of L.kanunkaensis listed by Springer (1987) and Black and Archer (1997) represent meristic differences along the tooth row rather than interspecific differences. Thus, L. kanunakensis is now considered to be a junior synonym of L. kutjamarpensis (Louys et al., 2007). FIGURE 2. Litokoala in dorsal view. Scale bar equals 1 cm. SYSTEMATIC PALEONTOLOGY Order DIPROTODONTIA Owen, 1866 Family PHASCOLARCTIDAE Owen, 1839 LITOKOALA KUTJAMARPENSIS Stirton 1967 (Figs. 2–5) Holotype—SAM P13845, RM1. Revised Diagnosis—In addition to dental apomorphies described elsewhere (Stirton et al., 1967; Black and Archer, 1997; Louys et al., 2007), Litokoala differs from Nimiokoala in having a larger superficial masseteric process, a more marked basioccipitalbasisphenoid flexion and more extensive posterior attachment of the pterygoid bones. Litokoala differs from Phascolarctos in possessing shorter paroccipital processes, wider mastoid processes, smaller postglenoid processes, keeled basioccipital and a shallower and more arched zygomatic arch; and in exhibiting a raised palate and molar occlusal plane relative to the glenoid fossa, a shorter glenoid fossa, larger medial pterygoid fossae, less inflated auditory bullae and a more prominent sagittal crest. Described Specimen—QMF 51382 is a partial skull of a young adult with all four molars fully erupted but lightly worn. Preserved is an almost intact basicranium (except for ventrally broken tympanic wings and postglenoid processes, and missing ectotympanics), a partial neurocranium with a relatively intact orbitotemporal region, one almost complete zygomatic arch, and a partial palate with complete left and right maxillary dentition (i.e. P3, M1-4). The dorsal portion of the cranium is broken away anterior of the glenoid fossa. Remarks—The genus Litokoala was erected by Stirton et al. in 1967 on the basis of an isolated M1 from the Kutjamarpu Local Fauna, Wipajiri Formation, Lake Ngapakaldi, South Australia. FIGURE 3. Litokoala in occipital view. Scale bar equals 1 cm. FIGURE 4. Litokoala in ventral view. Dashed lines represent suture lines, hatched areas represent broken bones. Abbreviations: RAHS, roof of the alisphenoid hypotympanic sinus; RSHS, roof of the squamosal hypotympanic sinus. 984 JOURNAL OF VERTEBRATE PALEONTOLOGY, VOL. 29, NO. 4, 2009 FIGURE 7. Nimiokoala ventral view. Scale bar equals 1 cm. FIGURE 5. Litokoala showing interior of neurocranium and cerebral face of left petrosal. Scale bar equals 1 cm. Abbreviations: SIPS, sulcus for the inferior petrosal sinus. P3, M1, left I1-3, C, P3, partial M1). The left side of the skull is missing the orbitosphenoid and auditory regions. DESCRIPTIONS Cranial Morphology of Litokoala kutjamarpensis General Cranial Configuration—QMF 51382 is approximately half the size of an adult Phascolarctos in linear dimensions and similar in size to an adult Trichosurus. It displays teeth slightly larger than Trichosurus but much smaller than Phascolarctos. The skull is broadest just anterior of the glenoid fossa, narrowing anteriorly along the zygomatic arch, a condition similar to Trichosurus, whereas in Phascolarctos the opposite condition is observed. The basioccipital is slightly smaller than the basisphenoid; these are approximately equal in Trichosurus, while Phascolarctos has a relatively larger basioccipital. Dentition—The dentition of this specimen was described in detail by Louys et al. (2007). Palate—Nasal, lacrimal, frontal and premaxilla bones are absent. The palate is relatively thin, preserved from the level of P3, with the posterior margin broken and only preserved as a thin plate containing the teeth. The posterior end of the tooth row lies well anterior to the basisphenoid-presphenoid suture. The palate is flat or slightly arched forward of M1, much less strongly arched than in Phascolarctos, but slightly more so than in Trichosurus. The maxillary-palatine vacuity is large and extends from the posterior margin of M2 to just behind the anterior margin of M4, terminating just anterior to the palatine-presphenoid suture. Interestingly, apparently homologous but relatively smaller paired vacuities are present in Phascolarctos, but lie entirely within the palatine; this finding has implications for studies that have treated the maxillary-palatine and palatine vacuities in marsupials as non-homologous structures (e.g. Voss and Jansa, 2003). Zygomatic Arch—The zygomatic arch (partially preserved only on left hand side, where it is missing the anterior part of the zygomatic process of the squamosal), is not as deep posteriorly in Litokoala relative to Phascolarctos, because the temporal crest (sensu Evans, 1993) of Litokoala is proportionally less extensive dorsally. The zygomatic process of the jugal is strongly arched in lateral view in QMF51382, and hence its ventral margin is very concave, whereas the ventral margin of the zygomatic process of the jugal is distinctly more horizontal in both Trichosurus and Phascolarctos. The jugal is relatively deeper posteriorly than in Trichosurus; it has a preglenoid process as in Phascolarctos, whereas this process is absent in Trichosurus (its jugal tapers to the rear). The superficial masseteric process extends ventrally below the level of the tooth row, level with the M2 paracone, as in Trichosurus (by contrast, this process terminates above the tooth row at Order DIPROTODONTIA Owen, 1866 Family PHASCOLARCTIDAE Owen, 1839 NIMIOKOALA GREYSTANESI Black and Archer, 1997 (Figs. 6–9) Holotype—QMF30482, partial cranium. Revised Diagnosis—In addition to dental apomorphies described elsewhere (Black and Archer, 1997), Nimiokoala differs from Litokoala in having a less marked basioccipital-basisphenoid flexion and less extensive posterior attachment of the pterygoid bones. Nimiokoala differs from Phascolarctos in possessing an uninflated rostrum, a flat, unarched palate, a low infraorbital canal and a narrow interorbital region; and in exhibiting a raised palate and molar occlusal plane relative to the glenoid fossa, a shorter glenoid fossa, larger medial pterygoid fossae, less inflated auditory bullae and a more prominent sagittal crest. Described Specimen—QMF 30483 is a partial adult skull approximately half the size of an adult Phascolarctos cinereus and approximately the same size as Litokoala kutjamarpensis and Trichosurus. The skull preserves part of the nasals, the premaxilla, the maxilla, lacrimal, frontal, palatine, squamosal (latter four bones preserved only on right side), parietal, alisphenoid, basioccipital, basisphenoid and interparietal. A loose petrosal was found in association with the skull. The dentition is partially preserved (right I1, FIGURE 6. Nimiokoala in dorsal view. Scale bar equals 1 cm. LOUYS ET AL.—FOSSIL KOALA CRANIA 985 FIGURE 8. Nimiokoala in lateral view. Dashed lines represent hypothesised suture lines. the level of M1 metacone in Phascolarctos). This process is predominantly composed of the maxilla, although the jugal makes a contribution laterally as in Trichosurus; by contrast, this process is formed entirely by the maxilla in Phascolarctos. Sphenopterygoid Region—Very little of the orbital fossa is preserved, but it is apparent that the orbital floor is not elevated relative to the anterior root of the zygomatic arch as it is in Phascolarctos. In lateral view, the alisphenoid is bound anteriorly by a small fragment of the palatine bone (preserved only on left hand side), which interposes between the alisphenoid and maxilla. In ventral view, the alisphenoid supports a broad, horizontally orientated pterygoid fossa posteriorly. The fossa itself is strongly rugose and is defined laterally by a strong ridge (the ectopterygoid crest) formed by the alisphenoid. The ectoptery- goid crest is strongly developed anteriorly, with a distinct hook present at its posteroventral margin, but it is not continuous posteriorly with the rim of the foramen ovale (unlike Nimiokoala). The transverse canal foramen is small and contained entirely within the posterior end of the pterygoid fossa; it is divided by a strut on the left hand side. The pterygoid bones are missing bilaterally but their extent is clear from rugose sutural zone along the margins of the basisphenoid. The sutural attachment for the pterygoids ends posteriorly level with the entocarotid foramina in Litokoala, as in Trichosurus, whereas the pterygoids extend well posterior to these foramina in both Nimiokoala and Phascolarctos. The glenoid fossa is rectangular and slightly curved parallel to the postglenoid process. It extends medially from the lateral edge of the skull, where it is anteropos- 986 JOURNAL OF VERTEBRATE PALEONTOLOGY, VOL. 29, NO. 4, 2009 either Trichosurus or Phascolarctos. Also unlike these taxa, Litokoala displays a narrower posterior margin of the skull relative to the width of the skull at the posterior root of the zygomatic arch. Like Trichosurus, Litokoala possesses an anteriorly directed concavity at the interparietal/parietal margin of the skull. The opposite condition can be seen in Phascolarctos, in which the interparietal protrudes slightly posteriorly, creating a convex posterior margin of the skull when viewed dorsally. The interparietal of Litokoala is exposed, most likely as a result of damage to the parietals; it is likely that the parietals covered at least the majority of the interparietal in life (this condition is found in at least one other marsupial, Monodelphis (Wible, 2003). The interparietal and parietal in Litokoala bear strong nuchal and sagittal crests, respectively. The squamosal is roughly rectangular. It is bordered medially by the tympanic and mastoid portions of the petrosal and dorsally by the parietal. The suprazygomatic canal exits dorsally, behind the root of the zygomatic arch. A large suprameatal foramen is present; it is divided internally, one branch joining the postglenoid foramen, the other opening into the endocranial cavity. Auditory Region—The hypotympanic sinus is less inflated anteriorly than in Phascolarctos, where the sinus extends deep into the root of the pterygoid process of the alisphenoid. It also differs from that of Phascolarctos in showing greater inflation anteromedially and less in the medial part of the squamosal. The roof of the tympanic cavity is comprised of both the alisphenoid and squamosal. The alisphenoid is the major element in the tympanic roof, and is the sole roofing element medially. The remainder of the tympanic roof, lateral to a bony canal within the alisphenoid that would have transmitted the mandibular branch of the trigeminal nerve prior to its exit via the foramen ovale, is formed by the squamosal. The tympanic floor is almost completely missing, but was probably composed largely of the alisphenoid, with only minor lateral contributions by the squamosal, medial to the glenoid cavity. Hence the bulla would have been partly bilaminar anterolaterally, with the squamosal contributing to the internal cavity and the alisphenoid enclosing it. This morphology is also present in at least some thylacoleonids (e.g. Wakaleo and Priscileo; Murray et al., 1987; pers. obs.) The squamosal is slightly more extensive medially than in Phascolarctos, where it stops well lateral to the canal for the mandibular branch of the trigeminal nerve. Basicranium—The basisphenoid is elongated and tapers anteriorly. The entocarotid foramen is large and perforates the basisphenoid well forward of the basisphenoid-basioccipital suture, but posterior to the pterygoid bone. The basioccipital shows strong midline ridging. The basioccipital and basisphenoid show minimal longitudinal flexion relative to each other. The paroccipital processes are relatively small but they show anterior ridging for attachment of the hyoid suspensory ligaments and muscles. Paired hypoglenoid foramina are present in a shallow depression immediately anterior to each occipital condyle; the anterolateral side of this depression is open, unlike Phascolarctos where a bony process of the basioccipital forms a bridge that extends to the base of the paroccipital process. Petrosal—In ventral view, it is obvious that the petrosal of Litokoala resembles that of most marsupials in being more horizontally inclined than it is in Phascolarctos; Phascolarctos is unusual in that the promontorium is orientated dorsoventrally. The promontorium of Litokoala is globular (whereas it appears relatively flatter and less inflated in Phascolarctos) but bears a well-defined, ridge-like rostral tympanic process positioned directly ventral to the fenestra cochleae. A crescent-shaped ridge, the crista promontorii medioventralis, curves anteromedially from the rostral tympanic process and defines the medial limit of the tympanic cavity. The fenestra cochleae is located at the posteromedial corner of the promontorium; the fenestra is tearshaped, with the pointed end of the tear located ventrally. The FIGURE 9. Nimiokoala right petrosal. Scale bar equals 1 cm. Abbreviations: SIPS, sulcus for the inferior petrosal sinus; RTPP, rostral tympanic process of the petrosal. teriorly longest, but is slightly narrower medially and terminates approximately 2 mm lateral to the lateral edge of the auditory bulla. It is of the complex form (i.e. with a distinct articular eminence anteriorly and a mandibular fossa posteriorly; see Aplin, 1987), with a well developed posterior groove and articular eminence. Laterally, the glenoid fossa is bordered by a well developed preglenoid process formed by the posterior end of the jugal. The postglenoid process is large and contains a deeply invasive zygomatic epitympanic sinus (postglenoid cavity). The postglenoid process is more transversely orientated in Litokoala than in Phascolarctos (where it is inflected posteromedially), and is less rugose (for attachment of the postglenoid ligaments) along its ventral margin. In Trichosurus, the ectotympanic is tightly fused to the posteroventral margin of the postglenoid process, whereas this is not the case in either Litokoala or Phascolarctos. As in other diprotodontians (Springer and Woodburne, 1989; Wroe et al., 1998), the large postglenoid foramen is located medial to the postglenoid process; it leads into a well developed bony canal that would have transmitted a branch of the postglenoid artery and accompanying vein across the epitympanic sinus to exit at the suprameatal foramen. Neurocranium—In dorsal view, the neurocranium is trapezoidal in Litokoala, wider posteriorly and narrowing slightly anteriorly. It is less trapezoidal than Trichosurus, but more so than Phascolarctos (in which the neurocranium approaches a rectangle). Its dorsal surface slopes anterodorsally, more steeply than LOUYS ET AL.—FOSSIL KOALA CRANIA fenestra vestibuli is located on the lateral face of the promontorium and is recessed within a small fossula; it is ovate in shape (with its major axis oriented anteroposteriorly) and is distinctly smaller than the fenestra cochleae. The surface of the promontorium between these two fenestrae is narrow and lacks the rounded ‘bulge’ seen in Nimiokoala and Phascolarctos. Immediately anterior to the fenestra vestibuli, the exit for the facial nerve is not floored by the petrosal, i.e. there does not appear to be a true secondary facial nerve foramen. This distinctive morphology is seen in the late Cretaceous stem-metatheriam ´ Deltatheridium (Rougier et al., 1998; Sanchez-Villagra and Wible, 2002) and, of perhaps more relevance, in vombatids ´ (Sanchez-Villagra and Wible, 2002). There is no evidence of a prootic canal or of the morphologically similar canal that is observed in Phascolarctos and which transmits a nerve rather than the lateral head vein (Aplin, 1990; see also Beck et al., 2008 Text S2 character 234). Posterolateral to the exit for the facial nerve and lateral to the fenestra vestibuli is the epitympanic recess, which is delimited anteriorly by a strong transverse ridge (the petrosal crest sensu Archer 1976; Muizon 1999; the lateral malleolar ridge sensu Aplin, 1990), anterior to which the petrosal makes a small contribution to the roof of the hypotympanic sinus. The anterolateral wall of the epitympanic recess is formed by the tuberculum tympani (sensu Wible, 2003) which also forms the anteromedial rim of the postglenoid foramen, while the posterolateral wall is formed by the squamosal. Immediately posterior and slightly ventral to the epitympanic recess, and separated from it by a distinct ridge, the fossa incudis is clearly identifiable; in this respect, Litokoala resembles Trichosurus and other ‘possums’, whereas in Phascolarctos (and also vombatids) there is no true fossa incudis, the crus brevis of the incus attaching to ‘the wall of the interosseous cleft between the petrosal and squamosal bones’ (Aplin, 1990). The medial wall of the fossa incudis in Litokoala is formed by the crista parotica. Immediately posterior to the fossa incudis, the tympanohyal is identifiable as a small but stout, ventromedially-directed process, posterior to which is the stylomastoid notch. Lateral to the epitympanic recess and fossa incudis, there are two large squamosal epitympanic sinuses: anteriorly a zygomatic epitympanic sinus invades deep into the postglenoid process, and its opening is almost a perfect circle; posterior to this, and separated from the first sinus by a lateral ridge, the second, smaller sinus forms a lateral trough in the squamosal and is partially floored by the posttympanic process of the squamosal. As in Phascolarctos, the caudal tympanic process of the petrosal of Litokoala is a ventrally extensive process clearly visible in ventral view at the posterolateral corner of the skull, bound laterally by the squamosal and medially by the paroccipital process; it is damaged on the right side of QMF51382. The caudal tympanic process of the petrosal forms the posterolateral wall of a deep but narrow fossa for the stapedius muscle, posterolateral to the promontorium. The anterolateral margin of the caudal tympanic process is scored by a deep stylomastoid notch for exit of the facial nerve, which is bordered anteriorly by the tympanohyal. This notch is not as well-marked in Phascolarctos. In Trichosurus, the facial nerve exits by a complete stylomastoid foramen formed by the caudal tympanic process of the petrosal posteriorly and the ectotympanic anteriorly. Posterior to the stylomastoid notch, the mastoid process of Litokoala is distinct but not hypertrophied, extending posterolaterally as a relatively low ridge towards the posterolateral-most point of contact between the mastoid exposure of the petrosal and the squamosal. The paroccipital process is nearly complete on right hand side of QMF51382, but is damaged on the left side, and was clearly proportionally much smaller than in Phascolarctos (in which this process is greatly hypertrophied such that it extends well below the level of the basioccipital and basisphenoid) and is also smaller than in Trichosurus. Litokoala resembles Phascolarctos in 987 that the paroccipital process is oriented essentially ventrally (and slightly anteriorly in Litokoala), whereas in Trichosurus it is directed posteroventrally. A distinctive feature of Phascolarctos is the presence of a vertically-directed groove in the base of the paraoccipital process immediately medial to the most medial point of the caudal tympanic process of the petrosal and lateral to the jugular foramen; this groove is absent in both Litokoala and Trichosurus. The cerebellar face of the petrosal shows a deep subarcuate fossa which greatly exceeds the internal auditory meatus in size. It has a thickened crus commune, the anterior tip of which slightly overhangs the internal auditory meatus. The gyrus of the anterior vertical semicircular canal is smoother and more rounded than in Phascolarctos, but as in Phascolarctos, there is a distinct depression within the sulcus of the sigmoid sinus (the mastoid sinus). However whereas the mastoid sinus in Phascolarctos invades deeply into the substance of the petrosal, in QMF 51382 it is no more than a shallow depression with several small foramina in its base. The hiatus fallopii for exit of the greater petrosal nerve from the petrosal is hidden by the alisphenoid, but the cavum epiptericum immediately anterior to the petrosal is well defined, and is floored almost entirely by the alisphenoid. The crista petrosa is well developed, forming a relatively tall crest along the lateral margin of the petrosal, as in both Phascolarctos and Trichosurus. Lateral to the subarcuate fossa, the petrosal makes no contribution to the lateral wall of the endocranial space, as in Phascolarctos but unlike Trichosurus. The ectotympanic is missing from both sides in QMF51382. However, the point of attachment of the anterior crus is marked by a depression between the postglenoid process and the tuberculum tympani of the petrosal. The posterior crus evidently had no bony attachment to either the postglenoid process or the mastoid exposure of the petrosal but was held in position by connective tissue, as in Phascolarctos, but unlike Trichsurus in which the posterior crus is fused to the mastoid exposure of the petrosal. Cranial Anatomy of Nimiokoala greystanesi General Cranial Configuration—Remaining portion of the skull indicates a cranium with relatively short, narrow rostrum and moderately flaring anterior zygomatic roots (though less so than Phascolarctos). The interorbital region is broad, as in Phascolarctos, whereas this region is narrow in Trichosurus. The neurocranium possesses strong temporal and sagittal cresting. The palate is flat anteriorly as in Trichosurus, not arched as in Phascolarctos. Molars in Nimiokoala are larger relative to the width of its palate than in Litokoala and Trichosurus, but similar in this respect to Phascolarctos. Although the posterior three molars are missing, if we assume a relatively flat tooth row, the glenoid fossa would probably have been located approximately 12-13 mm above the cheektooth row. The auditory bulla of Nimiokoala is only partially preserved but was clearly greatly inflated, less extensive ventrally than in Phascolarctos but more so anterolaterally, with the overall degree of inflation considerably greater than in Litokoala. Overall, the skull of Nimiokoala shows a morphology that is in many respects intermediate between Trichosurus and Phascolarctos: the anterior root of the zygomatic arch of Nimiokoala are more flared than those of Trichosurus, but the proportions and overall morphology of the neurocranium and rostrum of Nimiokoala are more similar to the general phalangerid condition than to that of Phascolarctos. However, the bullae of Nimiokoala are much more inflated than Trichosurus or any other comparably sized possum. Dentition—The dentition of QMF 30483 has been described in detail by Black and Archer (1997). Facial Region—Nasals relatively elongate (33 mm) and wide (23.5 mm) posteriorly, narrowing anteriorly (9 mm), dorsal 988 JOURNAL OF VERTEBRATE PALEONTOLOGY, VOL. 29, NO. 4, 2009 frontal by the alisphenoid, as in Trichosurus, whereas a short frontal-squamosal contact is present in at least some specimens of Phascolarctos examined (e.g. AR6509; J10023). Only a small sliver of the interparietal is preserved in Nimiokoala, but it is clear that this element was relatively wide but short, as in Trichosurus; in Phascolarctos the interparietal is generally more extensive anteriorly. The sphenorbital fissure is enclosed laterally by the alisphenoid. The smaller foramen rotundum is located immediately ventrolateral to the sphenorbital fissure and perforates the alisphenoid. A small sliver of the presphenoid-orbitosphenoid complex is visible anterior to the sphenorbital fissure. A broad, horizontally orientated medial pterygoid fossa is present on the alisphenoid immediately anterior to the auditory bulla; the fossa shows strong lateral ridging (the ectopterygoid crest) that extends posteriorly to merge with the ventromedial rim of the foramen ovale, and has well defined internal rugosity, reflecting the attachment of the medial pterygoid muscle. This condition contrasts markedly with the small, narrow and vertically orientated pterygoid fossae of Phascolarctos. The foramen ovale of Nimiokoala is located on the anterolateral margin of the auditory bulla, and represents the anterior opening of a long canal for the mandibular branch of the trigeminal nerve that passes above the roof of the tympanic cavity; the passage of this canal is visible as a bulge in the roof of the hypotympanic sinus, which is exposed due to breakage of the alisphenoid tympanic process. The endocranial cavity is clearly visible in QMF30483 due to breakage. The frontal forms the anterior portion of the endocranium, and it contacts the alisphenoid ventrally and the parietal dorsally. The parietal occupies the dorsal region of the endocranial cavity, posterior to the frontal. The alisphenoid makes up the majority of the floor of the endocranial cavity, and it extends into the lateral wall to contact the frontal anteriorly, the squamosal posteriorly and the parietal dorsally. The squamosal occupies the posterolateral corner of the endocranial surface; it is bound ventrally by a well developed sulcus along which lies the internal openings of the foramen ovale (posteriorly) and the foramen rotundum (anteriorly). Auditory Region—The squamosal clearly extends into the roof of the primary tympanic cavity, contrary to the observation by Black and Archer (1997:227) that QMF 30483 “exhibits the plesiomorphic diprotodontian condition wherein the alisphenoid forms the roof of the tympanic cavity”. In fact, the epitympanic wing of the squamosal forms almost half the bony roof of the tympanic cavity, lateral to the prominent gyrus marking the passage of the mandibular ramus of the trigeminal nerve as it extends anteriorly towards the foramen ovale at the anterior end of the auditory bulla. The alisphenoid is only partly preserved; it makes up the medial wall of the tympanic roof, as well as the preserved portion of the ventral auditory bulla. The degree of inflation of the auditory bulla, while not as extreme as that displayed by Phascolarctos, is substantial. The external surface of the bulla probably had a shallower, more rounded structure than in Phascolarctos, as there is no evidence of the anterolateral extension of the tympanic cavity or of the rugosity on the ventral surface of the bulla that is a hallmark of Phascolarctos. Outside the auditory bulla, the squamosal-alisphenoid suture lies anterior to the glenoid fossa and the root of the zygomatic arch. The zygomatic epitympanic sinus is developed between the posterior margin of the postglenoid process and the lateral wall of the squamosal epitympanic sinus. The surviving portion of the hypotympanic sinus is floored primarily by the alisphenoid and roofed by both an alisphenoid epitympanic wing and a squamosal epitympanic wing, a general anatomy very similar to Phascolarctos. Petrosal—The right petrosal is partially preserved as a loose element that lacks most of the pars mastoidea. It shows a moderately inflated promontorium, somewhat less globose than that of Litokoala but more rounded than that of Phascolarctos. portion flat in longitudinal profile as in Trichosurus, whereas there is slight flexion of this region in Phascolarctos; nasal-frontal suture obscured by gangue (matrix and/or glue), and nasals missing anterior tips. Rostrum uninflated; sides gently curved as in Trichosurus; flared dorsally in Phascolarctos, narrowing to palate. Premaxilla long, with C1 well separate from I3 and located within the premaxilla-maxilla suture; incisor arcade broadly U-shaped as in Trichosurus, as opposed to the narrower V-shape seen in Phascolarctos; incisive foramina are contained entirely within the premaxilla, their posterior margins just contacting the premaxilla-maxilla suture. In lateral view, the premaxillary suture extends posterodorsally from the ventral margin of the rostrum just anterior to the canines, meeting the nasal approximately two thirds of the way back along the length of the nasal, as in Trichosurus. The palate is flat (whereas it is distinctly arched in Phascolarctos), and the preserved portion is complete as far back as the inferred position of M2. The posterior part of palatine is missing. The anterior end of the infraorbital canal is positioned anterior to P3, 3 mm above the alveolar margin, in a similar position to Trichosurus; this contrasts with the more posterior and elevated position of the infraorbital canal in Phascolarctos, which is above the anterior root of M1 and located well above the toothrow. This reflects a great deepening of the maxillary body above the molars in Phascolarctos, a condition not developed in either Nimiokoala or Litokoala. Overall the facial skeleton of QMF30483 has a distinctly phalangeroid appearance. Orbital Mosaic—The medial wall of the orbital fossa is preserved on the right side of the specimen. The lacrimal is bounded posteriorly by the frontal, and ventrally and anterodorsally by the maxilla. The lacrimal is roughly tear-shaped, with the tip of the tear dorsal. The anterior margin of the lacrimal lies just anterior to the preorbital ridge. The orbital lamina of the maxilla is triangular, with its apex dorsal; its posterior edge makes broad contact with the frontal; and posterior to the maxillary foramen, contacts the palatine; anteriorly the maxilla contacts the lacrimal. The orbitosphenoid can be seen in the posteroventral corner of the sphenorbital region. The preorbital ridge is complete, and is perforated by paired lacrimal foramina. These consist of a larger ventral foramen (anterior) and a smaller dorsal foramen (inside orbit) and are separated by a small lacrimal process. The maxillary foramen (the posterior aperture of the infraorbital canal) is small, with a diameter of around 2 mm, similar in size to that of Trichosurus; the maxillary foramen lies at the anterior end of a well developed sulcus in Nimiokoala, as in both Trichosurus and Phascolarctos. The sphenopalatine foramen is positioned 7 mm posterior to the maxillary foramen and is situated on the palatine-maxilla suture. The dorsal margin of the palatine is visible posteriorly but is obscured anteriorly by matrix and glue. The ethmoid foramen is not clearly identifiable. Neurocranium—The interorbital region is bound by well developed temporal crests that clearly enclose a depressed area, as in Trichosurus. This contrasts with the flat surface present in Phascolarctos. The temporal crests merge to form a well developed sagittal crest at least 21 mm long which extends to the rear of the skull. The frontal is roughly rectangular, bound anteriorly by the lacrimal and maxilla, ventrally by the palatine, and posteriorly by the parietal and the alisphenoid (although the latter suture is largely missing due to damage). The parietal is archshaped. The alisphenoid-parietal suture is located either above or below the exterior pterygoid scar (some cracking in this region of the skull obscures the precise suture location). The parietal extends posteriorly to the posterior edge of the skull as in Phascolarctos, whereas it is excluded by the squamosal in Trichosurus. The medial border of the parietal is delimited by the sagittal crest, and it has a broad contact with the squamosal posterolaterally, although the squamosal-parietal suture is damaged and part of the squamosal lamina that overlies the parietal has broken away. The squamosal is widely separated from the LOUYS ET AL.—FOSSIL KOALA CRANIA A well developed rostral tympanic process of the petrosal originates anterior to the fenestra cochleae and extends anteromedially along the medial margin of the pars cochlearis; the rostral tympanic process is damaged anteriorly in QMF30483, but when intact it probably extended for most of the length of the pars cochlearis. The fenestrae cochleae and vestibuli are approximately the same size; the fenestra vestibuli is subovate and is recessed within a distinct fossula, while the fenestra cochleae is kidney-shaped and is not recessed. The tensor tympani fossa is not clearly marked on the lateral promontorium, but this region is partially obscured by matrix. Unlike Litokoala, but as in Phascolarctos, a complete secondary facial foramen is present and is located lateral and slightly dorsal to the fenestra vestibuli. The epitympanic recess is identifiable posterolateral to the secondary facial foramen, but damage to the petrosal posterior to this means that it is uncertain whether a distinct fossa incudis was present or not. It is also unclear whether a prootic canal (or the superficially similar canal that is present in Phascolarctos; Aplin, 1990) was present or not. The caudal tympanic process (as well as most of the rest of the pars mastoidea) is missing but the stapedius fossa appears to have been of similar dimensions to that of Litokoala. On the cerebellar side, the internal auditory meatus is well preserved, with both the medial foramen acusticum inferius (which transmitted the vestibuocochlear nerve) and the lateral foramen acusticum inferius (which transmitted the facial nerve) clearly identifiable. The subarcuate fossa is only partially preserved, with the mastoid and paroccipital margins broken away, but was clearly deep and well-excavated, unlike the very shallow condition in vombatids. The hiatus fallopii is visible in ventral view anterolateral to the promontorium i.e. it ´ opens in a ‘ventral’ position sensu Sanchez-Villagra and Wible (2002), as in Phascolarctos. The petrosal of Nimiokoala differs from that of Phascolarctos, in that a distinct rostral tympanic process is present in Nimiokoala but is absent in Phascolarctos ´ (Sanchez-Villagra and Wible, 2002) and the pars cochlearis of Nimiokoala appears more elongate anteroposteriorly (due at least in part to the dorsoventral orientation of the pars cochlearis in Phascolarctos). In turn, Nimiokoala differs from Litokoala in lesser inflation of promontorium, a slightly more elongate anterior process of pars cochlearis, presence of a true secondary facial foramen floored by the petrosal and a slightly larger fenestra vestibuli. Basicranial Region—The basisphenoid is elongate; it tapers anteriorly, bound laterally by sutural ridging for attachment of the pterygoid bone; a displaced fragment of the pterygoid is preserved on the right hand side, stuck to the underside of the auditory bulla. As in Phascolarctos, the suture for pterygoid bone attachment extends onto the basioccipital, behind the anterior entocarotid foramen. A large transverse foramen is located just lateral to the ridging, at the midpoint of the basisphenoid and at the posterior end of the pterygoid fossa. A small foramen, most likely for the nerve for the medial pterygoid muscle, is located anterolaterally to the transverse canal; on the right side it is well developed and lies within the ridging; on the left side it is small and occurs further laterally, within the pterygoid fossa. The small surviving fragment of the basioccipital bears a prominent midline ridge. The basioccipital and basisphenoid show distinct longitudinal flexion. The internal entocarotid foramen is relatively small; it lies just anterior to the rear of the basisphenoid and is largely obscured by the pterygoid ridging. DISCUSSION Comparisons between Litokoala kutjamarpensis and Nimiokoala greystanesi are hindered by the fact that much of what is preserved in one skull is generally absent in the other. However, the overall impression gained is that the two genera were quite similar in cranial anatomy, with the most obvious difference 989 being a larger superficial masseteric process in Litokoala and a more marked basioccipital-basisphenoid flexion and more extensive posterior attachment of the pterygoid bones in Nimiokoala. In the auditory region, where more detailed comparisons are possible, the two taxa share a common pattern of construction of the roof and floor of the primary tympanic cavity, and a relatively unspecialised morphology of the periotic. Both fossil koalas differ more substantially from Phascolarctos than from each other. Most obviously, both fossil genera are considerably smaller than even small adult specimens of Phascolarctos. Key cranial features of Phascolarctos that are absent in both fossil species include a lowering of the palate and molar occlusal plane relative to the glenoid fossa (brought about, at least in part, by a thickening of the palate), a wider glenoid fossa (reflecting a broader mandibular condyle), reduced medial pterygoid fossae, forward placement of the zygomatic process, more inflated auditory bullae and a less prominent sagittal crest. Additional features of Litokoala kutjamarpensis that distinguish it from Phascolarctos include short paroccipital process, wider mastoid processes, smaller postglenoid processes, keeled basioccipital and a shallower and more arched zygomatic arch. Additional features of Nimiokoala greystanesi which serve to distinguish it from Phascolarctos include an uninflated rostrum, a flat rather than arched palate, a low infraorbital canal, reflecting a shallow alveolar portion of the maxilla and a narrow interorbital region. Despite these differences, it should be noted that there exists a high overall similarity in basicranial anatomy between modern and archaic koalas, and further, that the similarities represent morphological derivations relative to primitive marsupial conditions. Indeed, based on basicranial characters alone, the question might be asked as to whether phascolarctids, with six genera currently recognised for 18 species, may have been oversplit at the generic level based on dental characters. Functional Interpretations of the Masticatory System Phascolarctos is characterised by an anisognathic chewing stroke (i.e. only one side of the mouth is in occlusion at any one time), directed labiolingually with a slight anteromedial shift (Lanyon and Sanson, 1986; Young and Robson, 1987; Davison and Young, 1990). The labiolingual stroke acts as shears, with the cristae and cristids of the molars cutting against one another (Lanyon and Sanson, 1986). The modern koala differs from Trichosurus in not showing any grinding or puncture movements (Young et al., 1990), as well as in the presence of a fused symphysis. The fused symphysis in Phascolarctos allows the transfer of forces generated by balancing side muscles to the working side of the jaw (Crompton and Lieberman, 2005). These features are presumably present to allow efficient mastication of eucalyptus leaves. The presence of broad, horizontally orientated pterygoid fossae on the skull in Litokoala, compared with the much more reduced state present in Phascolarctos, indicates the presence of larger internal and medial pterygoid muscles in archaic koalas. The internal pterygoid muscle is responsible for lateral translocation of the mandible and tooth row. Its fibres are oblique to the masticatory stroke and differ from other muscles of mastication in being slow-acting, powerful contractors. The medial pterygoids form a complex, often multipennate muscle mass that stabilises the jaw during opening and closing actions, and possibly provides protection of the mandibular symphysis in taxa that retain a mobile, unfused symphysis (Adams, 1919). In macropodids the well developed medial pterygoid muscle probably relates to anteromedial jaw movements and a mobile symphysis (Davison & Young, 1990). The presence of these enlarged muscles in Litokoala kutjamarpensis may indicate a more extensive horizontal shearing motion of the kind observed in typical placental ‘ungulates’. It might also relate to the presence of an 990 JOURNAL OF VERTEBRATE PALEONTOLOGY, VOL. 29, NO. 4, 2009 Although a detailed analysis of the auditory capacity of the koala has yet to been performed, we postulate that its large middle ear volume is associated with an increased sensitivity to low frequency sound. Among other groups of mammals, low frequency auditory specialization occurs most often in mammals that inhabit open environments, such as desert or semi-desert (Fleischer 1978; Webster 1962, 1966; Schleich & Vassallo 2003). Lower frequency sounds travel greater distances than higher frequencies without being attenuated below audibility (Fleischer 1978), hence increased sensitivity to low frequency sounds provides an animal with larger spatial capacity both for social communication and for detection of predators (Fleischer, 1978; Webster, 1962, 1966). Forest dwelling mammals rarely show such specializations. Increased sensitivity to low frequency sounds can be achieved within the middle ear apparatus by two means: modification of the ossicular chain itself and/or by increasing the middle ear volume to reduce ‘stiffness’ of ossicular function and possibly also to amplify low frequency sound through production of standing waves. An important consideration that is often overlooked is that sound has an actual physical size, hence the relationship between middle ear volume and auditory acuity is uncoupled from body size. For a small mammal to achieve low frequency acuity, the necessary increase in middle ear volume often leads to formation of grossly inflated auditory bullae which can radically transform their cranial configuration. However, in medium- to large-bodied mammals, the same physical enlargement of the middle ear cavity is typically accommodated without significant external inflation of the bulla. The koala is unusual in this regard as it shows significant ventral inflation expansion of the osseous auditory bulla (Fig. 1). Why this has occurred in the koala and not in vombatids (which possess a smaller hypotympanic middle ear cavity but an enlarged epitympanic sinus) is unclear but may relate to the need to simultaneously accommodate specializations both in their auditory and masticatory systems. Besides humans and dingos, which have arrived in Australia relatively recently, the koala has few contemporary natural predators (Smith, 1987; Martin, 1992). Prior to the late Pleistocene extinction of Australia’s megafauna (Barnosky et al., 2004), the marsupial lion (Thylacoleo carnifex) may have been its principal predator. Hyper-development of the koala’s auditory capacity thus may have evolved partly in response to the need to communicate over larger distances, but also for improved predator detection. Although the auditory bullae of the extinct Nimiokoala and Litokoala species are substantial, they are not as exaggerated as in the modern koala. The Oligo-Miocene environment of Riversleigh in which these animals lived is likely to have been rainforest (Archer et al., 1991; Bassarova, 2006). Higher frequency sound is deflected by vegetation and rapidly loses directionality, while lower frequencies pass through vegetation, thereby travelling further and maintaining directionality. The presence of relatively large bullae in the extinct forms was possibly a result of the need to achieve directionality of sound in a relatively closed forest habitat. Modern howler monkey communities may provide an appropriate analogue. However it should be noted that the auditory bulla is already well developed in both species of archaic koalas, suggesting its development occurred much earlier in the group. The increased volume of the auditory bulla in the modern koala compared to either Nimiokoala or Litokoala probably reflects the increasingly open nature of sclerophyll forests in Australia through the Neogene (Archer et al. 1991; Martin, 1998, 2006). To maintain functionality, the spatial range of koala communication presumably had to increase without loss of directionality. A decrease in the frequency of the koala’s bellows, enabling it to better travel these large open distances without being attenuated, would require increased sensitivity to these lower frequency sounds. unfused symphysis, with the pterygoid muscles providing a stabilising force during mastication, as in macropodids. The mandibular symphysis fuses quite early during cranial growth in Phascolarctos cinereus, prior to the attainment of full adult size. Fusion of the symphysis has been related to a dietary shift emphasising an increase of tougher foods in the diet (e.g. leaves or invertebrates), which require large occlusal forces (Beecher 1977, 1979). Scapino (1981) suggested that fusion of the symphysis in carnivores is related to body size, with larger animals requiring a fused symphysis in order to compensate for their inferior masticatory capacity. However, this hypothesis has been seriously questioned based on EMG studies of symphyseal fusion (Hylander et al. 2000). These studies suggest that symphyseal fusion is a result of two masticatory loading regimes – dorsoventral shear and lateral transverse bending (wishboning) (Hylander et al. 1998, Hogue and Ravosa 2001), with fusion acting to protect the mandible against anterior dislocation. All Nimiokoala dentaries currently recovered are cleanly broken along the symphysis, implying that this taxon also possessed an unfused symphysis (K. Black, pers. comm. 2008). With regards to other koala genera, Woodburne et al. (1987) suggest that the symphysis of Perikoala may have been fused, while that of Madakoala was not. An unfused symphysis in most archaic koalas might therefore reflect a different masticatory loading regime, with a smaller transverse component to the power stroke. It is possible that puncture-crushing movements of the type seen in Trichosurus, which subsists on a much broader diet, but absent in Phascolarctos, were present in archaic koalas. This raises the possibility that the diet of archaic koalas was different to that of Phascolarctos. The nature of the symphysis in archaic koalas can be tested by examining more fossils, including isolated teeth, as taxa with fused symphyses should possess molars with more horizontally orientated wear facets, anteroposteriorly elongate symphyses and relatively wider corpora (Hogue and Ravosa, 2001). Alternatively, it may be that the power stroke was more orthal in archaic koalas than in modern koalas. Considering the differences in masticatory musculature and the possible possession of an unfused symphysis in both archaic genera, it seems plausible the diet of ancient koalas were less specialised than that of Phascolarctos species, and that the almost exclusive association between eucalypts and koalas began after the split of Phascolarctos from Litokoala. This is not surprising – the dominance of eucalypts in Australian forests is a fairly recent phenomenon (major expansion of Eucalyptus forests in Australia did not occur until the late Miocene; Martin, 2006), and eucalyptus densities in Oligo-Miocene rainforests were unlikely to have been high enough to support a koala population (P. Adam, pers. comm. 2004). The dramatic changes in the cranial morphology of Phascolarctos, especially in the facial region, is therefore probably related to demand of increased transversely orientated masticatory loading of the cheekteeth, in response to a change to a tougher diet of eucalyptus leaves. Functional Interpretations of the Auditory System Modern koalas are extremely vocal animals that produce loud “bellows”, typically in the context of mating or agonistic behaviour (Smith, 1987; Mitchell, 1991a). Koala bellows are low pitched, and approach the optimum for long distance sound propagation (Mitchell, 1991a). Ranges of up to 800 m have been recorded, exceeding the home range limits of male koalas (Mitchell, 1991a, 1991b). Underpinning this vocal characteristic is a specialised hyoid apparatus (Aplin, 1990) that displays features in common with two other highly vocal groups of animals: the “roaring” Felidae (Pocock 1916) and the howler monkeys (Schon 1971, Hilloowala 1975, 1976). Compared to other Australian marsupials, middle ear volume in koalas is extremely large, approached only by vombatids. LOUYS ET AL.—FOSSIL KOALA CRANIA CONCLUSIONS With the drying out of the Australian continent and the increase of open sclerophyll forests, koalas, like most Australian marsupials, evolved innovative characteristics in order to adapt to this new environment. Unlike all other vombatiforms, which became largely terrestrial and adapted to open forests and grasslands, koalas took advantage of an ever increasing resource – eucalypt trees. This presumed increased dependence on eucalypts is reflected in the changes through time documented here in koala skulls. In order to adapt to a tougher diet, comprised almost solely of eucalyptus leaves, koalas may have compensated by a fusion of their dental symphysis and a deepening of their maxillae, thereby allowing greater masticatory forces. The late Cenozoic opening up of Australia’s forests would also have impacted on the social behaviour of the koala. The need to maintain social communication over increasingly long distances evidently provided an evolutionary incentive to increase auditory sensitivity to low frequency sounds, something that could be achieved through increasing the size of the auditory bullae. The combination of a specialised auditory system with a relatively unspecialised masticatory system in archaic koalas indicates that the need to achieve auditory sophistication predated dietary specialisation. Therefore, the relatively sedentary and vociferous existence of modern koalas may have already developed by early Miocene times, with the shift to a diet dominated almost exclusively by eucalyptus leaves occurring only later. In order to accommodate both the mechanical demands of this new diet, as well as maintaining their auditory sophistication, the koala underwent substantial changes to its cranial anatomy, in particular that of the facial skeleton. The unique cranial configuration of the modern koala is therefore the result of accommodating their masticatory adaptations without compromising their auditory system. ACKNOWLEDGMENTS The authors would like to thank K. Black and A. Gillespie for preparation of the fossil material described herein, and G. Louys for photography of fossil specimens. Vital support for research at Riversleigh has come from the Australian Research Grant Scheme (grants to M. Archer and S. Hand); the National Estate Grants Scheme, Queensland (grants to M. Archer and A. Bartholomai); the University of New South Wales; the Commonwealth Department of Environment, Sports and Territories; the Queensland National Parks and Wildlife Service; the Commonwealth World Heritage Unit; ICI Australia Pty Ltd; the Australian Geographic Society; the Queensland Museum; the Australian Museum; the Royal Zoological Society of New South Wales; the Linnean Society of New South Wales; Century Zinc Pty Ltd; the Riversleigh Society Inc.; the CREATE Fund of UNSW; and private supporters including P. Creaser, G. Johnstone, E. Clark, M. Beavis, M. Dickson, S. & J. Lavarack and S. & D. Scott-Orr. Vital assistance in the field has come from many hundreds of volunteers, staff and postgraduate students. We would like to thank P. Brewer, K. Black and two anonymous reviewers for their thoughtful and constructive comments on this manuscript. LITERATURE CITED Adams, L. A. 1919. A memoir on the phylogeny of the jaw muscles in recent and fossil vertebrates. Annals of the New York Academy of Science 27:51–166. Aplin, K. 1987. Basicranial anatomy of the early Miocene diprotodontian Wynyardia bassiana (Marsupialia: Wynyardiidae) and its implications for wynyardiid phylogeny and classification; pp. 369–391 in 991 M. Archer, (ed.), Possums and Opossums: studies in evolution. Surrey Beatty & Sons Pty Ltd and Royal Zoological Society, Sydney. Aplin, K. 1990. Basicranial regions of diprotodontian marsupials: anatomy, ontogeny and phylogeny. PhD dissertation, University of New South Wales, Sydney, New South Wales, Australia, 390 pp. Aplin, K., and M. Archer. 1987. Recent advances in marsupial systematics with a new syncretic classification; pp. xv–lxxii in M. Archer (ed.), Possums and Opossums: studies in evolution. Surrey Beatty & Sons Pty Ltd and Royal Zoological Society, Sydney. Archer, M. 1976. The basicranial region of marsupicarnivores (Marsupialia), interrelationships of carnivorous marsupials, and affinities of the insectivorous peramelids. Zoological Journal of the Linnean Society London 59:217–322. Archer, M. 1978. The nature of the molar –premolar boundary in marsupials and a reinterpretation of the homology of marsupial cheekteeth. Memoirs of the Queensland Museum 18:31–35. Archer, M., S. J. Hand, and H. Godthelp. 1991. Riversleigh. The Story of Animals in Ancient Rainforests of Inland Australia. Reed Books, Balgowlah, New South Wales, 264 pp. Archer, M., R. Arena, M. Bassarova, K. Black, J. Brammall, B. Cooke, P. Creaser, K. Crosby, A. Gillespie, H. Godthelp, M. Gott, S. J. Hand, B. Kear, A. Krikmann, B. Mackness, J. Muirhead, A. Musser, T. Myers, N. Pledge, Y. Wang, and S. Wroe. 1999. The evolutionary history and diversity of Australian mammals. Australian Mammalogy 21:1–45. Arena, D. 2004. The geological history and the development of the terrain at the Riversleigh World Heritage Area during the middle Tertiary. PhD dissertation, University of New South Wales, Sydney, New South Wales, Australia, 275 pp. ´ Asher, R. J., I. Horovitz, and M. R. Sanchez-Villagra. 2004. First combined cladistic analysis of marsupial mammal interrelationships. Molecular Phylogenetics and Evolution 33:240–250. Barnosky, A. D., P. L. Koch, R. S. Feranec, S. L. Wing, and A. B. Shabel. Assessing the causes of Late Pleistocene extinctions on the continents. Science 306:70–75. Bassarova, M. 2006. Taphonomic and palaeoecological investigations of Riversleigh Oligo-Miocene fossil sites – mammalian palaeocommunities and their habitats. PhD dissertation, University of New South Wales, Sydney, New South Wales, Australia, 233 pp. Beck, R. M. D. 2008. A dated phylogeny of marsupials using a molecular supermatrix and multiple fossil constraints. Journal of Mammalogy 89:175–189. Beecher, R. M. 1977. Function and fusion at the mandibular symphysis. American Journal of Physical Anthropology 47:325–336. Beecher, R. M. 1979. Functional significance of the mandibular symphysis. Journal of Morphology 159:117–130. Black, K., and M. Archer. 1997. Nimiokoala gen. nov. (Marsupialia, Phascolarctidae) from Riversleigh, northwestern Queensland, with a revision of Litokoala. Memoirs of the Queensland Museum 41:209–228. Cork, S. J., and G. D. Sanson. 1991. Digestion and nutrition–a review; pp. 129–144 in A. K. Lee, K. Handasyde, and G. D. Sanson (eds.), The Biology of the Koala. Surrey Beatty and Sons, Sydney, Australia. Creaser, P. 1997. Oligocene-Miocene sediments of Riversleigh: the potential significance of topography. Memoirs of the Queensland Museum 41:303–314. Crompton, A. W., and D. Lieberman. 2005. Biomechanics of the fused mandibular symphysis in placental and marsupial herbivores. American Journal of Physical Anthropology S40:89–90. Crosby, K., and C. A. Norris. 2003. Periotic morphology in the Trichosurin possums Strigocuscus celebensis and Wyulda squamicaudata (Diprotodontia, Phalangerida) and a revised diagnosis of the tribe Trichosurini. American Museum Novitates 3414:1–14. Davison, C. V., and W. G. Young. 1990. The muscles of mastication of Phascolarctos cinereus (Phascolarctidae: Marsupialia). Australian Journal of Zoology 38:227–240. Degabriele, R., and T. J. Dawson. 1979. Metabolism and heat balance in an arboreal marsupial, the koala (Phascolarctos cinereus). Journal of Comparative Physiology 134:293–301. Drummond, A. J., S. Y. W. Ho, M. J. Phillips, and A. Rambaut. 2006. Relaxed phylogenetics and dating with confidence. PLoS Biology 4:e88. Evans, H. E. 1993. Miller’s anatomy of the dog. W.B. Saunders, Philadelphia. Fleischer, G. 1978. Evolutionary principles of the mammalian middle ear. Advances in Anatomy, Embryology and Cell Biology 55:1–70. 992 JOURNAL OF VERTEBRATE PALEONTOLOGY, VOL. 29, NO. 4, 2009 Rougier, G. W., J. R. Wible, and M. J. Novacek. 1998. Implications of Deltatheridium specimens for early marsupial history. Nature 396:459–463. ´ Sanchez-Villagra, M. R., and J. R. Wible. 2002. Patterns of evolutionary transformation in the petrosal bone and some basicranial features in marsupial mammals, with special reference to didelphids. Journal of Zoological Systematics and Evolutionary Research 40:26–45. Scapino, R. 1981. Morphological investigation into functions of the jaw symphysis in carnivorans. Journal of Morphology 167:339–375. Schleich, C. E., and A. I. Vassallo. 2003. Bullar volume in subterranean and surface-dwelling caviomorph rodents. Journal of Mammalogy 84:185–189. Schon, M. A. 1971. The anatomy of the resonating mechanism in Howling monkeys. Folia Primatologica 15:117–132. Smith, M. 1987. Behaviour and ecology; pp. 56–78 in L. Cronin (ed.), Koala: Australia’s Endearing Marsupial. Reed Books, Sydney. Smith, M. 1980. Behaviour of the koala, Phascolarctos cinereus (Goldfuss) in captivity. 6. Aggression. Australian Wildlife Research 7:177–190. Springer, M. S. 1987. Lower molars of Litokoala (Marsupialia: Phascolarctidae) and their bearing on phascolarctid evolution; pp. 319–325 in M. Archer (ed.), Possums and Opossums: studies in evolution. Surrey Beatty & Sons Pty Ltd and Royal Zoological Society, Sydney. Springer, M. S., and M. O. Woodburne. 1989. The distribution of some basicranial characters within the Marspialia and a phylogeny of the Phalangeriforms. Journal of Vertebrate Paleontology 9:210–221. Stirton, R. A., R. H. Tedford, and M. O. Woodburne. 1967. A new Tertiary formation and fauna from the Tirari Desert, South Australia. Records of the South Australian Museum 15:427–461. Tedford, R. H., and M. O. Woodburne. 1987. The Ilaridae, a new family of vombatiform marsupials from Miocene strata of South Australia and an evaluation of the homology of molar cusps in the Diprotodontia; pp. 401–418 in M. Archer (ed.), Possums and Opossums: studies in evolution. Surrey Beatty & Sons Pty Ltd and Royal Zoological Society, Sydney. Voss, R. S., and S. A. Jansa. 2003. Phylogenetic studies on didelphid marsupials II. Nonmolecular data and new IRBP sequences: separate and combined analyses of didelphine relationships with denser taxon sampling. Bulletin of the American Museum of Natural History 276:1–82. Webster, D. B. 1962. A function of the enlarged middle-ear cavities of the kangaroo rat, Dipodomys. Physiological Zoology 35:248–255. Webster, D. B. 1966. Ear structure and function in modern mammals. American Zoologist 6:451–466. ¨ ¨ Wegner, R. N. 1964. Der Schadel des Beutelbaren (Phascolarctos cinereus Goldfuss 1819) und Seine Umformung Durch Lufthaltige Nebenohlen. Abhandlungen Der Deutschen Akademie Der Wissenschaften Zu Berlin 4:7–83. Weisbecker, V., and M. Archer. 2008. Parallel evolution of hand anatomy in kangaroos and vombatiform marsupials: functional and evolutionary implications. Palaeontology 51:321–338. Wible, J. R. 1990. Petrosals of late Cretaceous marsupials from North America, and a cladistic analysis of the petrosal in therian mammals. Journal of Vertebrate Paleontology 10:183–205. Wible, J. R. 2003. On the cranial osteology of the short-tailed opossum Monodelphis brevicaudata (Didelphidae, Marsupialia). Annals of Carnegie Museum 72:137–202. Woodburne, M. O., R. H. Tedford, M. Archer, and N. S. Pledge. 1987. Madakoala, a new genus and two species of Miocene koalas (Marsupialia: Phascolarctidae) from South Australia, and a new species of Perikoala; pp. 293–317 in M. Archer (ed.), Possums and Opossums: studies in evolution. Surrey Beatty & Sons Pty Ltd and Royal Zoological Society, Sydney. Wroe, S., J. Brammal, and B. N. Cooke. 1998. The skull of Ekaltadeta ima (Marsupialia, Hypsirymnodontidae?): an analysis of some marsupial cranial features and a re-investigation of propleopine phylogeny, with notes on the inference of carnivory in mammals. Journal of Palaeontology 72:738–751. Young, W. G., and S. K. Robson. 1987. Jaw movement from microwear on molar teeth of the koala Phascolarctos cinereus. Jounal of Zoology (London) 213:51–61. Young, W. G., C. K. P. Brennan, and R. I. Marshall. 1990. Occlusal movements of the brushtail possum, Trichosurus vulpecula, from microwear on the teeth. Australian Journal of Zoology 38:41–51. Submitted November 26, 2008; accepted January 17, 2009. Hilloowala, R. A. 1975. Comparative anatomical study of the hyoid apparatus in selected primates. American Journal of Anatomy 142:367–384. Hilloowala, R. A. 1976. The primate hyolaryngeal apparatus and herbivorous modifications. Acta Anatomica 95:260–278. Hogue, A. S., and M. J. Ravosa. 2001. Transverse masticatory movements, occlusional orientation, and symphyseal fusion in selenodont artiodactyls. Journal of Morphology 249:221–241. ´ Horovitz, I., and M. R. Sanchez-Villagra. 2003. A morphological analysis of marsupial mammal high-level phylogenetic relationships. Cladistics 19:81–212. Hume, I. D. 1999. Marsupial Nutrition. Cambridge University Press, Cambridge, 446 pp. Hylander, W. L., M. J. Ravosa, C. F. Ross, and K. R. Johnson. 1998. Mandibular corpus strain in primates: further evidence for a functional link between symphyseal fusion and jaw-adductor muscle force. American Journal of Physical Anthropology 107:257–271. Kirsch, J. A.W., F. J. Lapointe, and M. S. Springer. 1997. DNA-hybridisation studies of marsupials and their implications for metatherian classification. Australian Journal of Zoology 45:211–280. Lanyon, J. M., and G. D. Sanson. 1986. Koala (Phascolarctos cinereus) dentition and nutrition. II. Implications of tooth wear in nutrition. Journal of Zoology 209:169–181. Long, J., M. Archer, T. F. Flannery, and S. J. Hand. 2002. Prehistoric Mammals of Australia and New Guinea. University of New South Wales Press, Sydney, 244 pp. Louys, J., K. Black, M. Archer, S. J. Hand, and H. Godthelp. 2007. Descriptions of koala fossils from the Miocene of Riversleigh, northwestern Queensland and implications for Litokoala. Alcheringa 31:99–110. Luckett, W. P. 1993. An ontogenetic assessment of dental homologies in therian mammals; pp. 182–204 in F. S. Szalay, M. J. Novacek, and M. C. McKenna (eds.), Mammal Phylogeny: Mesozoic Differentiation, Multituberculates, Monotremes, Early Therians and Marsupials. Springer-Verlag, New York. MacIntyre, G. T. 1972. The trisulate petrosal pattern of mammals; pp. 275–303 in T. Dobzhansky, M.K. Hecht, and W.C. Steere (eds.), Evolutionary Biology 6. Plenum Press, New York. Martin, H. A. 1998. Tertiary climatic evolution and the development of aridity in Australia. Proceedings of the Linnean Society of New South Wales 119:115–136. Martin, H. A. 2006. Cenozoic climatic change and the development of the arid vegetation in Australia. Journal of Arid Environments 66:533–563. Martin, R. 1992. Of koalas, tree-kangaroos and men. Australian Natural History 24:22–31. Mitchell, P. J. 1991a. Social behaviour and communication of koalas; pp. 151–170 in A. K. Lee, K. Handasyde, and G. D. Sanson (eds.), The Biology of the Koala. Surrey Beatty and Sons, Sydney, Australia. Mitchell, P. J. 1991b. The home ranges and social activity of koalas - a quantitative analysis; pp. 171–187 in A. K. Lee, K. Handasyde, and G. D. Sanson (eds.), The Biology of the Koala. Surrey Beatty and Sons, Sydney, Australia. Muizon, C. de 1999. Marsupial skulls from the Deseadan (late Oligocene) of Bolivia and phylogenetic analysis of the Borhyaenoidea (Marsupialia, Mammalia). Geobios 32:483–509. Munson, C. J. 1992. Postcranial descriptions of Ilaria and Ngapakaldi (Vombatiformes, Marsupialia) & the phylogeny of the vombatiformes based on postcranial morphology. University California Publications in Zoology 125:1–99. Murray, P., R. Wells, and M. Plane. 1987. The cranium of the Miocene thylacoleonid, Wakaleo vanderleueri click go the shears – a fresh bite at thylacoleonid systematics; pp. 433–466 in M. Archer, (ed.), Possums and Opossums: studies in evolution. Surrey Beatty & Sons Pty Ltd and Royal Zoological Society, Sydney. Nagy, K. A., and R. W. Martin. 1985. Field metabolic rate, water flux, food consumption and time budget of koalas, Phascolarctos cinereus (Marsupialia: Phascolarctidae), in Victoria. Australian Journal of Zoology 33:655–665. Norris, C. A. 1994. The periotic bones of possums and cuscuses: cuscus polyphyly and the division of the marsupial family Phalangeridae. Zoological Journal of the Linnean Society 111:73–98. Pocock, R. I. 1916. On the hyoidea of the lion (Felis leo) and related species of Felidae. Annals and Magazine of Natural History 18:222–229. Ride, W. D. L. 1964. A review of Australian fossil marsupials. Journal of the Royal Society of Western Australia 47:97–131.