WARENDJA WAKEFIELD1, A NEW GENUS OF WOMBAT (MARSUPIALIA, VOMBATIDAE) FROM PLEISTOCENE SEDIMENTS IN McEACHERNS CAVE, WESTERN VICTORIA

2 By J. H. Hope 1 and H. E. Wilkinson

University, 1 Department of Prehistory, Research School o\ Pacific Studies, Australian National Canberra. 2 Geological Survey Office, Department of Minerals and Energy, Bendigo. Abstract Cave, Two mandibles and six isolated teeth recovered from Pleistocene sediments in McEacherns wombat. It combines western Victoria, represent a new genus and species of a morphologically primitive unlused symphysis, unrooted, slightly bilobed molars showing little curvature, with a slender ramus, weak development of the masseteric and pterygoid fossae, a low-set articular condyle, and a broad coro- such as noid process. The associated faunal assemblage includes typical later Pleistocene taxa, Modern taxa, Zvgomaturus triiobus, Sthenurus pilli, S. ef. occidentalis and Thylacoleo carnifex, forest dominated by Rattus fuscipes, Antechinus stuartii and Perameles nasuta, indicate wet sclerophyll conditions.

Introduction Systematica diagnoses are not intended to McEacherns Cave lies about 550 m south of The following at each taxonomic level, but the Glenelg River, in the Lower Glenelg Na- be exhaustive to those features which can tional Park, southwestern Victoria. In 1963, A. rather are restricted on the type and referred specimens C. Beauglehole and F. Davies discovered fossil be observed mammal bones in the cave and commenced of Warendja wakefieldi. excavations in the floor sediments. preliminary Superorder MARSUPIALIA Illiger, 1811 in 1964 and 1965 by This work was continued Order DIPROTODONTA Owen, 1866 who carried out ex- the late N. A. Wakefield, Family VOMBATIDAE Burnett, 1830 tensive excavations. In 1967, Wakefield pub- Diagnosis: Distinguished from all other Mar- lished a preliminary report describing the cave the presence of four bilobed, faunal re- supialia by and its sediments, and listing the hyposodont, labially curved and open-rooted mains from the initial excavations in 1963 and molars and open-rooted 11 and P3. 1964. The bulk of the material, only partly sorted, was deposited in the National Museum of Victoria in 1972, after Wakefield's death. In Warendja gen. nov. the 1975, J. H. Hope commenced sorting Type species; Warendja wakefieldi sp. nov. Wakefield Collection, and followed this by Known distribution: Pleistocene, western Vic- further excavations at McEacherns Cave in toria. November-December 1977. In May 1978, the Diagnosis; Distinguished from all other vom- remains of a previously undescribed genus of batids by small, sub-rectangular, minimally wombat were found in the material from the bilobed molars, which show little longitudinal 1964-5 excavation. The specimens have been curvature; smooth, unfused mandibular sym- registered in the palaeontological collection of physis; the combination of a broad ascending National Museum of Victoria. the ramus and well developed coronoid process Terminology and Measurements with poorly-developed masseteric and ptery- goid fossae; and the close approximation of the Mandibular terminology follows Stirton the plane of the mandibular Archer (1978). articular condyle to (1967), and dental morphology permanent cheek teeth tooth row. In this latter system, the Woiwuro and M5. Etymology: In the language of the of vombatids are P3, M2, M3, M4 people of the Melbourne district, warendj Measurements were made with a Mitutoyo dial means wombat (Hercus, 1969). caliper, to 0.1 mm.

109 Memoirs of the National Museum Victoria, No. 43, 1982. 110 J. H. HOPE AND H. E. WILKINSON

Warendja wakefieldi sp. nov. Table 1

Plate 3, Figure 2 Measurements (mm) of the mandible of Warendja wakefieldi Holotype: NMV P48980, right mandibular fragment with P 3 , M 2 -5, and alveolus for Ii. P48980 P48982 The articular condyle is present but the upper portion of the coronoid process is broken away. Length of mandible 92.5 est. 100 Referred specimens: NMV P48982, right man- Greatest width of mandible 21.5 21.2 Depth of ramus below mid M 4 20.3 24.2 with -5 and the basal dibular fragment P 3 , M 2 Length of diastema 16.8 16.9 portion of Ii . The articular condyle and part of Length from alveolus of incisor to alveolar margin of 51.2 57.6 the coronoid process are not preserved. NMV posterior M 5 Width of ascending ramus 33.0 P48981, isolated right M 5 . NMV P165428, Condyle — transverse width in 1 3 vertical orientation 12.9 isolated left I . NMV P165429, isolated left P . 4 maximum transverse width 14.1 NMV P165430, isolated left M . NMV maximum length in an- 4 PI 65431, isolated right M . NMV PI 65432, teroposterior direction 7.8 3 Height from ventral border of isolated left M . ascending ramus to sigmoid notch 33.1 Type locality: McEacherns Cave, Lower Glenelg National Park, western Victoria. The concavity beneath the anterior root of the specimens were among material in the National ascending ramus and a convexity beneath M 4 . Museum of Victoria derived from the sediments In both mandibles the symphysis is elliptical in identified as Pleistocene by Wakefield (1967). shape. Its major axis is about 34 mm long in P48980 was collected 21 December 1964, from P48982 (the posterior end is broken in P48980), level 'R\ by N. A. Wakefield. P48982 was col- and in both specimens dips posteriorly at an lected 4-5 September 1964, from a section 24 to angle of 30° with respect to the dorsal edge of 27 feet northwest of the cave entrance, to 1 the horizontal ramus. The length of the minor inch below 'ML\ by N. A. Wakefield. P48981 axis is 9.4 mm in P48980 and 10.8 mm in was collected 22nd May 1964, from level *B2' in P48982. The surface of the symphysis is only a section 18 to 21 feet northwest of the cave en- slightly roughened in both specimens, and there trance, by N. A. Wakefield. P165428-32 were is no indication of symphyseal fusion between collected from level 'Q\ by N. A. Wakefield. left and right mandibles. Diagnosis: That of the genus until other species This condition could be taken as evidence are described. that both mandibles were juvenile, but P48980 Etymology: Named in honour of the late Nor- is certainly fully adult. A fortuitous break be- the full man Arthur Wakefield. tween P 3 and M 2 allowed inspection of length of these teeth, which have almost iden- Description tical top and bottom dimensions. In juvenile Mandible: (Table 1): The mandible is wombats, all teeth are distinctly conical, with remarkably smaller and more delicately built proximal dimensions greater than distal (wear than in all other known vombatids. In P48980 surface). The disparity reduces progressively especially, and to only a slightly lesser degree in until adulthood is reached. P48982 is an older P48982, this is emphasised by the swept-back adult specimen, showing a greater amount of appearance of the ascending ramus, the tooth wear. Therefore the lack of symphyseal anterior border of which is inclined at a lesser fusion is a valid diagnostic character. angle than in any other described species. The The diastema is simple in morphology and lightly built appearance of the mandible is due relatively short. In P48980 the diastemal to the presence of very shallow masseteric and margin is a sharp crest, folded inwards slightly, pterygoid fossae, again a feature not found in above a longitudinal groove on the labial sur- any other wombat species. face of the ramus. P48982 is similar, but is not When viewed laterally, the ventral border of as well preserved. The mental foramen in the mandible forms a very shallow curve with a P48982 is large, and lies below the anterior NEW WOMBAT GENUS II

alveolar edge of P3 . In P48980 the mental P48980 it probably did not extend as far as the foramen lies anterior to P3 , below the diastema. most posterior point on the condyle. A second, smaller foramen lies posterior to There is no digastric fossa present on either this, below the posterior half of P-$. mandible. Instead, the lingual surface of the In both mandibles the anterior root of the horizontal ramus forms a smooth convexity, ascending ramus is opposite the posterior lobe with its highest point immediately behind the of Ms. Its anterior border sweeps back in a posterior limit of the symphysis. This bulge is gentle convex curve. Although the top of the caused by the large open root of the lower in- coronoid process is not preserved in either cisor, which in P48982 can be seen through a specimen, it is clear that it, and the ascending break in the bone. ramus, are larger relative to the postero-ventral width of the ramus than in any other wombat. Dentition (Tables 2, 3): The total length of the molar row in the holotype P48980 is 32.5 Angle a as defined in Figure 1 is 56° in mm (occlusal measurement), and the corresponding P48980, and 59° in P48982. The condyle is figure for the referred specimen P48982 is 37.8 preserved only in P48980. It is small and ovoid, These figures are significantly smaller than its transverse width about twice its length mm. for any other wombat currently accepted as a antero-posteriorly. It is set very low, close to valid species. All the teeth are open rooted. As the plane of the tooth row, as indicated by a in Lasiorhinus there is no measurement of 18° for angle b as defined in Vombatus and distinct cemento-enamel junction. The cemen- Figure 1 . The articular surface of the condyle is tum continues over the enamel up the column on the lingual side, approaching the line of the so in most tooth row. of the tooth to the occlusal surface, cases enamel is visible only as a rim at the oc- The masseteric fossa is well-preserved in both clusal surface. With isolated teeth, the distribu- mandibles. It is very shallow in P48980, deeper tion of enamel can be determined not only by in P48982. The deepest area of the concavity its appearance at the occlusal surface, but by lies immediately behind the anterior border of the presence of brown and white transverse the ascending ramus, indicating that the main markings, indicating zones of enamel forma- area of insertion of the masseteric muscle was tion, on the columns of the tooth, towards the higher and more anterior than in other wom- root. bats. The posterior masseteric eminence is well developed in P48982, with a sharp up-turned U: The incisor is represented by a broken basal edge; in P48980 this outer flange is smooth and portion in P48982, and by an almost complete weakly developed. There are two foramina alveolus in P48980. In P48982 it is a simple, within the masseteric fossa in P48982, but only ovate tooth, much deeper than wide at its one in P48980. broken end. The alveolus in P48980 is similar in The pterygoid fossa is also very shallow for a shape. The enamel is restricted to the ventral root ex- vombatid. In P48980 it forms a shallow lateral and lower labial surfaces. The open concavity with no ventral floor, due to the lack tends back to beneath the mid-point of Mj« flange the of development of the lingual of P s : The premolar is sub-triangular in occlusal fossa. In P48982, this flange is better outline, weakly bilobate, and has its long axis developed, and the fossa is deeper with a flat parallel to the antero-posterior axis of the man- floor ventrally. The height of the ventral rim of dible. It is slightly offset lingually to the molar the mandibular foramen above the ventral row, and the posterior apex of the triangular border of the horizontal ramus is 9.8 mm in occlusal outline is adpressed against the antero- In P48980, and 10.1 mm in P48982. P48980 lingual corner of M 2 . In P48980 the tooth is this foramen is simple and rounded; in P48982 nearly vertical, but in P48982 it is curved centre of is restricted it is part of a large depression in the antero-posteriorly. Enamel to the the pterygoid fossa, and the bone posterior to it anterior rim of the anterior lobe of Pi in angular is fenestrated. In both mandibles the P48980; in P48982, enamel occurs only on the preserved process is broken but in the better anterior labial surfaces of both the lobes. There 112 J. H. HOPE AND H. E. WILKINSON

Table 2 Table 3 Measurements (mm) of lower teeth of Measurements (mm) of isolated teeth of unless Warendja wakefieldi (occlusal, unless Warendja wakefieldi (occlusal, otherwise stated) otherwise stated)

P48980 P48982 I' (PI 65428) depth at proximal end of wear facet 5.9 37.8 width at proximal end of wear facet 4.0 Total length P 3 -M 5 32.3 40.8 depth at open root 5.4 Total length P 3 -M 5 t 35.5 width at open root 3.7 P length 5.3 6.6 3 length of wear facet 9.2 width of anterior lobe 2.4 2.9 width of wear facet 3.9 width of posterior lobe 3.1 3.7 3 6.4 height above alveolar margin of P (PI 65429) length 4.0 posterior crest 4.0 6.1* width of anterior lobe 4.5 height above alveolar margin of width of posterior lobe notch 2.2 5.3 M 4 (PI 65430) length 7.7 difference in elevation between width of anterior lobe 6.5 notch and posterior crest 1.8 0.8 width of constriction 5.2 M; length 7.4 8.2 width of posterior lobe 6.2 width of anterior lobe 3.9 4.7 M 4 (PI 65431) width of constriction 3.5 4.6 width of anterior lobe 6.5 width of posterior lobe 5.1 5.6 3 (P165432) length 7.8 height above alveolar margin of M 4.4 9.3 5.6 crest of anterior lobe M 5 (P48981) length height above alveolar margin of width of anterior lobe 4.3 posterior lobe 2.5 6.2 width of constriction 4.1 difference in elevation between width of posterior lobe 4.9 upwards projection of anterior lobe, and occlusal surface of the posterior lobe 2.3 3.1 distinct bilobation. The lobes are almost Mi length 7.4 8.6 quadrangular in cross-section, with rounded width of anterior lobe 4.4 5.5 corners. A thin rim of enamel extends around width of constriction 4.2 5.2 lingual and labial edges of the occlusal sur- width of posterior lobe 5.5 6.2 the face of the teeth. It is absent anteriorly and M.i length 6.8 8.7 width of anterior lobe 5.0 5.8 posteriorly on each tooth. The anterior lobe of width of constriction 4.0 5.0 is the M 2 is narrower than the posterior, which width of posterior lobe 4.8 6.0 most nearly cylindrical of any lobe of the four 7.4 M 5 length 6.0 width of anterior lobe 4.2 5.1 molars. The anterior lobe of M 2 is distinctly width of constriction 3.7 4.8 smaller in both length and width. It is also width of posterior lobe 4.0 4.8 distinguished by a peculiar wear pattern, 1, width 5.2t 5.5 related to that of the premolar. Whereas the oc- depth 8.3t 8.0 _ posterior clusal surfaces of M 3 5 , and of the * Not strictly comparable to P48980 because alveolar lobe of M 2 , are nearly horizontal, the occlusal margin appears to be more broken away. surface of the anterior lobe of M 2 rises an- t Alveolar measurements. teriorly at a sharp angle. When viewed laterally,

is a distinctive pattern of wear on both the anterior lobe of M 2 in P48982 protrudes as premolars. In P48980 this has resulted in a deep a sharp, high crest from the nearly horizontal notch on the occlusal surface running dia- occlusal surface of the tooth row. In P48980 the gonally from antero-lingual to postero-labial, equivalent high crest is made up of both the

leaving the anterior and posterior rims of the anterior lobe of M 2 and the posterior lobe of

tooth standing as sharp crests. In P48982, at a P 3 .

later stage of tooth-wear, the occlusal surface is In common with other vombatids, M 3 is the saddle-shaped, with the enamel-less posterior widest molar, its posterior lobe being the widest rim left as a sharp point. part of the molar row. The lobes are more

M2-5: The occlusal outline of all the molars is nearly equi-dimensional than those of M 2 , and sub-rectangular, with slightly developed, but the occlusal outline is almost subrectangular. NEW WOMBAT GENUS 13

Fig. 1. Schematic diagram of the mandible of Warendja (dorsal) and labial faces. There is an elongate, illustrating the method of measuring in- wakefieldi ovate wear surface on the posterior face of the clination of anterior border of ascending ramus (angle a) and elevation of condyle (angle b). tooth. The occlusal surface is slightly asym- metrical, being flattened on the lingual side, The anterior lobe in both specimens retains a where it abuts the right incisor. 3 trace of a transverse loph on the lingual side, P : A single tooth, P165429, is tentatively iden-

and this is also visible on the anterior lobes of tified as a left upper premolar. It is similar to the lower premolars in being longitudinally, M 4 and M 5 in P48982. In M 4 the two lobes are nearly equal in size and the tooth is the most rather than laterally curved (as the molars are), nearly rectangular of all the molars. The last and in having an extensive exposure of enamel on the convex (anterior) face. It is, however, molar, M 5 , is the smallest, and tapers markedly posteriorly. The isolated tooth, P48981, ap- larger and more ovate than the lower premolars. It is bilobed, with the labial groove pears to be a right M 5 , in showing the bilobed form characteristic of the molars, but with one deeper than that on the lingual face. The depth lobe clearly smaller than the other. It is most of enamel on the anterior face is 5.5 mm below the occlusal surface. Enamel extends onto the similar to M 5 of P48980. Wear on M3-5 is faces transverse and has left the lingual rim of the labial and lingual of the anterior lobe, and molars standing higher than the labial rim. also occurs on the antero-labial face of the The remaining five isolated teeth are inter- posterior lobe. preted as representing elements of the upper The upper premolar should occlude with

. In dentition. both P 3 and the anterior lobe of M 2 both

1 lower dentitions available, differential wear has I : A left upper incisor, P165428, is attributed the anterior teeth, leaving the to Warendja because of its open root, but occurred on smaller size and different morphology from any anterior lobe of M 2 standing as a high crest. This suggests that the upper premolar occludes other wombat. The tooth is longitudinally that there is curved, laterally compressed and ovate in cross- mainly with the lower, and minimal

wear on the anterior lobe of 2 , except at the section. It tapers slightly from the distal to the M 2 proximal end. Enamel occurs on the anterior posterior of this lobe, where it meets M . The 114 J. H. HOPE AND H. E. WILKINSON

/-_ n

a

i !*+.

3 cm _L

Fig. 2. Isolated teeth of Warendja wakefieldi. NMV each upper molar is labial, as in other vom-

1 PI 65428, left l a, labial view; b, occlusal view. , batids, then the labial and lingual orientations NMV P165429, ? left P\ c, labial view; d, occlusal 4 are easily determined. The allocation of each view. NMV P165430, ? left M , e, occlusal view; f, posterior view. NMV P165432, ? left 3 oc- M , g, tooth to left or right, or to specific positions in clusal view; h, posterior view. NMV P48181, right the tooth row, is tentative, and has been based i, M 5 , occlusal view; j, posterior view. Orientation of occlusal views; a-anterior, 1-lingual. on the direction of curvature of the roots, by comparison with Vombatus and Lasiorhinus. occlusal surface of PI 65429 shows a saddle- These molars are distinguished by the asym- shaped wear pattern similar to that of the lower metry of their medial constriction. The lingual premolars, but less extreme. The wear is consis- (convex) face is barely constricted, with a tent with the argument that P 3 occludes mainly, shallow indentation running the length of the or only, with P 3 , but a complete maxillary root. The lingual enamel surface is virtually tooth row, with associated mandible, would be uninfected. In contrast, the labial (concave) needed to verify this. surface is more deeply constricted, to the degree Upper molars: Three isolated molars are at- found on both labial and lingual faces of the tributed to Warendja on the basis of size, and lower molars. their minimal degree of bilobation and cur- As in the lower molars, enamel is restricted to vature. They are larger than the lower molars, the labial and lingual surfaces of the teeth. In have an asymmetric occlusal outline, and are the only unbroken specimen, P165430, the more strongly curved, so are interpreted as labial edges of the anterior and posterior lobes upper molars. form sharp points, although this is more the On the assumption that the concave face of result of wear than of the structure of the tooth. NEW WOMBAT GENUS 115

Wear has left the labial rim of the tooth stand- in being sub-triangular in occlusal outline ing higher than the lingual rim. In P 165430 rather than ovoid to sub-rectangular. There is traces of transverse lophs can be seen on the oc- little affinity with the ovoid, more or less ob- clusal surface of the anterior and posterior liquely set lower premolars o\' Lasiorhinus and lobes. Vombatus. The sub-rectangular, sub-bilobate molars Discussion differ from all other wombats in their distinc- tive of The taxonomy of wombats at the generic occlusal outline and small degree longitudinal curvature, but are closest to level has been in a state of flux for many years. Phascolonus in general ap- One of us (HEW) presented a phylogeny of the and Ramsayia pearance. most noticeable differences be- family at the 1971 Brisbane ANZAAS, but cir- The are the cumstances have so far prevented publication tween Phascolonus and Warendja relatively small degree of medial constriction of this work, which was based on a comprehen- and the less rounded angles of the molars in the sive study of virtually all fossil wombat latter. It is easy to envisage an evolutionary material available in Australia at that time. The trend from the Warendja molar type to that of most recent published consideration of wombat Phascolonus by an increase in medial constric- phylogeny is that of Dawson (1981), who tion, and an increase in rounding of the lobes to recognised the following Pleistocene and give a more cylindrical appearance. There is no modern taxa: Vombatus ursinus, Phascolonus close affinity to the rounded triangular lobes of gigas, Ramsayia magna, Lasiorhinus medius, Lasiorhinus molars, and even less to the sharply and Lasiorhinus tatifrons, and suggested that angled lobes of Vombatus molars. the recently described Phascolonus temleyi The single tooth, considered to be a Pj, Archer & Wade 1976 should be referred to representing the Miocene genus Rhizophas- Ramsayia. In addition to the above, Wilkinson colonus, differs from all known wombats, in- (in prep) recognises Vombatus bucket ti, cluding Warendja, in having closed roots. Lasiorhinus krefftii and Phascolonus angus- Although not directly comparable with Waren- tidens as valid species of true wombats, and dja, as illustrated in Stirton et al. (1967), it docs regards the generic status of medius as proble- show some similarity to the molars of Warend- matical. Rhizophascolonus crowcrofti is the ja in the rounded cylindrical form of the lobes. only described vombatid genus restricted to the The simple cylindrical lower incisor of ovoid Tertiary (Stirton et al. 1967), but this differs cross-section is very similar to that of from all other vombatids in having a rooted Phascolonus angustidens (Wilkinson, in prep.), premolar. and is also similar in both shape and the extent Dawson's phylogenetic scheme is at con- of surface enamel to Phascolonus gigas, and siderable variance with that put forward by especially to Ramsayia (Dawson 1981, Wilkin- Wilkinson (1971), in prep), on which the discus- prep.). The upper incisor is also simple alter- son in sion below is largely based. However, the and is quite distinct from the broad flattened native views put forward by Dawson are also upper incisor of Phascolonus gigas. referred to. The larger of the two mandibles (P48982) is Generic status for Warendja wakefieldi is fully adult, so Warendja wakefieldi is con- justified by the so far unique combination of siderably smaller than any of the described the open-rooted, hypsodont teeth, with the wombat species. The nearest in size is the small gracile mandibular morphology. The teeth, Bass Strait Islands wombat, Vombatus ursinus, clearly vombatid in form, differ from all known but this is a scaled down version of the common vombatids in their minimal curvature, and their wombat of southeastern Australia, Vombatus weakly bilobed morphology. The lower pre- hirsutus, which is regarded by some authorities molar shows some similarity to those of (e.g. Dawson 1981) as a sub-species of ursinus. Phascolonus and Ramsayia, particularly in the Even if ursinus is accepted as a valid species, it antero-posterior orientation, incipient biloba- larger than W. wakefieldi. differs is still tion and size, relative to the molars, but 116 J. H. HOPE AND H. E. WILKINSON

The detailed morphology of the mandible of Table 4 Warendja reflects this small size. The slender Comparative measurements of the angle of in- jaw has shallow masseteric and pterygoid clination of the ascending ramus (angle a) and fossae, and a small articular condyle, compared of the elevation of the condyle (angle b). Angles with the deep pterygoid fossae, deep to very approximate only deep masseteric fossae, and wide, robust ar- ticular condyles in other vombatids. In Taxon a b Phascolonus and Ramsayia the mandible is VOMBATIDAE very robust, but deep and comparatively nar- Warendja wakefieldi row. In Lasiorhinus and Vombatus the man- NMV P48980 56° 18 c 59° dible is very broad, with thick, relatively NMV P48982 Vombatus ursinus 70° 30° shallow rami, which support very powerful Lasiorhinus latifrons 65° 35° Musculature. Warendja differs from all of Phascolonus gigas 1 75° 40° these in the delicate build of the mandible and DIPROTODONTIDAE 2 62° <34° the lesser development of the masseteric and Raemeotherium yatkolai Diprotodon optaturn 105° 65° pterygoid fossae. While this light build is to PHASCOLARCTIDAE reflection some extent probably just a of small Phascolarctos cinereus 75° 38° size, two features of the mandible are especially PHALANGERIDAE un-wombatlike. The size and smoothness of the Trichosurus vulpecula 65° 28° Phalanger orientalis 58° 10° mandibular symphysis indicate that the left and right mandibular rami never fused, as happens PETAURIDAE Pseudocheirus peregrinus 75° 38° in other wombats. The extremely low position MACROPODIDAE of the articular condyle relative to the plane of Bettongia leseuer 55° 25° 57° 22° the molar row is unique in vombatids, and in- Aepyprymnus rufescens 90° 40° deed unusual compared with most other Wallabia bicolor Macropus giganteus 75° 40° diprotodont families (Table 4). THYLACOLEONIDAE In general, the articulation of the jaw tends Thylacoleo carnifex 3 35° 10° to be in line with the tooth row in carnivores, PERAMELIDAE Isoodon obesulus 60° 30° but is much higher than the tooth row in her- Perame/es nasuta 48° 30° bivores. The high position increases the lever DASYURIDAE arm of the masseter muscle. The low angle in Dasyurus maculatus 60° 12° 62° 10° Warendja is more comparable to that found in Sarcophilus harrisi carnivores, such as Sarcophilus, Dasyurus and 1 Measured on Plate XL, Stirling (1913). Thylacoleo (Finch 1971), rather than her- 2 Figures quoted from Rich el at. 1978.

3 bivores. The only diprotodont groups which ex- Relative to occlusal plane of P 3 . hibit a low-set condyle are the phalangerids and potoroines. The jaw structure in rat-kangaroos unknown skull of Warendja. By analogy with such as Aepyprymnus and Bettongia differs in the skulls of Trichosurus and Phalanger, other ways from that in Warendja. The Warendja is likely to have had a more rounded phalangerids, Phalanger and Trichosurus, are cranium, with greater areas for insertion of the quite similar, however, not only in the position temporal muscles than in other wombats, and of the condyle, but in the relative proportions possibly some development of a sagittal crest. of the masseteric and pterygoid fossae and the In Vombatus and Lasiorhinus, the structure of ascending ramus. A comparative study of the the zygomatic arch is distinctive, with no jaw mechanics of these taxa might elucidate the development of the masseteric process, but

feeding habits of Warendja. rather a broad area excavated beneath it for the These characteristics of the mandible of insertion of the masseter muscle. Given the Warendja indicate a very different system of probable lesser development of the masseter in jaw musculature from other wombats, and this Warendja, the structure of the zygoma may has implications for the structure of the as yet have been more like that of the phalangerids, NEW WOMBAT GENUS 117

with the anterior zygomatic surface vertical lematical) is the link between Phascolonus- rather than horizontal (as in other wombats), Ramsayia and LasiorhinuSy from which Vom- and with a masseteric process. In summary, the batus has in turn been derived. In the light of skull of Warendja will probably look more like this, the provenance, probable age and faunal that of Trichosurus than any wombat. This is association of Warendja are of especial not to imply any special relationship between interest. phalangerids and Warendja\ these characters are probably plesiomorphic for diprotodonts Age and Palaeoenvironment generally. McEacherns Cave is formed in the Gambier

Dawson (1981) carried out a cladistic analysis Limestone, of Oligocene-Miocene age. It is ap- of the five vombatid taxa she recognised, using proximately 60 m long, along a NW-SE axis, a set of seven dental and palatal characters. The averages 3 m in width and 3 m in height and is resulting cladogram linked Vombatus ursinus connected to the surface by a vertical shaft 13 m with Phascolonus gigas, and Ramsayia magna deep and 1.5 m in diameter. The cave fill con- with Lasiorhinus fatifrons and L. medius. For sists of sandy fossiliferous units and extensive most of Dawson's dental characters, Warendja roof-fall. The cave itself and the early excava- exhibits the plesiomorphic state. The one excep- tions of the fossil-bearing sediments, in the tion is the depth of the lower incisor, where the northwest chamber of the cave, are described

apomorphic state, I 2 being deeper than wide, is by Wakefield (1967) and Link (1967). found in Warendja, Warendja is intermediate Wakefield recognised a series of stratigraphic with regard to the structure of the lower units in the sandy floor sediments. The oldest of premolar, but Dawson concluded that con- these, which he called the 'Pleistocene* vergence had occurred in this character among sediments, consist of a block of compacted grey vombatids. Dawson's cladistic analysis does not sands laid down on limestone boulders, part of clarify the relationship of Warendja to other an earlier roof-fall. Overlying this block is a

vombatid taxa, since it is plesiomorphic in most 'calcarenite' deposit, about 5-10 cm thick, con- observable characters. sisting of finely laminated layers of calcium car-

Warendja fits rather better into the as yet un- bonate beneath 5-20 cm of grey clay. Above published phylogenetic scheme of Wilkinson and on each side of the block of grey sand with (1971; in prep). As shown above, it is linked to its calcarenite capping are successive layers of the Phascolonus-Ramsayia group by premolar white, red and black sands, Wakefield's 'Recent' and molar morphology, and more particularly sediments. Laterally the white sands are in fact by the morphology of the lower incisor. It has at a lower level than the grey block, filling a almost nothing in common with either cavity formed by slumping and collapse. Lasiorhinus or Vombatus, apart from family The new excavations by Hope at McEacherns characters. The suggestion by Dawson that Cave confirm this basic stratigraphy, although these latter genera have evolved separately, and some details are at variance with Wakefield's gained similarity by convergence, cannot be report, and the stratigraphy of his 'Recent sustained when skeletal characters are con- sands' is probably more complicated than he sidered. The skulls of Vombatus and Lasior- recognised. hinus are basically similar, but readily distin- Wakefield obtained only one radiocarbon guished, as are the skeletons of K ursinus and date from the site based on 373.4 g of mammal the uppermost layers of L. latifrons. However, the skeleton of L. bones extracted from about 0.5 cm below the krefftii blurs the generic boundary to a remark- the grey block, layer. This material gave a date of able degree, which is more readily explained by calcarenite common ancestry than by convergence. 15,200±320 years before present (GaK-509). Warendja probably represents an ancestral This date for the top of the grey sand, along absence of any element of the extinct vombatid type, from which the Phascolonus- with the Ramsayia group evolved. Thascolomys' Pleistocene fauna from the overlying white and sands (except as reworked material) was in- medius (whose generic assignment is prob- red 118 J. H. HOPE AND H. E. WILKINSON

terpreted by Wakefield as indicating that the A considerable period of time must be older grey sand was late Pleistocene in age, allowed for the almost complete filling of the while the remaining sediments were all cave by the older sediments, the deposition of Holocene in age. He suggested that the the laminated calcarenite layer, and for the calcarenite layer indicated a period of aridity, subsequent removal of much of these attributed to 'a mid-Holocene warm-arid sediments, before the younger phase of deposi- period'. Link (1967), however, argued that the tion began, sometime before 15,000 years ago. calcarenite might have been formed by frost Further, given the relative positions of the rem- brecciation of the walls of the cave during an nants of the older sediments, the excavated arid and possibly cold phase about 8-11,000 block of grey sediment may in fact date from years before present. the earlier stages of the older depositional The results of the 1977 excavations at event. An absolute age for the grey sediments is McEacherns Cave suggest that the sedimentary likely to prove impossible to obtain, but a sequence is in fact much older. Preliminary hypothetical age for the calcarenite deposit and radiocarbon dates indicate that the younger the sediment removal can be suggested, on the sands were deposited between 15,000 and 2000 assumption that these are related to changes in years before present. A sample of the laminated regional watertable. calcarenite layer has also been dated, and this During the last interglacial maximum gave an age of 28,580 ±850 years (ANU-2030). episode, from about 135-120,000 years ago, the

It was expected that this date might be similar maximum sea level relative to the New South

to Wakefield's original date of 15,200 years Wales coast is considered to have been 5± 1 m before present on bone from immediately under above present sea level (Chappell and Thorn the calcarenite layer, on the grounds that since 1978). In western Victoria, high sea level stands the bone date was likely to be affected by con- of approximately the same magnitude, 7.5 m tamination, the calcarenite and bone car- and 4 m (the later regarded as a stage during bonates might have equilibrated. the retreat of sea level), are also attributed to

It has proved difficult to obtain suitable char- this time period (Gill and Amin 1975). In- coal from the older grey sand for dating. One terstadial high sea levels younger than 118,000 sample has given a preliminary date of about years are thought to have been at least 8 m 24,000 years before present (ANU-2372), but lower than the 135-118,000 year levels (i.e. given the reversal between this and the below present sea level) (Chappell and Thorn calcarenite date, it is likely that some as yet 1978). unclear source of contamination is affecting McEacherns Cave lies approximately 550 m organic materials in this unit. south of the Glenelg River, which is tidal at this

There is some circumstantial evidence to sug- point, and the surface of the younger cave gest that the grey sands may be considerably deposit is approximately 6 m above the present older than the last Pleiostocene glaciation and river level. Therefore during the last interglacial thus probably beyond the range of radiocarbon maximum, the lower levels of the sediments in dating. There is clear evidence in the excavated the cave were likely to have been affected by a section that the grey block is in effect a residual, higher regional watertable, related to the higher

left stranded after a period of collapse, and sea level. subsequently surrounded by the younger Corroborative evidence may come from a sediments. Elsewhere in the cave, remnants of study of the calcarenite sediments, whose origin similar sediments containing a similar fauna, is unclear at the moment. Wakefield suggested including extinct species, are plastered on the that they were of aeolian origin, while Link cave walls up to 3 m above the top of the grey proposed frost action; a further possibility is block. The exact relationship of these sediments that they were laid down under standing water. to the grey block cannot yet be determined, but If in fact the formation of the calcarenite layer they may relate to the same event of cave fill, and the subsequent roof-fall and removal of most of which has now disappeared. much of the older fill were related to the rise Nl \Y WOMMM (it NHS 11^ and then lowering of the watertable, then these In the older Pleistocene sediments the launal events may have occurred during and at the end assemblage is dominated bj Rattus fuscipes, of the last interglacial, 135-118*000 years ago. Antechinus stuariii, Perameles rtasuta and the The grey sediments themselves would he even extinct kangaroo Sthenurus gilli, Othei extinct older. Pleistocene species present are SarQOphilus The specimens of Warettdjo wakefieldi were laniariuSt Zygomaturus trilobus, Thylacoleo

<:(. thriiientalts found among the material From Melaeherns carnifc\ t Sthcrmm.s and Pro Cave in the N. A. Wakefield Collection at the temnodon c\. brehus. I 'ornbafus sp. also occurs National Museum of Victoria. Some of tins in the assemblage, but only isolated teeth have material is poorly documented and is difficult to so far been recovered. Notable by their absence attribute with absolute confidence to a pai are many elements of the modern woodland ticular stratigraphic layer. In 1964-5 Wakefield and heath communities, such as Antechinus excavated five 3 ft wide sections from wall to swainsonii, Stninthopsis cf. leucopus* Isoodon wall across the width of the cave. These obesulus, Trichosurus vulpecula, Macropus and together extended from 15 ft to 30 ft from the rufogriseuSt Conilurus albipes Pseudomys cave entrance, and were numbered accordingly shortridgei, all o\' which aie abundant in the in the cave. Preliminary in 3 ft blocks, i.e. 15-18 ft, 18-21 ft etc. Most of younger sediments analyses o\ the material collected by Wakefield the material collected during I his phase of the excavation was labelled with the section in 1964-65, and by Mope in 1977 support these numbers, and also with the depth (torn the sur- faunal lists for both the older and younger face, but no record was kept of the lateral posi sediments. launal tion across the cave. Part o\' the collection, Wakefield interpreted the association however, was labelled using a code oi letters or from the older Pleistocene sediments as in numbers. Bones so labelled can be allocated a dicative Of wet sclerophyll forest, which todaj rain stratigraphic position only by comparison of occurs in areas o\' Victoria with an annual or more, such as the ( )t colour, preservation and faunal composition fall of 1000 1200 mm

highlands. I he with better documented material. way Ranges and the east -central in faunal composition between One specimen of Warendja, P4K982, is dramatic change sediments also supports clearly provenanced within the system. This is the older and youngei long hiatus in sedimentation labelled '24'-27\ 0"-l" below Ml.' (ML middle the hypothesis of a the cave. layer -ealearenite layer). Unfortunately it is within not possible to allocate the remaining Conclusion specimens to specific positions in the excava tion. The colour and degree o\ mineralisation Warendja wakefieldi is an enigma. The open- is clearly vombalid, but it of all specimens, and o\' the rest o\' the bone rooted dentition material with which they are associated, in- combines some characters that could be seen as dicates that they are derived from the older plesiomorphic, such as the mimimally bilobed Pleistocene unit. form of the molars, and the extent of enamel on such as the grooving on p and Wakefield (1967) lists the faunal assemblages I,, with others, 3 I,, which could be regarded ;»s recorded from the older Pleistocene sediments, the depth o\ The mandible is unique in vom and from the younger Pleistocene red and black apomorpluc. bears more resemblance to that of sands, based on material collected during the bat ids, and in phalangerids. It probably represents a initial excavations in 1963 and 1964. He plesiomorphic diprolodonl morphology. tcrpreled the faunal sequence from the younger have been no surprise if Warendja sands (now known to be from more than 15,000 It would as in- wakejieldi had turned up in a Miocene site, to about 2000 years before present) where its small size and primitive features dicating a change from a 'mallee' fauna, would have conveniently fitted it foi the role of through a fauna including some woodland

of ancestral wombat. Warendja, howevei , has species, to the modern dry sclerophyll forest been found in terrigenous sediments wilhm a (he district. 120 J. H. HOPE AND H. E. WILKINSON

cave eroded in Miocene limestone, in associa- Chappell, J. and Thom, B. C, 1978. Termination of last tion with typical Pleistocene and modern interglacial episode and the Wilson Antarctic surge hypothesis. Nature. 272: 809-810. species. Although circumstantial evidence sug- Dawson, L., 1981. The status of the taxa of extinct giant gests that the sediments from which Warendja wombats (Vombatidae: Marsupialia), and a considera- comes may be older than 120,000 years, there is tion of vombatid phylogeny. Aust. Mammalogy. 4: no reason to suppose they are significantly 65-79. Finch, E., 1971. Thylacoleo, marsupial lion or marsupial older, say, of early Pleistocene age. So far, sloth? Aust. Nat, Hist. 17(1): 7-11. Warendja is the only taxon recognised in the Gill, E. D. and Amtn, B. S., 1975. Interpretation of 7.5 site which has not been found elsewhere in Vic- and 4 metre Last Interglacial shore platforms in toria in a late Pleistocene context. Southeast Australia. Search. 6: 394-396. The palaeoenvironment suggested by the Hercus, L. A., 1969. The languages of Victoria: a late survey. Australian Institute of Aboriginal Studies, No. faunal assemblage associated with Warendja is 17. wet sclerophyll forest. The suite of modern Link, A. G., 1967. Late Pleistocene-Holocene climatic fluc- species in the assemblage occurs in such forest tuations; possible solution pipe-foiba relationships;

in Gippsland today. Warendja, with its and the evolution of limestone cave morphology. Zeits. Geomorph. 11: 117-145. primitive morphology, may be a conservative Rich, T. H., Archer, M., and Tedford, R. H., 1978. survivor from the forests of the Tertiary, linger- Raemeotherium yatkolai, gen. et sp. nov., a primitive ing into the Pleistocene in the forested areas of diprotodontid from the medial Miocene of South southeastern Australia. Australia. Mem. Nat. Mus. Vict. 39: 85-91. Stirling, E. C, 1913. On the identity of Phascolomys (Phascolonus) gigas, Owen, and Sceparnodon ram-

sayi, Owen, with a description of some parts of its Acknowledgements skeleton. Mem. R. Soc. S. Aust. 1: 127-178. The M. A. Ingram Trust supported the 1977 Stirton, R. A., 1967. The Diprotodontidae from the Ngapakaldi Fauna, South Australia, pp. 1-44 in Stir- excavations at McEacherns Cave, which were ton, R. A., M. O. Woodburne, and M. D. Plane, Ter- carried out by permission of the National Parks tiary Diprotodontidae from Australia and New Service of Victoria. A great many people have Guinea. Bull. Bur. Min. Resour. Geol. Geophys. Aust. assisted with the excavations in McEacherns 87. Stirton, R. A., Tedford, R. H., and Woodburne, Cave and with the subsequent processing of the M. O., 1967. A new Tertiary formation and fauna material, among them Ken Aplin, who also from the Tirari Desert, South Australia. Rec. S. Aust. provided some useful criticisms of this paper. Mus. 15: 427-462. Dragi Markovic took the photographs and Win Wakefield, N. A., 1967. Preliminary report on McEachern's Cave, S.W. Victoria. Vict. 84: Mumford drew Figures 1 and 2. Nat. 363-383. Permission for H. E. Wilkinson to publish Wilkinson, H. E., 1971. Speciation in the family Vom- was given by the Acting Director of Geological batidae (Marsupialia). Aust. N.Z. Assoc. Adv. Sci. Survey, Victoria. Congr. Abstract. Sec. 3, p. 8. Wilkinson, H. E., in prep. The identity of the fossil wombat, Phascolomys angustidens De Vis. Wilkinson, H. E., in prep. Speciation in the family Vom- batidae (Marsupialia). References Archer, M., 1978. The nature of the molar-premolar Explanation of Plate boundary in marsupials and a reinterpretation of the PLATE 3 homology of marsupial cheekteeth. Mem. Qd. Mus. Warendja wakefteldi. 18: 157-164. NMV P48980, holotype, right man- dible, McEacherns Cave, Victoria, fig. 1, labial view, x 1; Archer, M. and Wade, M., 1976. Results of the Ray E. fig. 2, lingual view, x 1; fig. 5, occlusal view, x 1. NMV Lernley Expeditions, Part 1 . The Allingham Formation P48982, right mandible, McEacherns Cave, Victoria, fig. 3, and a new Pliocene vertebrate fauna from Northern labial view, x 1; fig. 4, lingual view, x 1; fig. 6, occlusal Queensland. Mem. Qd. Mus. 17: 379-397. view, x 1. MEM. NAT. MUS. VICT. 43 PLAIT 3