MEMOIRS OF THE QUEENSLAND MUSEUM BRISBANE l VOLUME 41 30 JUNE 1997 PART 2 THE PALAEONTOLOGY AND GEOLOGY OF DUNSINANE SITE, RIVERSLEIGH DERRICK A. ARENA Arena, D.A. 1997:06:30: Palaeontology and geology of Dunsinane Site, Riversleigh. Mem- oirs of the Queensland Museum 41(2): 171-179. Brisbane. ISSN 0079-8835, The stratigraphy, fossil assemblages and models of formation and post-depositional history of Dunsinane Site, on the western ‘arm’ of southern Gag Plateau, are described. The fossil vertebrate fauna most resembles the late Oligocene (System A) fauna of White Hunter Site. Rich, exceptionally preserved invertebrate and plant (twigs, leaves, reproductive organs) assemblages occur in nodules of iron oxide-rich flourapatite and as fragments derived from these nodules and appear to have been preserved by early diagenetic microbially-mediated phosphatisation. The sediment is most likely lacustrine and may directly overlie Precambrian sediments. Erosion of overlying sediments and severe chemical weathering may have led to development of the surface lag of insoluble residue which includes phosphatised bone, nodules and other fossil fragments. Relationships of the fossiliferous nodules to the vertebrate fauna and surrounding sediments are not fully determined. Derrick A, Arena, School of Biological Science, University of New South Wales, New South Wales 2052, Australia; received 10 November 1996. Dunsinane Site was discovered in 1990 during exploration of southern Gag Plateau at Riversleigh, NW Queensland (Fig. 1A). In one area the surface was strewn with fragments of fossil bone, fossil wood and nodules of rock that contained leaves, wood, other plant material and invertebrates. Dunsinane Site is the only fossil site at Riversleigh containing plants and one of the few yielding arthropods (Archer et al., 1994). The site was also distinctive in that the fossils were associated with a soft, apparently un- consolidated sediment, rather than embedded in solid limestone. Issues to be resolved regarding this site included the nature and provenance of the fossilised plant and arthropod material, and the relationships of this material to other Riversleigh sediments. CONCEPTS AND TERMINOLOGY Palaeontological concepts, biostratigraphy and taxonomic classification follow Archer et al. (1994) and Creaser (1997). The terms ‘Dunsinane limestone’, ‘Dunsinane deposit’, “Dunsinane sediment’ and ’Dunsinane calcrete’ are used here informally. STRATIGRAPHY AND GEOLOGY Representative rock types from Dunsinane Site have been examined by powder X-ray diffrac- tion, SEM and thin-sectioning techniques. A sample of 4 nodules was used for destructive analysis. The area around Dunsinane Site is dom- inated by 4 main rock types: Precambrian quartz- ite, ferruginised deposits, overlying Tertiary limestone and the Dunsinane limestone. PRECAMBRIAN QUARTZITE The grey Precambrian quartzite is a massive, thick tabular- bedded, crystalline pure quartz sed- iment. This and a laminated chert constitute the terrain around Gag Plateau and form the base- ment underlying the Dunsinane deposit. FERRUGINISED DEPOSITS Outcrops of ferruginised deposits generally do not exceed 15m in diameter. The ferruginisation is apparently related to localised groundwater activity. This type of deposit occurs throughout the Riversleigh area, particularly along geologi- cal boundaries. Around Dunsinane Site such de- posits occur in the Precambrian quartzite, overlying Tertiary limestone, and at a point at the junction of the Precambrian quartzite and the Dunsinane sediment, apparently post-dating all of these sediments. In general they consist of pisolitic iron oxide, and iron oxide- enriched al- terations of the sediments in which they occur. OVERLYING TERTIARY LIMESTONES The hard, micritic, Tertiary limestones directly overlie the Dunsinane limestone. These lime- stones exhibit vertical changes in colour and fre- quency of molluscs and calcareous mud clasts, all of which may be construed as primary bedding features. Occasional vertebrate bone fragments occur with molluscs which are very common and well-preserved. Contact of the Tertiary Lime- stone with the underlying Dunsinane sediment is MEMOIRS OF THE QUEENSLAND MUSEUM FIG. 1. A, Southern Gag Plateau and Dunsinane Site Local area map. Study area in B outlined. Contours at mAHD. (Southern Gag Plateau map inset from Megirian 1992). B, Dunsinane Site study area indicating the positions of Dunsinane, Bernie’s Cooking Pot (BCP), Custard Tart (CT) and Sue’s Rocky Road (SRR) Sites. Stipling=Dunsinane limestone. Contours metres from arbitrary field datum. apparently undulatory, and may be unconform- able. DUNSINANE LIMESTONE The Dunsinane limestone outcrop is at least 150 m across. It includes the fossil sites Bernie’s Cooking Pot (BCP), Custard Tart (CT) and Sue’s Rocky Road (SRR) (Fig. 1B), which have pre- viously been regarded as separate sites. The limestone has been severely weathered, resulting in replacement of most of the original sediment and its sedimentary structures by soft calcrete. Relict unweathered outcrops occur as sandy limestone with apparent flat bedding and minute bone fragments. Because the overlying limestone is closely associated with these relict outcrops, it appears that the weathering event post-dates the deposition and erosion of the over- lying sediment. Because the fine, well-sorted particles and ap- parent flat-bedding does not indicate high energy flow, the Dunsinane sediment was probably de- posited in a pond or lake. This relatively arena- ceous sediment was probably deposited under near shore conditions in a quartz-rich Precambr- ian terrain. Fossil bones and nodules are preserved with brown iron oxide-rich flourapatite (fluorapat- ite=calcium fluoride phosphate; Cas(PO4)3F). Energy dispersive X-Ray spectrometry demon- strated that iron oxide appears to be included within euhedral flourapatite crystals. Flourapatite has replaced the majority of structures within nodule matrices. Bones are impregnated with iron-oxide rich flourapatite, particularly around the inner parts. Bones and nodules occur in relict in situ sediment and in the surface layers across the area of the outcrop. Nodule fragments and fragments of fossil wood derived from weathered nodules are common. There are also amorphous concretions of flourapatite and sparry calcite deep within the calcrete (Im depth) in some places. MODEL OF POST-DEPOSITIONAL HIS- TORY Weathering is a striking feature of the Dunsin- ane sediment. The calcrete is most likely autoch- thonous because of the relationship between relict unweathered outcrops of Dunsinane sedi- ment and overlying rock. Local groundwater ac- tivity has resulted in ferruginisation in the vicinity, although this has not occurred at Dunsin- ane Site itself. Meteoric and surface water are the most likely agents responsible for chemical weathering of the original Dunsinane sediment. Severity of weathering has been influenced by permeability of the original Dunsinane sediment, impermeability of the Precambrian quartzite and Tertiary limestone, the situation of Dunsinane Site at the junction of 3 drainage channels, and its position near the top of the local drainage net- work. Water focussed on Dunsinane Site by the local drainage system was channelled into the perme- able Dunsinane sediment before draining away DUNSINANE SITE into the lower catchment. This caused the chem- ical weathering of the limestone responsible for: l, apparent subsidence, slumping and surface lowering of the sediment; 2, accumulation of a surface lag of fossils mineralised with insoluble flourapatite; 3, partial dissolution and precipita- tion of flourapatite bodies deep in the original sediment; and 4, a thin calcareous duricrust which has cemented surface debris. FLORA AND FAUNA VERTEBRATES. Due to weathering and poor condition of the fossilised bone, identifiable ver- tebrate fossils from Dunsinane Site are relatively uncommon. However, there is marked diversity implied by the number of groups present, In con- trast with many Riversleigh sites, fish and bird remains have not yet been found at Dunsinane Site. The Vertebrates includes: Class Reptilia Order Testudines Family Chelonidae genus & sp. indet. Order Crocodilia Family Crocodylidae ?Baru sp. Class Mammalia Order Marsupialia Suborder Diprotodontia Family Wynyardiidae ?Namilamadeta sp. Family Diprotodontidae Subfamily Diprotodontinae ?Bematherium sp. Subfamily Zygomaturinae ?Neohelos sp. Family Macropodidae Subfamily Balbarinae Nambaroo sp. nov. genus & sp. indet. Subfamily Balungamayinae ?Wabularoo sp. Order Placentalia Suborder Chiroptera Family Hipposideridae ?Brachipposideros sp. Neohelos, Brachipposideros, large crocodil- ians and chelonid turtles which are known from Riversleigh sediments of various Oligocene and Miocene ages are thus not useful for more precise biocorrelation. At Riversleigh wynyardiids are known only from System A and B faunas (Archer et al., 1994). Primitive balungamayine and balbarine kangaroos such as ?Wabularoo and 173 Nambaroo respectively (the latter appears to be conspecific with a very primitive form from Sys- tem A White Hunter Site (Cooke, 1997)), are known from Riversleigh System A and B faunas. With the exception of Pleistocene Diprotodon optatum , diprotodontines are restricted to System A at Riversleigh (Black, 1997). Thus we regard the fauna as most likely a System A fauna. INVERTEBRATES Insect fragments are often quite small, usually around Imm and include a variety of beetle elytra and beetle prothoraces which show similarities to those of curculionids (weevils) and buprestids (jewel beetles). Curculionids have been found in the Upper Site assemblage (Archer et al., 1994), although the long history of this group makes them of little biostratigraphic use. Termite re- mains may also be present. A partial gastropod shell has been identified as a probable terrestrial camaenid (W. Ponder pers. comm. 10/95), Camaenids which are known from the Mesozoic occur in System A (Archer et al., 1994) and are likely to occur in Riversleigh sed- iments of all ages. FLORA The plant assemblage consists of small pieces of twig-like wood (some stems may originally have been 3-4cm in diameter), leaves, seeds and reproductive organs. The various types of wood imply high diversity. Angiosperm wood is present as well as probable gymnosperm wood. Leaf fragments are invari- ably broad and serrate-margined. Leaf cuticles are intact in a number of specimens. There are several pneumatophore-like organs up to 2cm in diameter with a central aerenchyma of elongate cells and a narrow epidermal layer with structures resembling lenticels. Groups recognised include Proteaceae, Casuarinaceae, Myrtaceae and possibly Epacridaceae (R. Hill pers. comm. 9/95, 10/96). Proteaceae are evergreen trees and shrubs, and are known from the Early Cretaceous to Holocene (Hill, 1994). The generally xeromorphic Casuarinaceae have a record from the Paleocene to Holocene (Hill, 1994). Myrtaceae are known from the mid-Paleocene to Holocene (Martin, 1994). The Epacridaceae which are prominent in extant scleromorphic flo- ras have existed in Australia since the Late Cre- taceous (Jordan & Hill, 1996). Serrate leaves are not good indicators of a trop- ical closed forest origin (R. Hill pers. comm, 174 MEMOIRS OF THE QUEENSLAND MUSEUM rear Dh n LE Se DUNSINANE SITE (75 9/95). Apparent growth rings and false growth rings are evident in some Woad samples. Some rings are broad, inferring a Jong growth season which seems to be terminated abruptly (R. Hill pers. comm. 10/95). In general the associated floristics and the timing of fossil occurrences of epucnds throughout the Tertiary coincide with temperate climatic conditions, and the nature of the macrofossil record is inconsistent with mod- em tropical or sub-tropical rainforest Jordan & Hill, 1996), Pollen has been recovered fram the nodules but not yet analysed. PALAEOENYVIRONMENTAL IMPLICATIONS Biocorrelation of the Dunsinane Site fauna with that of White Hunter Site (via Nambaroa und ?Bematheriun) allows tentative age assessment. White Hunter Site has been correlated with Ktadunna Faunal Zone D (Ngama LF. Lake Pal- ankarinna) on the ilariid, Kuterintja ngama (Myers & Archer, 1997). Sediments associated with this South Australian fauna have been pal- veumagnetically dated ut 24.7-25 Ma (Late Oligocene; Woodburneetal,, 1994), This interval coincides with an ‘icehouse’ event, climatic con- ihtions normally characterised by cooler, drier. seasonal climatic conditions (Frakes et al,, 1987) The ‘greenhouse’/ ‘icehouse’ climatic Auctua- tions of the Tertiary are reflected in the sedimen- tary and terrestrial fossil records of Australia (Frakes et al., 1987; Archer et al., 1995). The characteristics of the Dunsinane Site fora so far observed may indicate ‘icehouse’ conditions. TAPHONOMY PHOSPHATISATION OF FOSSIL. ASSEM- BLAGE Early diagenetic phosphatisution, usually asso- ciated with microbial activity, has been identified as the mode of preservation of phosphatic nod- ules and exceptionally preserved fossils (i.e. Bal- son. 1980; Müller, 1985: Pinna, 1985; Seilacher el al., 1985; Soudry & Lewy, 1988; Allison, 19884, b, c; Martill, 1988, 1989, 1990; Lucas & Préval, 1991; Briggs & Kear, 1993; Briggs et al., 1993), As confirmed by laboratory experiments (i.e. Prévat & Lucas, 1986; Hirschler et al., 1990; Briggs et al, 1993: Briggs & Kear, 1993), micrabially mediated phosphatisation can occur within or adjacent to bacteria, and can result in the formation of globular apatite microstructures that faithfully preserve the structure of organ- isms, sometimes at the microscopic level. Condi» lions under which flourapalite replacement of organic tissue and carbonates may occur are (Lucas & Prévôt, 1991); 1) a concentration of organic phosphorous is required in the systém (ie. the sediment); 2) anoxic conditions which support bacteria capable of precipitating apatite; 3) acidic conditions that destabilise carbonates; these circumstances commonly occur in the inter- sua of phosphate-rich sediments, These condi- tions may be enhanced by ‘closure’ by a thin film of sediment or bacterial slime, or enclosing struc- tures of organisms or sediment (i.e. Curapaces, pore spaces) that contain the optimal environ- ment for apatite précipitation (Krajewski, 1984; Seilacher el al., 1985; Martill, 1988, 1989,1990; Soudry & Lewy, 1988; Hirschler et al., 1990; Lucas & Prévét, 1991; Wilby & Martill 1992; Briggs ct al., 1993; Briggs & Kear, 1993). Characteristic flourapatile microstructures very similar ta those found in fossils from clse- where in the world thought to be preserved in this way Occur in nodule material examined by SEM (Fig. 2G.H). The lack of distortion or crushing ol tissues, preservation of the cellular structures of leaves (see below) and presence of organic male- rial in the Dunsinane nodules indicate carly dia- genene, pre-compaction mineralisation. This process is suggested us the mode of preservation for the 3D arthropod fossils from Upper Site {Duncan & Briggs, 1996). Fragments of algal layers in nodule material may indicate thin micro- bial films that ‘sealed’ ussues, enclosing condi- tions favourable to preservation and encouraging and accelerating mimeralisation, The Dunsinane nodule material may have minvralised in a waler PUG. 2. A, BDunsinane Site nodules illustrating the variety of shapes and sizes. Scale bar=Sem. B, Some nodules have protusions (in this case a piece of fossil wood). Scale=20mm. C, Leaf fragment, D, QMF31 306, a 3-dimensional fruit tentatively assigned to the Epacridaceae. E, Thin section of nodule. entirely phosphansed organic material. Note the undistorted arthropod with cuticle at lower-right, Plane polarised light. Scale=SO0jem. F, Transverse section of d leaf lamina. The vertically elongated cells are palisade cells, beneath them is intact spongy mesophyll tissue. Plane polarised light. Seale=200,.m. G, SEM of nodule material showing, the chanscteristic microsimcture of early diagenetically precipitated flourapatite replacing organic tissue, Scale=20.um, H, Globose Nonrapatite microstructures. The pseudo-hexayonal crystals are flourapatite. Seadle=Sjum. 176 MEMOIRS OF THE QUEENSLAND MUSEUM FIG. 3. Models of formation of the Dunsinane Site nodules. A-C showing models of the arrival/for- mation of nodules. A, Whilst there is little evidence of high-energy activity in the sediments, erosion and transport cannot be ruled out as arounding mechanism. B, Nodules may have formed by partial mineralisation of a mat of organic material. C, Nodules may be de- rived from a weathered overlying deposit in which they either formed or into which they were trans- ported. This model is regarded as relatively unparsimonious consid- ering the partially phosphatised vertebrate assemblage is unlikely to have been reworked and that nodules occur in in situ Dunsinane site limestone. Once incorporated into the sediment, the ?unconform- ably overlying younger Tertiary limestone was deposited (D,E). Development of a karst terrain (F) and erosion of overlying sediment allowed weathering of the Dunsin- ane deposit (G). Calcretisation (medium stippling) and surface lowering resulted in the observed distribution of outcrops of relict and weathered Dunsinane lime- stone, overlying Tertiary limestone and a residual lag of insoluble fos- sils. DUNSINANE SITE body that had stagnated possibly because of a lack of freshwater input; a proximal acidic anoxic sediment such as peat or influxes of nutrients causing microbial blooms that used up available oxygen and increased available organic phospho- rous. As partially decaying organic matter accu- mulated on and in the substrate, reducing conditions would develop leading to accumula- tion of iron hydroxides and destabilisation of carbonates (Allison, |988b) and a proliferation of anaerobic microbes. Phosphatisation may have occured in issues in the substrate and close to il, umil halted by exhaustion of available phosphate supplies, or dilution of the phosphate-rich me- dium perhaps by an influx of freshwater supply- ing oxygen and dissolved calcium carbonate PLANT ASSEMBLAGE Plant orgim assemblages composed of Lwigs, leaves and reproductive organs such as those at Dunsinane Site (Fig. 2C.D) are generally re- garded as accumulations with limited lateral wind iransport (Collinson, 1983; Spicer, 1980, 1989). Plant assemblages accumulated at the site of growth usually contain roots, wood fragments and plant bases; catastrophic accumulations are poorly sorted and contain components of all types (Collinson, 1983; Spicer, 1989), Leaves most likely to he preserved in excellent condition are those that fall directly into water, since contact with the ground usually drastically reduces the chances of leaf preservation in an aquatic environment (Ferguson, 1985, Spicer, 1989. T991). Dissolution of unti-lungal, anti-mi- crobial and structural compounds within leaves begins immediately after immersion or contact with the ground, increasing susceptibility to mi- crobial attack and structural collapse within 24 hours (Ferguson, 1985; Spicer, 1989, 1991), Leaf litter is generally dispersed within 50m of the perent vegetation (Ferguson, 1985; Spicer, 1989, 1991), Hill & Gibson (1986) found that the ma- jority of leaves collected from a lake bed were from species oveurring within 50m of the shore. Leaves in the Dunsinane assemblage with limited immersion damage. such as the exceptionally preserved leal with intact cellular detail (Fig. 2F), may have been derived from vegetation that occurred within 50m of the site of preservation, Since most Ieaves take hours or days to sink (Ferguson, 1985; Hill & Gibson, 1986; Spiver, 1980, 1989, 1991). buoyanlorgans such ws twigs and seeds are prevalent inthe assemblage and are often in quite good condition showing little abra- sion or decomposition, and the insect material is su plentiful and diverse these factors may point tò the accumulation bemg in shallow nearshore water. The plant material thus appears to be part of a proximal assemblage, possibly occurring us an immersed mat of vegetation. The process of mineralisation that preserved the plant material must have occurred in the early stages of degra- dation since leaves have limited structural dam- age and intact cellular structure, in particularly spongy mesophyll tissue which is generally the most susceptible to breakdown (Spicer. 1989). The pneumatophore-like organs in the nodule assemblage could indicate anoxic condifiuns close to the site of preservation. PROVENANCE AND FORMATION OF THE PHOSPHATISED NODULES The Dunsinane Site nodules (Fig. 2A,B) may have formed in a variety of ways. Their sub- rounded shape is possibly caused by transport but this i$ not supported by the Jack of any other evidence of high-energy Now. More lightweight fresh clods of organic material may have been sub-rounded by transport and subsequently de- posited and phosphatised, However, it is debat- able whether the nodule material had been subject to a sufficient compression und consolidation. given the freshness and lack of distortion of some fossil structures. The nodules may have formed al Dunsinane Site, as parts of a mat of fresh Organic material resting on the sediment surface The radiation of mineralisation front points within the mat may have produced the somewhat rounded, but otherwise variably sized and shaped nodules. Roundness of nodules may also be al- tributed to weathering whilst embedded in the sediment and after exposure, resulting in contin- uous exfoliation of outer layers. MODELS FOR ACCUMULATION AND PROVENANCE AND FORMATION OF NOD- ULES (Fig, 3). Allochthony of the nodules mist be considered because of their peculiar mincral- ogy. Intraformational nodules may have Formed in an overlying sediment which was weathered, resulling in the descent of sub-rounded nodules to the surface of the Dunsinane sediment anil bunal by younger Tertiary lacustrine sediment (hence the unconforming contact). However this is highly unlikely considering that nodules and other flourapatite concretions occur within in sitt Dunsinane sediment. The shared mineralogy of the nodules and ver- tebrate fossils is compelling when considering 178 contemporaneity. The phosphatisation process can tend to target only those issues which have a high organic phosphorous content (Balson, 1980; Allison, 1988 a,c), suggesting that Dunsinane Site bones, which show varying degrees of phos- phate enrichment, may have been subject to this process when they were fresh and still retained some Organic content, This is suggested by the greater enrichment of the inner parts of bone Which would have contained high concentrations of orgame material. The extreme damage to the Dunsinane Site bone, but lack of evidence of reworking or trampling (particularly in the case ol a partially phosphatised, extremely damaged "Bematherium skull with dentaries articulated) may indicate damage by the highly acidic condi- tions in which the plant material was preserved. This would have softened and dissolved bone in the nearby substrate. Nevertheless, the evidence Is ambiguous and al his stage, CONCLUSIONS Dunsinane Site contains a probable System A vertebrate fauna which may be tentatively dated at 24.7-25 Ma. The fossils from Dunsinane Site are preserved with iron oxide-rich fMourapatire, which appears to have precipitated as the result of early diagenetic microbially mediated phos- phatisation, The Dunsinane sediment probably formed in a low-energy environment, and has been severely weathered, resulting ina lag of insoluble fossil material on the surface. Relation- ships of the flora and invertebrate fauna to the other components of the site are as yet unre- solved. ACKNOWLEDGEMENTS Amongst the many people to whom Lowe grat- itude for discussion, advice, technical informa- lion and access to resources are: Michael Archer, Henk Godthelp, Suzanne Hand, Robert Hill, Gregory Jordan, Jane Heath, Boh Mesiboy, Al- berto Albani, Bernard Cooke, Philip Creaser, Robert Jones, Winston Ponder, Anna Gillespie. Stephan Williams, Mel Dickson, Helene Martin, Peter Atherden, Rad Flossman, Michacl de Mol und fellow students. LITERATURE CITED ALLISON, P.A, [988a. Taphonomy of the Eocene London Clay biota, Palaeontology 31: 1079-1100. 1988b. Phosphatized soft-bodied squids from the Jurassic Oxford Clay. Lethaia 21: 403-410, MEMOIRS OF THE QUEENSLAND MUSEUM 1988c. Rorserval-Lagerstutien: cause and chessili- cation. Paleobiology 14; 331-344, ARCHER, M., HAND, S.J. & GODTHELP, H. 1994. Riversleigh. 2nd ed. (Reed:Sydhey), 1995, Tertiary environmental ond biotic change in Australia. Pp, 77-90. In Vrha, E.S., Denton, G.H.. Partridge, T.C, & Burekle, L.H. (eds), Paleacli- mate and evolution, with emphasis on human ongins. {Yale University Press: New Haven), BALSON, P.S, 1980. The origin and evolution of Ter- tiary phosphorites from eastern England Journal of the Geological Society of London 137: 723- 729. BLACK, K 1997. Diversity and biosiratigraphy of the Diprotodontoidea of Riversleigh, northwestem Queensland, Memoirs of the Queensland Museum 41: 187-192. BRIGGS, D.E.G, & KEAR, AJ. 1993. Fossilization of soù tissue inthe laboratory, Science 259: 1439- 1442. BRIGGS, DE.G,, KEAR, AJ, MARTILL, D.M. & WILBY,P.R, 1993. Phosphauzation of soft-tissue in experiments and fossils. Joumal of the Geoto- gical Society of Landon 150: 1035-1038. COLLINSON. M.E. 1983, Accumulations of fruits and seeds in three small sedimentary enyironments in southern England and their palaeoecological im- plications, Annals of Botany 52: 583-592, COOKE, B.N. 1997. Biostratigraphic implications of the fossi) kangaroos at Riverlseigh, northwestem Queensland. Memoirs of the Queensland Museum 41; 295-302, CREASER, P, 1997. Oligocene-Miocene sediments of Riversleigh: Ihe potential significance of Wpogra- phy. Memoirs of the Queensland Muscum 41: 303-314, DUNCAN, LJ. & BRIGGS, DEG. 1996. Three-ci- mensionally preserved insects. Nature 381: 30-31. FERGUSON. D.K. 1985. The origin of lenf-assen- blages - new light on an old problem. Review of Paleobotany and Palynology 46; 117- 188. FRAKES, L.A., MCGOWRAN, B, & BOWLER, IM, 1987, Evolution of Australian environments, Pp, 1-16. In Dyne, G.R. & Walton, D.W. (eds), Fauna of Australia, vol 1 A, General Articles. (Australian Government Publishing Service: Canberra). HILL, R.S. 1994. The history of seleeted Australian taxa. Pp.390-419. In Hill, R.S. (ed.), History of the Australian vegetation: Cretaceous to Recent, (Cambridge University Press; Cambridge). HILL, R.S, & GIGSON, N, 1986. Distribution of poten- tial macrotossils in Lake Dobson, Tasmania, Jour- nal of Ecology 74: 373-384. HIRSCHLER, A., LUCAS, J. & HUBERT, J, 1990). Bacterial involvement in apatite genesis. FEMS Microbiology Ecology 73: 211-220. JORDAN. G.J. & HILL, R.S. 1996, The fossil record of the Epacridaceae. Annals of Bolany 77: 341-346, KRAJEWSKI, K.P, 1984, Early diagenetic phosphate cements in the Albian condensed glaticonitic lime- DUNSINANE SITE stone of the Tatra Mountains, Western Car- pathians, Sedimentology 31: 43-470. LUCAS, J. & PREVOT, L. 1991. Phosphates and fossil preservation, Pp. 389-409. In Allison, P.A. & Briggs, D.E.G. (eds), Taphonomy. (Plenum Press: New York). MARTILL, D.M. 1988. Preservation of fish in the Cretaceous Santana Formation of Brazil, Pal- aeontology 31: 1-18. 1989, The Medusa effect: instantaneous fossiliza- tion. Geology Today 5; 201-205, 1990, Macromolecular reolution of fossilized mus- cle tissue from an elopomorph fish. Nature 346: 171-172. MARTIN, H.A. 1994. Australian Tertiary phylogeog- raphy: evidence from palynology. Pp. 104-142. In Hill, R. (ed.), History of the Australian vegetation: Cretaceous to Recent. (Cambridge University Press: Cambridge). MEGIRIAN, D, 1992. Interpretation of the Miocene Carl Creek Limestone, northwestern Queensland. The Beagle, Records of the Northern Territory _, Museum of Arts and Sciences 9: 219-248, MULLER, K.J. 1985, Exceptional preservation in cal- careous nodules, Philosophical Transactions of the Royal Society of London B31 1; 67-74. MYERS, T. & ARCHER, M. 1997. Kuterinja ngama (Marsupialia, Illariidae) from the Late Oligocene of Riversleigh, northwestern Queensland. Mem- oirs of the Queensland Museum 41: 379-392. PINNA, G. 1985. Exceptional preservation in the Jurassic of Osteno, Philosophical Transactions of the Royal Society of London B311: 171-180. PREVOT, L. & LUCAS, J. 1986, Microstructure of apatite-replacing carbonate in synthesised and 179 natural samples. Journal of Sedimentary Petrol- ogy 56: 153-159, SCULTHORPE, C.D. 1967, The biology of aquatic vascular plants, (Edward Amold: London). SEILACHER, A., REIF, W.E. & WESTPHAL, F. 1985. Sedimentological, ecological and temporal patterns of fossil Lagerstatten. Philosophical Transactions of the Royal Society of London B31 1: 5-23. SOUDRY, D., & LEWY, Z. 1988. Microbially influ- enced formation of phosphate nodules and megafossil moulds (Negev, southern Israel), Pal- aeogeography, Palaeoclimatology, Palaeoeco- logy 64: 15-34. SPICER, R.A. 1980. The importance of depositional sorting to the biostratigraphy of plant megafossils. Pp. 171-184, In Dilcher, D.L. & Taylor, T.N. (eds), Biostratigraphy of fossil plants. (Dowden, Hutchinson & Ross: Stroudsberg). 1989. The formation and interpretation of plant fos- sil assemblages, Advances in Botanical Research 16; 96-191. 1991, Plant taphonomic processes. Pp. 72-115. In Allison, P.A. & Briggs, D.E.G. (eds), Tapho- nomy. (Plenum Press: New York). WILBY, P.R. & MARTILL, D.M. 1992. Fossil fish stomachs: a microenvironment for exceptional preservation. Historical Biology 6: 25-36. WOODBURNE, M.O., MACFADDEN, B.J., CASE, J.A., SPRINGER, M.S., PLEDGE, N.S., POWER, J.D., WOODBURNE, J.M., SPRINGER, K.B. 1994. Land mammal biostratigraphy and magnetostratigraphy of the Etadunna Formation (Late Oligocene) of South Australia. Journal of Vertebrate Palaeontology 13: 483-515. A NEW SPECIES OF PALORCHESTIDAE (MARSUPIALIA) FROM THE LATE MIDDLE TO EARLY LATE MIOCENE ENCORE LOCAL FAUNA, RIVERSLEIGH, NORTHWESTERN QUEENSLAND KAREN BLACK Black, K.. 1997:06:30. A new species of Palorehestidue (Marsupialia)from the late middle to early late Miocene Encore Local Fauna, Riversicigh, nonhwestern Queensland. Memoirs of the Queensland Museum 41(2): 181-185, ISSN 0079-8835. A single palorchestid M! from the Encore Local Fauna, Riversleigh, northwestern Queens- Jand is described as Palorchestes anilus sp, nov. In size and morphology, it is intermediate between the M! of middle Miocene Propalorchestes novaculacephalus Irom System C deposits. Riversleigh and the Bullock Creek Local Fauna, Northern Territory, and that of Palorchesies painei from the late Miocene A)coota Local Fauna, Northern Territory. These relationships support an early late Miocene uge for the Encore Local Fauna and confirm that Propalorehestes is the sister-group of Palerchestes, Consequently, the monophyly of Palorchestidae 1s cast further in doubt, Species of Ngupakaldiu and Pitikantia may be more appropriately regarded as plesiomorphic members of Diprotodontidae. (JPalorchestidae, Palarchestes, Propalorchestes, late Miocene, Riversleigh. Karen Black, School of Biological Science, Universin of New South Wales, New South Wales 2052, Australia; 4 November 1996. Arecently discovered upper molar from Encore Site on the Gag Plateau, Riversleigh has in- creased the late Oligocene to late middle Miocene material of Palorchestinae to 9 specimens. This paucity of material, which prior to 1986 consisted exclusively of the highly derived Palorchesies, has made resolution of relationships within the family difficult, Although Palorchestes annulus sp. nov, is only known from an isolated M!, it adds substantially to phylogenetic understanding within the family. On the basis of vertebrate stage-of-evolution biocorrelation the Encore Local Fauna is cur- rently regarded as late middle tò early late Miocene (approximately 10Ma; Archer et al.. 1995). Taxa from Encore Site are more derived than those characteristic of Riversleigh’s. upper System C assemblages yet plesiomorphic relative 19 related taxa of the late Miocene Alcoota Local Fauna, Northern Territory (Archer et al., 1995), The species described below supports an early Jate Miocene age. Institutional abbreviations used here are as fol- lows: QMF, Queensland Museum palacontologi- cal collection; CPC, Commonwealth Palaeontological Collection at the Australian Geological Survey Organisation. Canberra; NTMP, Art Gallery and Museum of the Northern Territory palacontological collection; SAMP, South Australian Museum; UCMP, University of Culifornia, Berkeley. Cusp nomenclature follows Archer (1984) and Rich et al. (1978) except that their hypocone of upper molars is the metaconule following Tedford & Woodburne (1987). Molar number homology follows Luckett (1993). Higher level systematic nomenclature follows Aplin & Archer (1987), SYSTEMATIC PALAEONTOLOGY Order DIPROTODONTIA Owen, 1866 Suborder VOMBATIFORMES Woodburne, 1984 Infraorder VOMBATOMORPHIA Aplin & Archer, 1987 Superfamily DIPROTODONTOIDEA Archer & Bartholomai, 1978 Family PALORCHESTIDAE Tate, 1948 emend. Archer & Bartholomai, 1978 Palorchestes Owen, 1873 TYPE SPECIES. Palarhestes azac/ Owen, 1873. OTHER SPECIES. P. parvus De Vis, 1895; P. painei Wondburne, 1967; P. selestiae Mackness, 1995, Palorchestes anulus sp. nov. (Figs 1-2, Table 1) MATERIAL, Holotype, QMF30792, a right M Hhišs- ing the posterior cingulum and anterior and posterior roots from the late middle Miocene to early lare Miocene Encore Local Fauna, on the Gag Plateau. Riversleigh. ETYMOLOGY. Latin nulus, link; refers toits being a structural link between Propalorchestes and Pal- 182 TABLE |. Measurements (mmjot palorehestid M^. — MEMOIRS OF THE QUEENSLAND MUSEUM parvus and P. azael: in lacking the second medial forelink; in haying a less well- devel- Species | No. | Length ae P Elenor oped midlink which is deeply V-shaped, its - “ y respective anterior and posterior crests P gar ae QMF30792 130 | meeting lower in the transverse median val- NTM P895-1 Pr. pontioulus QMF30883 ley (and more buccally) than the well-deyel- oped structure in bhoth P. parvus and P. azael; in lacking the second buccal midlink, and having only a poorly-developed acces- Pr. nava- | culacephatas | NTMP862-27 sory crest extending anteriorly from a me- dial point on the metalophy in having a poorl y-developed lingual cingulum; in lack- P.selestiae | QMF12455 ing a buccal cingulum; and in haying well- UCMP 70553 R developed postparaconal and [UCMP 70553 L P. pone UCMP 70550 | postmetaconal crests extending posteriorly 4 from the apices of the paracone and metacone respectively. Palerchestes anulus differs fram P. selestiae: in having a short crescenuc lin- gual cingulum; in lacking the anterolingual forelink; in lacking the secondary midlink and (he minor lingual midlink; and in having less-crenulated enamel at the hase of the protoloph and metaloph. Palorchestes anulis differs from P. parvus in having a straighter, less crescentic metuloph and in having a Jess crenulate transverse median valley- Palorchestes anulus differs from P. azael UCMP 66521 | 17.8 | CPC6752 18.2 QMF 784 | 207 1 MF12476 al E) P pariis. |_QMF2963 19.3 QMF3719 19.3 P- azael P31370 P3137] PF31372 orchestes and to the distinct midlink, a character of Palarchestes. COMPARISON. Palorchestes anulus differs from P. painei in being proportionately narrower anteriorly and posteriorly, in ats poorly developed lingual cingulum, more Open transverse median valley lingually, less tightly V-shaped transverse median valley in lingual view and Jess well-de- veloped hindlink. Palorchesres anulus differs from P. parvus, P. selesttue and P. azael in bemg smaller; in having generally Jess well-developed links; in having a shallower, more open transverse median valley; in having a more buccally positioned midlink: in having a less well-developed, lower, less bucc- ally extensive (i.e. in lacking the anterobuccal cingulum) anterior cingulum (compared with the high, loph-like anterior cingulum m both P. parvus and P. azael) and consequently, in lacking the deep valleys formed between the anterior cingulum and the anterior base of the protoloph. Palorchestes anulus differs from both P. in lacking the well- developed lingual mid- link; in having a better-developed posterior cingulum; and in having a hindlink devel- oped. DESCRIPTION. Tooth rectangular, bilophod- onl, consisting of an anterior protoloph connect- ing the protocone with the paracone, and a posterior metaloph connecting the metacone with the metaconule. Protoloph antenorly convex: metaloph slightly more linear, with its lingual end detlected posteriorly, Metaconule highest cusp; the paracone and protocone subequal in height: metacone lowest cusp (taking into account slight wear on the apices of the major cusps). Anterior cingulum well-defined but low on the anterior base af the crown, extending lingually trom the anterobuccal tooth margin to the anterolingual base of the protocone. Lingual cingulum short, poorly defined, connecting the posterolingual base of the prolocone to the anterolingual base of the metaconule, Posterior cingulum not pre- served (but suggested by the short crest at the posterolingual base of the metaconule). Forelink well-developed, extending anteriorly MIOCENE PALORCHESTID, RIVERSLEIGH ER FIG: |. Palorchestesanulus sp. nov, Holotype, QMF30792: Occlusal stereopair of right M} Bar indicates 10mm Irom the apex of the protoloph at a point slightly lingual to the paracone apex, meeting the anterior cingulum at the parastylar corner of tooth. Two accessory crests (or incipient links) poorly-de- fined: one originating ut the paracone and fading down the anterobuccal face of the crown; the second originating from the protoloph at à point shghtly lingual to the main forelink, extending anteriorly and slightly buceally, along the longi- tudinal axis of the tooth, terminating in the valley between the anterior base of the protoloph and the anteriorcingulum, Single midlink formed by the junction of respective anterior and posterior crests from the metaloph and protoloph meeting low in the transverse median valley (making the link sharply V-shaped in lateral view) approxi- mately 4mm from the buceal tooth margin. An additional moderately-developed crestextending anteriorly from the apex of the metaloph into the transverse median valley but without a connect- ing crest from the protoloph. Hindlink well-de- veloped, extending posteriorly and slightly lingually from the metaloph, approximately Smm lingual to the buccal tooth margin. A thickening in the enamel (the posterior metaconule buttress) posterior to the metaconule apex but probably not developed into a crest. A similar buttress on the posterior llank of the protocone. DISCUSSION, Palorchestids are rare. fragmen- tary components Of Tertiary fossil assemblages Until recently, the family consisted of only the primitive, generalised, lute Oligocene Neapakal- dia and Pinkanta, and the derived, highly specialised late Miacene to late Pleistocene Pul- orchestes. The large temporal and morphological gaps separating these groups has made relation- ships within the family difficult to resolve. Stirton (1967) recognised 4 subfamilies within the Diprotodontdae und included Neapakaldia and Pitikanria in the Palorchestinae (later raised to family status) based on similarities in basicranial morphology to Palorehesres. However, these supposed apomorphies are also shared with the Diprotodontinae and have since (e.g. Archer 1984) been intepreted as symplestomorphic within Vormnbatomorphia. Consequently, Archer (1984) concluded the Palorchestidae was nol mo- nophyleuc, a view later confirmed by Murray (1986;1990), with his description of Pre palorchestes dentitions and cranial fragments trom the middle Miocene Bullock Creek Locul Fauna. Northern Territory, and several Oligo- Miocene sites at Riversleigh, Murray (1990) con- cluded thal Prepalorchesres is the plessomorphic sister-taxon of Pafarchesres and demonstrited a structural transition from the selenodont wynyardiid molar pattern to the bophodontpal- orchestid molar pattern. He further concluded that Ngapakaldiq and Pitikentia, haying sup- pressed their selenodont herilape, show closer affinities to the fully bilophodont diprotadontids than pulorchestids. Preliminary analyses of late Oligocene and Miocene diprolodontids and pal- orchestids from Riversleigh further suggest that Neapakeldia and Pitikantia should be regarded as primitive members of Diprotodontidac. Palorchestes anulus supports a Propalor- chestes/Palorchestes sister-group relationship and confirms douhts (Areher & Bartholomai, 1978; Archer, 1984: Murray. 1990: Mackness, 1995) about the monophyly of the family. The M! of P. anulus is intermediate in a number of key features beween the middle Miocene Pro- palorchestes novaculacephalis from the Bullock Creek Local Fauna, and System C deposits al Riversleigh, and Palorchestes painei from the late Miocene Alcoota Local Fauna, The Encore M! consistently groups with Propalorehestes novecilacephalus and P. painei tailing within the 154 À SG Pr. ponwtlus 24 < Pr. nowcularephalus T: og eg = P. anulus Pro 20 o Pame! Q £ + P. selestiae D = 5 2 c ; absence of a well-developed lingual margin on upper molars; presence of upper ca- nines (except Neohelos); absent or small 194 MEMOIRS OF THE QUEENSLAND MUSEUM FIG. 1. Silvabestius johnnilandi gen. et sp. nov., holotype, QMF30504. A, right lateral view. B, ventral view, C, dorsal view. Bar = 20mm. hypocone on P3 (except Alkwertatheriumwebbi); and Nimbadon by: its small size, molar gradient more anteriorly convex upper molar lophs (ex- that does not appreciably increase posteriorly. It cept for Nimbadon), It is distinguished from other is distinguished from other zygomaturines except zygomaturines except Raemotherium yatkolai Nimbadon and Neohelos by: the anterobuccal PRIMITIVE ZYGOMATURINE SILVABESTIUS GEN. NOV 195 FIG. 2. Silvabestius johnnilandi gen. et sp. nov., holotype, QMF30504. Occlusal stereopair of upper cheektooth dentition. Bar = 10mm. blade from the parametacone on P3; absence of a well-developed buccal cingulum or metastyle on P3, It can be distinguished from other zygomaturines except Neohelos, Nimbadon and A. webbi by an undivided parametacone on P3. ETYMOLOGY. Latin silva, forest and bestia, beast; for its inferred habitat; masculine. Silvabestius johnnilandi sp. nov. (Figs 1-6) MATERIAL. Holotype QMF30504, a juvenile skull and associated mandible with completely unworn left and right cheek tooth rows. The basicranium, palate and nasals are incomplete. Both dentaries are missing the coronoid process. 12-3 is missing on each side. P3and M3-4 on each side were unerupted at the time of 196 FIG. 3. Silvabestius johnnilandi gen. et sp. nov., type QMF30504. Occlusal stereopair. Bar = 10mm. death. Paratype QMF30505, a virtually complete adult skull; its frontals are broken and slightly depressed; a large fracture runs diagonally from the left orbit to the dorsal surface of the premaxilla and continues ventrally through the right canine alveolus and onto the palate. The right palatine bone is fractured and depressed. The left mastoid/paroccipital process is broken. The supra- occipital is incomplete. Both types from VIP Site which is 3m below D Site; the latter contains diprotodontoids some of which belong to genera that occur in the late Oligocene in the Etadunna Foration (Woodburne et al., 1994) and thus we assign VIP Site to the late Oligocene. Although a juvenile the holotype has perfectly unworn dentition which maximises information about crown morphology and both the skull and lower jaws, whereas the paratype is only a skull. ETYMOLOGY. For Professor John Niland, Vice Chancellor of the University of New South Wales who MEMOIRS OF THE QUEENSLAND MUSEUM has been a strong supporter of the Riversleigh Project and helped collect at Riversleigh. RELATIONSHIP OF THE HOLO- TYPE TO THE PARATYPE. We conclude that QMF30504 was a de- pendent pouch-young living on its mother’s (QMF30505) milk be- cause: 1) both specimens were in the same deposit, within 1m of each other; 2) the lack of wear on the teeth of QMF30504 most of which were still erupting and the lack of fusion of any of its cranial bones; 3) when the upper and lower cheek teeth of the juvenile are brought into occlusion, the upper and lower inci- sors do not meet, but are separated by a gap of approximately 6 mm, a gap appropriate to suit a mother’s nipple. 4) LI of the adult skull, which was thought to be missing, was found adjacent to the nose of the juvenile skull suggesting that the adult and juvenile died nose to nose, suggesting an emotional rela- tionship. DIAGNOSIS. This species differs from S. michaelbirti in the follow- ing combination of features: it is much larger; it has an expanded parastylar region on P? and more right dentary of holo- distinct parastyle; it has a more pos- terior parametacone on P3; it has a larger protocone on P3 that is also more distinctly separated from the base of the parametacone by a deep fissure; it has a small but distinct hypocone on P: and it has less steeply sloping surfaces on the protolophs and metalophs of ML- DESCRIPTION. Upper incisors. Land RI! in the Holotype; L, RI!-2 and LI? in the Paratype. I! large, curved, with a thick, rounded base tapering to its tip. Enamel confined to the anterolateral surfaces and the upper third of the distal surface. L and RI! convergent at their tips. I? small, sub- ovale, tapering anteriorly, with narrow amtërior tip contacting the posterolateral face of I!, with crown dominated by a large, ovate wear facet. P posterior and slightly lateral to IP, smaller than I’, with a triangular occlusal surface, with apex ori- ented anteromedially, with a wear facet on most of the occlusal surface. PRIMITIVE ZYGOMATURINE SILVABESTIUS GEN. NOV 197 FIG. 4. Silvabestius johnnilandi gen. et sp. nov., right dentary of holotype QMF30504. A, lingual view. B, buccal view. Bar = 10mm. Upper canines. Adult skull with small canine alveolus just posterior to the premaxillary/maxil- lary suture; obscured in the juvenile skull. P3. Four cusps: a large parametacone; a well- developed protocone; a small distinct parastyle; and a poorly-developed hypocone. Widest across the protocone. Emargination between the bases of the protocone and lobate parastylar region defining the anterior and posterior moieties. Para- type premolar proportionately longer than that of the Holotype and the anterior and posterior moi- eties are more distinct. Large, undivided para- metacone the tallest cusp on the premolar, situated along the midline of the tooth, slightly posterior to its transverse axis. Protocone large, lingually opposite the parametacone. Hypocone small but distinct, on the lingual cingulum at the posterolingual base of the protocone. Paratype P? with extremely reduced and non-cuspate hypocone. Parametacone pyramidal in occlusal view with steep buccal, anterolingual and posterolingual faces. Each face defined by a distinct blade: an anterior preparametacrista, a posterior postparametacrista and an anterolingually di- rected blade. Preparametacrista linear, extending anteriorly (and slightly buccally) from the parametacone apex, continuous with a short, ar- cuate, posterobuccal blade from the parastyle. The anterolingually-oriented blade descending the anterolingual face of the parametacone, ter- minating in the fissure that separates the bases of the protocone and parametacone, just prior to meeting its counterpart, a short, linear, an- terobuccally-directed protocone crest. A similar condition is found in Nimbadon lavarackorum (Hand et al., 1993a) and some Neohelos speci- mens from the Oligocene-Miocene of Rivers- leigh. In the paratype, the anterolingual parametacone blade is continuous with the an- terolingual cingulum, a feature previously re- garded as an autapomorphy of Nimbadon (Hand et al., 1993a). A slight swelling at the junction of the anterolingual blade and the lingual cingulum may represent a small protostyle. 198 TABLE |. Measurements (mm) of Silvabestius. H=Height of crown; A/P=anterior-posterior length, Th=Thickness; W=width; IMP° = implantation angle; #=broken specimen; *=dimensions of alveoli. A. | MEASUREMENTS Species No _| Side H QMF | L $ 7.1 30504 johnnilandi R 8.3 QMF | L 16.0# | 8.7 6.9 | 30505 | R | 20.14 | 83 6.9 . a MF | L 18.9# 7.2 7.0 michaelbirti Q \ 20493 | R | 18.34 | 81 6.8 B. MEASUREMENTS | r . ; MF | L 5.4 6.4 5.0 ohinnilandi OME: + j 30505 | R | 54 63 | 51 2 * j * michaelbirti OME Liy 3:9 4.5 Etat [r| s7 | 60 | 47 Cr MEASUREMENTS ; ; ; MF | L 4.7 5.3 4.1 johnnilandi Q 30505 | R Tp 4 ; . Zt. MF | L - 5.3* 4.8* michaelbirti Q 20493 [TR | - 52% | 43* | D. UPPER CANINE MEASUREMENTS : ; . | QMF | L - 30r p 2.2* johnnilandi 30505 R 3 ue TT ` R MF L 6.3 5.1 3.5 haetbien VO michaelbirti i 70493 [r| 62 | 52 xe E. LOWER INCISOR MEASUREMENTS ~ J | L Th WwW IMP° vp. | QMF |L | 70 47 | 30 johnnilandi 30504 | rR | ai | 46 | 30 J Postparametacrista well-developed, blade-like, continuous with the posterolingual cingulum, Small, erect parastyle at the anterolingual cor- ner, slightly buccal to the tooth margin, domi- nated by a distinct, arcuate blade continuous with the preparametacrista. Parastyle in paratype re- duced to a slight swelling at the anterior tip of the preparametacrista. Short, ill-defined, an- terolingual parastyle crest continuous with the anterolingual cingulum. Shallow anterolingual basin between the bases of the parastyle, parametacone and protocone, better defined in holotype RP?, but extremely shallow on the crown of the paratype. Short, posteriorly-directed, apical protocone crest terminating 1/3 the way down its posterior surface. Two short, linear crests extending from the apex of the hypocone; one extending anteri- orly and terminating at the posterior base of the protocone; other extending posteriorly, continu- MEMOIRS OF THE QUEENSLAND MUSEUM ous with the posterolingual cingulum. Buccal cingulum poorly-developed in the Holotype. Buccal surface of the parametacone with uni- form, steep gradient to the base of the tooth crown. Well-developed buccal cingular shelf in the Paratype. Upper molars. Proportionately similar to Nimbadon whitelawi from the mid-Miocene Bul- lock Creek Local Fauna (Hand et al., 1993a). M!. Sub-rectangular, low-crowned, trans- versely-bilophodont, with an anterior protoloph and posterior metaloph. Protoloph and metaloph short, anteriorly-convex crests, former slightly more arcuate than the latter; protoloph consisting of a buccal paracone and a lingual protocone; metaloph consisting of a buccal metacone and a lingual metaconule; metaconule the tallest cusp, followed by the paracone, metacone, then pro- tocone. Parastyle small, at the anterobuccal end of the anterior cingulum, connected to an anterior paraconal crest by a short, posteriorly-oriented blade. Paratype parastyle poorly developed, slight swelling at the junction of the anterior paraconal crest and the anterior cingulum, Postmetacrista distinct, extending posteriorly from the metacone apex, continuous with the posterior cingulum. Metastyle a slight swelling at the posterobuccal margin, most reduced in the Paratype. Posterior cingulum terminating at the posterolingual base of the metaconule. Lingual cingulum in Holotype poorly developed, in Para- type short, crescentic, joining bases of the pro- tocone and metaconule, blocking the lingual exit of the interloph valley. Short, indistinct, postparaconal crest descending the posterior face of the paracone, terminating in the interloph val- ley, before meeting its counterpart, an indistinct premetacrista, Short, linear, apical blade on the metaconule. Paratype with facets extending posterobuccally from the apices of the protocone and metaconule, fading down the posterior flanks of their respective cusps. M?*+. Similar to M'except: Larger, proportion- ally wider anteriorly, with molars trapezoidal in occlusal view; paracone tallest cusp, protoloph higher than metaloph; parastyle and metastyle and their associated crests extremely reduced, as are the postparaconal crest and the premetacrista; lophs relatively longer, with protoloph longer than the metaloph, becoming more anteriorly convex in the more posterior molars; metaloph severely reduced, more obliquely oriented rela- tive to the protoloph, with posterior moiety se- verely reduced. Molars increasing in size through M!*, decreasing through M*4. M+ reduced. lack- PRIMITIVE ZYGOMATURINE SILVABESTIUS GEN. NOV. 199 FIG.5. Silvabestius johnnilandi gen. et sp. nov., paratype QMF30505, A, right lateral view. B, ventral view, C, dorsal view. Bar = 20mm. 200 MEMOIRS OF THE QUEENSLAND MUSEUM FIG. 6. Silvabestius johnnilandi gen. et sp, nov., paratype, QMF30505. Occlusal stereopair of upper cheektooth dentition. Bar = 10mm. ing enamel. Adult cheektooth row with an in- creasing molar gradient from M! to M4, Lower dentition. A single pair of lancealate, procumbent lower incisors in the Holotype. Enamel only on the lateral and ventral surfaces of the crown. A longitudinal groove occupying the internal dorsal surface, extending from the base of the incisor to the tip, approximately 2mm below and lingual to the dorsolateral tooth mar- gin. P3. Short, subovate, longer than wide, tapering anteriorly, dominated by a high, steep protoconid, with a short, steep, arcuate blade extending pos- teriorly from the protoconid apex, terminating in a small cuspid on the posterior cingulid. A poorly defined crest descending the anterior face of the protoconid, terminating in a slight swelling at the anterior tooth margin. Posterobuccal cingulum poorly developed. Lingual cingulum curving around the posterolingual tooth margin, from the apex of the posteromedial cuspid to a point mid- way between the protoconid apex and posterior tooth margin. A vertical flanking crest descend- ing the steep buccal face of the protoconid to a point just anterior to the terminus of the lingual cingulum. TABLE 2. Measurements (mm) of Silvabestius. L=length; AW=anterior width; PW=posterior width; *estimates. A. UPPER CHEEK DENTITION 9 d Species No. | Side m o Ma M3 Me L | wife |aw| pw] L | aw! pw] L | aw] pw L | aw| pw QMF | L | 12.2] 10.5 | 13.5 | 104 | 10.2 | 14.8 | 12.4 | 11.2 | 145/119] - | 12.0] 11.4] 9.0 johnnilandi (20204 | R [11.7 | 10.7 | 13.6 | 10.6 | 10.2 | 14.5 | 12.5 | 11.4 | 14.3 | 12.9 | 10.8 | 1.5 | 11.8 | 9.1 QMFL [13.1] 11.0] 144] 11.9 115 | 147/132 [118 [152/132 [114 [153 [12.8] 98 30505] r | 12.6| 10.9 | 13.6 | 11.7 | 11.4 | 15.0 | 12.3 | 11.2 | 15.3 | 12.8 | 10.8 | 15.5 | 14.1 | 10.7 | QMF|_L | 9.6 | 7.5* |10.4*| 9.4* | 9.0* | 12.9 |10.1*| 9.2* | 12.0 | 11.4*| 9.8* | 12.0 | 10.2 | 8.2 michaelbirti 20493 20431 r | 9.6 | 83 | 11.3] 9.7 | 93 | 110] 104 | 96 | 11.9 | 10.3 | 87 | 11.3 | 102] 8.1 sp. CPC | R [125] 93 | 135] 103] 107] - |124| - B. LOWER CHEEK DENTITION | QMF| L |108| 7.1 | 13.5| 81 | 89 |146| 9.6 |100 |146 | 11.3 | 9.6 |127| - | 92 Johnnilandi 30504 ; - r |103| 66 |137| 81 | 87 |153| 9.6 | 9.9 | 13.7] 11.0 | 10.2 | 12.9 | 10.8 | 9.3 PRIMITIVE ZYGOMATURINE SILVABESTIUS GEN. NOV. 201 FIG, 7. Silvabestius michaelbirti sp. nov., holotype, QMF20493. A, right lateral view. B, ventral view. C, dorsal view. Bar = 20mm. 202 MEMOIRS OF THE QUEENSLAND MUSEUM FIG. 8. Silvabestius michaelbirti sp. nov., holotype dentition. Bar = 10mm. Mj. Subrectangular, anteriorly-tapering, trans- versely bilophodont. Anterior protolophid short, steep, high, running parallel to a steep, low, pos- terior hypolophid. Protoconid higher than metaconid; entoconid higher than the hypoconid. Paracristid anteriorly directed, blade-like, linking the protoconid to the anterior cingulum, slightly buccally convex. Anterior cingulum curving lin- gually and downward, to the base of the crown, meeting a relatively indistinct preprotocristid. Deep anterior basin defined by the junction of these crests and the anterior base of the pro- toconid. A short, vague crest fading down the posterobuccal base of the protoconid. Similar, better-defined crest fading down the posterior face of the metaconid. A well-developed, an- terobuccally oriented prehypocristid terminating approximately 1/3 the way down the anterior face of the hypolophid. Lingual cingulum discontinu- ous, comprised of 2 ridges extending posterobuccally and posterolingually from the bases of the protoconid and hypoconid, respec- tively, with both ridges terminating just short of meeting each other in the interlophid valley. Sim- ilar short ridge at the posterobuccal base of the metaconid. Posterior cingulum short, not extend- ing around the bases of the entoconid and hypoconid. Transverse valley U-shaped in lin- gual view. , QMF20493. Occlusal stereopair of upper cheektooth Mp. Similar to Mj in all aspects except: slightly larger; protolophid subequal to its hypolophid; protolophid and hypolophid transversely wider; paracristid and prehypocristid extremely re- duced; all ridges associated with the main cuspids reduced; anterior face of the protolophid steeper; anterior cingulum better-developed; anterior basin absent. M3-4. Only partially erupted, similar to M2 in all aspects except: M3 less elongate, but wider than Mz; Mg reduced relative to M2-3; hypolophid re- duced relative to the protolophid, slightly offset relative to the protolophid, more posteriorly con- vex; interlophid valley more open and broadly U-shaped. REMARKS. Variation in the hypocone is com- mon among Oligo-Miocene zygomaturines. In fact, hypocone-less variants have been recorded for species of Neohelos (Peter Murray pers. comm.) and Nimbadon (Hand et al., 1993a), The development of a hypocone has, in the past, served as a synapomorphy of the Zygomaturinae. We have used this feature in the phylogenetic systematic analysis presented below because, al- though variable, it is generally a good indicator of phylogenetic relationships. PRIMITIVE ZYGOMATURINE S/ILVABESTIUS GEN. NOV. FIG. 9. Silvabestius sp., CPC7337. Occlusal view of right maxillary fragment with P?-M? from Site D Locality, Riversleigh. (After Tedford, 1967). Silvabestius michaelbirti sp. nov. (Figs 7-8) MATERIAL. Holotype QMF20493, a relatively com- plete cranium from late Oligocene Hiatus (South) Site (Creaser, 1997) with left and right cheek tooth rows; a longitudinal fracture runs through LM!-3; RM! is frac- tured diagonally from the anterolingual tooth corner to the posterobuccal tooth corner; L and RI! are present TABLE 3. Character state polarities for selected zygomaturine genera using species of Ngapakaldia (Palorchestidae) as an outgroup. A=absent; P=pres- ent; S=small; D=distinct,; W=weak; ST=strong; SQ=square; R=reduced; ==equal or sub- equal.PLESIO=plesiomorphic; APO=apomorphic. CHARACTER | PLESIO- APO- 1. Parastyle development on p? | A P 2. Hypocone development on p° | A/S D 3. Division of parametacone into two distinct cusps on P A F 4, Well-developed diagonal crest on arametacone af pe : A P 5. Link between protocone and arametacone on P 6. Mesostyle retracted towards cingulum 7. Size of P’ relative to equal molars S 8. Elongate P? A 9, Buccal cingulum on P? ST 10. Loss, incorporation or suppression of stylar cusps C and D with respect to A lophs 11. Small parastyle on M! 12. Occlusal shape of M' 13. Short posterior moiety on P4 14. Paracristid on M; L Y jomo 203 but incomplete; left 1? and L and RP are missing; the cranium is in two parts, the division occurring approx- imately 15mm posterior to the molar rows; the nasals, frontals and palate are incomplete; the basicranium is relatively intact; the posterior region of the neurocran- ium is missing on each side. ETYMOLOGY. For the former Vice Chancellor of the University of New South Wales, Professor Michael Birt, who assisted in the collection of specimens and helped meet the cost of preparation.. DIAGNOSIS. Differs from S. johnnilandi by: being smaller; having canines; having reduced parastyle and parastylar region on P? and less elongate P3; having a more central parametacone on P?; having a smaller protocone on P3 that is less distinctly separated from the base of the parametacone; lacking a hypocone on P3; and having steeper protoloph and metaloph faces on M14, Differs from Silvabestius sp. in: being higher crowned; lacking a well-developed parastylar region on P?; and having a reduced posterolingual cingular shelf on P?. DESCRIPTION. This species is described in so far as it differs from other species of the genus. Incisor arcade poorly-preserved with only the L and R I! (both of which are incomplete) and the right I? remaining. Incisors as in S. johnnilandi. Canines small, pointed, on the premaxillary- maxillary suture whereas in S. johnnilandi the tiny canine alveoli lieapproximately 2mm behind the premaxillary-maxillary suture. Left cheektooth row badly fractured with a large fissure extending across the buccal margin of M!, diagonally through M? and across the lingual region of M?, terminating at the anterior base of the hypolophid of M?. Right M! fractured diagonally from the anterolingual tooth corner to the posterobuccal tooth corner. P* smaller, lacking a distinct cuspate parastyle, with parastylar region reduced; vague pre- parametacrista continuous with the anterior cin- gulum; parametacone more posteriorly along the longitudinal axis, with postparametacrista more steeply inclined; protocone smaller, less dis- tinctly separated from the base of the parametacone; hypocone absent; protocone with- out crests. Upper molars. Lower-crowned, with more ex- tensive wear facets, with more steeply inclined posterior surfaces of the protoloph and metaloph; with a molar gradient not increasing posteriorly. 204 FIG. 10. Comparison of left P? of Silvabestius: A-A’, occlusal stereopair of S. johnnilandi, holotype, QMF30504. B-B’, occlusal steropair of S. johnnilandi paratype QMF30505. C-C’, occlusal ste- reopair of S. michaelbirti holotype QMF20493. Bar Silvabestius sp. (Fig. 9) MATERIAL. CPC7337, a right maxillary fragment with P3, M! and the anterior half of MŽ from late Oligocene Site D Locality, Riversleigh. MEMOIRS OF THE QUEENSLAND MUSEUM DESCRIPTION. See Tedford (1967). REMARKS. The teeth are missing most of their enamel so our understanding is based on dentine core morphology. Compared to Silvabestius Johnnilandi this specimen is lower-crowned (though the extent to which this is an artefact of enamel loss is not known); the protocone on P3 is smaller and is less distinctly separated from the base of the parametacone; and a hypocone is (most probably) absent on P?. This specimen is not designated a new species because of its poor preservation. The lack of enamel on the crown makes it difficult to interpret the stage of development of the parastyle and hypocone. Variation between adult and juvenile dentitions of S. johnnilandi, suggests the above mentioned differences in dental morphology may not warrant specific distinction. PHYLOGENETIC ANALYSIS Aplin & Archer (1987) and Marshall et al. (1989) recognised the Palorchestidae, divided into Palorchestinae and Ngapakaldinae, and the Diprotodontidae, divided into Diprotodontinae (sensu Archer, 1977) and Zygomaturinae. Archer & Bartholomai (1978) united these two families in the Diprotodontoidea. The monophyly of these groups is under question as well as the basis for their higher level association (Archer, 1984; Aplin & Archer, 1987; Murray, 1990a). Conse- quently, it is difficult to select an appropriate outgroup for phylogenetic analysis of the Diprotodontidae and to determine character state polarities within the group (Murray, 1990b; Hand et al., 1993a). We choose the palorchestids TABLE 4. Distribution of character states in representative diprotodontoids. 0= plesiomorphic state; 1=apomorphic state; 2=most apomorphic state; a=multistate character; ?=missing data. Taxon l 2 3 4 5 6 7 8 9 10 11 12 13 14 Neapakaldia 0 0 0 0 0 0 0 | 0 0 0 0 0 0 0 Diprotodontines 0 0 0 0 0 0 0 0 0 1 0 0 0 l Zygomaturus 2 1 1 0 1 1 a 0 0 1 2 l l 0 Kolopsis 2 l 1 0 1f I l 0 0 l 2 l l 0 Neohelos 2 1 0 0 a 1 l 0 0 1 2 1 1 0 Kolopsoides 2 2 1 0 l 0 l 1 l l 1 1 1 0 Nimbadon 2 A eg ao eile ae foo | ae a bean alo Plaisiodon 2 1 0 O | 1 0 l 1 l 1 a l 1 0 Alkwertatherium 2 0 0 0 0 0 1 0 0 1 0 l l l Silvabestius l 0 0 l 0 0 l 0 0 1 a 0 1 0 Raemeotherium ? ? ? ? ? ? ? ? ? ? ? ? ? 0 PRIMITIVE ZYGOMATURINE SILVABESTIUS GEN Ngapakaldia and Pitikantia as outgroup for anal- ysis of the Zygomaturinae due to their plesiomorphic position in the Diprotodontoidea. CHARACTER ANALYSIS. Fourteen dental characters considered taxonomically useful in zygomaturine intergeneric relationships are em- ployed in a cladistic analysis (Table 4). Selection of characters follows previous phylogenetic anal- yses by Murray (1990b) and Hand et al. (1993a). Character state polarities (Table 3) were deter- mined by outgroup analysis using Ngapakaldia and Pitikantia. Character state polarities for Nimbadon were scored only for Ni. lavarackorum and Ni. whitelawi, for reasons given in the discus- sion. Raemeotherium yatkolai was excluded from the phylogenetic analysis because 13 of the 14 characters are unknown for this species. A Wagner analysis was performed using the PAUP (version 3.1.1) computer program (Swofford, 1993). Multistate characters were treated as polymorphisms. Character optimisa- tion was performed using both the accelerated (ACCTRAN) and delayed (DELTRAN) transfor- mation algorithms. CHARACTERS EXCLUDED FROM THE ANALYSIS. For many characters examined from previous phylogenetic analyses, intraspe- cific and interspecific variation was found to be high. Similarly high variation is characteristic of the Oligo-Miocene zygomaturines Neohelos and Alkwertatherium (Murray, 1990b). Characters with a high degree of intraspecific variation ex- cluded from the analysis include: the degree to which the parastyle of P? is posteriorly inclined or ‘hooked’; shape of the lingual cingulum on P3; presence or absence of a protostyle on M!; extent of mpetaloph reduction on M*-*; extent of reduc- tion of M+. ‘Proportional similarity’ of P? as a synapomorphy for a Plaisiodon/ Nimbadon/ Neohelos/ Kolopsis/ Zygomaturus clade, used by Murray (1990b) and Hand et al. (1993a), is a character complex relating to the development of the parastyle, protocone and hypocone. Accord- ingly, it is not included as a discrete character. RESULTS. Wagner analysis using the branch and bound algorithm generated a single most parsimonious tree (Fig. 11) involving 23 steps with aconsistency index (CI) of 0.880, a retention index (RI) of 0.880 and a rescaled consistency index (RC) of 0.774. The high consistency index probably reflects the limited number of charac- ters used in the analysis. Tree topology was iden- . NOV. 205 3 » £ = z3 T m 3 5 2 B n = = 3 = = E S 5 = = è a = S 3 3 8 KA = = S = 5 = 8 Š = d = oo Š = Ea à È = S Š 5 = & & & a= S = E = S z S z a a =< zs. A 2 4 N %1 0-2 0->1 0>1 Ejo 0->1 FIG. 11. Hypothesis of zygomaturine relationships, based on dental characters (Table 3) used by Murray (1990b) and Hand et al. (1993) in cladistic analyses of the Diprotodontidae. Apomorphies are represented by boxed numbers. Arrows indicate character state transformations. Symbols: 0, plesiomorphic character state; 1, derived character state; 2, most derived char- acter state. tical irrespective of whether ACCTRAN or DELTRAN character optimisation was em- ployed. DISCUSSION Silvabestius is the most plesiomorphic zygomaturine known with the possible exception of Raemeotherium yatkolai. Silvabestius is basal to the diprotodontoid radiation. Although similar to the primitive palorchestid Ngapakaldia in den- tition (Tedford, 1967) and the middle ear, Silvabestius is assigned to the Zygomaturinae because it possesses a number of zygomaturine 206 synapomorphies including a parastyle on P? and a ventral alisphenoid tympanic process. Stirton et al. (1967) and Murray (1990b) sug- gested that zygomaturines evolved from primi- tive diprotodontine-like forms in which the parastyle on P? was small. Trends in premolar morphology in Silvabestius support this hypoth- esis. Transition from the simple, bicusped diprotodontine P? to the plesiomorphic, 3-cusped zygomaturine premolar of S. michaelbirti is achieved through expansion of the parastylar re- gion (resulting in a more elongate premolar) and the posterolingual cingular shelf. Subsequent transition to the more typical 4-cusped zygomaturine premolar of S. johnnilandi is achieved through development of a distinct parastyle on the enlarged parastylar region and development of a distinct hypocone on the en- larged posterolingual cingular shelf. Silvabestius is the plesiomorphic sister group to all other zygomaturines (Fig. 9). Plesiomorphic features of the dentition include a poorly-devel- oped parastyle, a poorly-developed hypocone, an undivided parametacone on P?, and a well-devel- oped paracristid on Mı. Monophyly of Zygomaturinae is supported by: a parastyle on P?; P? equal or subequal to the molars in length; and an enlarged posterior moi- ety on P3. The zygomaturine sister-group of Silvabestius is united by synapomorphic a mod- erate to large parastyle on P? and square vs elon- gate molars (in occlusal view). A distinct hypocone on P? and a well-developed parastyle on M! are shared by Nimbadon, Kolopsoides, Plaisiodon, Neohelos, Kolopsis and Zygo- maturus. The latter genera, with the exclusion of Nimbadon, is united by a linking crest between the apices of the protocone and parametacone on P3. Kolopsoides and Plaisiodon are united by an elongate P? and a weak buccal cingulum on P’. Neohelos, Kolopsis and Zygomaturus form a clade rationalised by a mesostyle that is retracted towards the cingulum on P* and by a large parastyle on M!. Kolopsis and Zygomaturus are united by a parametacone on P? that is divided into two distinct cusps. Topology of the cladogram (Fig. 9) is consis- tent with that found in Hand et al. (1993a) and Murray (1990b). However, it differs in the rela- tive positions of species of Nimbadon, Plaisiodon and Kolopsoides. These differences result from the exclusion of certain characters used in previ- ous analyses which are of low taxonomic value. A hooked or posteriorly inclined parastyle on P? has been used by Murray (1990) and Hand et al. MEMOIRS OF THE QUEENSLAND MUSEUM (1993a) as a synapomorphy uniting Plaisiodon and Nimbadon. In this analysis the character was found to be highly variable and hence of low taxonomic value. No special affinity was found between Nimbadon and the Kolopsoides/ Plaisiodon clade. Further, Hand et al. (1993a) united Nimbadon, Plaisiodon, Neohelos, Kolop- sis and Zygomaturus to the exclusion of Kolopsoides by the similarly proportioned P?. This character was excluded from the current analysis because it was found to be dependent on the degree of development of the parastyle, pro- tocone and metacone on P’. The position of Kolopsoides cultridens within the Zygomaturinae is difficult to determine, partly because of the paucity of knowledge about late Miocene diprotodontoids. Stirton et al. (1967) suggested that K. cultridens is most closely related to Kolopsis on the basis of a di- vided parametacone on P?. Flannery (1994) sug- gested that all New Guinea diprotodontids, including K. cultridens, are closely related and probably stemmed from a Kolopsis-like ancestor. K. cultridens is here linked with the late Miocene Plaisiodon centralis, from the Alcoota Local Fauna, a relationship first proposed by Archer (1984). If this relationship is accepted a divided parametacone has developed within the Zygo- maturinae in the Kolopsoides/ Plaisiodon lineage and again, convergently, in the Neohelos/ Kolop- sis! Zygomaturus lineage. Homoplasious division of the parametacone into 2 distinct cusps would then have developed during the late Miocene perhaps in response to increasing dryness. With plant materials becoming less succulent and more abrasive, there may have been strong selective pressure on all lineages to increase molariform features as well as the surface area of premolars. We suggest that Nimbadon scottorrorum is more appropriately assigned to Neohelos. Fea- tures of the dentition that align Ni. scottorrorum with Neohelos are: larger, squarer molars; more robust cingula on the cheekteeth; greater parastyle development on M*?; a relatively erect parastyle on P3; and an anterolingual crest ex- tending from the parametacone apex that meets a buccal crest from the apex of the protocone (a feature common in species of Neohelos) rather than continuing on to meet the anterolingual cin- gulum as in Ni. lavarackorum and Ni. whitelawt. Material referrable to a small species of Neohelos from Riversleigh’s System A Local Faunas, and forms intermediate between this small Neohelos species and N. tirarensis from Riversleigh’s Sys- tem B Local Faunas, are relatively indistinguish- PRIMITIVE ZYGOMATURINE SILVABESTIUS GEN. NOV. able interms of morphology and dimensions trom the Holotype and only specimen of Ni seattorrorum, QMF23157, Consequently, char- acter polarities for species of Nimbadon m this analysis were determined using Ni. lav- areckorum and Ni. whitelawi only, Similarities in dentitions of Silvabesrits johnnilandi and Nimbadon lavarackorum, both from Riversleigh, and Ni. whitelawi, from the Bullock Creek Local Fauna include: molar di- mensions; structure of the parastylar region; pro- tocone; hypocone; buccal cingulum; and anterolingual parametacone blade on P“. The last feature was suggested (Hand et al., 1993a) as an autapomorphy for Nimbadon but it is variably expressed in Neohelos und well-developed in Silvabesrius, Dentitions of S. johnnilandi and Nimbadon differ mainly in the degree of devel- opment of the parastyle on F? and the extent to which it is separated Irom the base of Lhe paru- metacone, Plesiomorphic features of Nimbaden include a small parastyle and undivided paru- melacone on P*, elongate vs relatively squarish molars, and a well-developed paracristid on Mj. Murray (1990p) suggested that Nimbadon may be a basal zygomaturine with affinity to Neohelos and Plaisiodan; he suggested 2 minor zygomatur- ine lineages, Nimbadon, Neohelos and Kolopsis, united largely through plesiomorphic features and Alkwertatherium, Plaiyiodon and Kolops- pides, united through phenctic similarities of the dentary and lower incisors. He also suggests that these lineages may be united through common ancestry in Nimbadon. In our analysis Nimbadon occupies a more plesiomorphic position than in mher analyses, Some doubt remains about whether Raemerotheriam yatkolai is appropriately in- cluded within Diprotodontidae. Rich et al. (1978) noted that it exhibited no apomorphic character states that precluded the possibility that it was a primitive macropodoid but, equally. neither does it exhibit any character states that are undoubted nicropodoid synapomorphies, The relative phy- logenctic positions of Silvabestius and Raemea- therimcannot be resolved until upper dentitions are discovered for the latter, Differences in the lower dentition of S. johnnilandi compared with R. yatkelai include; il is larger, lophids on the lower molars are less anteriorly convex; the ante- rios moiety of Mi is broader; the premetieristid of M_ is better developed as is the anterolingual basin at the anterior base of the protolophid; the pratasiylid (Rich etal., 1978) al the buccal base of the protoconids on My is absent; and the pru- 207 tolophid on M; is more postenorly inclined, Both species exhibit an Mı: Ma ratio (equivalent to M3; My in Rich et al., 1978) approaching 1, a condi- lion regarded as primitive among dipreto dontoids. There are no apomorphic features of the lower dentitions that enable either to be regarded as unambiguously more plesiomorphic than the other. Silvabestivs johnniland( is derived relative to §, michaelbini as evidenced by the following fea- tures of its P*:.a more distinct parastyle; a larger, lobate parastylar region: a better defined prepara- metacrista; und a smal) but distinct hypocene, The relative position of Silvabestius sp. is more difficult to determine because of poor preservi- lion, However, the follawing features of the den- tine core suggest that Silvabestins sp, as structurally intermediate between S. michaelbirti and S, jolhnnilandi, parastylar region more exten- sive than §. michaelbirti but reduced in compar- ison to $. johnnilandi; hypocone absent in 5. michaelbirti, a more extensive posterolingual cingular shelf in $, sp. a feature that anticipates the development of a hypocone in $. johnnilandi. Silvabestius sp, hus no autapomorphic features in the dentition that would preclude it being the ancestor of S. johmnilandi. Alternatively, Silvabestius sp. may represent at extreme end of an S. johnnilandi morphocling, If this is the case, the simple dental structure of S. michaelbirti and the absence of any aulapomorphic features of the dentition, suggest this species may represent the direct ancestor of S. johnnilandi and the structural antecedent of all other zygomaturines. ACKNOWLEDGEMENTS We thank Peter Murray and Richard Tedford for critically reading a dratt of this manuscript. We acknowledge support Irom: Australian Re search Grant Scheme, The University of New South Wales, National Estate Grants Scheme (Queensland), World Heritage Unit of the De- partment of Environment, Sports and Territories, Queensland National Parks and Wildlife Service (particularly Paul Sheehy), Waanyi People and Carpentarian Lund Council, JCT Australia, Ans- tralian Geographic Society, Queensland Mu- seum; Australian Museum, Century Zinc (particularly Doug Fishburn), Mount Isa Mines. Mount Isa City Council, Surrey Beatty & Sons, Riversleigh Society Ing., Blame Clark. Sue & Jim Lavuruck, Sue & Don Scou-Orr, Margaret Beavis and Martin Dickson; research colleagues notably Henk Gouthelp, Suzanie Hand, Bemie Cooke, 208 Alan Bartholomai, Phil Creaser, Peter Murray, David Ride. Anna Gillespie, Virginia O'Donoghue, Cathy Nock, Syp Praseuthsouk, Stephan Williams and Gill Goode, and numerous postgraduate students working on Riversleigh fossil materials, Particular thanks are due to Alan Rackham for assistance in the field. LITERATURE CITED APLIN, K. & ARCHER, M, 1987. Recent advanges n marsupial systematics wilh a new synererie clas- sification, Pp. xv-Ixxii. In M. Archer (ed), Pos- sums and opossums: studies in evolution, (Surrey Beatty & Sons and Royal Zoological. Society of NSW; Sydney). ARCHER, M. 1977. Ongins and subfamilial relation- ships of Diprotodon (Diprotodontidae, Marsupialia). Memoirs of the Queensland Mu- scum 18; 37-39. 1984, The Australian marsupial radiation, Pp. 633- ROR, In Archer, M. & Clayton, G, (eds), Verte- brate zoogeography and evolution in Australasia (Hesperian Press: Perth), ARCHER, M. & BARTHOLOMAIL, A, 1978. Tertiary mammals of Australia; a synoptic review, Al- cheringa 2: L-19. ARCHER, M. & BLACK, K. 1995. A moment of motherhood: a reconstruction of anancient drama, Riversleigh Notes 25: 4-5. ARCHER, M.. GODTHELP, H.. HAND, S.J. & MEGIRIAN, D. 1989. Fossil mammals of Riversleigh, northwesterm Queensland: prelimi nary overview of biostraugraphy, correlation and environmental change, The Australian Zoologist 25: 35-69. ARCHER, M., HAND, S.J. & GODTHELP, H. 1994. Riversleigh, 2nd Ed, (Reed Books: Sydney). FLOWER, W.H. 1867. On the development and suc- cession of teeth in the Marsupialia. Philosophical Transactions of the Royal Society of London 157: 631-641. HAND, 5.7, ARCHER, M., GODTHELP, H., RICH, T.H, & PLEDGE, N.S. 1993a, Niinbadon, a new genus and three new species of Tertiary zygomulurines (Marsupialia. Diprotodontidue) from northem Australia, with a reassessment of Neohelos. Memoirs of the Queensland Museum 33: 193-210, HAND, $.J., ARCHER, M., GODTHELP, H. 1993b. ?Ruemeotherium whitworthi, a new primitive zygomaturine (Marsupialia, Diprotodontidac) trom Tertiary limestones on Riversleigh Station, northwestern Queensland. (unpublished). MEMOIRS OF THE QUEENSLAND MUSEUM LUCKETT, W P. 1993. An ontogenetic assessment of dental homologies in therian mammals. Pp, 182- 204, In FS, Szalay, MJ. Novacek & M.C. Me- Kenna (eds), Mammal phylogeny: Mesozoic differentiation, maultituberculates, monolremes, early therians and marsupials. (Springer-Verlag: New York). MARSHALL, L.G., CASE, LA. & WOODBLIRNE, M.O. 1989, Phylogenetic relationships of the fam- ilies of marsupials, Current Mammalogy 2: 433- 502 MURRAY, P. 1990a. Primitive marsupial tapirs Pro- palorchesies nevaculacephalus Murray and P. ponticulus sp, nov.) from the mid-Misvene of north Australia (Marsupialia: Palorchestidae), The Beagle, Records of the Northern Teritory Museum of Arts and Sciences 7: 39-51. 1990h. Alkwertatheritom webbi, a new zygomatur- ine genus and species from the late Miocene Alcoota Local Fauna, Northern Territory (Marsupialia: Diprotodontidae). The Beagle, Re- cords of the Northern Territory Museum of Arts and Sciences 7: 53-80. MYERS. T, & ARCHER, M. 1994. A cautious correla- tion. Riversleigh Notes 23: 11, RICH, T.H., ARCHER, M, & TEDFORD, R.H. 1978. Raemeotherium yatkolai gen, et sp. nav, a primi- tive diprotodontid from the medial Miocene of South Australia. Memoirs of the National Mu- scum of Victoria 39: 85-9], STIRTON, R.A., WOODBURNE, M.O. & PLANE, M.D. 1967. A phylogeny of the Tertiary Diprotadontidae and its significance in correlu- tion, Bulletin of the Bureau of Mineral Resources, Geology and Geophysics, Australia $5: 149-160. TEDFORD, R. H. 1967. Fossil mammal remains from the Carl Creek Limestone, northwestem Queens- land, Bulletin of the Bureau of Mineral Resources, Geology and Geophysics. Australia 92: 217-237, TEDFORD, R.H, & WOODBURNE, M.O. 1987. The Hlariidae, anew family of vambatiform marsupials from Miocene strata of South Australia and an evaluation of the homology of molar cusps in the Diprotodontidac. Pp. 401-418. In M. Archer (ed,), Possums and opossums: studies in evolution. (Surrey Beatty & Sons and Royal Zoological So- ciety of NSW: Sydney), WOODBURNE, M.O., MCFADDEN, B.J.. CASE, J.A., SPRINGER, M.S., PLEDGE, N.S., POWER, J.D., WOODBURNE, J.M. & SPRINGER, K.B, 1994, Land mammal biostratigraphy and magnetostratigraphy of ihe Eiadunna Formation (Late Oligocene) of South Australia. Journal of Vertebrate Paleontology 13: 483-515. NIMIOKOALA GEN. NOV, (MARSUPIALIA, PHASCOLARCTIDAE) FROM RIVERSLEIGH, NORTHWESTERN QUEENSLAND, WITH A REVISION OF LITOKOALA K. BLACK AND M. ARCHER Black, K. & Archer, M., 1997:06:30. Nimiokeala gen. nov,( Marsupialia, Phascolarctidac) from Riversleigh, northwestern Queensland, with a revision of Lirokogla. Memoirs of the Queensland Museum 41(2); 209-228, Brisbane, [SSN 0079-8835. The early to middle Miocene phascolarctid Nimiokoala greystanesi gen. et sp. nov. is described from Riversleigh, northwestern Queensland. Nimiokoala sp. is recognised from a late Oligocene deposit in South Australia. The complex molar morphology and plesiomorphic basicranial morphology of Nimiokoala are indicative of a more diverse infraorder of phascolarctomorphians. Similarities in molar morphology between Nimiokoala and ektopodontids are noted and may reflect similar ecological niche. Litokoala kanunkaen- sis 1S described from the Miocene of Riversleigh extending its range and distribution. Cladistic analyses of dental characters of all living and extinct taxa support the intrafamilial relationships proposed hy Woodbume et al. (1987a). Litokoala is identilied as the sister group of the modern genus and Nimiokoala is mast closely related to the Litrokoala/Phascolarctos clade. C Phascolaretidae, Nimiokoala, Litokoala, Oligocene, Miocene, Riversleigh. K, Black & M, Archer, School of Biological Science, University af New South Wales 2052, Australia: received 4 November 1996. A partial skull, representing a small, highly specialised phascolarctid was collected from Boid Site East, System B deposits (Archer et al., 1989; 1991), Riversleigh, northwestern Queens- land, Based on a superficial analysis of dental morphology, the skull was incorrectly figured by Archer et al. (1991) as a new species of the Miocene phascolarctid Litekoala. Two species of Litekodla were known: L. kutjamerpensis (Stirton et al., 1967) from the Kuyamarpu Loca! Fauna, Wipajiri Formation, Lake Ngapakalui, South Australia (known from a single upper molar) and L kanunkaensis (Springer, 1987) Irom the Kanunka North Local Fauna, Etadunna Formation, South Australia (known from 2 iso- lated teeth and some molar fragments). N. greystanest gen. et sp. nov. ts similar to Litokeala but its generic separation is substanti- ated by comparison with new material (an M? and a lower dentition) of L. kKanunkaensis from Sys- tem C (Archer et al.,1989; 1991) at Riversleigh. A further new species of Nimiokvala is recognised from South Prospect B Locality, Lake Namba. Frome Downs Station, South Australia. The moderate abundance of N. greystanesi in Riversleigh deposits contrasts with the paucity of material of other phascolarctids which are known from isolated teeth or, at best, dentitions. This paper describes dentition of the new genus and analyses phascolarctid phylogeny based on den- tal characters. Suprageneric nomenclature follows Aplin & Archer (1987). Dental terminology follows Archer (1978b) except that what was then under- stood to be the hypocone of upper molars is now atcepted to be the metaconule (Tedford & Woodburne, 1987). Cheek tooth homology is that proposed by Luckett (1993), Biostratigraphic no- menclature follows Woodburne et al. (1993), Archer et al. (1989) and Archer et al. (1991). Specimens referred to are housed in the following repositories: Queensland Museum (QMB), South Australian Museum (SAMP), University of Cal- ifornia at Riverside (UCR), Measurements were made using a Wild MS microscope with digital length measuring set. SYSTEMATICS Order DIPROTODONTIA Owen, 1866 Suborder VOMBATIPFORMES Woodburne, 1984 Infraorder PHASCOLARCTOMORPHIA Aplin & Archer, 1987 Family PHASCOLARCTIDAE Owen, 1839 Nimickoala gen. nov. TYPE SPECIES. Nimiokoald greystanesi sp. nov. ADDITIONAL SPECIES. Nimiokoula sp. DIAGNOSIS. Nimiokeala differ from all other phascolaretids in: being smaller (except species 210 of Litokeala), Waving a large. cuspale parestyle which is pyramidal in occlusal view (except Litokoala kutjaimarpensis), a well-developed, erescentic paraconule and neometaconule on M- 4 the latter of which is double cusped in the more posterior molars; a more reduced metacone and metaconule on M* and a more rounded posterior margin of that tooth. Robust, ridge-like erenulations; strong posterobuccal ridges from the apices of the pro- tocone and metaconule; a discontinuous neometaconule subdivided into two or more parts and a (variably) discontinuous paraconule; a posteralingual cusp on Ps; a well-developed posterolingual erest from the apex of the pro- tostylid on Mi; a weaker metastylid; a well-de- veloped neomorphic cuspid occupying the trigonid basin between the metavonid and pro- toconid an Mza; an anterobuccally directed pre- enlucristid (as opposed to anterolingually directed in other phascolarctids); a well-devel- oped cuspate entostylid ridge; and a more posterobuccally directed postprotostylid cristid. Nimiokeala greystanesi differs from Koubor int being higher crowned, lacking the bicusped P3; having strong posterolingual paracristae and posterolingual metacristae: having a pratostyle; and in having reduced stylar cusps and Irom Madakoala in: being higher crowned; more cren- ulated; in having a more cuspate, isolated posterolingual cusp on P* and in lacking the lin- gual cingular ridge of that tooth (in M. sp. cf. M. wellsi); having wider, less elongale upper molars, a more buccal junction of the premetacristae and postparacristae; u well-developed protostyle; rel- atively smaller buccal surfaces of the paracone and metacone; lacking the transverse connection between the metacone and metaconule on M%; having a less elongate Py that consists of three main apices (as opposed to four in Madakoala), a small posterolingual cuspule and a proportion- ately larger, more isolated posterobuccal cusp an that tooth; lacking the well-developed buccal und lingual crests from the main apices of Py; having a larger, more cuspate protostylid on Mi; a more ingually situated protoconid and a more lingual junction of the eristid obliqua and posiprotocristid in the Mj; a well-developed an- terobuccal ridge off the entoconid apex: and in lacking the hypoconid-entoconid crest on M24, Nimiokoala difters from Perikoala in being less crenuilated) lacking the lingual shelf basal to the metaconid and entoconid on the lower molars; and in those features listed for Madakeala. Nimiokoala differs from Litokeala, MEMOIRES OF THE QUEENSLAND MUSEUM Phascolarctos ahd Cundokoala yorkensis in: having a bulbous, less trenchant p; lacking the bulbous cuspule at the anierolingual base of the metaconule of M! (M! unknown for L. hanunkaensiy) and the resultant pocketdeveloped between the premetaconule erista, posiprotocrista and the lingual margin ; lacking the anterolingual protocone crest on M! lacking a metastylid fold: lacking buccal ribs on the pro- toconid, metaconid and entocomid; having a dis- continuous lingual ectolophid wherein the postmetacristid and preentocristid do not meet: having a weaker entostylid; and m having less separation between the protoconid and paraconid on MiMi unknown for L kutjamarpensis). Nimiokoula greystanesi differs Irom L. hutjamarpensis in lacking the buccal extension connecting the paraconule to the buccal margin, Nimiokoala differs from L. Aanmunkaensis in: hav- ing a posterolingual cuspid on P3!.a greater sepa- ration of the anterior and posterior apices ol Ps and a more cuspate anterior apex of that tooth; lacking the lingual and buccal ribs from the main apices on Ps; lacking an entoconid lingual shelf (postemostylid cristid) and metaconid lingual shelf; and in lacking posterolingual protoconid ridge on My, Nimiokoala differs from Phascolarvtos and Cundekoala yvorkensis in: having a large posterolingual cusp on P? and strong anterior, buccal and lingual crests that extend trom the anterior apex of that tooth; lacking the lingual cingular ridge of P3; having proportionatcly wider upper molars; lacking the pocket at the anterior base of the protocone on M!; having a proportionately less elongate Px having a large posterobuccal cusp on P3 and strong anterior and buceal crests extending from the most anterior apex of that tooth; having a preprotostylid cristiá on Mı that is not continuous with the anterior cingulum; lacking the transverse connecuon be- tween the apices of the metaconid and protoconid on Mz; lacking the columnar stylids of the metaconid and entoconid on M}.4; and in lacking the lingual ridge connecting, the bases of the protoconid and hypoconid on M2.4. REMARKS. Nimiokoala sp. is known only from a dentary fragment with M2.4. Comparisons in- volving the upper teeth of this genus are therefore confined to features of N. greystanesi sp. nov. ETYMOLOGY. Latin nimia, excessive: refers w the complex molar morphology relative to other phascolarctids. MIMIOKO4L4 GEN. NOV,, NEW KOALA FROM RIYERSLEIGH Nimiokoata greystanest sp. nov, (Figs 1-3; Table 1) ETYMOLOGY. For Greystanes High School, winner of the Sydney Morning Herald) Riversleigh competi- lian, MATERIAL. Holotype QMF30482, crunial trigment with parts of the left and right maxillae, juguls and palatine, part of left premaxilla and partof right frontal. The left alveoli of 1-3, PS, M1 and right PS, MI, from Neville’s Garden Site, System B (Archer ct al., 1989), early to middle Miocene. Other material: Boid site cast- OMF30483 partial skull with the leftand right premaxillae, nasils, palatine, part of the lett and right Jugals, right frontal, part of right parietal, part of basisphenoid and basioccipital, part of right tympanic bulla, left 1-3, C}, P3 and anterior halfof M! right H, alveoli of I3, alveolus of C!, P3 (missmg buccal margin) ind M* (which is missing the enamel fromthe parastylar corner); QMF30484, right M24; OMP30485., lett M2; QMP30486, left MŽ fragment with posterior part of metuconule and the neometaconule preserved; QMF30487, right dentary fragment with alveolus of l, broken P3, M14. Neville's Garden Site - QMF30512, right M$; QMF23026, lefi M?; QMF24232, left M2; OMF24257, right M2; QMF24233, left M4: QMF20901. len M4 aml right M*: QMF23027, left M* and right M9; OMF29624, right dentary with P3, M1-3; OMF30488, lel) Mi; QMF30489, right M1; QMF24266, right M1; QMF24265, right M2; QMF20903, left M2, Camel Sputum Site - QMF30490, left P3; QMP30491, left dentary with tj, alveolus of P3 and alveolus of Mj-2; QMF30492, broken right M3; QMF24351, right P3. Inaheyance Site - QMF30493, left dentary fragment with T, P3, Mj-2, M3 missing the anterobuccal comer, Ma missing the buccal margin of the protoconid and the posterobuccal tooth comer. Dirk's Towers Site - QMF245 16, molar fragment; QMF24291, left M? frag- ment; Rat Vomit Site - OMF30494, left dentary frag- ment with P3, Mj-3 (all missing the lingual margin), and alveolus of M4. Upper Site - QMF30495, lef P>; QMF30496, right P3; QMF30497, right MÌ. RSO Site ~ QMF30498, left M3; OMF30499, talonid of M3, DIAGNOSIS. NM. preysranesi differs from Nimiokoala sp. in having: less rounded lower nywlars; less lingually sloping surfaces of the metaconid and entoconid; a slightly larger en- lostylid ridge; a larger neomorphic cuspid in the ingonid basin in Mo.4) and a lesser reduction of the talonid in My, DESCRIPTION. incisors. Left and right T! of QMF30483 short, pointed, with convergent tips, with enamel restricted tò the anterior tooth lace, with posterolingual faces dominated by large, triangular wear facets, Left 2 small, subowate in all occlusal view, Gepering anterolingually, without enamel on the occlusal surface. Lefi 1° small, pointed, peg-like, contucting posterobuccal cor- ner of the letl 1°, with enamel preserved on the buccal face. Canines. Left canine short, peg-like, appruai- mately 4mm behind the upper incisor arcade, P?. LP relatively robust, bulbous, wider poste- Hourly than anteriorly, with 5 majors cusps, 3 of which lie along a Slightly crescentic longitudinal crest with an additional posterobuccal and posterolingual cusp, The anterior most cusp 1.61mm posterior to the anterior tooth niurgin, Prominent crests extending anteriorly, posteri- orly and buccally from its apex; anterior crest bifurcating into anterior and posterolingual spurs; spurs extending towards the base of the crown, Medial cusp separated from the anterior cusp by a deep crevice, connected to the posterior cusp by a short longitudinal crest, Apex of the medial cusp 2.32mm posterior to the anterior tooth mar- gin, Well-developed crest extending anterobuce- ally from the medial apex, fading into the base of the crown. Posterior cusp 3.19mm behind ante- rior tooth margin, with well-developed posterior crest extending from its apex: bifurcating at the base of the cusp, with one arm continuing poste- riorly and fading into the base of the crown; other cingulum-like arm curving around the lingual tooth margin, connecting to the posterior crest of the well-developed posterolingual cusp- Posterdlingual cusp apex opposite and slightly posterior to the medial cusp, with an an- terolingual ridge fading into the lingual base of ihe crown from the apex. Posterobuccal cusp opposite, bul slightly anterior to the posterior apex, with a well-developed crest extending posterobuccally from the apex, and fading into the base of the crown. An additional small cuspule at ihe anterior base of the posterolingual cusp, RES similar to LP?, but with less worn, medial cusp, smaller posterolingual cusp having weaker posterior crest extending from below the cusp apex and more distinct posterobuccal cusp, Upper premolars (QMF30490, QMF30495, QMF3(M9A) resemble P’? of the holotype excep! for: buccal and lingual crests of the anterior cusp, the anterobuceal crest of the medial cusp and the posterobuccal crest of the posterobuceal cusp are reduced in QMF30495; undivided posterior cresi of the posterior cusp fading tito the anterior hase of the crown in QMF30495, QMF30490 and QMF30496; small cuspule at the anterior base of ihe posterolingual cusp absent in QMF30495 und N N MEMOIRS OF THE QUEENSLAND MUSEUM FIG. 1. Partial skull of Nimiokoala greystanesi gen. et sp. nov., QMF30483: A, ventral view; B, right lateral view; C, left lateral view. Bar = 10mm. QMF30496; enamel missing from most of the duced buccal crest of the anterior cusp, terminat- occlusal surface of larger QMF30490 with re- ing before reaching the base of the crown. M!. Buccal margin of left M! slightly concave, NIMIGROALA GEN. NOV. NEW KOALA FROM RIVERSLEIGH sloping anterolingually opposite the metacone; apices. of the metacone and paracone slightly overhanging their lingual bases: lingual bases of paracone and metacone more lingually sloping in QMF30483 and QMF30512 than in the holotype; Mmetacone laller than paracone; metaconule taller than protocone, parastylar corner, posteriot anil anterior base of metacone, posterior base of paracone, buccal bases of protocone and metaconule, lingual bases of paraconule and neometaconule and transverse valley crenulated. Yaraconule well-developed, crescentic, cuspale, occupying the longitudinal valley between paracone and protocone, bifurcate anteriorly, main arm extending anterabuceally and connect- ing to the anterolingual base of the parastylar comer, with two shorter, anteriorly directed spurs terminating at the base of the anterior cingulum and meeting the anterior cingulum an QMF30483. Posteriorly paraconule connecting to a strong posterolingual paracrista and a well- developed crest extending posterobuccally trom the protocone apex: this connection more ad- vanced in LM?*, Paraconule not hifureate poste- riurly in RM! of holotype (or in QMF30483) bul connecting to the posterolingual paracrista. Pos- leriorly bifurcate paraconule in QMF30512 with- out connection to the posterobuceal protocone crest. Large, crescentic, bicusped neometaconule oecupying longitudinal valley between metacone and metaconule in M'. Neometaconule in LM?* divided into Iwo distinct cuspate V-shaped struc- tures. Neometaconule highly variable. In LM! of holotype connected to premetaconule crista by shorLanteriorspur (notin RM!), with second spur extending anlerobuccally from the neometaconule and terminating in the transverse valley. In QMF30512 anterobuccal arm continu- ous with well-developed transverse ridges occu- pying the transverse valley of that tooth. Neometaconule connecting to anterolingual metacrista in RM! of QMF30483. Bifurcating posteriorly in LM! of the holotype: Short posterolingually directed spur meeting a well-de- veloped crest extending posterobuccally from the metaconule apex (these structures not meeting in RM! of the holotype); main arm of the yeometaconule extending posterobuccally, meet- ing both the posterolingual metacrista and the posterior cingulum. Neometiconule not meeting posterolingual metaerista or posterior cingulum in QMF30483 and OMF305 12. Slightly crescen- lic preparacrisia extending unterobuceally from the paracone apex to the buccal margin, hifurcal- ing, with one arm Continuing amerobuceally to meet the postparastylarerista; second arm exter ing posterobuceally, connecting with stylar cusp B. Slightly longer postparacrista bifurcatinyg: al the buccal margin, with one arm continuing posterobuccally to meet the premetacrista at the buccal margin; with second arm extending un- terobuccally (a meet stylar cusp C at buccal tooth margin. Paracone buccal basin closed on M! of holotype, variably closed in other specimens. Metacone with stylar cusps D and E redeceu (stromgeron QMF305 12); with buccal basin shal- low and open. Height of the stylar cusps decreas- ing progressively trom the purastyle to E. Strong posterolingual paracristae, anlerolingual metacrisiae and posterolingual metacrista in the upper molars of the holotype. Anterolingual purucrisiae in QMF30512, QMF20901 apd OMF23027, Protacone apex opposite but slightly posterior to the paracone apex, slightly lingual relative to the metacanule apex. Creseenjic postprotocrista and premetaconule crista meeting in the transverse valley well buccal to the lingual tooth margin. A series of short spurs extending buceally from this junction. terminate in the transverse median valley. Protostyle well- devel- oped, at the anterobuccal base of the protocane. extending posteriorly from the preprotocrista, ter- minating at the buccal base of the protoconc opposite the apex of the paraconule. Surluce enamel crenulations in the transverse valley ol the LM! aligned into two parallel discontinuous) Iransverse crests, contrasting with the reticulare pattern in RM! and more posterior molars of holotype, Distinet transverse crests occupying the transverse valley in QMPF30483 and QMF30498 , M?**. Similar in most aspects to LM! except for- teeth tapering more posteriorly, triangular buccal surfaces of the paracone and metacone sloping more steeply 10 the buccal margin; buccal margin more concave, buccal margin of the metacone sloping more posterolingually, patastyle absent; more reticulate crenulations, reduced stylar cusps; a deep pocket formed by the junction of the premetaconulecrista, postprotocrista, the posterobuccal crest ol the protlocone and the posterolingual arm of the paraconule; protostyle reduced, connecting to the paraconule: neometaconule reduced, divided into 2 distinet V-shaped structures; neometaconule mecting un- terolingual metacrista, a vague unterolingual paracrista in LM? M? (Fig, 2B), Paracone, paraconule and pro- tocone at similar stage of development to M*of the holotype. Metacone and metaconule ¢x- 214 tremely reduced to small cuspules on the posterobuccal and posterolingual margins, re- spectively. Posterior margin crescentic, consist- ing of a series of ridges extending anteriorly into the transverse valley. Premetacrista forming part of the posterior margin, curving anterobuccally around the tooth margin to meet the postparacrista at the buccal exit of the transverse valley. Neometaconule reduced, indistinguisha- ble from the enamel crenulations. Right dentition of the holotype similar to the left except for: smaller RM!; less convex buccal margin; paracone and metacone sloping more gently to the buccal margin; deeper and less cren- ulated transverse valley lacking parallel trans- verse ridges; protocone higher than the metaconule; paraconule and neometaconule taller and less crescentic; paraconule apex closer to the protocone, not bifurcate anteriorly, without connection to the posterobuccal crest of the pro- tocone; neometaconule not bifurcate anteriorly; protostyle weaker; anterolingual paracrista vaguely developed in the right M!. LOWER DENTITION (Figs 2C, 3A-C). I, Inci- sor narrow, lanceolate, with enamel covering ventral and buccal surfaces of the anterior half, with fine shallow longitudinal groove linearly along the buccal margin terminating approxi- mately Imm posterior to anterior tip of tooth. P3. LP3 small, short longitudinally, of 3 aligned main cusps (anterior, medial and posterior), a large posterobuccal cusp and a small posterolingual cusp; lingual surface sloping stee- ply to the lingual tooth margin; shallow crevice separating the anterior cusp from the medial cusp; medial apex connected to the posterior cusp by a short longitudinal crest; prominent anterolingu- ally directed crest and posterobuccally directed crest extending from the anterior apex and fading into the base of the crown; prominent crest con- necting posterior cusp to posterior cingulum; apex of the posterobuccal cusp opposite and slightly posterior to the posterior cusp. Short an- terobuccally and posterolingually directed crests extending from the apex of the posterobuccal cusp. Lower molars distorted, with metaconid and entoconid anteriorly displaced relative to the pro- toconid and hypoconid, respectively. Mı. Subrectangular, anteriorly attenuated; hypoconid largest cusp on the talonid; protoconid largest cusp on the trigonid; well-developed pro- tostylid at the buccal margin, opposite and slightly posterior to the protoconid apex. Pro- MEMOIRS OF THE QUEENSLAND MUSEUM tostylid and hypoconid pyramidal in occlusal out- line in QMF30493 but more rounded in QMF30494. Preprotocristid extending an- terolingually from the protoconid apex to a rela- tively small paraconid at the anterolingual corner (in QMF30488 paraconid connected to the pre- metacristid by a short longitudinal crest). Pre- metacristid poorly developed. Linear postprotocristid extending posteriorly into the central basin, meeting the crenulated cristid ob- liqua slightly lingual to the longitudinal tooth axis; postmetacristid extending posterolingually towards the weakly developed metastylid at the lingual tooth margin, with a well-developed posterolingually directed ridge branching off and extending into the transverse valley separating the talonid and trigonid. Crescentic pre- protostylid cristid curving anterolingually along the anterior tooth margin, connecting to the an- terobuccal base of the protoconid, this connection slightly more lingual in QMF30488 and slightly more buccal in QMF30489. Postprotostylid cristid extending towards the lingual margin, ter- minating at the anterior base of the hypoconid; less developed in QMF30488 and QMF30494, fading into the posterior base of the protostylid. Well-developed posterolingually developed crest extending from the protostylid apex, terminating at the posterior end of the longitudinal crevice separating the protostylid and protoconid, more crescentic and with an additional short lingual ridge extending from the protostylid apex in QMF30488. Preentocristid extending an- terobuccally (as opposed to anterolingually in other phascolarctids) into the transverse valley separating talonid from trigonid. A second crest extending anterobuccally from the entoconid apex, running parallel to the preentocristid. An additional buccally directed crest extending from the entoconid apex into the valley separating the bases of the entoconid and entostylid ridge. Postentocristid extending posteriorly along the lingual margin to a poorly developed entostylid at the posterolingual corner of the tooth. Large cuspate ridge in the longitudinal valley between the entoconid and hypoconid analogous to, al- though more strongly developed than, the en- tostylid ridge in Litokoala and Phascolarctos. Entostylid ridge bifurcating anteriorly; two an- terobuccally directed ridges extending from its apex, meeting the cristid obliqua. Bifurcating posteriorly: two posterolingually directed ridges terminating at the base of the posterior cingulum. Entostylid not bifurcate in QMF30494. M2-3. M2 and M3 similar to Mı except for: NIMIOKOALA GEN. NOV., NEW KOALA FROM RIVERSLEIGH 21 Ww FIG. 2. Nimiokoala greystanesi gen. et sp. nov. A-A’, QMF30482, holotype, occlusal stereopair of rostral region. B-B’, QMF23027, right M4, occlusal stereopair. C-C’, QMF30487, right dentary fragment. occlusal stereopair. Bar = 5mm. protostylid absent; paraconid absent; preproto- cristid continuous with anterior cingulum; postprotocristid linear, extending posterolingu- ally towards its junction with the cristid obliqua; premetacristid more prominent, continuous with the anterior cingulum; entostylid reduced result- ing in a more rounded posterolingual tooth cor- ner; entostylid ridge reduced, not bifurcate; entoconid reduced in M3; hypoconid reduced; preentocristid and posterobuccal metacristid joining, closing the transverse valley well buccal to the lingual tooth margin. A neomorphic cuspid in the longitudinal valley between the metaconid and protoconid in M2-4, with a well-developed posterolingually directed ridge and a weaker an- terobuccally directed ridge extending from its apex, more strongly developed and more sepa- rated from the metaconid in M24 of QMF30487 than of QMF30493. My. My similar to Mo.3 except for; entostylid ridge further reduced, connecting posteriorly to the posterior cingulum; neomorphic cuspid re- 216 MEMOIRS OF THE QUEENSLAND MUSEUM FIG. 3. Nimiokoala greystanesi gen. et sp. nov. A-C, QMF30493. A-A’, occlusal stereopair of left dentary. B, buccal view. C, lingual view. Bar = 10mm. duced; postprotocristid and cristid obliqua not meeting end to end but parallel each other (as in L. kanunkaensis and some P. cinereus); talonid severely reduced resulting in a greater anterior displacement of the entoconid and metaconid (relative to the hypoconid and protoconid, respec- tively) (as in L. kanunkaensis and some pseudocheirids). REMARKS. QMF30482 (Fig. 2A) was chosen as NIMIOKOALA GEN. NOV., NEW KOALA FROM RIVERSLEIGH 217 TABLE 1. Dentition measurements (mm) for Nimiokoala greystanesi sp. nov. and Nimiokoala sp. (last line SAMP19952 only). L= length; W=width; AW, anterior width; PW, posterior width; *=estimate. A. UPPER CHEEK DENTITION Spec. "A r3 Ml Ma L|w| L Jaw L [aw] pw QMF30482 | L | 4.50 | 3.98 645 | 5.94 i Holotype | R | 4.48 | 3.51 | 5.99 | 5.74 5. omrsodes lasi [395] ~ [539 : [| R | 4.63 | 3.83 | 6.30 | 5.47 | 5. QMF30484 | R 4.60 | 4.48 | 2.12 QMF30490 | L | 4.87 | 4.15 [ QMF30495 | L | 4.52 | 3.68 QMF30496 | L | 4.46 | 3.53 QMF30512 | L 6.81 | 5.99 QMF24267 | R QMF24232 | L QMF30485 | L QMF30498 | L | QMF23026 | R 5.52 QMF30497 | R 5.28 | 5.28 | 4.72 QMF24233 | L 4.12*| 3.96 | 1.91 Pt ie = 4.47 | 4.24 | 2.65 R 4.67 | 4.32 | 2.62 amo E | 4.68 | 4.69 | 2.81 4.29 | 4.16 | 2.30 B. LOWER CHEEK DENTITION QMF30487 |_R_| 4.10 | 3.00 | 5.70 | 3.15 | 3.69 | 5.77 | 3.28 | 3.62 | 5.56 | 3.19 | 3.28 | 5.07 | 3.16 | 3.00 QMF30493 3.35 | 2.44 | 5.95 | 3.54 | 3.40 | 5.85 | 3.40 | 3.43 | 5.50] - [3.04]s10/ - | - QmF30494 | L | 3.60 | 2.46 /5.37| - | 3.41 | 5.86 | 3.22 | 3.28 | 5.57 | 3.17 | 3.05 QMF29624 | L | 3.86 | 2.54 | 5.91 | 3.07 | 3.48 | 5.75 | 3.26 | 3.34 | 5.53 | 3.30 | 3.12 QMF24351 | R | 3.21 | 2.25 VL tet QMF24266 | R 6.22 | 3.29 | 3.66 QMF30488 | R 6.19 | 3.42 | 3.66 | QMF30489 | R 5.96 | 3.45 4.04 | O MF20903 | L | | 6.31 | 3.55 | 3.92 | QMF24265 | R | | [ 5.84 | 3.27 | 3.30 el | SAMP19952| L | | 5.04 | 2.77 | 3.00 | 5.18 | 2.95 | 2.88 | 4.32 | 2.53 | 2.19 holotype because it contains most of the upper dentition well-preserved and relatively unworn. The holotype is thought to represent a juvenile because of the shorter premaxilla in comparison with the adult skull, QMF30483. Nimiokoala sp. (Fig. 4, Table 1) MATERIAL. SAMP19952 left dentary with alveoli of P3 and M1, M2-4 intact from late Oligocene to middle Miocene South Prospect B locality, Lake Namba, Frome Downs Station, South Australia. REMARKS. Enamel is missing from the lingual margins of the metaconids and entoconids and the buccal margins of the protoconids and hypoconids. Features of the M2 are difficult to discern because of extreme wear in that tooth. Only points of difference from N. greystanesi are noted here. Lower molars are proportionately smaller (by approximately 15%) and more rounded, espe- MEMOIRS OF THE QUEENSLAND MUSEUM FIG. 4. Nimiokoala sp., SAMP19952, from South Prospect B Locality, Lake Namba, South Australia. Occlusal stereopair of right dentary, Bar = 10mm. cially the talonid. The neomorphic cuspid be- tween the metaconid and protoconid in M>-4 is weaker, as is the entostylid ridge. The lingual surfaces of the protoconid and metaconid slope more lingually. The talonid of Mg is further re- duced. This taxon is left in open nomenclature in the absence of the upper dentition. The features men- tioned above as differences between the central Australian and Riversleigh materials vary signif- icantly between individuals of N. greystanesi and may not be reliable species indicators. Litokoala Stirton, 1967 TYPE SPECIES.Litokoala kutjamarpensis Stirton, 1967. OTHER SPECIES. Litokoala kanunkaensis Springer, 1987. DIAGNOSIS. Litokoala differs from other phascolarctids in having a posterobuccal crest extending from the apex of the metaconid on M2-4 (although this crest is reduced on M4); a well-developed posterolingually directed crest from the protoconid apex of Mg. Lirokoala differs from other phascolarctids except Phascolarctos and Cundokoala yorkensis in: having a neomor- phic cuspule at the anterior base of the meta- conule of M!; an anterolingual protocone crest on M! and consequently a much squarer an- terolingual margin of that tooth; a metastylid fold wherein the postmetastylid cristid is continuous with the preentocristid; an anteriorly bifurcate preentocristid on the lower molars; and internal ribs on the metaconid and hypoconid of Mj. Litokoala differs from all other phascolarctids except Nimivkoala in having: a well-developed crescentic paraconule and neometaconule on upper molars; an anteriorly displaced entoconid NIMIOKOALA GEN. NOV., NEW KOALA FROM RIVERSLEIGH 219 FIG. 5. Litokoala kanunkaensis, Henk’s Hollow Site, Riversleigh Station. A-A’, QMF30502, RM? occlusal stereopair. B-D, QMF30500, RP3. B-B’, occlusal stereopair. C, buccal view. D, lingual view. Bar = 5mm. (relative to the hypoconid) on M4; an anteriorly displaced metaconid (relative to the protoconid) on M4; and a parallel arrangement of the postprotocristid and cristid obliqua on M4 wherein these cristids do not meet end to end. Litokoala differs from Koobor in: being smaller; having less rounded bases of the paracone, metacone, protocone and metaconule; having less elongate molar proportions; reduced stylar cusps; a paraconule on M?; and a neometaconule on M!»?. Litokoala differs from Madakoala and Peri- koala in: being smaller; higher crowned; more selenodont; having a larger neometaconule; a more crescentic, less linear paraconule; an en- tostylid ridge in the lower molars; a more lingual junction of the cristid obliqua and postprotocristid; a larger developed protostylid and a more lingual protoconid on Mj; a less strongly developed, more buccally positioned paraconid on Mj); and a more strongly developed entostylid. Litokoala differs from Perikoala in: lacking the well-developed stylar border of the paracone and metacone; lacking the additional stylar cusp an- terior to stylar cusp E; having a weaker entoconid lingual shelf; and a weaker premetacristid. Litokoala kutjamarpensis differs from Madakoala and Perikoala in having a buccal extension of the paraconule which connects to the buccal margin of M!. Litokoala kanunkaensis differs from Madakoala and Perikoala in: lack- ing the M2 protostylid ridge; having a less elon- gate P3; a more prominent posterobuccal cusp; and having 3 apices on the longitudinal crest (4 in Madakoala devisi; 5 in M. wellst) on P3. Litokoala differs from Nimiokoala in: having less steeply (more buccally) sloping buccal sur- faces of the paracone and metacone; lacking the division of the neometaconule in the more poste- rior molars; lacking the well-developed posterobuccal crest from the apex of the metaconule; having a weaker posterobuccal crest from the protocone apex; lacking the posterolingual ridge off the apex of the pro- tostylid on Mj; having a less cuspate, more pos- teriorly positioned entostylid ridge; having an anterolingually (as opposed to anterobuccally) directed preentocristid; a more posterolingually directed postprotostylid cristid; a lingual en- toconid shelf; and a more strongly developed entostylid. Litokoala kanunkaensis differs from species of Nimiokoala in: lacking the posterolingual cusp on P3; in having a weaker anterior cusp on P3; buccal and lingual ribs which extend from the main apices of the longitudinal crest in the P3; and in lacking the neomorphic cuspid between the metaconid and protoconid of M>.-4. Litokoala differs from Phascolarctos and Cundokoala yorkensis in: being smaller; having well-developed anterolingual metacristae; hav- ing proportionately less elongate lower molars; a less pronounced entoconid basal shelf; and in lacking the ridge connecting the buccal bases of the protoconid and hypoconid of M2-4. 22 MEMOIRS OF THE QUEENSLAND MUSEUM Litokoala kanunkaensis Springer, 1987 (Figs 5-6, Table 2) MATERIAL. Holotype SAMP32397 (=[UCR21926) a right M2 from latest Oligocene UCR Locality RV-8453 Kanunka North Local Fauna, Etadunna Formation, west side of Lake Kanunka, South Australia. Other material: from Kanunka North Site: UCR21945, aright M4; UCR21980, a metacone of a right M3; UCR21979, a metacone of LMI; late Oligocene. From Henk’s Hollow Local Fauna: QMF30500, right P3; QMF 13079, right dentary fragment with posterior half of P3, M1-2; QMF30502, right M3; from Gag Site: QMF30501, right M1; from Gotham Site: QMF30503, MX fragment containing paracone and buccal half of protocone. From JC9 Site: QMF20809, right M3; Sys- tem C, Riversleigh, middle Miocene (Archer et al.. 1989; 1991). DIAGNOSIS. Litokoala kanunkaensis differs from L. kutjamarpensis in: having a short longi- tudinal spur connecting the paraconule to the anterior cingulum; lacking the bulbous cuspule at the anterolingual base of the metaconule; lacking the protostyle (the latter two features may be result of differences along the tooth row rather than interspecific differences); lacking the dis- tinct posterolingual and anterolingual paracrista; having a strong anterobuccal crest from the metaconule apex. COMPARISON WITH THE HOLOTYPE. Prior to this study L. kanunkaensis was known from two isolated lower molars, an M2 and Mg. Refer- ral of the Riversleigh material to L. kKanunkaensis is based on the similarity of the M2 of QMF13079 to the holotype. Some small differences are: M2 of QMF13079 is slightly larger; posterobuccal metaconid strut is poorly defined and bifurcates in the trigonid longitudinal valley, one spur fad- TABLE 2. Measurements (mm) for right lower denti- tion of Litokoala kanunkaensis from Riversleigh’s System C deposits. Abbreviations as for Table 1. | Py L QMF Ww AW| PW No. 30500} 4.12) 3.10) 30501 13079] —_|— 20809 3.57 5.48 339/340 ing buccally into the protoconid base and the other spur continuing posteriorly then turning sharply lingually before becoming part of the molar crenulation pattern; the entostylid ridge is less cuspate and arises from the posterobuccal base of the entoconid (unlike the holotype in which it arises from the postentocristid); and the transverse median valley, the posterior base of the protoconid, the cristid obliqua and the transverse valley separating the bases of the protoconid and hypoconid are more crenulated than in the holo- type. QMF 13079 M2 is significantly more worn than the holotype which may account for the poorer crest definition. Analysis of intraspecific varia- tion in Phascolarctos cinereus and Nimiokoala greystanesi indicate that these differences do not warrant specific distinction. Most variable fea- tures in the modern species include: molar size; pattern and degree of molar crenulations; cristid obliqua; entostylid ridge. Although the entostylid ridge may be phylogenetically significant in phascolarctids, it is highly variable. It may be a well-developed cuspate structure, relatively in- distinguishable as a series of discontinuous cren- ulate ridges, completely isolated at the base of the entoconid or variably connected to the postentocristid, the base of the entoconid or the. posterior cingulum. Hence differences between the holotype and QMF13079 are within intraspe- cific variation. DESCRIPTION. P3 (Fig. SB-D). Short, semi-rec- tangular, of 4 major cusps anteriorly, medially, posteriorly and posterobuccally on the crown. Anterior, medial and posterior cusps on a longi- tudinal crest extending the length of the crown. Anterior cusp tallest, 1/3 of way along longitudi- nal crest, above the anterior root of P3. Anterior, buccal and lingual crests extending from its apex; anterior crest fading into the anterolingual base of the crown; lingual crest fading down the lin- gual tooth margin, curving posterolingually at its tip; buccal crest short, fading into the buccal tooth margin. Anterobuccal, anterolingual and posterolingual crests extending from the poste- rior cusp apex; anterobuccal crest poorly devel- oped, extending into the valley between the posterior and posterobuccal cusps, meeting a short lingual ridge from the posterobuccal cusp apex; anterolingual crest better developed, fading down the lingual margin; posterolingual crest curving anterolingually along the base of the crown, becoming cingulum-like, fading into the base of the posterior cusp. Posterobuccal cusp NIMIOKOALA GEN. NOV., NEW KOALA FROM RIVERSLEIGH 221 FIG. 6. Litokoala kanunkaensis. A-A’, QMF13079, right dentary fragment with posterior half of P3, M1-2, occlusal stereopair. B-B’, QMF30501, RM}, occlusal stereopair. C-C’, QMF20809, LM3 occlusal stereopair. Bar = 5mm. opposite but slightly posterior to the posterior cusp, with a short posterolingual ridge extending from its apex and terminating at the posterior end of the valley between the posterior and posterobuccal cusps, with short, anterobuccal crest extending from its apex and fading towards the base of the crown. M; (Fig. 6B). Semi-rectangular, tapering slightly anteriorly. Protoconid most worn of the major cusps, with well-developed protostylid op- posite but slightly posterior to apex. Short, linear cristid extending posteriorly from the apex of the protostylid into the transverse valley, bifurcating into posterolingual and posterobuccal arms, both arms terminating in the transverse valley between the talonid and trigonid. (In QMF13079 postprotostylid cristid more crenulated, bifurca- tion less distinct and more a part of the crenula- tion pattern). Anterolingual cristid from protostylid apex crescentic, well defined, contin- uous with the anterior cingulum extending to the anterolingual corner of the tooth. Lingual rib from the protostylid apex short, fading into its base. Crescentic, cingulum-like ridge extending posterobuccally from the posterobuccal base of the protostylid, fading into the anterior base of the hypoconid. Preprotocristid extending an- terolingually to the paraconid, a moderately de- veloped, slightly cuspate structure at the anterolingual tooth corner, with a posterolingu- ally directed ridge fading down the lingual mar- gin towards the base of the crown. Anterobuccal crest continuous with the anterior cingulum/pre- protostylid cristid. Postprotocristid extending posteriorly meeting cristid obliqua slightly lin- gual to the longitudinal axis of the tooth. Postmetacristid extending posterolingually from the apex of the metaconid to the lingual margin, swelling at this point to form the metastylid. Posterolingual ridge from the metastylid short, fading down the lingual tooth margin, better de- veloped in QMF13079. Postmetastylid cristid a short, buccal ridge, continuous with the pre- entocristid, resulting in the metastylid fold. En- toconid sub-pyramidal. Preentocristid extending anteriorly towards the transverse axis, curving lingually, bifurcating with one spur extending lingually to meet the postmetastylid cristid and other extending anterobuccally into the trans- verse valley to lingual base of the junction of the postprotocristid and cristid obliqua. QMF13079 (Fig. 6A-A’) with a short anterolingual spur ex- 2 tending from the junction of the postprotecristid and cristd obliqua to the anterobuccal spur of the preentocristid. A well-developed anterohuccal ridge extending from the entoconid apex, termi- nating at Lhe anterobuccal base of the entocomd, just lingual to the anterior arm of the entostylid ridge, Postentocristid curving posterolingually from the entoconid apex to the posterolingual loofh corner, swelling at ths point to form asmall entostylid. A short anterolingual ridge lading down the lingual tooth margin from the entostylid apex, better developed in QMF13079. A short posterobuceal ridge extending from the èn- tostylid apex, comlinuous with the posterior cin- gulum. Well-developed cuspute entostylid ridge at the posterobuccal base of the protoconid, fad- ing anteriorly along the longitudinal tooth axis, with a shorter posterior spur terminating before meeting Lhe posterior cingulum. QMF13079 with entostylid ridge extending buccally from the postentocrisnd, similar in this respect to the ho- lotype. Hypoconid largest of the main cusps. Posthypocristid extending posterolingually along ihe posterior tooth margin, becoming continuous with the posterior cingulum, Cristid obliqua long, well-developed, crenulated, relatively linear, ex- tending anterolingually from the hypoconid apex, meeting the postprotocristid on the transverse valley slightly lingual to the longitudinal axis. Talonid basin, transverse valley between the talonid and trigonid and buccal base of the en- toconid and protoconid crenulated. Lingual base of the hypoconid and protostylid lightly crenu- lated, Protostylid occupying the anterobuecal corner of the tooth, lying well buccal to the lon- giiudinal axis. Short lingual ribs extending from apices of the entoconid, hypovonid, protoconid and protostylid. Short buccal ribs off the pro- toconid and metaconid apices, less prominent in the holotype, M: (Fig. 6C-C") is similar to the holotype èx- cept for: being slightly larger, having buccal bases of the protoconid and hypoconid more rounded, the lingual shelfof the metaconid better developed, the cristid obliqua and postprotocristid meeting at a more lingual posi- lion. the anterobuecal entoconid ridge and the posterobuceal metaconid ridge poorly developed (possibly due to wear) and the entostylid ridge extending buecally from the postentosty lid cristid rather than [rom the postentoeristid. M? (Fig, 5A-A‘), Length 5.6mm, anterior width 5.22mm: posterior width 4.2mm, Sub- quadrate, tapering slightly posteriorly, with shal- low surface enamel crenulations in the transverse MEMOIRS OF THE QUEENSLAND MUSEUM valley and posterior base of paracone, Buccal bases of the protocone and metacone large, rounded, with lingual bases of all major cusps sloping gently lingually. Triangular buccal sur- face of paracone reduced relative to the metacone (less marked in QMF30503). Buccal margin of the paracone sloping amterolingually; buccal mar- gin ofthe metacone sloping slightly posterolingu- ally in occlusal view. Buccal basins of the paracone and metacone shallow and open. Buccal basin of the paracone deeper than that of the metacone, Stylar cusps B and C well-developed on the paracone, Stylar cusps on the metacone poorly developed, stylar decrease progressively from B to E. In QMF30503 stylar cusp C more strongly developed, almost closing the paracone buceal basin, Preparacrista relatively short crest, extending anterobuccally to the anterobuccal tooth corner, bifurcating, with one arm curving bucvully and becoming continuous with the ante- rior cingulum, the other short ridge extending posterobuccally, fading along the paracone buc- cal Margin to a shght swelling representing stylar cusp B, Longer, relatively linear postparacrista extending posterobuccally to the buccal margin, mecting premetacrista, closing the buccal exit of the transverse valley. Stylar cusp Da slight swell- ing at the buccal tip of the premctacrista. Postmetacrista linear, extending posterobuccally from the metacone apex q0 the posterobuccal tooth corner, meeting the posterior cingulum. Paraconule at the anterolingual base of the paracone, slightly cuspate at this point, bifureat- ing anteriorly, with main linear arm extending anterobuccally, terminating at the anterolingual base of the paracone before meeting the anterior cingulum, witha minor arm connecting the ante- tior cingulum. Linear posterior arm of the paraconule extending posteriorly along the longi- tudinal valley between the paracone and pro- locone, biturcaling at a point just anterior to the transverse median valley: A shon posterobuccal arm mecting the postéerolingual paracrista; sec- ond arm continuing posteriorly into the trans verse median valley, A cuspate, anteriorly hifureate neometaconule at the anterolingual base of the metacone, with the main crenulated arm extending anterobuceally, meeting the an- lerolingual metacrista atthe base of the metacone, A shorter arm fading lingually from the neometaconule apex into the longitudinal valley ag the buccal base of the hypocone. Posteriorly, the linear arm of the neometaconule fading into the longitudinal valley between the metavone und metaconule. Buceal surfaces of the protocone and NIMIOKOALA GEN. NOV., NEW KOALA FROM RIVERSLEIGH Koobor Madakoala Pertkoala Nimiokoala Litokoala Phascolarctos AES Bhb»ı [9 Jos1 [io]0->1 [i3]o>1 [16]0>1 [4 Jos1 [6 ]o+1 [s]o>1 [ii] 1>2 1->2 [pa] o->1 [7] 0>1 7 jos1 [is] o+1 ft Jos1 [5 Jo>1 [i2jo>1 [u]o->1 FIG. 7. Phylogeny of the Phascolarctidae based on cladistic analysis of dental characters (Table 3). Apomorphies are represented by boxed numbers. Ar- rows indicate character state transformations. 0 = plesiomorphic character state; | = derived character state; 2 = most derived character state. metaconule sloping gently buccally. Pre- protocrista long, crescentic crest, extending an- terobuccally, curving more buccally into the longitudinal valley, meeting the anterior spur off the paraconule, becoming continuous with the 223 anterior cingulum, The slightly more crescentic postprotocrista extending posterobuccally to- ward the transverse valley, turning sharply bucc- ally, meeting the short, slightly crescentic prehypocrista at the buccal end of the transverse valley between the metaconule and protocone. Prehypocrista bifurcating just posterior to the transverse valley, with one arm continuing an- terobuccally to the postprotocrista, with the other crenulated arm extending buccally, then an- terobuccally into the transverse median valley. Junction of the postprotocrista and prehypocrista effectively sealing off the lingual exit of the trans- verse valley well-in from the lingual margin. Two short ridges at the posterolingual base of the protocone and the anterolingual base of the metaconule. Posterolingual base of the protocone slightly crenulated and cingulum-like. The long, slightly crescentic posthypocrista extending posterobuccally, becoming continuous with the posterior cingulum. A relatively well-developed anterobuccal ridge from the metaconule apex, fading towards its base. Posterobuccally directed ridge from the protocone apex fading into the posterobuccal base of the protocone. Posterolingual paracrista bifurcating at its base, one arm meeting the anterobuccal arm of the paraconule and continuing into the transverse median valley, with the other extending posterobuccally into the transverse median val- ley. Anterolingual metacrista bifurcating at the anterolingual base of the metacone, one arm con- tinuing anterolingually to meet the anterobuccal arm of the neometaconule, the other extending anterobuccally into the transverse valley between the paracone and metacone. PHYLOGENETIC ANALYSIS Seventeen dental characters which provide phylogenetically useful data on phascolarctid re- lationships are analysed (Tables 3, 4). Character polarities were determined using selenodont ilariids and subselenodont wynyardiids as out- groups. CHARACTER ANALYSIS Thirty potentially useful dental characters were reduced to 17 following analysis of variation in the living species. The Pleistocene Cundokoala yorkensis Pledge, 1992 is not significantly differ- ent in dental morphology from Phascolarctos; generic distinction is not warranted. It is here regarded as a separate species of Phascolarctos. TABLE 3. A list of character state polarities (PLES= plesiomorphic state; APO= apomorphic state) for the Phascolarctidae using ilariids and wynyardiids (Marsupialia: Vombatimorphia) as oulgroups. A=ab- sent; P=present; LI=linear; C=crescentic; LA=large; R=reduced; W=weak; S=small; M=moderate; B=buccal; LIN=lingual third of trigonid CHARACTER PLES | APO | |.Bladed premolar | A P 2.Trenchant premolar A P 3. Posterobuccal cusp on pP P A 4. Structure of paraconule | LI C 5. Development of neometaconule JA | P 6. Protostyle development | A P 7. Stylar cusp development LA | R 8, Posterolingual paracristae | A/W | P 9. Neomorphic cuspule at base of A p metaconule 10. Anterolingual protocone crest A | P 11. Parastyle development on M! S M/L 12. Protostylid development on Mı _| A P 13. Metastylid fold mpm A TE 14. Entostylid ridge A P BE Position of protoconid on M, 4 B LIN 16. Buccal ribs on lower molars j| A P - 7 - - 1 17.Anteriorly displaced entoconid relative to hypoconid and anteriorly displaced A P metaconid relative to protoconid in My l. Premolar cusp pattern: P3 in most phascolarctids (e.g. Madakoala, Perikoala, Nimiokoala, Litokoala, Phascolarctos) has a ten- dency to be bladed, as opposed to bicusped and weakly bladed in Koobor. Woodburne et al. (1987a) regard the bicusped P3 as plesiomorphic among phascolarctids but its occurrence in other vombatiform groups (e.g. ilariids and wynyardiids) suggests that it is apomorphic in phascolarctids. 2. Trenchant P* Phascolarctos has a trenchant MEMOIRS OF THE QUEENSLAND MUSEUM P? in contrast to a bulbous P? in Koobor, Madakoala, Perikoala and Nimiokoala. Nariids and wynyardiids have a bulbous, non-trenchant P? suggesting the trenchant P? in Phascolarctos is apomorphic. 3. Posterobuccal cusp of P The small posterobuccal cusp on P? in Koobor, Madakoala, Perikoala and Nimiokoala is lacking in Phascolarctos. It is absent in wynyardiids and variable in ilariids. Its occurrence in most phascolarctids and ilariids suggests that it is plesiomorphic among phascolarctids. 4. Crescentic paraconule: The paraconule var- ies among phascolarctids from a small, linear structure in Madakoala and Perikoala to moder- ately developed in Phascolarctos (although not uniform), to a large crescentic paraconule in Litokoala and Nimiokoala occupying the longitu- dinal valley between the paracone and protocone. It is a moderate swelling at the anterolingual base of the paracone in M! of Koobor; however, it is suppressed on M?* as in ilariids. A paraconule is absent in wynyardiids but a small swelling at the anterolingual base of the paracone in ilariids. Hence, absence of a crescentic paraconule is re- garded as plesiomorphic among phascolarctids. 5. Neometaconule: Absent in Koobor , weakly developed (or absent) in Madakeoala and Peri- koala, variable in Phascolarctos, large and cres- centic in Litokoala and Nimiokoala (although reduced in the more posterior molars) and double cusped in the latter. Its absence is regarded as plesicmorphic in koalas because of its absence in ilariids and wynyardiids. 6. Protostyle: Present in Phascolarctos, Nimiokoala and Litokoala although it generally diminishes in size from M! to M4. It is absent in Madakoala, Perikoala, Koobor, ilariids and wynyardiids which suggests its presence is apomorphic. 7. Stylar cusp development: Large stylar cusps are thought to be primitive among TABLE 4. Character state distribution within Phascolarctidae. Abbreviations: 0, plesiomorphic state; 1, apomorphie state; 2, more derived apomorphic state; ?, signifies missing data. Tilz|[3 false rz} | 9 wt 14 | is [16 | 17 Phatoeniaell 1 j i 1 0 | l | | l l l 1 2 l 1 EA 0 Litokoala | 1 C t h MUSI PINAL [at a P zaia a O A Nimiokoala l ojojiı i | l 1 | 0 IK PPIE 1 | o | 1 Perikoala | ıļojojojıfjoļıjojojojojıjojojıjojo Madakoala | 1,0 | 0 0 l 0 0 : Oo | 0 | 0 | 0 l | 0 | 0 0o 0 | 0 Koobor |01010];/o0]/o0lololololfoj1i1l[7>}/o]7] 2/7] 2 NIMIOKOALA GEN. NOV., NEW KOALA FROM RIVERSLEIGH diprotodontians (Rich & Archer, 1979), reflected in the enlarged stylar cusps of ilariids and wynyardiids. Large stylar cusps occur in Koobor and Madakoala. Although Perikoala has reduced stylar cusps, the stylar region is represented by a longitudinal crest that appears to subsume rela- tively large stylar cusps. Those of Nimiokeala, Litokoala and Phascolarctos (excluding the parastyle of M!) are suppressed relative to all other phascolarctids. Because of the large stylar cusps in ilariids and wynyardiids, reduction in some phascolarctids is regarded as apomorphic. 8. Posterolingual paracristae: Well-developed and aligned with the preparacristae in Nimiokoala, Litokoala and Phascolarctos but ab- sent (or weakly expressed) in Koobor, Madakoala, Perikoala, ilariids and wynyardiids. Absence is regarded as plesiomorphic. 9. Cuspule at the anterolingual metaconular base of M': A bulbous cuspule lies at the an- terolingual base of the metaconule on M! of L. kutjamarpensis (M24 unknown). It is variable in Phascolarctos but is absent in all other phascolarctids and ilariids and wynyardiids sug- gesting that absence is plesiomorphic. 10. Anterolingual protocone crest of MY An anterolingual crest extends from the apex of the protocone towards the base of the crown on M! of L. kutjamarpensis and Phascolarctos resulting in a relatively square protocone base. This crest is absent in all other phascolarctids, in wynyardiids and in ilariids. Absence is regarded as plesiomorphic among phascolarctids. 11. Parastylar development on M" The parastyle (‘stylar cusp A’) is regarded as a sepa- rate character (distinct from character 7) because it undergoes independent evolutionary change relative to stylar cusps B through E. Small on M! of Madakoala and Perikoala; moderately devel- oped in Koobor, Litokoala and Phascolarctos; large in Nimiokoala. The small parastyle of ilariids and wynyardiids suggests that this is the plesiomorphic condition. 12. Protostylid: The protostylid on M; has de- veloped independently in pseudocheirids (Woodburne et al., 1987b), macropodoids (Archer, 1978a) and ilariids (Tedford & Woodburne, 1987). In koalas, the protostylid ranges from weak (Madakoala, Perikoala) to very large (Nimiokoala, Litokoala, Phascol- arctos). Lower molars of Koobor are unknown. Because a protostylid is absent in wynyardiids (Pledge, 1987a) and probably ilariids (Tedford & Woodburne, 1987 consider both alternatives), a protostylid is interpreted here as derived. N N n 13. Metastylid fold: An autapomorphy of the Phascolarctidae. Phascolarctos and Litokoala have a metastylid fold in which the post- metastylid cristid is continuous with the pre- entocristid. Perikoala and Madakoala lack the metastylid fold which is represented by a swell- ing at the posterior tip of the postmetacristid. Nimiokoala lacks a metastylid fold and the metastylid is reduced relative to other phascolarctids. Because it is absent in ilariids and wynyardiids it is difficult to determine polarity, However, considering that discrete metaconids and entoconids is the plesiomorphic condition in all marsupial groups, the condition within the Phascolarctidae in which these cusps are linked by blades, in this case via a metastylid, is inter- preted as derived. 14. Entostylid ridge: Present in L. kanunkaensis and Phascolarctos and a well-developed cuspid in an homologous position in Nimiokoala, Its absence in the lower molars of ilariids and wyn- yardiids suggests that absence is plesiomorphic. 15. Position of the protoconid on Mi: Woodburne et al. (1987a) used this to determine koala intrafamilial relationships, with the more plesiomorphic koalas having a less lingual pro- toconid because of weaker protostylid develop- ment. The protoconid is within the lingual third of the trigonid of M; in Perikoala, Nimiokoala, Litokoala and Phascolarctos, Considering the buccal protoconid in wynyardiids and ilariids, the more lingual position in some phascolarctids is considered to be derived. 16. Ribs on the conids of lower molars: Internal ribs on the protostylid, metaconid, protoconid, entoconid and hypoconid of lower molars is apomorphic and shared by Litokoala and Phascolarctos. These ribs are absent in all other koalas. Presence is considered derived. 17. Anteriorly displaced entoconid and anteri- orly displaced metaconid of Ma: Torsion of Mg such that the entoconid is displaced anteriorly relative to the hypoconid and the metaconid is displaced anteriorly relative to the protoconid occurs in Litokoala and Nimiokoala but no other phascolarctids, This condition is absent in ilariids and wynyardiids; absence is interpreted plesiomorphic. RESULTS Wagner analysis using the branch and bound algorithm produced a single most parsimonious tree involving 22 steps with a CI of 0.864 and a RC of 0.746 (FIG. 7). DISCUSSION Recent classifications (e.g, Woodburne, 1984, Aplin & Archer, 1987) group the 6 fossil and living genera (14 species) of koalas in the Phascolarctidae; however, they also predict a large morphological range of kKoalu-like animals, Woodbume (1984) erected the Superfamily Phascolarctoidea and Aplin & Archer (1987) the Infraorder Phascolaretomorphia, cach containing only the Phascolarctidae. Complex molar mor- phology and plesiomorphic basicranial morphol- ogy of Nimiekeala indicate a more diverse infraorder of phascolarctomorphians. In some aspects of dental morphology, Ninuokoala appears to have converged on pseudocheirids and ektopodontids. Slight torsion is evident in lower molars of Nimiokeala and the consequent arrangement of the metaconid and entoconid is similar to some pseudocheirids (e.g. Psendochiraps archeri). Springer (1987) alsa noted more similarities in the My of L. Kanunkeensis to pseudocheirids than to any other phascolarctid. This i$ also the case with the Mg of Nimiokoala. Well-developed parastyle, large paraconule and large ncometaconule of the upper molars of Nomokoala and the entostylid ridge and neomorphic cuspid on the lower molars are rem- miscent of the cuspate Joph(id)s of cktopodontid molars; in particular the lower molars of Darcius dugg Rich, 1986 from the Pliocene Hamilton Local Fauna, Victoria, Archer (1976) suggested that the ¢ktopodontid molar pattern may have evolved through an alignment of the wrinkles, conules and crenulations in the molars of seleno- dont forms and in particular, the koalas. The high crowned, highly ¢renulate, complex molars of Numiekoala support this view, Recent clussilications (Archer & Aplin 1987; Marshall et al, 1989) support an ekfopadontid/phalangenid affinity and place the Ektopodontidae within the Phalangeroidea based on the shared strongly angulate enistid obliqua and reduced metaconid on the lower molars and the loss of P1. Consequently, similarities in phascolarctid and ektopodontid molars suggest similar ecological niches, Pledge (1982) sug- gested several possible dietary preferences of ektopodontids, ranging from browsing and fru- fivoraus to granivorous and inseclivorous, Sim- ilar dietary specialisations could explain the complex denutions of Nimiokeala, Since the first fossil koala was deseribed in 1957 there have been two different attempts to MEMOIRS OF THE QUEENSLAND MUSELIM analyse intrulamilial relationships. Archer (19784) separated phascolarctids into Litekoala, Koobor and Perikoalal Phascolarctos lineages. Litrokeala kutjamarpensis was considered to be the most plesiomorphic phascolarctid based on ihe metaconule and lingual buttresses on the metacone of the upper molar (Archer, 1978a). Perikoala and Phascolarctos were interpreted to be more closely related to cach other than either was tu L kutjamarpensis, However, Archer (1978a) also noted a large structural distance separating these groups. Pertkoala exhibited a number of plesiomorphie features relative to Phascolarctos and several autapomorphic leu- tures that precluded it being antecedent to Phascolarctos. Molar morphology of Keeber led to its interpretation by Archer (19784) as the sister-group of a combined Phascolarctos! Pert koala clade. Woodburne et al.'s (1987a) cladisuc analysis followed a reassessment of character state morphoclines for phascolaretids. Relative posi- tions of Phascolarctos, Litokoala, Perikoala and Madakoala are confirmed herein (Fig. 7). Rela- tionships of Lirokoala have been unclear due to its poor fossil record, Archer (1978a) regarded it as the plesiomorphic sister group of all phascolarctids based on a single upper molar. Re-unalysis of that tooth by Woodburne et al. (1987a) and Springer’s (1987) 2 isolated lower molars and upper molar fragments of L. kanunkaensis Suggested Litokeala was more closely related to Phascolarctos. Litokeala male- rial from Riversleigh supports Woodbume etal,’s (1987a) hypothesis. Features now found to support the Litokeala and Phascolarctos relationship include the neomorphic cuspule at the anterolingual base of the metaconule on M! (and M2" in Phascolarctos). (he metastylid fold in which the postmetastylid cristid is a continuous fold with the preentocristid and the internal ribs on the cuspids of lower molars, Features, regarded hy Springer (1987) as autapomorphies of L. kanunkaensis but which ure in fact variably pres- ent in Phascolarctos include anteriorly bifurcate pre-entocristid, reduced talonid on Mg and paral- lel arrangement of the cristid obliqua and posiprotocristid on My wherein these cristids do not meet end to end, The latter two features are also variably present in Nintiokoala. Litokaala is a derived relative to Phascolarctos in having a more crescentic paraconule and neometaconule in the upper molars, a well-developed metacomd NIMIGKOALA GEN. NOV . NEW KOALA FROM RIVERSLIIGH posterobuccal crest on M2.4and in having a well- developed posterolingual protocristid On Ma. Néniokoala, Liiokoala and Phascolarctos form a glade by sharing a protostyle in (he upper mo- ines, strong poOsterolingual paracristac, a large protostylid on M; and a well-developed en- tostylid ridge. Litokoala and Nimiokoala are de- rived relative to Phascolarctos in the shared erescenuc paraconule and neumetaconule in M'4 and a well-developed parastyle on M!. Lirokoala and Nimiokoala synapomorphies, interpreted as convergences (Fig.7) are: well-developed, highly erescentic paraconule and neometaconule, large, pyramidal parastyle and extreme torsion on My such that the entoconid and metaconid are dis- placed anteriorly relative to the hypovonid and protoconid, respectively, Nimiokoala is derived relative to other phascolarctids in having strong posterobuccal crests from the apices of the pre locone and metaconule, a discontinuous neometaconule which 1s subdivided into two cus- pate parts, a posterolingual cuspule on F°. a well- developed posterolingual protostylid cris}id on Mi. weak metastylid in the lower molars, an anterobuceally directed preentocristid (an- terolingually directed in other phascolarctids) and a neomorphic cuspid occupying the trigonid busin between metaconid and protuconid on Mp. There are doubts about the inclusion of Koobor in Phascolarctidae, Pledge (1987b) suggested that Koobor is allied to ilariids such as Kuterimtja ngama, Figure 7 may support the hypothesis that it tics outside the Phascolarctidae. Clarification of Keobor's relationships requires discovery of lower molars because these differ significantly in ilariids and phascolarctids, Except for 2 partial skulls from Riversleigh, extinct phascolarctids are represented by isolated teeth or dentitions, As a result, current under- standing of the basicranial region is based on the modern species. Phascolarctos cinereus has an autapomorphic bilaminar bulla wherein the tym- panic cavity is roofed by both an alisphenoid tympanic process and the squamosal epilympanic wing (Aplin, 1987), The partially preserved basi- cranial region of Nimiokoala greystanesi (Fig, LA-C) exhihits the plesiomorphie diprotodontian condition wherein the alisphenoid forts the roof al the tympanic cavity, suggesting a large struc- tural distance between these koalas. The plesiomorphi¢ basicranium of N. greystanesi sup- purts the hypothesis that phascolarctids diverged fram near the base of the diprotodontian tree (Archer, 1976). Intragencric relationships ol Lirekovwle are dit t PJ J ficult to interpret due to a lack of comparable material between the species. Litokodld kar unkaensiy appears to be plesiomerphic relalive to L, kutjumarpensis because it lacks a proto style and the bulbous cuspule at the anterolingual base of the metaconule on M3, However, these features may be a result of changes along the tooth ruw (ie. an artefact of comparing an M* with an M’) rather than interspecitic differences. Litokaala kanunkaensis Yrom early-iniddle Miocene System C (Archer et al., 1989; L991) is the same age or slightly younger than, the Kutjamarpu Local Fauna. ACKNOWLEDGEMENTS Wethank Neville Pledge and Mike Woodbume who read a draft of this paper, Linda Gibsim, Australian Museum for access 16 collections and Ross Arnett for assistance with photography. Riversleigh research is supported by the Austri- lian Research Grant Scheme; National Estate Grants Scheme (Queensland); University of N.S.W.; Commonwealth Department of Environ- ment, Sports and Territories; Queensland Na- tional Parks and Wildlife Service; Commonwealth World Heritage Umit; ICI Aus- tralia; Australian Geographic Society; Royal Zoological Soviety of N.S.W., Linnean Society of N.S.W. Century Zine) Riversleigh Society; Elaine Clark; Margaret Beavis; Martin Dickson; Sue & Jim Lavarack; and Sue & Don Sceott-Orv- Field support came from hundreds of volunteers and staft and students of the University of N.S.W Skilled preparation of Riversleigh material hes been cared oul by Anna Gillespie, LITERATURE CITED APLIN, K. 1987, Busicranial anatuniwy of the early Miocene diprotodontian Wyeyardia bassinn (Marsupialia; Wynyardiidae) and ls implications for wynyardiid phylogeny und classification. Pp. 369-91. In M. Archer (ed.), Possums and opos- sums! studies in evolution, (Surrey Beally & Sos and Royal Zoological Society of NSW: Sydney), APLIN, K. & ARCHER, M, 1987. Recent advances in marsupial systematics with anew syneretic clas- sifieation. Pp. xv-lxxii. bo M. Archer (ed), Pos- sums and opossums: studies in evolution. (Surrey Beatty & Sons und Royal Zoological Society of NSW! Sydney). ARCHER, M. 1976, Phascolarctid origins and the po- tential of the selenodont molar in the evolution of diprotodont marsupials, Memoirs ot (he Qucens- land Museum 17; 367-371. 19783 Koalas (phascularciids | and their sigrili- cunce m marsupial evolouion, Pp, 20:8. In T.L Berguv. (ed.), The Koala, (Zoological Parks. Board N.S.W. Sydney), 1978b. The nature of the molar-premolar boundary in marsupials and a reinterpretation uf Ihe homol- ogy of marsupial cheek teeth, Memoirs ot the Qucenslund Museum 18: 157-164. 1987, Possums and opossums: studies in evolution. (Surrey Beatty & Sons and Royal Zoological Society of NSW: Sydney). ARCHER, M., GODTHELP. H., HAND, S.1..& MEG- IRIAN. D. 1989. Fossil mammals of Riversleigh. northwestem Queensland. preliminary overview Of biostratigraphy, correlution and env ronmental change, Australian Zoologist 25: 27-65. ARCHER, M., HAND, 8, & GODTHELP, H. T991, Riversleigh, (Reed: Sydney), LUCKETT, W.P. 1993, An ontogenetic assessment of dental homologies in thean mammils, Pp, 182- 204. In Szalay, FS,, Novacek, MJ, & Mtr Kenni, M.C. (eds), Manimal phylogeny: Mesozore difter- enualion, mulituberculates, monotremes, early tenang and marsupials. (Sprmger Verlag, New York). MARSHALL, L.G.. CASE, J.A. & WOODBURNE, M.O. 1989, Phylogenetic relationships of the fam- ilies OF marsupials. Current Mammology 2; 433- 502. PLEDGE. N.S. 1982, Enigmatic Ekopodon: A cause history of palaeontological interpretation. Pp. 477-488. In PV. Rich & E.M. Thompson (eds), The fossil vertebrate record of Australasia. (Mor ash University Offset Printing Unit: Claytoni 19873, Muramura williamsi, à new genus and spe cies of Twynyardiid (Marsupialia: Vombatoidea) from the middle Miocene Etadunna Formation of South Australia. Pp. 393-400 In M. Archer (ed), Possums ahd opossums: studies in evolution, (Surrey Beatty & Sons and Royal Zoological Society of NSW. Sydney). 1987b. Kareringe ngamu , anew genus and species ob primiive vonibatoid marsupial from the me- did Miocene Ngama Lond Fauna of South Aus tralia. Pp. 419-22. In M. Archer (ed.), Possums and opossums; studies in evolution, (Surrey Beatty & Sons und Rayal Zoologicul Society of NSW: Sydney). 1992. The Curramulka Local Fauna: a new Late Tertiary fossil assemblage from Yorke Peninsula, South Australia. The Beagle: Records of the Northern Territory Museum of Arts and Sciences 9 115-142. RICH, T. H. 1982. Monotremes, placentals and marsu- pals: their record in Australia and its biases. Pp. 385-477. In PV. Rich & EM. Thompson (eds), The fossil vertebrate récord of Austfalasta. (Monash University Olfset Printing Unit: Chay- lon), L986. Darcits duggani, a new ektopodontd (Marsupialia, Phalangeroidea) from the early Pliocene Hamilton Loval Fauna, Australia. Uni- MEMOIRS OF THE QUEENSLAND MUSEUM versity of Califor Publications in the Geologi- eal Sciences 141. 08-74, RICH, T.H.V. & ARCHER, M. 1979, Namilamadete snideri, a new dipretodontan from the medial Miocene of South Australia. Alcheringy 3: 197- 208. SPRINGER, M.S. 1987. Lower molars of Litokoala (Marsupialia:Phascolarctidac) and their bearing om phascolarctid évolution. Pp. 319-25, In M. Archer (ed.), Possums and opossums: studies in evolution. (Surrey Beatty & Sons and Royal Zou- logical Society of NSW: Sydney). STIRTON, R.A, 1957. A new koala from the Pliocene Pilankarinna Fauna of South Australia, Records of the South Australian Museum 13; 71-78, STIRTON, R.A., TEDFORD, R.H., & WOODBURNE, M.O. 1967. A new Tertiary formation and fauna from the Tirari Desen, South Australia. Records ol the South Australian Museum 15; 427-402. YTEDFORD, R.H.& WOODBURNE, M.O. 1987, The llanidae, anew family of yornbatiform marsupials from Miocene strata of South Australia and an evitlualion of the homolgy of molat cusps in the Diprotodontidac. Pp, 401-418. In M. Archer (ec), Possums and opossums; studies in evolution, (Surrey Beatty & Sons and Royal Zoological So- ciety of NSW: Sydney), WOODBURNE, M.O. 1984. Families of marsupials: relanonships, evolution and biogeography. Pp. 48-71. In T, W. Broadhead, (ed), Mammals: notes fora short course, (Palvoniological Society; Boul- der), WOODBURNE, M.O., TEDFORD, RH., ARCHER, M.. TURNBULL, W.D., PLANE, M.D. & LUNDELIUS, E.D. 1985, Biochronology of the continental mammal record of Australia and New Guinea, Special Publication, South Australian De- partinent of Mines and Energy 5: 347-363. WOODBURNE, M.O, TEDFORD, R.H., ARCHER, M. & PLEDGE, N.S. 19878, Mudekoela, a new genus and two species of Miocene koalas (Marsupialia: Phascolarctidae) trom South Aus- tralia. and a new species of Perikoala. Pp. 293- 317 In M, Archer (ed.), Possums and opossums: studies in evolution. (Surrey Beatty & Sons and Royal Zoological Society of NSW: Sydney). WOODBURNE, M.O,, TEDFORD, R.H & ARCHER, M, 1987b. New Miocene ringtail possums (Marsupialia: Pseudocheiridae) from South Aus- tralia. Pp. 639-79, In M. Archer (ed,), Possums and opossums“ studies in evolution. (Surrey Beatty & Sons and Royal Zoological Society of NSW; Syd- ney). WOODBURNE, M.O,, MACFADDEN, B.J., CASE, J.A., SPRINGER, M.S.. PLEDGE, N.S., POWER, LD, WOODBURNE, JLM, & SPRINGER, K.B. 1993, Land mammal biostrat- igraphy and magnetostratigraphy of the Etadunnu Formation (ate Oligocene) of South Australia, Journal of Vertebrate Paleontology 13: 483-515. A KINGFISHER (HALCYONIDAE) FROM THE MIOCENE OF RIVERSLEIGH, NORTHWESTERN QUEENSLAND, WITH COMMENTS ON THE EVOLUTION OF KINGFISHERS IN AUSTRALO-PAPUA WALTER E. BOLES Bales, W.E. 1997 06 30; A kingfisher (Halcyonidae) from the Miocene of Riversleigh, northwestern Queensland, with comments on the evolution of kingfishers in Australo-Papua, Memairs of the Queensland Museum 41(2)> 229-234, Brisbane. ISSN 0079-8835. A Miocene kingfisher from Riversleigh, norlhwestern Queensland, represented by «a com- plete carpometacarpus, is the earliest record of the Haleyonidae from Australasia. Tl shares similarities with several modem genera, but a positive generic identification cannot be made. Although it can be distinguished from extant species, this skeletal element is insufficient to ereet a new genus, A processus dentiformis in Tanysiprera and Melidera and its absence in Todiramphus and other genera suggest that the former genera are among the more primitive of the Australo-Papuan kingfishers. The less developed processus deptiformis in the Riversleigh specimen is consistent with it being an earlier member of the Todiramphus lineage, Of living Kingfishers examined, all that retain the processus dentiformis are inhab- itants of rainforest, [|Kingfisher, Halcyonidae, Riversleigh, Miocene, evolution. Walter E. Boles, Australian Museum, 6 College Street, Sydnes N.S.W. 2000, Australia; received 4 November 1996, The Kingfishers (Alcedinidae s.l.) are subdi- vided into 3 subfamilies. DNA-DNA hybridisa- lion studies (Sibley & Ahlquist, 1990) suggested that these should be recognised as families. Cerylidae do not occur in Australasia, Al- cedinidae (‘river kingfishers’) and Halcyonidae [= Dacelonidae auct] (‘tree kingfishers’) are rep- resented in Australo-Papua by 5 species in | genus and 21 species in 5-6 genera, respectively (Beehler et al., 1986; Fry et al.. 1992; Christidis & Boles, 1994). There are no named Tertiary forms from out- side Australasia (Olson, 1985). Mourer- Chaurviré (1982) listed this farnily (Alcedinidae s.l,) from Eocene-Oligocene deposits at Quercy, France, and Olson (1985) noted that he had ex- amined specimens close to this family originating trom the lower Eocene of North America and the medial Eocene of Germany. All Australian Qua- ternary kingfisher material is referable lo modern taxa: Alcedo azurea, Dacelo novaeguineae, Todiramphus pyrrhepygia and To, sanctus (Baird, 1991). No Tertiary kingfishers are known from Australasia (Fordyce, 1991; Vickers-Rich, 1991). Described herein is a Miocene kingfisher from Riversleigh, northwestern Queensland. METHODS Measurements (Steadman, 1980) were made with vernier calipers accurate to 0.05mm and rounded to the nearest 0.1mm, Terminology of bones largely follows Baumel & Witmer (1993). Todiramphus and Syma are considered distinct from Haleyon, following Christidis & Boles (1994), Institutional prefixes are AM (Australian Museum), ANWC (Australian National Wildlife Collection), MV (Museum of Victoria), QM (Queensland Museum) and USNM (United States National Museum). SYSTEMATIC PALAEONTOLOGY Family HALCYONIDAE Although the Halcyonidae includes some of the largest kingfishers in the world, size is not a valid character for family allocation of osteologicul material. Australia’s 2 Alcedinidae, Alcedo pusilla and A, azurea, are the country’s smallest kingfisher species (wing lengths 55mm and 75rnim. respectively). but the closely related A. websteri of New Britain has a wing lengih of 90mm, overlapping in size the smaller haleyanids (e.g., Todiramphus macleayii, wing length 90mm). The carpometacarpus of the Halcyonidac can be distinguished from that of the Alcedinidae and Cerylidae (Table 1) and on this basis the Riverslcigh fossil is assigned to the Haleyonidae. Halcyonid gen. indet. Fig. IC MATERIAL, QMF29719. right carpometacurpus with only minor ahrusion to some surfaces from middle MEMOIRS OF THE QUEENSLAND MUSEUM TABLE 1. Characters for separating the carpometacarpus of the Alcedinidae, Cerylidae and Halcyonidae. Character Alcedinidae Cerylidae Halcyonidae proximal border of dorsal carpal trochlea more angular more rounded more angular development of os_metacarpalis alulare more gracile more robust more robust orientation of os metacarpalis alulare more caudal more caudal more proximal tip of processus extensorius rounded expanded, slightly rugose rounded position of processus intermetacarpalis more proximal more proximal more distal width and distal extension of sulcus interosseus narrower, not as extensive distad broad, extending almost broad, extending almost to facies digitalis minor to facies digitalis minor plane of synostosis metacarpalis distalis and distal ends of os metacarpalis major and os metacarpalis minor os metacarpalis minor depressed below plane os metacarpalis minor depressed shghtly below plane flat, coplanar Miocene to early late Miocene Last Minute Site from System C (Archer et al., 1989, 1994). This site is interpreted to represent shallow pools or even emer- gent accreting surfaces, and is dominated by terrestrial vertebrates including the possums Djilgaringa gillespieae Archer et al., 1987 and Strigocuscus reidi Flannery & Archer, 1987. Other avian taxa from this site are a range of passerines, including the logrunner Orthonyx kaldowinyeri Boles, 1993. DESCRIPTION, Length 15.8mm. Length of spatium intermetacarpale 61% of length of car- pometacarpus. Processus dentiformis low and pointed, located about midway between the distal edge of facies articularis alularis and the cranialmost point of facies articularis digitalis major. Os metacarpale majus of about equal thickness for entire length, in ventral view. Spat- ium intermetacarpale gradually becoming wider distally. Processus intermetacarpalis far proxi- mally in spatium intermetacarpale. REMARKS. Among extant Australo-Papuan Halcyonidae, Tanysiptera and Melidora have a low, flat processus dentiformis, Syma has an al- most non-existent processus dentiformis as a barely raised roughened area, and Todiramphus and Dacelo, as well as Afro-Asian Halcyon, lack it (Boles unpubl. data), X-ray photographyot study skins showed Actenoides and Lacedo have a low processus dentiformis, but Halcyon (Pelargopsis) and Clytoceyx do not. The x-rays, while sufficient for determining this process, are not adequate for detailed comparisons of these taxa. Other than size, there are not substantive differences between the carpometacarpi of Todiramphus, Dacelo (and presumably Clytoceyx); Syma differs only in its low processus dentiformis. Because Dacelo and Clytoceyx are considerably larger (D. novaeguineae 32.5- 35.6mm), they are not considered further. Subse- quent comparisons involve Todiramphus, Syma, Tanysiptera and Melidora. The fossil (length 15.8mm) is in the size range of Tanysiptera (sylvia 14,0-15.2mm; galatea 15.2-16.5mm) and Todiramphus (sanctus 14.8- 15.1mm; macleayii 14,5-15.7mm; pyrrhopygia 15.7-16.5mm; chloris 16.5-17.8mm). It is larger than Syma (torotoro 13.5mm; megarhyncha 14.4mm) and smaller than Melidora macrorrhina (18.9 mm). It differs from Tanysiptera and Melidora and resembles Syma and Todiramphus by being more slender and by having spatium intermetacarpale longer relative to the length of carpometacarpus and the dorsal rim of trochlea carpalis extending less distally relative to the ventral rim. The fossil differs from Todiramphus and resembles Tanysiptera, Melidora and Syma by having processus dentiformis present. This process, however, is narrow and pointed, rather than broad and flat as in these genera (and larger than in Syma), and is situated more proximally relative to the spatium intermetacarpale and pro- cessus intermetacarpalis than in Tanysiptera and Melidora, and more distally than in Syma. From all 4 genera, the Riversleigh specimen differs by having the caudal edge of os metacarpale majus straighter and less caudally concave, making os metacarpale majus thicker and spatium inter- metacarpale proportionally narrower relative to the width of the carpometacarpus. The significance of these differences is uncer- tain, They are individually minor, yet within the Halcyonidae the amount of variation in this bone is little so that these differences may assume greater importance. Variation in the carpometa- carpus, however, is not representative of that of the remainder of the skeleton or indeed the rest of the morphology (Boles unpubl. data). The fossil cannot be assigned to Tanysiptera, Melidora, Syma or Todiramphus and can be dif- frerentiated from extant species of these genera. A MIOCENE KINGFISHER FROM RIVERSLEIGH 231 FIG. 1. Carpometacarpi of Australo-Papuan halcyonid kingfishers, in ventral view. All right side except for B. A, Melidora macrorrhina (ANWC CORS-53). B, Tanysiptera sylvia (AMO.60926). C, Riversleigh Halcyonidae gen. indet. (QMF29719). D, Todiramphus sanctus (AMO.57182). E, Dacelo novaeguineae (AM 0.59217). Scales = 5mm. Overall it has the greatest resemblance to species of Todiramphus in size and morphology, despite the presence of a small processus dentiformis. Given the relatively limited importance of carpo- metacarpal variation in the halcyonids and the limited fossil material it is imprudent to recognise anew genus at this time. DISCUSSION To place the Riversleigh fossil in perspective within halcyonid kingfisher evolution, the prim- itive members of the family must be identified. Fry (1980a, b) employed 3 criteria for this. Prim- itive Kingfishers have 1) generalised diets and relatively unspecialised modes of foraging (i.e.; sitting and pouncing, non-fishing); 2, stable hab- itats and not of recent origin (rainforest), within which they may be discontinuously or relictually distributed; and 3, oligotypic genera (i.e. with few species) without close relatives. Fry (1980a) con- cluded that kingfishers (all families as recognised here) arose in Malesia, the area between Indo- China and the Coral Sea. Prominent among the primitive forms were halcyonid kingfishers, many of which live in Malesian rainforests. Fry (1980a) speculated that ‘the Daceloninae [= Halcyonidae] have a history of evolution in eastern equatorial rainforests al- most as ancient as the mid-Cenozoic origin of the Alcediniformes [sensu Feduccia (1977)]’. By the early Miocene, the present geographical config- uration of Malesia had been reached. This, in Fry’s opinion, provided ‘ideal circumstances for the multiplication of species, resulting in a fauna of forest-dwelling, non-fishing kingfishers ... At some more recent time, perhaps about the mid- Pliocene, this fauna gave rise to a lineage, Hal- cyon [s.1.], adapting to more open habitats’. Under this suggested sequence of events, the rainforest-dwelling Tanysiptera, Melidora, Acten oides and Lacedo would be among the more primitive genera. Todiramphus, included by Fry (1980a) in his open habitat Halcyon, would be more derived. Presumably Syma (also included in Halcyon sensu Fry [1980a]) also was considered by Fry (1980a) to be a more derived taxon. Al- though entering open country adjacent to forest, the two species of Syma are essentially rainforest inhabitants, particularly in New Guinea (Coates, 1985). In addition to sharing the primitive characters proposed by Fry (1980a, b), species of Tan- ysiptera and Melidora (as well as Actenoides and Lacedo) have a low but well-defined processus dentiformis on the carpometacarpus. This is ab- sent in the other halcyonid genera examined, the Alcedinidae and Cerylidae, and some other cora- cijtorm families (c.g. Momotidae, Coraciidac). If any of these families ps used for outgroup com- parisons, the suggested polarity of the processus dentiformis is that its presence is derived. A sim- ilar comparison using other coraciiform families (Todidae, Phoenjculidae, Upupidae, Bucerotidae), in which the structure is present, gives the opposite conclusion: presence of the processus dentiformis is primitive, its absence derived. Within the Meropidae, the siructure is present in some species (Merops ornatus) and absent in others (M. apiaster). The significance of this variation is unknown, as are the functional aspects of the processus dentiformis. Thus the polarity of this Character’ s presence is not known, (This, of course, assumes that the processus dentiformis i$ homologous across the order, Whether this is so, and what relationship it has to the similar process found in the majority of the Passeriformes, is unknown.) Within the halcyonid kingfishers the presence of the proces- sus Jentiformis exhibits a strong correlation with Fry's (1980a,b) primitive criteria. Superficially there seems to be litlein common externally between Melidera and Tanysiptera beyond the hasic kingfisher similarities. Each has specialised generic characters: a hooked bill in Melidora (Hooked-billed Kingfisher) and elon- gated, spatulate central rectrices in Tanysiprera (paradise kinglishers). Melidora macrorrhina 1s arather drably coloured species, Other than blue scalloping on the crown, the plumage is a combi- nation of browns and white. The underside is white, while the back, rump, tail, and wings are dark brown with paler brown scalloping, This plumage is quite unlike that of the paradise-king- fishers Tanysiptera, adults of which are strikingly patterned in unmarked blues und blacks, and usti- ally either white or buff/rose. The juvenile plum- age of Tanysiptera, however, is brown with scalloping, and Fry (1980h) was ‘impressed by its |Melidora's| plumage resemblance to the distine— tive juvenile of Tanysiptera galarea’. That these two genera might be closely related was sug- gested by Fry (19806), who thought it ‘possible that Melidora and Tanysiptera are of immediate common descent and the former is a specialised derivative that has retained, in the adult, the an- cesiral juvenile plumage’. The presence of sim- ilar plumages is also evident in female Lacedo and sume species of Actenoides. notably A princeps and A, lindsay/ of all ages. This resem- blance between Actenoides and Tanysiptera and between Lacedo and Melidere was commented on by Fry (1980hb). MEMOIRS OF THE QUEENSLAND MUSEUM Bell (1981) also considered that Melidora was closest to Tanysiptera. He noted that the call notes of these two species were similar and the distress notes identical. They have similar skele- tal proportions, particularly in the relative length of the legs when compared to Todiramphus, Hal- cyen or Dacelo (Boles, unpubl, data), These two genera thus have more similarities than might be immediately obvious, They also share habitat preferences. Although Melidora macrorrhina and Tanysiptera species will enter mangroves, teak plantations and drier adjacent country, they are primarily occupants of rainforests. Lacedo pulchella and the 6 species of Acrenoides inhabit rainforest, preferably in a primary, undisturbed stale Species of Todiramphus are uniform in plum- age, Most have a variation of the basic pattem of green or bluc upperparts and white or light orange underparts and collar, Several subgroups can be discerned, but they still show only small diver- gences from this general form. These species are almost all sit and pounce feeders, Habitat prefer- ences among species are more varied, ranging Irom rainforest to open country, Within Australo- Papua, however, few occur in rainforest and there is a decided bias towards open forest, woodland, mangroves, clearings and open country. Most rainforest inhabiting forms are found on islands of the southwest Pacific, The processus dentiformis in the Riversleigh kingfisher is smaller than that in primitive mod- em forms. This could indicate that it has been undergoing reduction since the split of ils lincuge from that of Tanyyiptera-Melidara. In this re- spëctitis consistent with what would be predicted fora primitive species of Todiramphis. The bone exhibits a largely Todiramphuy character while retaining this more primitive halcyonid feature. The identification of the Riversleigh fossil as a haleyonid is compatible with Fry's (1980a) inter- pretation, as is considering the presence of the processus dentiformis as primitive. According to Fry's scenario, an early Miocene kinglisher should be a primitive form, His criteria, however, are noLuselul in this situation. The foraging meth- ods of the fossil cannot be determined, nor can its systématic isolation be ascertained. The Riversleigh ħabitat is considered (Archer et al., 1992) to have been rainforest. but this cannot be used as a character for making a taxonomic de- Jermination, Although this fossil permits identification as a haleyonid kingfisher, itis not clear whether this species belongs to un existing genus or should be A MIOCENE KINGFISHER FROM RIVERSLEIGH allocated to a new one, The presence of a feature found in living primitive genera and the kingfisher's occurrence in what is considered to have been rainforest suggest that il, too, was it more primitive form. This is consistent with the sequence of evolutionary events suggested by Fry (19808), ACKNOWLEDGEMENTS For access tọ specimens 1 thank Wayne Longmore (Queensland Museum), Jerry van Tets and Richard Schodde (Australian National Wild- hfe Collection), Les Christidis and Rory O’ Brien (Museum of Victoria), and Storrs Olson (United States National Museum). The Australian Mu- seum provided a venue in which to work and funds lo support this research. The Riversleigh material was collected via an ARC Programme Grant 10 M, Archer; support [rom the University of New South Wales; a grant from the Depart- ment of Arts, Sport, the Environment, Tourism and Territories to M. Archer, S. Hand and H. Godthelp; a grant from the National Estate Pro- gramme Grants Scheme to M, Archer and A. Bartholomai; and grants in aid to the Riversleigh Research Project from Wang Australia Pty Ltd, ICI Australia and the Australian Geographic So- city. REFERENCES ARCHER, M,, GODTHELP, H, HAND, SJ, & MEGIRIAN, D. 1989. Fossil mammals of Riversleigh, northwestern Queensland: prelimi- nary overview of biostratigraphy. correlation and environmental change. Australian Zooogist 25: 29-65. ARCHER, M., HAND, S.J. & GODTHELP, H. 1991. Riverslcigh. (Reed: Sydney). 1994. Riversleigh, 2nd edition. (Reed: Sydney). 1995, Tertiary environmental and biotic change in Australia, Pp. 77-90, In Vrba, E.S., Denton, G.H.. Partridge, T.C.& Burkle, L.H. (eds) Paleoclimate and evolution, with emphasis on human origins. (Yale University Press: New Haven). ARCHER, M., TEDFORD, R.H. & RICH, T.H. 1987. The Pilkpildridae, a new family and four new species of ?pelaund possums (Marsupialia: Phalungerida) from the Australian Miocene. Pp. 607-627, In Archer, M. (ed) Possums and opos- sums: Studies in evolution. (Surrey Beatty. Chup- ping Norton) BAIRD, R.E. 1991. Avian fossils from the Quaternary of Australia. Pp. 809-870. In Vickers-Rich, P.V., Monaghan, J.M., Baird, R.F. & Rich, T.H. (eds) Vertebrate palaeontology of Australasia. (Pioneer Design Studio: Methoume). BAUMEL, J.J. & LM. WITMER, 1993. Osteologia. Publications of the Nuttall Ornithological Club 23: 45-132, BEEHLER, BM. PRATT, TK & ZIMMERMAN. D.A. 1986. Birds of New Guinea, (Princeton Un» versity Press: Princeton). BELL, H.L. 1981. Information on New Guinea king- fishers, Alcedinidae. Ibis 123: 51-61. BOLES, W.E. 1993, A logrunner Orthonyx from the Miocene of Riversleigh, northwestem Queens- land. Emu 93: 44-49, CHRISTIDIS, L. & BOLES, W.E. 1994, Taxonomy and species of birds of Australia and its territories. RAOU Monograph 2 (Royal Australasian Omit- thologists Union: Melbourne). COATES, B.J. 1985. The birds of Papua New Guises. vol 1, (Dove: Brisbane): FLANNERY, T.F. & ARCHER, M. 1987, Strigocuscus reidi aid Trichosurus dicksoni, Wo new tossil phalangerids (Marsupialia: Phalangeroidea) from the Miocene of northwestern Queensland. Pp. 527-536. bi Archer, M. (ed) Possums and opos: sums: Studies in evolution. (Surrey Beatty: Chip- ping Nonon). FORDYCE, R.E. 1991. A new look at the fossil verte- brate record of New Zealand. Pp. 1191-1316. In Vickers-Rich, P.V., Monaghan, J.M., Baird, R.F. & Rich, T.H. (eds) Vertebrate palacontology ot Australasia, (Pioneer Design Studio; Melboume), FEDUCCIA, A. 1977, A model for the evolution of perching birds. Systematic Zoology 26: 19-31. FORSHAW, J.M. 1987. Kingfishers and related birds. vol. 2 Alcedinidae: Halcyon to Tanysipiera, (Lansdowne: Sydney). FRY, C.H. 1980. The origin of Afrotropical kinghsh- ers. This 122; 57-72. 1980h. The evolutionary biology of Kingfishers (Al cedinidae), Living Bird 18: 113-160, FRY, C.H, PRY, K. & HARRIS, A. 1992. Kinglishess, bee-caters and rollers, (Croom Helm: London). MOURER-CHAURVIRE, M. 1982 Les oiseaux fossiles des Phosphoriles du Quercy (Bocéne Supérieur a Oligocene Supérieur): implications Sai akc, aie Geobios, mém, spéc 6; -426. OLSON, S.L. 1985. The fossil record of birds. Pp- 79-238, In Farner, D.S., King, JR. & Parkes, K.C. (eds) Avian biology, vol. 8, (New York: Aca- demic Press). SIBLEY, C.G. & AHLQUIST. J.E. 1990, Phylogeny and classification of birds! A study in molecular evolution, (Yale University Press: New Haven), STEADMAN, D.W. 1980. A review of the ostealogy and paleontology of turkeys (Aves; Meleagridinae).Contributions to Science from the Natural History Museum of Los Angeles County 330: 131-207. TEDFORD, R.H. 1967. Fossil mammals from the Carl Creek limesione, northwestem Queensland. Bul- letin of the Bureau of Mineral Resources Geology and Geophysics 92: 217-236. 234 MEMOIRS OF THE QUEENSLAND MUSEUM VICKERS-RICH, P. 1991. The Mesozoic and Tertiary © VICKERS-RICH, P.V., MONAGHAN, J.M., BAIRD, history of birds on the Australian plate. Pp. 721- R.F. & RICH, T.H. (eds) (1991). Vertebrate pal- 808. In Vickers-Rich, P.V., Monaghan, J.M., aeontology of Australasia. (Pioneer Design Stu- Baird, R.F. & Rich, T.H. (eds). Vertebrate pal- dio: Melbourne). aeontology of Australasia. (Pioneer Design Stu- dio: Melbourne). HINDLIMB PROPORTIONS AND LOCOMOTION OF EMUARIUS GIDJU (PATTERSON & RICH, 1987) (AVES: CASUARIIDAE) WALTER E. BOLES Boles, W.E. 1997 06 30: Hindlimb proportions and locomotion of Emuarius gidju (Patterson & Rich, 1987) (Aves: Casuariidae). Memoirs of the Queensland Museum 41(2): 235-240. Brisbane. ISSN 0079-8835. Using proximal and distal fragments, the length of the tarsometatarsus of Emuarius gidju is estimated and compared to that of other hindlimb elements. From these proportions and other hindlimb morphology, the inferred locomotory mode of E. gidju is compared with Recent casuariids. Emuarius gidju appears to have been more cursorially adapted than Casuarius and dwarf Dromaius, suggesting at least some open habitat in the Riversleigh palaeoenviron- ment. Using the relationship between weight and least circumference of the femur and tibiotarsus in Recent birds, the weight of E. gidju is suggested to have been 19-21kg. [] Emuarius, Aves, hindlimb, locomotion. Walter E. Boles, Australian Museum, 6 College Street, Sydney NSW 2000, Australia; received 4 November 1996. Emus, Dromaius (Dromaiinae), form a promi- nent element of Australia’s avifauna. The closely related cassowaries, Casuarius (Casuariinae), are more restricted in distribution. These two groups occupy very different modern habitats, and loco- motory adaptations correlated with these differ- ent habitats are obvious in the relative pro- portions of the lower limb bones. Because of the relationship between habitat and limb propor- tions, fossil emus and cassowaries are potentially good palaeoenvironmental indicators. Emus have a better fossil record (Patterson & Rich, 1987), than cassowaries (Vickers-Rich, 1991). Patter- son & Rich (1987) described lower limb elements from the Miocene of central Australian as a small emu, Dromaius gidju. Boles (1991) erected Emuarius for this species and considered it closer to emus than cassowaries, but nearer their dichot- omy than any other described taxon. The type material of Emuarius gidju is part of the Kutjamarpu Local Fauna, recovered from the Leaf Locality (UCMP V-6213) on the E shore of Lake Ngapakaldi in the E Lake Eyre Sub-Basin, South Australia. Much £. gidju material occurs at Riversleigh NW Queensland. Archer et al. (1989, 1994) considered the Riversleigh pal- aeohabilat as rainforest, based on mammal re- mains. This paper is to 1) estimate the lower limb proportions of E. gidju; 2) compare these to mod- ern emus and cassowaries and, by implication, to their style of locomotion; 3) make an initial esti- mate of the weight of E. gidju; and 4) interpret the possible palaeohabitat at Riversleigh where E. gidju occurs. MATERIALS AND METHODS Osteological terminology follows Baumel & Witmer (1993). Measurements were made with vernier Calipers accurate to 0.05mm and rounded to the nearest 0.1mm. Institutional acronyms are AM (Australian Museum), QM (Queensland Mu- seum) and SAM (South Australian Museum). The type specimen of E. gidju (SAMP26779) is an associated distal tibiotarsal fragment, prox- imal tarsometatarsus including much most of the shaft, and complete set of pedal phalanges (Patterson & Rich, 1987) and numerous speci- mens of Emuarius are known from Riversleigh (Table 1). To compare changes in relative proportions of the casuariid hindlimb, the following measures were calculated for the bone lengths of the 3 extant species of Casuarius and the 1 living and 2 recently extinct dwarf species of Dromaius: _ Tibiotarsus x 100 TRYEMR = Femur TM Tur = a x 100 TBI yr = Tibiotarsus x 100 Tarsometatarsus ` Because the purpose was to find general, rather than detailed, directions of change, measure- ments were taken from the literature (Table 2) and rounded to the nearest mm. Means were used where available. The sample sizes were often small, sometimes comprising single individuals. Predicted body weight of Emuarius gidju was calculated (Campbell & Marcus, 1991) from AM F78585 (near complete femur) and QMF16827 236 (distal tibiotarsal fragment). Tape was wrapped around the bones at their least circumferences, marked at the point of overlap, straightened and measured with calipers. These results were used in the equation logio(weight)=a«log\o(circumfer- ence)+b, where the values of a (slope) and b (interecept) were those determined by Campbell & Marcus (1991) for all birds for the respective elements (femur: a=0.41 1, b=—0.065; tibiotarsus: a=2.424, b=0.076). RESULTS Increased cursoriality in these birds is associ- ated with an increase in the lengths of the tibio- tarsus and tarsometatarsus relative to that of the femur (Howell, 1944). Relative proportions of the hindlimb contributions of the long bones in Casuarius, Dromaius and Emuarius (Fig. 2) show that that of the tibiotarsus remains more or less consistent in all taxa; that of the femur de- creases with increased cursoriality, whereas that of the tarsometatarsus increases. The relationship between these changes is consistent for Recent species: TBT/FMR = 0.45 [TMT/FMR] + 113; r=0.84. These changes appear independent of overall size when compared between genera, but within genera the smaller members have the greater TBT/TMT and smaller TMT/FMR. The TBT/FMR in Casuarius remains constant. It sug- gests from these figures that the smaller species in each genus are the least cursorial members. The Kutjamarpu femur (AMF78585; Boles, 1991) of E. gidju is 194mm long, which, because of abrasion, is a few mm less than its original length, of about 198mm and very likely not 200mm. Complete tibiotarsi are unknown for E. gidju (Table 1). Nevertheless, it is possible to predict the size and relative proportions of this bone from other hind limb elements. Tar- sometatarsi are represented by the holotypical proximal end and shaft, and several distal frag- ments. An estimate of the tarsometatarsal length was made by using the proximal end and shaft and a distal fragment. The proximal tarsometatarsal fragment is 276mm long; the longest edge of its shaft is straight and shows no evidence of flaring outward to trochlea metatarsi II. The distal piece is 64mm long; the small portion of shaft remain- ing is just proximal to the flaring of trochlea metatarsi II. Because there is little, if any, overlap between these two pieces, minimum length of the tarsometatarsus is 340mm (Fig. 3). MEMOIRS OF THE QUEENSLAND MUSEUM TABLE 1l. Specimens that have been referred to Emuarius gidju. *=specimens described by Boles (1991). SITE REG, NO. ELEMENT Lake Ngapakaldi Holotype: associated SAM distal tibiotarsus, P26779 proximal tarsometarsus, complete pes AMF78585* Femur Riversleigh Systa aora QMF16827* Tibiotarsus S Fee suing QMF29720 Vertebrae QMF29721 Vertebrae | QMF16828* Femur QMF16829* Femur AMF78586* Tibiotarsus omron? Tibiotarsus QMF29723 Tarsometatarsus QMF29724 Tarsometatarsus QMF29725 Tarsometatarsus QMF29726 | Tarsometatarsus MF29727 | Tarsometatarsus Upper MF16830* Rostrum QMF16831*| — Scapulocoracoid QMF29728 Vertebrae S Een ?B, Dirks QMF29729 | Tarsometatarsus System C QMF16832* Femur ag | AMF78587* | Tarsometatarsus Using a tarsometatarsal length of 340mm and a femoral length of 198mm, gives a TMT/FMR of 172, greater than that of any Recent casuariid except Dromaius novaehollandiae. Using these values with the regression equation for the family (Fig. 4) gives a predicted TBT/FMR of 190, which corresponds to a tibiotarsal length of 376mm, These 3 values give a combined length of 916mm, virtually the same as the hindlimb of Casuarius casuarius, although the proportional contribution of each bone to this length is differ- ent (Fig. 2). This figure must be used with caution. The fragments on which it is based represent different individuals from different localities. As such, there are several sources of potentially significant variation between components. Geographical variation may not have been of major importance, living D. novaehollandiae exhibits little differen- tiation across its range. There is much greater size variation across this species’ chronological HINDLIMB PROPORTIONS OF EMUARIUS GIDJU than any known species of emu (illustrated in Patterson & Rich, 1987). In emus, phalanx ungualis of digit III is longer than that of digit I, whereas in cassowaries it is reversed, with phalanx ungualis of digit II ex- tended into a long spike several times the length of phalanx un- gualis of digit II. Emuarius does not have phalanx ungualis of digit II developed into a spike, but it is still longer than that of digit II. Cassowaries have digits H and IV relatively long com- pared with digit III (ratios of IL: and IV:IIN, respectively, without phalanges unguales 0.82, 0.76). Both are substan- FIG. 1. Leaf Locality femur of Emuarius gidju compared with femora of tially reduced in relative length Recent casuariids, in cranial view. From left, Casuarius Casuarius, C. bennettii, Emuarius gidju, Dromaius novaehollandiae and D. ater. range, with mainland fossils usually smaller than modern birds. Patterson & Rich (1987) suggested that it ‘may have been at any one time in the Pleistocene both larger or smaller than at present’. Considerable intraspecific variation in living D. novaehollandiae is probably related to age and sex, as well as individual differences (Marchant & Higgins, 1990), For example, among 22 modern specimens, Patterson & Rich (1987) found a range in tarsometatarsal length of 332-422mm (mean 383mm: s.d. 18.0). Miller (1963) described Dromaius ocypus from the middle to late Pliocene Palankarinna Local Fauna from Lake Palankarinna, northern South Australia. This species was intermediate in size between Emuarius gidju and living Dromaius novaehollandiae. The tarsometatarsus of D. oc- ypus is markedly shorter relative to its width than is that of D. novaehollandiae, but less so than in Casuarius. Vickers-Rich (1991) interpreted this as suggesting a less cursorial lifestyle for D. ocypus than for D. novaehollandiae. Because Patterson & Rich (1987) did not give comparative figures for relative widths, it is difficult to quan- titatively compare E. gidju with these species. Visually, Emuarius appears to be proportionally thinner for its length than is D. ocypus, but not to the degree of D. novaehollandiae. Patterson & Rich (1987) pointed out that, al- though, the foot of E. gidju is more like that of emus than cassowaries, it is more cassowary-like in emus (ratios as above - 0.55, 0.63). Both digits II and IV of Emuarius are comparatively longer than those in emus (digit IV to a lesser degree), but not to the extent seen in cassowaries (0.57, 0.68). Patterson & Rich (1987) suggested that the reduction of digit IV and particularly of digit II in E. gidju, compared with the highly cursorial D. novaehollandiae, appears to parallel similar reductions in other groups of terrestrial birds and mammals. In comparison with the pedal phalanges of cas- sowaries, those of emus are dorsoplantarly com- pressed. Emuarius is somewhat intermediate, with its phalanges substantially more dorsoplantarly compressed than those of casso- waries, but less compressed than (but more sim- ilar to) the condition in emus. Campbell & Marcus (1991) stated ‘the least shaft circumference of either [femur or tibiotar- sus] can be a reliable indicator of the weight of a fossil bird’, From their equation, and measure- ments of E. gidju bones, predicted weights of this species were 21kg based on the femur and 19kg based on the tibiotarsus. The least circumference of the tibiotarsus is almost always at or distal to the midpoint of the bone; in most birds it is in the distal third (Campbell & Marcus 1991). No tibio- tarsal specimen of E. gidju is complete, although the length of that measured is about a third of the predicted length for this bone. It is possible that the least circumference occurs on the missing section of tibotarsus, and the value given here would have to be adjusted. The predicted N wo oo TABLE 2. Measurements (mm) and measures of relative proportions of hindlimb bones of Recent emus and cassowaries and of Emuarius gidju. For calculation of predicted measurements of E. gidju and of measures of MEMOIRS OF THE QUEENSLAND MUSEUM relative proportions, see text. watercourses. It was not until at least the middle Miocene (c, 15 mya) that drying of the climate began and open habitats started TBT/|TMT|TBT/| tO appear on a large scale. Species FMR | TBT | TMT Source FMR\/FMR|TMT | There is no evidence of grass- | -—| lands before the end of the late Dromaius Patterson & Miocene to Pliocene (Martin, novaehollandiae 203 401 383 Rich, 1987 198 | 189 | 105 this volume). Dromaius 2 Morgan & 2 The graviportal locomotion baudinianus 164 | 293 234 Sutton, 1928 179 | 143 | 125 of E AEA is associated Dromaiusater | 178 | 331 | 274 | Morsan&. | 186 | 154 | 121 | with movement through the - t ea dense vegetation of the Aus- ease er 236 | 388 | 306 |Richetal., 1988| 164 | 130 | 127 | tralo-Papuan rainforests. In- PP —- creased cursoriality seems sonaiiting 232 | 384 | 300 |Richetal., 1988| 166 | 129 | 128 | correlated with the appearance = 3 = - of more open habitat, in which Casuarius bennetti | 221 365 | 229 |Rich etal., 1988| 165 | 104 | 159 sustained running could take Emuarius gidju 98 | 376 | 340 This paper 190 | 172 | 111 place. (While able to run if weights from the two bones are close enough to give an acceptable first estimate for E. gidju. DISCUSSION Throughout the early Tertiary, much of Aus- tralia was covered in closed forest, and the cli- mate was considerably more humid than at present (Frakes et al., 1987). The dominant veg- etation type over much of the continent was rainforest. Nothofagus-dominated rainforest covered the Lachlan River valley during the late Eocene to late Oligocene-early Miocene (Martin, 1987). Even where closed forests were not pres- ent, gallery rainforest probably occurred along Casuarius bennetti Casuarius casuarius Emuarius gidju Dromaius ater Dromaius novaehollandiae TARSOMETATARSUS required, cassowaries have limited opportunities in such habitat to work up and sustain a reasonable degree of speed because sufficiently open areas are restricted.) Morpho- logical correlates with cursoriality include pro- portional lengthening of the tibiotarsus and tarsometatarsus, and a reduction in the relative size of digit II. Conditions between the extremes of the states found in Casuarius and Dromaius are suggestive of levels of cursorial ability inter- mediate between theirs, although at what point along the scale cannot be determined. This, in turn, suggests the possibility that the amount of open habitat might be also somewhere between that available to cassowaries and emus. TIBIOTARSUS SOOM Poe KKK ND ISOIN ELIIS Oo OO OS SOS SSS SO s SOSS hSOST FIG. 2. Comparative proportions of bones in the hindlimbs of Emuarius gidju, two species of Casuarius and two species of Dromaius. Note that while the tibiotarsal proportion remains relatively constant, the proportion comprising the femur decreases as that of the tarsometatarsus increases. HINDLIMB PROPORTIONS OF EMUARIUS GIDJU FIG. 3. ‘Combined’ length of Leaf Locality proximal tarsometatarsus and Riversleigh distal tarsometatar- sus compared with tarsometatarsi of Recent casuariids, in cranial view. From left, Dromaius novaehollandiae, D. ater, Emuarius gidju, Casuarius casuarius and C. bennettii. Casuarius has been considered more primitive than Dromaius on the basis of distribution, habi- lal preferences and hindlimb. Schodde & Calaby (1972) and Schodde (1982) cited the cassowaries as elements of the Tumbunan avifauna, which represents the earliest lineages of extant Aus- tralo-Papuan birds. The emus were placed by Schodde (1982) in the autochthonous Eyrean fauna, which evolved in response to the opening of the Australian habitat. Boles (1991) consid- ered Emuarius to be closer to the dichotomy of cassowaries and emus than any other known taxon, but too cursorially adapted and Dromaius- like to have been the common ancestor. It appears to mark in the Casuariidae a stage in the transition from a graviportal to a more cursorial locomo- tion. Although some characters of the hindlimb of E. gidju are more similar to either Casuarius or Dromaius, many are intermediate between liy- ing members of these genera (Boles, 1991). Boles (1991) drew attention to the fact that the lower limb bones were more similar to those of Dromaius, whereas the femur, whose proportions are more dependent on the bird’s mass than its locomotion (Prange et al., 1979), more closely resembled Casuarius. There are alternative explanations that accom- modate both the cursorial hindlimb proportions of E. gidju and the absence of open spaces. One possibility is that the ground cover of the rain forests was sufficiently open for cursorial ani- mals to move rapidly without obstruction. For example, modern Nothofagus forests are fre- quently open below the canopy, without the dense understory of some other rain forest types (pers. obs.). Neville Pledge (pers. comm.) suggested that a situation may have existed similar to that which occurred on Kangaroo Island when the dwarf emu Dromaius baudinianus was extant. Much of the island’s vegetation was very thick and would have prevented rapid passage of a large animal such as the emu. Large mammals, however, forced runways through the vegetation, permit- ting them to move with comparable, albeit re- stricted, ease. The emus apparently took advantage of these runways for their own prog- ress. Likewise, the large mammals known from Riversleigh may have opened similar pathways through the thick undergrowth, which could have been used by E. gidju. Nonetheless, D. baudinianus was also noted for being very fast and virtually uncapturable in open areas. A small Miocene dromaiid from Alcoota, NT, is known from 2 phalanges and 3 unassociated trochleae (Patterson & Rich, 1987). These are of comparable size with E. gidju. There are slight differences in phalangeal morphology, and Patterson & Rich (1987) retained these speci- mens as Dromaius sp. indet. until more complete material is available. The Alcoota palaeohabitat has been interpreted as a lake, bordered by sedge or grassland, grading to woodland and gully for- est (Murray & Megirian, 1992). This is a different environment than that interpreted for the older Riversleigh deposits. Even if the Alcoota dromaiid proves to be E. gidju, it would have limited relevance to reconstructing the Riversleigh habitats because of the age differ- ences of the deposits. It would, however, suggest that E. gidju was preadapted for the more open Alcoota environment. Models of this species’ locomotory mode de- pend on extrapolations from living, non-conge- neric relatives. These are speculative, and must be treated as such, while palaeoenvironmental reconstructions based on them require an even greater degree of caution. 240 FEMOROTIBIAL INDEX “n iha ta an tae 10 ur FEMOROMETATAASAL INDEK FIG. 4, Relationship of TMT/FMR and TBT/FMR of Recent species of Casuarins (open cireles) and Dromaius (closed circles). Species numbered us fol- lows: 1. Casuarius benneiti; 2. C. casuarius, 3. C, ynappendicularus: 4, D. baudimenys; 5, D, ater: 6, D. novuehollandiae. Regression line is TBT/FMR = 0.45 [TMT/FMR |+ 1 13; r=0,84. The line vertical line represents the intersection with this line of the TMT/FMR (172) for Emuurius gidjn as caleulated in the text. Predicted TBT/FMR js 190. ACKNOWLEDGEMENTS In addition to acknowledgements in Boles (1991) T thank Neville Pledge for discussion on the Kangaroo Island habitat, Lynne Ho and Mau- rice Ortega for figures, and Anna Gillespie for information on Æ. gidju, The Riversleigh mate- rial was collected via an ARC Programme Grant to M. Archer; a grant from the Department of Arts, Sport, the Environment, Tourism and Ter- ritories to M. Archer, S, Hand and H. Godthelp; support trom the University of New South Wales; a grant from the National Estate Programme Grants Scheme to M. Archer and A. Bartholomai; and grants in aid to the Riversleigh Research Project from Wang Australia Pty Ltd, ICI Aus- tralia and the Australian Geographic Society. REFERENCES ARCHER, M., HAND, S.J. & GODTHELP, H, 1994. Riversleigh, 2nd ed, (Reed: Sydney). ARCHER, M., HAND, 5.. GODTHELP, H. & D. MEGIRIAN. 1989. Fossil mammals of Riversieigh, northwestem Queensland. prelimi- nary overview of bjostratigraphy, correlation and environmental change. Australian Zoologist 25: 29-65. BAUMEL, J.J. & L.M, WITMER. 1993. Osteologia, Publications of the Nuttall Ornithological Club 23: 45-1372. MEMOIRS OF THE QUEENSLAND MUSEUM BOLES, W-E. 1991. Revision of Dromaiis gidju Patter- son & Rich, 1987, with 4 reassessment of its generic position. Natural History Museum of Los Angeles County Science Series 36: 195-208. CAMPBELL, K.E., JR. & MARCUS, L. 1991. The relationship of hindlimb bone dimensions to body weight in birds. Natural History Museum of Los Angeles County Science Series 36: 395-412, FRAKES, L.A., McGOWRAN, B. & BOWLER, J.M 1987. Evolution of Austrahan environments. Pp, 1-16, In Dyne, G.R. & Wallon, D-W. (eds), Fauna of Australia vol. 1A, General articles. (Australian Government Publishing Service: Canberra). HOWELL, A,B, 1944, Speed in animals, Their special- ization for running and leaping. (University of Chicago Press: Chicago). MARCHANT, S. & HIGGINS, P.J, (coordinators). 1990, Handbook of Australian, New Zealand und Antarctic birds, 1. (Oxford University Press: Syd- ney). MARTIN, H-A. 1987. Camozoic history of the vegeta- tion.and climate of the Lachlan River Region, New South Wales. Proceedings of the Linnean, Society of New South Wales 19: 213-257. MILLER, A.H. (963. Fossil ratite birds of the late Tertiary of South Australia. Records of the South Australian Museum 14: 413-420, MORGAN. A.M. & SUTTON, J. 1928. A critical de- scription of some recently discovered bones of the extinct Kangaroo Island Emu (Dromaius diemenianus). Emu 28: l-19. MURRAY, P. & MEGIRIAN, D. 1992. Continuity and contrast in middle and late Miocene vertebrate communities trom the Nonhen Territory. The Beagle 9: 915-218. PATTERSON, C, & P.V. RICH. 1987. The fossil his- tory of the emus. Drometius (Aves: Dromatinae), Records of the South Australian Museum 21: §5-117. PRANGE, H.D., ANDERSON, J.F. & RAHN, H. 1979, Scaling of skeletal mass to body mass in birds and mammals, American Naturalist 113; 103-122. SCHODDE, R. 1982, Origin, adaptation und evolution of birds in arid Australia. Pp. 191-224, In Barker, W.R, & Greenslade, P.J.M. (eds), Evolution of the flora and fauna of arid Australia. (Peacock Publi- cations: Frewville). SCHODDE, R. & CALABY, J.H. 1972. The biogeog- raphy of the Australo-Papuan bird and mammal faunas in relation to Torres Strait. Pp. 257-300. In Walker, D. (ed.). Bridge and hirien the natural and cullural heritage of Torres Strait, (Australian National University Press; Canberra). VICKERS-RICH, P. 1991. The Mesozoic and Tertiary history of birds on the Ausiralian plate Pp. 721- 808. In Vickers-Rich, P.V., Monaghan, J.M.. Baird, RF. & Rich, T.H, (eds), Vertebrate pal- geontology of Australasia. (Pioneer Design Stu- dio: Melbourne). RIVERSLEIGH BIRDS AS PALAEOENVIRONMENTAL INDICATORS WALTER E. BOLES Boles, W.E. 1997 06 30: Riversleigh birds as palaeoenvironmental indicators. Memoirs of the Queensland Museum 41(2): 241-246. Brisbane. ISSN 0079-8835. Fossilised birds from Riversleigh are used to make a palaeoenvironmental reconstruction. Difficulties that hamper this attempt are discussed. From the early to early late Miocene deposits, a range of taxa demonstrate aquatic situations; four others are indicative or al least suggestive of rainforest; one hints at at least some open spaces; and six are ambiguous because of insufficient morphological variation between taxa with different ecological preferences or insufficiently known palaeobiology. The only species thus far identified from Pliocene Rackham’s Roost Site points to conditions similar to those at Riversleigh today. [] Riversleigh, birds, palaeoecology. Walter E. Boles, Australian Museum, 6 College Street, Sydney New South Wales 2000, Australia; received 4 November 1996. Modern birds are excellent habitat indicators. Potentially, the Tertiary avifauna of Riversleigh could serve a valuable role in interpreting the palaeoenvironment. Numerous bird remains have been recovered, but few have been studied (Rich, 1979; Boles,1991, 1993a, b, c, 1995, 1997a, b). Based primarily on the diverse mammal re- mains, Archer et al. (1989) interpreted the vege- tation of Riversleigh during the early to early late Miocene as ‘dense, species-rich gallery rain- forests probably similar to those that persist today in mid montane New Guinea’. Archer et al. (1994) concluded that the surrounding Pliocene habitat was ‘a dry sclerophyll forest or woodland with a grassy understorey, probably not too un- like the environment that dominates Riversleigh today’. This paper reviews available information about Riversleigh birds as it might contribute to interpretations of Tertiary environments. Material (Table 1) is lodged in the Queensland Museum (QM) and Australian Museum (AM), The geology and geography of the Riversleigh deposits are available elsewhere (Archer et al., 1989, 1994, 1995; Megirian, 1992). INTERPRETATION OF HABITATS: BASIC TENETS One of the most striking features of the Riversleigh avifauna is the large proportion of small terrestrial forms compared to other middle Tertiary sites in Australia, which are largely dom- inated by waterbirds and larger, flightless forms, both of which are also present at Riversleigh. Among modern Australian terrestrial (non- aquatic) birds, the prevailing pattern is consid- ered to be that the more primitive (least mated hindlimb proportions of £. gidju with those of Recent Dromaius and C asuarius. The tibiotar- sus and tarsometatarsus of E. gidju were rela- specialised) members of lineages are found in montane and subtropical rainforest and contigu- ous wet forests, whereas the more derived taxa occur in open habitats. Schodde & Faith (1991) considered that ‘rainforest-inhabiting members, particularly in montane New Guinea and subtrop- ical Australia, represent ancestral forms from which those in scleromorphic vegetation have been derived’. These ancestral components have been recognised as the Tumbunan fauna (Schodde & Calaby, 1972; Schodde, 1982; Schodde & Faith, 1991). The Tumbunan avifauna is now largely centred on ‘a Nothofagus-myrtle- podocarp-dominated forest of the type once widespread across Australia through the mid Tertiary’ (Schodde, 1982), although some ele- ments of this fauna now extend well beyond this habitat. Although the Tumbunan-type habitat is now mostly restricted to higher elevations, a dis- tribution that is relictual, it was once more wide- spread through lower altitudes. Schodde & Faith (1991) suggested that ‘the subtropical Tumbunan avifaunas now present in montane New Guinea were widespread in Australia’ in the mid-Terti- ary. Thus the habitat of Riversleigh during the Miocene was probably similar to that retained in these present Tumbunan refuges. This apparent relationship between level of specialisation and habitat has been adopted as a basis for palaeoenvironmental interpretation. A fossil taxon is tentatively considered a likely rainforest inhabitant if it is primitive in its lin- eage. It may also be considered to occur in rainforest if its affinities are to a group that is today confined to rainforest. If all living species occur in a certain habitat then that habitat is considered to most likely for the fossil form. lower leg permits these birds to capture prey from hollows and recesses inaccessible to other preda- tory birds. Modern species occur in a range of There seems little problem with freshwater aquatic birds. Within a family, there may be some variation in the type of aquatic habitats preferred, although the range of differences is substantial in only a few instances. Modern habitat preferences are used as an indication of the fossils’ pal- aeohabitats, unless otherwise indicated. Unless there is some outstanding morphologi- cal feature that signals a major shift in its biology, a fossil bird is considered to have similar ecolog- ical characteristics as its modern counterparts. Similarities in morphology between fossil and living forms are interpreted to share similar func- tions, unless there is evidence to the contrary. SYSTEMATIC LIST Dromornithidae This extinct family comprises 8 species in 5 genera (Rich, 1979), These were large, flightless birds with major, ‘ratite’-grade modifications to an entirely terrestrial lifestyle, The extent of these masks any relationships to other known orders of birds, although this family is no longer consid- ered closely allied to living ratites (Olson, 1985). Barawertornis tedfordi and Bullockornis planei occur at Riversleigh as conspicuous faunal elements at some sites because of their abundance and size. Despite this, they are limited pal- aeoenvironment indicators because little is un- derstood of their biology. The 2 monotypic Riversleigh genera are among the least known. At Riversleigh, the 2 species occur together, often, but not always, with other large animals (e.g., D-Site). At several sites they occur with aquatic animals such as lungfish, turtles and croc- odiles. Whether this is an indication of a water- side association or a taphonomic artefact 1s not known. Nothing in the foot structure is obviously modified for entering water, nor has there been previous suggestion of an aquatic association. Relative proportions of the hindlimb bones can be a useful indication of locomotory mode. In few dromornithid species, however, are complete specimens known for all major leg elements, and these rarely represent the same individual. Re- constructions must necessarily be based on the better known forms, particularly Genyornis new- toni. Estimates of leg proportions are based on measurements given by Rich (1979). The hindlimb proportions of most dromornithids are very different from those of Casuarius or Dromaius, with only [/bandornis lawsoni having proportions approaching those of living emus; Rich (1980) and Vickers-Rich (1991) considered MEMOIRS OF THE QUEENSLAND MUSEUM this the most cursorial species. Where known, the tarsometatarsus of other species is short relative to the other long bones, more like the moas . The moas (Worthy, 1991) provide some sug- gestions about aspects of dromornithids’ life- style. Most moas were forest dwellers, almost exclusively herbivorous. None seemed adapted to cursorial locomotion. Moa remains are often recovered in large numbers, indicating that these were gregarious birds. Many Australian drom- ornithid sites yield large numbers of specimens, indicating aggregations. Such a concentration of animal biomass is analogous to moas and sug- gests that dromornithids were herbivorous (Vickers-Rich, 1991), The dromornithid bill was much heavier and deeper than the moas’ (Olson, 1985, fig. 3; Vickers-Rich, 1991, pl. 4). The skull was larger and more robust, with scars indicating broad at- tachments for the jaw muscles (P. Vickers-Rich pers. comm.). Regardless of what dromornithids ate, they were equipped to handle more substan- tial food items than were moas. Their bills are not hooked or otherwise suggestive of a predatator. Unlike moas, no remains of dromornithid food have been found. Large accumulations of dromornithid gastroliths (gizzard stones) occur at Riversleigh (Archer et al., 1994:79) and other sites (Stirling & Zeitz, 1900; Vickers-Rich, 1991). By the late Miocene, both graviportal and cur- sorial species lived in northern Australia. This was taken to indicate both forest and open country by Rich & Baird (1986), who did not consider dromornithids to have been particularly success- ful in invading grasslands, The Riversleigh fossil material does not show cursorial modifications. It can be tentatively concluded that Barawer- tornis and Bullockornis were forest dwellers. Casuariidae - emus and cassowaries Emuarius Boles, 1991 occurs at Riversleigh. Living emus and cassowaries Casuarius are quite different in their habitat preferences and locomo- tory styles. The latter is reflected in the hindlimb, suggesting potential for inferring palaeohabitat from a comparison of E. gidju with living forms. Patterson & Rich (1989) found that the phalan- ges of E. gidju were between those of Dromaius novaehollandiae and Casuarius in morphology, although more similar to the former in relative lengths and in degree of dorsoplanar compres- sion. The fossil form D. ocypus Miller, 1963 had a tarsometatarsus that is markedly shorter relative to its width than that of D. novaehollandiae, but longer than Casuarius. Vickers-Rich (1991) in- RIVERSLEIGH BIRDS AS PALAEOENVIRONMENTAL INDICATORS TABLE 1. Bird families in Riversleigh Tertiary deposits and sites from which these have been recovered. A=Dromornithidae; B=Casuartidae; C= Phalacrocoracidae; D=Ciconiidae; E=Anatidae; F=Accipitridae; G= Rallidae; H=Cacatuidae; I= J=Apodidae; K=Halcyonidae; L= Passeriformes (! =Menuridae; 2 =Oriolidae; * = Orthonychidae). tively longer than in any other casuariid, except D. novaehollandiae. The structure of the phalanges, rela- tive width of the tarsometatarsus and proportions of the hindlimb suggest Psittacidae; We that £. gidju was more cursorial than terpreted this to mean that D. ocypus was less cursorial than D. novaehollandiae. Emuarius gidju approaches D. novaehollandiae in tarso- metatarsal length:width ratio more closely than does D. ocypus (Boles, this volume a). Increased cursoriality in the Casuariidae is characterised by an increase in the lengths of the tibiotarsus and tarsometatarsus relative to the femur. Boles (this volume a) compared the esti- mated hindlimb proportions of E. gidju with those of Recent Dromaius and Casuarius. The tibiotar- sus and tarsometatarsus of E. gidju were rela- SITE A|B|CID|E|FIG/H|1|3|K|L Casuarius and may have approached SYSTEM A the ability exhibited by D. D-Site x Ls a ent E. gidju may pay abitat preferences resemblin - DST EQUIVALENT ——| those of i aviacholloniine. ase [Sticky Beak x X largely open country, although some SYSTEM A OR B rainforest types could possibly offer a White Hunter X [x] x|x|x X | sufficiently open understorey. SYSTEM B ; Camel Sputum x lx] x x x x | Phalacrocoracidae - cormorants Helicopt x A distal carpometacarpal fragment pter iiai F m comes from a large cormorant. Be- j t—+ | yond signaling a lacustrine situation, Panorama X | itis uninformative. RSO X X xX X | Upper X |X x!| Ciconiidae - storks Wayne’s Wok x x x Stork remains comprise | proximal Wayne’s Wok Il ‘Tx | and 2 distal tarsometatarsal frag- Neville’: Garden m x2 ments, 1 quadrate and a partial skull. SYSTEMIE The tarsometarsi do not belong to - Ephipphiorhynchus, the only living Bitesantennary X | genus in Australia, and are probably Dirks Towers X X | referable to Ciconia, a genus now Microsite top x | found in Eurasia, South America and SYSTEM C Africa. All living storks have associ- Archie's Absence | ] y | ations with shallow, slow moving Hinde's tdiow x water, although they are not restricted + to aquatic habitats; they do not enter Gag x X | heavily forested areas. All eat small Gotham X| animals, including vertebrates, and Jim’s Carousel |X | some (Leptoptilos) consume carrion. Last Minute | X x$] 4 Ringtail x X x | Anatidae - waterfowl Two Trees T X Several specimens have been allo- cated to this family but no further FEE determination has been made. Sub- Rackham’s Roost x X | groups of living waterfowl have cir- cumscribed habitat preferences. The Riversleigh specimens indicate aquatic, probably lacustrine, situations. Accipitridae - diurnal birds of prey Pengana robertbolesi, a large bird of prey, with hyperflexible tarsal joint (Boles, 1993a) is con- vergent with the living Polyboroides (Africa) and Geranospiza (South America). Mobility of the lower leg permits these birds to capture prey from hollows and recesses inaccessible to other preda- tory birds, Modern species occur in a range of There seems litle problem with freshwater 244 habitats and do not pennitany meaningful extrap- olation to Riversleigh. A femoral fragment of this lamily is of comparable size to Pergane, but can be referred only tentatively to this taxon. Rullidae - rails There is much rail material, representing most forelimb and hindlimb elements, from several siles, but most abundantly at White Hunter site, possibly from a single individual, All specimens represent a medium-sized rail about the size of living Gallinulatenebrosa. From the shape of the carpometacarpus and the relative sizes of the wing and leg elements, ių appears 10 have been flightless, This rail is probably related to the native-hens 7rbonyy, NOW usually merged as a subgenus of Gallinula. The native-hens comprise two living endemic Australian species, one of which is flightless, and a flightless Pleistocene species endemic to New Zealand. Although largely remaining, in the vicinity of water, both Australian species freely enter adjacent open country, The flightless Tasmanian merrierii én- ters cultivated paddocks, and mainland ventralis may move some distance from water in semiarid und arid regions. Species of Gellinula are pregar- ious, Which 15 consistent with the number of fossils found al some sites, The Riversleigh rail indicates the proximity of wetlands, but little else about the local environment. Cacatuidae - cockatoos A rostrum has been referred to the modern white vockatoos, Cacatua (Boles, 1993h), Within Australia, these species occur from rainforest lringes through open forest and woodland toopen arid country. The Riversleigh bird is considered to have been similar to the group of white cock- atous with small bills and rounded, uncoloured crests, such as the corellas. These species exhibit a considerable range of habitat preferences, from central Australian arid zone (C. pustineter) to rainforests on some islands e.g., Solomon Ishinds (C..ducorpsii). This range of habitats occupied by modern species renders the Riversleigh specimen of hte value in palacohabitat reconstruction. Psittacidae - parrots Two carpometacarpi and a tarsometatarsus from Rackham's Roost come from the living Budgerigar Melopstttacus undulatus, This is a good indication that the Pliocene habitat at Riversleigh was open and Jighily timbered. Today this species occurs in the arid, semi-arid this the must cursorial species. Where know, the 4 ` ” t Do eee MEMOIRS OF THE QUEENSLAND MUSEUM and subhumid zones, including Kiversleigh, but never fur from water. Apodidae - swifts Humer, a coracoid and. tarsometatarsus of a medium-sized swift are close toa large species of swiltlet Collocalia, the only genus thal breeds in Australia at present. These species nest in caves, bur, despite the number of apparent Riversleigh deposits originating from cave floors, surpris- ingly few remains have been found at Riversleigh thus far. One of the bones is that of a young bird, strong indication that at least some level of local breeding was taking place. Swifts are not good environmental indicators, because they ane weriul feeders, capturing flying insects above the hahitar canopy, irrespective of what that habitat may be. Halcyonidae - forest kingfishers The single specimen available is assigned to the Halcyonidae, possibly close to Todiramphus (Boles, 1997b) and similar to more primitive living halcyonids ( Tanysiptera, Melidara, Syma), which are rainforest inhabitants (Fry, 1980a,b), The Riversleizh fossil resembles what would be predicted for an early member of the Todiramphus lineage. Australian Todiramphus occur outside rainforest, but species of the genus live in rainforests in other parts of Australasia. Passeriformes - songbirds These birds are good habitat indicators al spe- cific or generic level, Songbirds are known from about 100 specimens from at least 20 sites (Boles,1995); however, only 3 specimens thus far provide useful habitat indications. Orthonyx kaldowinyėri, of Which a femur has been reported (Boles. 1993c), belongs to a genus with the 2 living species confined to rainforest of the east coast and New Guinea, occasionally entering dense bordering vegetation (¢.g.. Lantana). Green Waterhole Cave, SE South Australia, the site tor O, Aypsilapltus, never had ramforest, but there 1s evidence for a thick cover of Leptospermum (Baird, 1985), which would prob- ably have provided adequate cover, The lyrebird Menura tvawanotles is repre- sented by a carpometacarpus (Boles, 1995). The two living species Occupy rainforest and; in the case of M. novaehollandiae, contiguous forest. The habitat preferences of the Riversleigh species of Orthonyx und Menura are assumed to be sim- ilar to those of their modern congeners. A large lower mandible from Neville's Garden Site (unpubl. data) ts from the Oriolidae which RIVERSLEIGH BIRDS AS PALAEOENVIRONMENTAL INDICATORS includes the frugivorous forest birds Oriolus and Sphecotheres viridis; many occur in closed forest. The fossil suggests rainforest, but is not diagnos- tic. TAPHONOMY None of the aquatic or semi-aquatic species, except the Gallinula-type rail, show any evidence of unusual causes of death or accumulating agents. The rails occur in greater numbers at a handful of sites, possibly due to a gregarious habit rather than taphonomy. Dromornithid remains being very common at some sites may also be a result of gregariousness or their corpses could have been accumulated during flooding. They are often found with large aquatic taxa (e.g., lungfish, turtles, crocodiles), and these may represent thanatocoenoses. Most other terrestrial species are too infrequent to provide any clues, but may be best considered chance survival of the remains of animals that died for a variety of reasons. Exceptions are the small terrestrial forms, par- ticularly passerines, which appear to have been killed by ghost bats (Macroderma). The living M. gigas is a predator of small vertebrates, which it captures from the ground or perch. Prey are eaten in the roost caves by chewing through the pectoral region, manifested in the skeleton as extreme damage to the sternum (Boles unpubl. data); dis- tal wings and legs are discarded. This is consis- tent with the elements that predominate at former Macroderma-accumulated sites at Riversleigh (carpometacarpus, tibiotarsus, tarsometatarsus). Unlike northern Australia today, where only a single species exists, Riversleigh is known to have supported many species of ghost bats during the Tertiary (Hand in press). Several sites have been identified as the remnants of Macroderma roost caves, such as the Miocene Gotham Site and the Pliocene Rackham’s Roost Site. PROBLEMS IN INTERPRETATION There are several reasons why the Riversleigh birds do not offer the same depth of environmen- tal data as living forms. Many specimens are yet to be studied. Ordinal level identification (Table 1) is frequently not fine enough for meaningful habitat inferences, especially for terrestrial spe- cies. There may be insufficient morphological differences between related forms occupying dif- ferent habitats, Much of the biology of extinct groups remain unknown. Basic assumptions could be wrong; primary ones employed here are that fossil birds had similar ecological character- istics to living counterparts and within a lineage more primitive species occur in rainforests, more derived ones in more open habitats. PALAEOENVIRONMENT The birds show that both aquatic and terrestrial habitats were prominent through the early to mid Miocene at Riversleigh. Because of the broad spectrum of wetland situations in which they occur, the cormorant, ducks and rail do not pro- vide any clues to the detailed nature of these systems. Based on modern habitat preferences, the stork indicates shallow, slow moving, lacus- trine situations somewhere in the area. Several of the better taxonomically resolved specimens are consistent with a closed forest. The passerines Orthonyx and Menura belong to fam- ilies which are today almost exclusively re- stricted to rainforest. Living halcyonid king fishers occur through most Australian habitats; the Riversleigh form, however, is suggestive of more primitive, rainforest-inhabiting taxa. Possible support for a more open habitat in some places comes from Emuarius. Its hindlimb proportions are thought to resemble those of Dromaius novaehollandiae, a highly cursorial animal. If the assumption can be made that the similarities in morphological proportions of the hindlimbs reflect similarities in function, this im- plies an advanced level of cursoriality in Emuarius as well. The Pliocene environment of Riversleigh, based on Rackham’s Roost site, was quite differ- ent from the Miocene. The Pliocene fossils are all small forms, mostly passerine but including a small extant parrot, Melopsittacus undulatus. This species is widespread, in arid and semiarid woodlands and scrublands, including the Riversleigh area. It requires proximity of water. Thus it suggests that Riversleigh in the Pliocene was probably very much like it is today. The Riversleigh early to mid Miocene environ- ment probably included shallow water at a num- ber of different sites, with some surrounding rainforest and some more open forrests. ACKNOWLEDGEMENTS Comparative material has been made available by the curatorial staff of the Australian National Wildlife Collection, Museum of Victoria, Queensland Museum, South Australian Museum, United States National Museum and University of Kansas Museum of Natural History. Valuable discussions were provided by R.F. Baird, the late G.F. van Tets, P. Vickers-Rich, N. Pledge, M. Archer and A. Gillespie. Support came from an ARC Grant; the University of New South Wales; Department of Arts, Sport, the Environment, Tourism and Territories; the National Estate Grants Scheme; Wang Australia; ICI Australia and the Australian Geographic Society. LITERATURE CITED ARCHER, M., HAND, S.J. & GODTHELP, H. 1994. Riversleigh 2nd edition. (Reed: Sydney). 1995, Tertiary environmental and biotic change in Australia. Pp. 77-90. In Vrba, E.S., Denton, G.H., Partridge, T.C. & Burkle, L.H. (eds), Paleocli- mate and evolution, with emphasis on human origins. (Yale University Press: New Haven). ARCHER, M., HAND, S.,GODTHELP, H. & MEGIR- IAN, D. 1989. Fossil mammals of Riversleigh, northwestern Queensland: preliminary Overview of biostratigraphy, correlation and environmental change. Australian Zoologist 25: 29-65. BAIRD, R.F. 1985. Avian fossils from Quaternary de- posits in ‘Green Waterhole Cave’, south-eastern South Australia. Records of the Australian Mu- seum 37: 353-370. BOLES, W.E. 1991. Revision of Dromaius gidju Patter- son and Rich, 1987, with a reassessment of its generic position. In Campbell, K.E., Jr (ed.), Pa- pers in avian paleontology honoring Pierce Brodkorb. Natural History Museum Los Angeles County Science Series 36; 195-208 1993a. Pengana robertbolesi, a peculiar bird of prey from the Tertiary of Riversleigh, northwestern Queensland, Australia, Alcheringa 17: 19-25. 1993b. A cockatoo (Aves: Psittaciformes: Cacatuidae) from the Tertiary of Riversleigh, northwestern Queensland, Australia, with com- ments on the value of the rostrum to the system- atics of parrots. Ibis 135: 8-18. 1993c. A logrunner Orthonyx from the Miocene of Riversleigh, northwestern Queensland. Emu 93: 44-49. 1995. A preliminary analysis of the Passeriformes form Riversleigh, northwestern Queensland, Australia, with the description of a new species of lyrebird. Courier Forschungen-Institut Senckenberg 181; 163-170. 1997a. Hindlimb proportions and locomotion of Emuarius gidju (Patterson and Rich. 1987) (Aves: Casuariidae).Memoirs of the Queensland Museum 41; 235-240. 1997b. A kingfisher (Halcyonidae) from the Miocene of Riversleigh, northwestern Queens- land, with comments on the evolution of king- fishers in Australo-Papua.Memoirs of the Queensland Museum 41: 229-234. MEMOIRS OF THE QUEENSLAND MUSEUM FRY, C.H. 1980a. The origin of Afrotropical king fish- ers. Ibis 122; 57-72. 1980b. The evolutionary biology of kingfishers (Al- cedinidae). Living Bird 18:113-160. HAND, S.J. In press. Macroderma malugara, a new Miocene megadermatid (Microchiroptera) from Australia, with broader comments on megadermatid evolution. Geobios. MEGIRIAN, D. 1992. Interpretation of the Miocene Carl Creek Limestone, northwestern Queensland. The Beagle 9: 219-248. OLSON, S.L. 1985. The fossil record of birds. Pp. 79-238. In Farner, D.S., King, J.R. & Parkes, K.C. (eds), Avian biology, vol. 8. (Academic Press: New York). PATTERSON, C. & P.V. RICH. 1987. The fossil his- tory of the emus, Dromaius (Aves; Dromaiinae). Records of the South Australian Museum 21: 85-117. RICH, P.V. 1979. The Dromornithidae, an extinct fam- ily of large ground birds endemic to Australia, Bulletin of the Bureau of Mineral Resources Ge- ology & Geophysics, Australia 184: 1-196. 1980. The Australian Dromornithidae: a group of extinct large ratites. Natural History Museum of Los Angeles County, Contributions to Science 330: 93-103. RICH, P.V. & BAIRD, R.F. 1986. History of Australian avifauna. Current Ornithology 4: 97-139. SCHODDE, R. 1982. Origin, adaptation and evolution of birds in arid Australia. Pp. 191-224. In Barker, W.R. & Greenslade, P.J.M. (eds), Evolution of the flora and fauna of arid Australia, (Peacock Publi- cations: Frewville). SCHODDE, R. & CALABY, J. 1972. The biogeogra- phy of the Australo-Papuan bird and mammal faunas in relation to Torres Strait, Pp. 257-300. In Walker, D. (ed) ‘Bridge and Barrier: The natural and cultural history of Torres Strait.’. (Australian National University Press: Canberra). SCHODDE, R. & FAITH, D.P. 1991. The development of modern avifaunulas, Pp. 404-412. In Bellet al. (eds), Acta XX Congressus Internationalis Or- nithologici. (New Zealand Ornithological Con- gress Trust Board: Wellington), STIRLING, E.C. & ZEITZ, A.H.C. 1900. Fossil re- mains of Lake Callabonna. I. Genyornis newtoni, A new genus and species of fossil struthious bird. Memoirs of the Royal Society of South Australia 1: 41-80. VICKERS-RICH, P, 1991. The Mesozoic and Tertiary history of birds on the Australian plate. Pp. 721- 808. In Vickers-Rich, P.V., Monaghan, J.M., Baird, R.F. & Rich, T.H. (eds), Vertebrate pal- aeontology of Australasia. (Pioneer Design Stu- dio: Melbourne). WORTHY, T.H. 1991. An overview of the taxonomy, fossil history, biology and extinction of moas. Pp. 555-562. In Bell et al. (eds), Acta XX Congressus Internationalis Ornithologici. (New Zealand Orni- thological Congress Trust Board: Wellington). A NEW OLIGOCENE-MIOCENE SPECIES OF BURRAMYS (MARSUPIALIA, BURRAMYIDAE) FROM RIVERSLEIGH, NORTHWESTERN QUEENSLAND J. BRAMMALL AND M. ARCHER Brammall, J. & Archer, M, 1997 06 30: A new Oligocene-Miocene species of Burramys (Marsupialia, Burramyidae) from Riversleigh, northwestern Queensland. Memoirs of the Queensland Museum 41(2): 247-268. Brisbane. ISSN 0079-8835 Burramys is abundant m the Oligocene-Miocene at Riversleigh, northwestern Queensland. Burramys brutyi sp. nov, 18 represented by over 150 dentary and mavillary fragments and isolated teeth from 22 sites. Burramys appears to be morphologically conservative, with only minor metrical variation between specimens of B. Arutyi from different sites and relatively few features distinguishing Miocene, Pliocene and Recent species. Phylogenetic analyses suggest that B. briayi is the plesiomorphic sister-group to all other species of Burramys. with B. wokefieldi s\ster-group to the clade comprising B. triradiatus and B. parvis. i Burramyidae, Burramys brutyi, Riversleigh. Oligocene. Miocene. J. Brammall & M, Archer, Schoal of Biological Science, University af New Sanh Wales, New South Wales 2052, Austrilia; received 4 November 1996, Burramys was represented only by Pleistocene fossils of B. parvus trom Wombeyan Caves, NSW (Broom, 1896) and Pyramids Cave. Victo- ria (Wakefield, 1960) unti] 1966 when the Moun- tain Pygmy-possuni, B. parvus, was discovered alive at Mount Hotham, Victoria (Anon,, 1966; Warneke, 1967). Two more fossil species of Burramys have been identified: early Pliocene B. triradiatus from Hamilton, Victoria (Turnbull et al., 1987) and B. wakefieldi from late Oligocene (Woodburne et al.. 1993) Ngama Local Fauna, South Australia (Pledge, 1987). Discovery of Miocene Burranys at Riversleigh extends the geographic range far north and provides the first sizeable Tertiary sample (150 specimens). A met- nic analysis of this sample aims to determine taxa present and to assess variabon, Burramys brutyi sp. nov. is used as the basis tor an evaluation of intrageneric phylogenetics of Burramyy, Dental homology follows Flower (1867) tor premolar numbering and Luckett (1993) for pre- molar/molar boundary and molar number. Tooth positions given without super- or subscript num- hers refer to both upper and lower teeth; thus M? and My are individual teeth bul M4 refers to both. Molar cusp nomenclature follows Archer (1984) not Pledge (1987). Pledge’s paraconid is our pro- toconid: his protoconid is not recognised. Higher systematic nomenclature follows Aplin & Archer (1987). System nomenclature 1s based on Archer et al. (1989) and Creaser (1997), Mu- terial referred to is housed in the Queensland Museum, Brisbane (QMF) or Museum of Victo- ria, Melbourne, (NMVP). Measurements in millimetres (mm) are to the nearest 0.0) mim using a Wild MMS235 Digital Length-Measuring Set attached to a Wild MSA stereomicroscope. Molar lengths and widths and molar row lengths were measured as the maximum dimensions of the enamel-covered crown(s) with the teeth in occlu- sal view, with lengths taken along the an- teroposterior axis of the tooth and widths measured perpendicular to that axis. For P3 in dorsal view and P? in ventral view, maximum length was measured parallel to the apical blade edge, and anterior, posterior and maximum widths were measured perpendicular to the blade edge; buceal and lingual heights were measured from the base of the enamel at the saddle between the roots, to the median apical edge, parallel to the posterior edge of the tooth. Statistical analy- ses were performed using SYSTAT and Kaleida- Graph data analysis and graphics applications. METRIC ANALYSIS Despite overall uniformity, Riversleigh Burramys material shows some varialion in rela- tive and absolute premolar and molar sizes. Met- ric analysis of dental features attempted to identily patterns which might indicate sexual di- morphism, specific or subspecific separation or differentiation of populations from different sites. Univariate and bivariate distributions and princi- pal components analysis were employed. Cheektooth dimensions (Table 1) for Recent B, parvus populations refer to left dentition ex- cept where the right dentition was more complete, Standard error (SE) is used rather than standard deviation (SD) hecause it better indicates reliabil- ity of the mean estimate, The coeffietent of vari- ation (CV )= SD divided by mean x 100. 248 TABLE 1. Cheektooth dimensions of Burramys species. Results given as: Mean MEMOIRS OF THE QUEENSLAND MUSEUM + Standard Error (No. Specimens) Coefficient of Variation (CV%). CV not given where n 2. L = length, AW = anterior width, PW = posterior width, MW = maximum width, LH = lingual height, BH = buccal height. = al e Burramys Lower teeth CV 1.81 +0.01 (38) 1.03 +0.01 (38) 1.22 +0.01 (38) 1.27 +0.01 (29) 1.44 +0.01 (37) 1.73 +0.02 (37) 1.24 +0.01 (32) 0.78 +0.01 (32) 0.95 +0.01 (32) 1.09 40.01 (32) 0.88 +0.01 (34) 0.96 +0.01 (34) 0.93 +0.02 (10) 0.84 +0.01 (10) 0.85 +0.02 (10) 0.66 (1) 0.64 (1) 0.50 (1) 2.30 +0.02 (27) 3.24 +0.05 (8) 3.83 (1) 3.37 3.91 Upper teeth pe L P3 AW 2.01 +0.02 (17) 0.93 +0.01 (17) 1.20 +0.01 (17) 1.25 +0.01 (17) 1.58 +0.02 (17) 1.65 +0.02 (17) 1.12 40.02 (14) 1.16 £0.02 (14) 1.17 £0.01 (14) 1.39 +0.01 (14) 0.98 +0.01 (8) 1.16 +0.01 (8) 0.93 +0.01 (8) 0.86 +0.03 (3) 0.93 +0.02 (3) 0.72 +0.03 (3) 0.67 +0.02 (3) 5.68 0.71 +0.02 (3) 4.97 0.46 +0.03 (3) 10.80 2.11 £0.03 (8) 3.40 2.98 +0.08 (3) 4.55 3.55 40.03 (3) 1.33 4.89 4.32 4.93 4.41 4.67 4.44 4.86 5.25 4.14 3.90 3.45 2.84 2.40 6.43 2.84 7.86 CV is less than 11 throughout and usually less than 6 (Table 1). Following Simpson et al. (1960), this degree of variation indicates an unmixed sample, although Gingerich (1974) cautions against uncritical application of this absolute CV criterion and recommends greater emphasis on relative variabilities of different teeth. In approx- iee CV CV CV 2.58 +0.05 (4) 1.04 +0.01 (4) 1.49 +0.06 (4) 1.67 +0.06 (4) 2.02 +0.04 (4) 2.44 +0.05 (4) 2.17 +0.01 (21) 1.96 0.85 +0.02 (21) 10.51 1.32 +0.01 (21) 1.39 +0.02 (21) 1.92 +0.01 (19) 2.22 +0.02 (20) 1.57 +0.01 (21) 1.00 +0.01 (21) 1.25 +0.01 (21) 1.57 +0.01 (21) 1.21 +0.01 (21) 1.32 +0.01 (21) 1.23 +0.01 (19) 1.06 +0.01 (19) 1.07 +0.01 (19) 0.68 +0.01 (14) 0.66 +0.01 (14) 0.52 +0.01 (14) 3.13 £0.01 (21) 4.34 +0.01 (19) 4.93 +0.02 (14) 1.55 (1) 1.23 (1) 1,32 (1) 1.32 +0.04 (2) 1.13 +0.00 (2) 1.17 +0.01 (2) 4.29 0.00 0.61 1.29 2.59 +0.02 (2) 0.91 +0.02 (2) 1.63 +0.05 (2) 1.63 +0.05 (2) 2.32 +0.02 (2) 2.16 +0.01 (2) 1.09 2.32 4.79 4.79 1.22 2.27 +0.01 (19) 2.60 0.75 +0.02 (19) 10.39 1.13 40.01 (19) 3.82 1.24 +0.01 (19) 2.44 1.92 +0.01 (16) 2.76 2.06 +0.02 (18) 3.08 1.51 +0.01 (19) 1.94 1.40 +0.02 (19) 7.08 1.45 +0.01 (19) 4.01 - 1.68 +0.01 (18) 2.77 1.22 (1) 1.45 +0.01 (19) 1.53 1.34 (1) 1.56 +0.01 (19) 2.20 1.10 (1) 1.27 +0.01 (19) 3.37 1.09 +0.02 (18) 8.17 1.19 +0.03 (18) 10.83 0.88 +0.02 (18) 10.71 0.77 +0.01 (13) 4.86 0.74 +0.02 (13) 8.81 0.51 +0.01 (13) 9.49 2.96 +0.01 (19) 1.85 4.07 +0.02 (18) 1.63 4.77 +0.03 (13) 1.86 imately 80% of measurements B. parvus has a lower CV than the Riversleigh sample, but the interspecific differences in CV are generally not great. CVs for B. triradiatus fall within approxi- mately the same ranges as those for the Riversleigh and Recent specimens, but are de- rived from very few specimens and are therefore NEW OLIGOCENE-MIOCENE BURRAMYS FROM RIVERSLEIGH 249 ai k- — = 7 12 A 5 B 6 5104 E | ¢ 5 510- o | T E g] E 4 Ë, fa} ey | 9 | 2 e o Oo 6 a 3 uw y 2 n o i [=] o 2 = - =- z z 2 14 1 o E, i EA 02 ja saad .ä) Ri 110 1.16 4,22 128 134 140 1.00 1.08 112 116 1.20 2. M; length Mp length \eq— o uer Ay - 10; 10D 7 |E F u $ | u 6-4 w B 5 Bi. | 8 £ E 4- E 21 E 6 Oo g 44 g fe] 2 5 2 S olm a «| | eo | i 1.60 1.70 7.80 1.90 200 2,10 105 1412 119 1.26 183 1.40 145 1.56 1,67 1.78 1.89 2.00 Ps length P, posterior width P; buccal height = System A BB = Systems fE = System C FIG. 1, Frequency histograms for some lower tooth measurements of Riversleigh Burramys specimens. All Measurements inmm. not considered reliable. Total variation (as indi- cated by CVs) suggests 1 species of Burramys in the Riversleigh sample spanning greater variation than the sample af Recent B, parvus. Where variation between taxa is small (as 1s likely with small-bodied taxa), it may be ob- scured by epigenetic morphological variation, by tooth wear or by measurement error; metric dif- ferences between closely related taxa are most likely to be detected by examining structures with the lowest levels of such variation, Ma, in the centre of the P4 -Mg tooth row, is in that sense the most functionally integrated of these teeth; it may therefore be expected to be least variable (Gingerich, 1974). Similarly, total molar row lengths may be more tightly controlled than the Jengths of individual molars. M2 dimensions and molarrow measurements (ineluding partial molar row measurements such as M,.2 length) are gen- erally the least variable measurements in B. parvus and the Riversleigh sample; P3 length is also relatively constant (Table 1), Thus analysis ol the Riversleigh sample was focused on P3 and M).2, although all other measurements were ex- amined, Frequency histograms for some measurements are bimodal, while others ure either unimodal or perhaps incipiently bimodal. Mı-> length (Fig. 1C), with CV=3.37 is bimodal. Ma length (Fig. IB; CV=3,72) and P; buccal height (Fig. 1E; CV=5.64) are considered bimodal, though not with certainty. Mı length (Fig. 1A; CV=3.71) may represent a bimodal distribution but could equally be a sample trom a unimodal. normal distribution; P3 length (Fig. 1D; CV=4.00) and P3 posterior width (Fig. 1E; CV=5.07) distributions could each be described either as having 2 or 3 peaks, or as representing single normal distribu- tions, Kolmogoroy-Smirnoy Lilliefors tests indi- cate that some of the univariate distributions differ significantly trom normal (Table 2) and comparison with Table | shows that these include several with low variation. Thus univariate Ire- quency distributions hint that the sample repre- sents more than one population. bul do not provide a basis for subdivision. Bivariate plots (Fig. 2) suggest no clear divi- sions other than those evident in the univariate distributions, such as the apparent bimodality of M: length (Fig. 1B, 2B). They show that specj- mens from Systems B and C have overlapping distributions, but thal for some measurements, specimens from System C sites are, on average, smaller than specimens from System B sites, This is so for Mı length and P3 length (Fig. 2A, 2C, 2D) and to 4 lesser extent for Mo length (Fig. 2B), 250 MEMOIRS OF THE QUEENSLAND MUSEUM 1.105 1.155 A g B = o ? £ | 3 1.004- 1 e ees... S A 3 1.054 spayed 3 c o i Sse “ce eu dace: oun) ie = 0.954 m e “ing ‘ Š 1.00 4 an ee E- ae E A | m a | e f A" 3 eo Q 0.90 fave Ae: e E P A T. 200-95 ee a ee onl Ce = o e° S eo ob to | 0.80 + T T T T 0.85-+ T T T 1 1.10 1.15 1.20 1.25 1.30 1.35 1.40 1.02 1.04 1.06 1.08 1.10 1.12 1.14 1.16 1.18 M, length Mz length 2.00 2.05 u E m 4 e @ = 2 1.85 He it DESEE AN ak 2 2 E e9 a ce oo oe oe ied 1.704 A reat | (E) | 1.50 4— ++ +> —1—_ + 1.65+—L. r i - : i 1.65 1.70 1.75 1.80 1.85 1.90 1.95 2.00 2.05 1.10 1.15 1.20 1.25 1.30 1.35 1.40 P3 length M, length EH = System A @ = System B O = System C FIG. 2. Bivariate plots for some lower tooth measurements of Riversleigh Burramys specimens. All measurements inmm. TABLE 2. Column 1, Kolmogorov-Smirnovy Lilliefors tests for normality. Probability (P) values below 0.05 indicate a difference from normality significant at the 95% level. Columns 2 and 3, mean values for Systems B and C respectively. Column 4, Students t-tests; P indicates a significant difference between Systems B and C. Abbreviations as for Table 1. [| | LLilliefors Test (P) P3 L 0.088 T.82 1.79 0.266 0.175 0.455 0.678 0.215 0.904 0.002 0.193 0.072 0.009 0.009 0.585 0.593 0.177 0.038 0.054 0.001 1.03 1.22 1.26 1.43 1.72 1,25 0.79 0.96 1.09 0.87 0.95 0.93 0.85 0.85 2.32 3.25 1.05 1.22 1.29 1.44 1.76 1.21 0.76 0.95 1.09 0.91 0.98 0.98 0.81 0.81 2.27 3.16 0.474 0.928 0.211 0.923 0.254 0.071 0.268 0.625 0.903 0.041 0.111 0.458 0.394 0.458 0.226 0.529 NEW OLIGOCENE-MIOCENE BURRAMYS FROM RIVERSLEIGH 251 @ =SystemB [O = SystemC A = C+ (Encore Site) FIG. 3. Specimens of Burramys from various sites at Riversleigh plotted on principal component axes ob- tained using 11 measurements from P3 and M}-2. Eigenvectors recorded in Table 3. X = first principal axis, Y = second principal axis, Z (perpendicular to page) = third principal axis. Solid line encloses spec- imens from System B sites. Dashed line encloses specimens from System C sites, including Encore Site. Dotted line excludes from System C ‘aberrant’ specimen QMF30104, indicated by arrow. but appears not to be the case for P3 buccal height (Figs 2C). M2 posterior width shows the opposite trend (Fig. 2B), whereby System C specimens are on average larger than System B specimens. Student’s t-tests show these differences to be non-significant at the 95% level (Table 2), but a principal components analysis employing dimen- sions of P3 and Mj.2 (Fig. 3, Table 3) confirms that total variation is explained partly by System C specimens being smaller than system B speci- mens. Eigenvectors for component 1 are all pos- itive (Table 3), indicating that this is a general ‘size component’; specimens scoring high on the first component (i.e. falling further towards the positive, or right-hand side of the X-axis in Fig. 3) are larger than those to the left. Although there is considerable overlap between Systems B and C, the centre of mass of the System B distribution is further to the right than that for System C. Specimens from Encore site (younger than Sys- tem C, ?early late Miocene) cluster at one ex- treme of the System C distribution, with the exception of a single large aberrant specimen QMF30104 from Gag Site (Fig.3). In the System B-System C continuum (Fig. 3) the cluster of Encore Site specimens falls on the ‘older’ (Sys- tem B) end of the System C spectrum. Despite the apparent trend of mean difference between specimens from Systems B and C, spec- imens from both Systems are present in each of the apparent peaks of the univariate distributions (Fig. 1A-F). This suggests that the underlying structure of the sample is not simply anagenetic change tracked from the older System B sites to younger System C sites, though such may have occurred. The bimodality of several of the fre- quency histograms may reflect sexual dimor- phism and/or 2 roughly contemporaneous taxa. This suggestion is also supported by data plotted against sites arranged in estimated stratigraphic order (Fig. 4A-F.) Although samples from indi- vidual sites are inadequate to compare within- and between-site variation statistically, variation between sites is only a little greater than that within Upper site, provenance of the largest sam- ple. Caution is therefore necessary when interpre- ting apparent between- or across-site trends (such TABLE 3. Results of principal components analysis using 11 measurements of P3 and Mj-2 of Burramys specimens from Riversleigh. Abbreviations as for Table 1. 88.799 10 93.318 98.466 99.797 _ 100.000 252 MEMOIRS.OR THE QUEENSLAND MUSEUM P posterior width $1904F .-__. 4 ++ £18044 - 3°30: —_ a H 8 $ è og H 9 1.704 = OF E H: 3 ° è t QO m 1.60... š Fe am s e Site number bet e a ee |e || A Low 8 $ HGHB oLowc moc | Cr MDB HGHC H = SystemA @-=SystenB [O = SystemC FIG, 4. Riversleigh Burramys: size measures against sites in stratigraphic sequence (Archer et al, 1989). Distances on horizontal axis arbitrary. Sites: l=White Hunter, 2=Creaser’s Ramparts: 3=Outasite; 4=RSO, as size decline over time) as being significant. Although site 8 (Fig.4) includes Ten Bags Site, Mike’s Potato Patch and Upper Site, most speci- mens are from Upper Site and these span the Tange of Variation at site 8. If two morphotypes ure present they are both represented in Upper Site (Fig.4), Muirhead (1994) demonstrated size-guilding comparable to that in Recent mammal communities among 7 Upper Site bandicoot species separated on size; this is possibly due to competitive displacement of taxa that eat size-variable foods such as seeds or insects Within a single community, Thus, Upper Site probably represents a single diyerse community and the 2 morphotypes of Burramys could be sexual dimorphs or sympatric taxa. We reject sympatry because of morphological consis- tency of specimens falling near the peaks in the Mı- length distribution; the size difference be- tween those peaks is too small (Roth, 1981) to represent 2 species in different niches at the same level of a food web. The ratio of the second peak to the first (Fig. 1A-C) is 1.06, short of the often- cited cutoff value of 1.3. (Roth (1981) showed that the ‘constant ratio rule’ is empirically unsub- slanuiated, but suggested that character displace- ment is unlikely to be indicaled by ratio values lower than 1.3.) Challenging the likelihood of either sexual di- morphism orsympatry is the fact that some spec- imens are in the higher peak of apparently bimodal distributions. for some measurements, but the lower peak for others; whereas other specimens remain in one peak or (he other tor all or most measurements. There appears to be no combination of features that can be used to sub- divide the sample; this is supported by the multi- variate analysis (Fig. 3) which fails to divide the sample, A general trend to declining size through Systems B and C (Figs 2,3,4A,C) is evident, but some Encore Site specimens suggest reversal of the trend. SUMMARY Riversleigh Burramys specimens may repre- sent two populations. Patterns of variation also suggest a cline of dereasing size through time; however, small sample sizes and uncertainty of relalive ages limit the reliability of this observa- 3=Wayne s Wok; 6=Camel Sputum, Neville's Garden and Dirks Towers; 7=Inabeyance; 8=Ten Bags, Mike's Potato Patch and Upper Site; 9=Kangaroo Jaw; 10=Gag; 1J=Last Minute) 12=Main Site: 13= Jim's Jaw; 14=Wang,; |5=Encore. NEW OLIGOCENE-MIOCENE BURRAMYS FROM RLVERSLEBIGH 253 tion. If two populations have heen sampled, mag- nitude and distribution of variation suggest that these are males and females of | species, Extant populations of B. parvus are not dentally dimor phic (Brummall, unpubl,), but their alpine habitat is far removed from the Miocene rainforest envi- ronmental Riversleigh (Archer ctal., 1989, 1991) so it is not possible tò infer that Recent and Miocene Burramys share population structures, We recognise a Single new species. SYSTEMATIC PALAEONTOLOGY Class Mummalia Linnacus, 173% Supercohort Marsupialia Tigern 18} 1 Onler Diprotodontia Owen. 1866 Superfamily Burramyardea Broom IRY8 Family Burramyidae Broom. 1898 Burramys Broom. 1896 Burramys brulyi sp. nov. (Figs 5-9: Tables 1,4) ETYMOLOGY For the late Arthur Bruty who, to- ether with his daughter Blame Clarke, helped colleet many specimens (ind discovered Bruty & the Beast Site on the Gag Phatewu. MATERIAL. Holotype QMFHIIO2 (Fig. 5), a lef dentary (DEN) with Ij, P13, My.2 and alveoli for I2 and M3-4. The tip of fy i$ missing, as ure the condylar, angularund coronoid processes, Paraty pes QMF301 76 (Piz. 6), R DEN with P3-3,M [-4, broken anterior ty Po and missing (he ascending ramus and condylar, angular and eoronoid processes, QMF3009) (Fig. 7), L maxilla with Pè, M!4 and palate medial to cheektecth. Types from early to mid Miocene Upper Site on Godihelp Hill, DSite Plateau, Other material: SYSTEM A - White Hunter Site. QMF23344, RM2; QMF23500, DEN with RP3, SYS- TEM B - Camel Sputum Site, QMF20732, DEN with RM, P3: QMF20735, R DEN: QMF20736, TM? QMF30000, maxilla with LPM), OMF30107, DEN with LI), P2-3,M1-2, OMF301 10, DEN with RI, Poa, Mj-2, Inabeyanee Site: QMF3I0079, DEN with LPs. Mi-3. Mike's Potito Patch Site; QMP20759, DEN with LM2: OMF20760, LM! QMF20761, P3 or P3. Neville's Garden Site: QMP207 18. DEN with RP3, My, QMP 20748, LMa; OMF20902, DEN with RI, Ps, M1; QMF23349, DEN with LPs. My-2; QMF23376, DEN with RPs, Mp, QMF23511, DEN with RP3: QMF24261. maxilla wilh RP; QMP30089, maxilla with RP, MiA, QMF30092, maxilla with RP?" MEt (MF30113, DEN with RP3, M1; QMP30114, Ups. QMF30132. DEN with LP}, M1-3; QMF30271. RM, Quase OMP20769, t URN; OMPS0080, DEN with LIL P3, Mi-a, RSO Site: OMF3008], DEN wilh LPa, M14; QMP30084. DEN with RI]. Pa, Mi- QMF30094, maxilla with RP% OME30140. LPS OMF30İ41, LP; QMF30142, RP®, Ten Bags Site: OMF23502, DEN with LPa, Mj. Upper Site QMF20774, DEN with Rij; QMF20775, DEN with Lin P3 OMF20776, DEN wilh RP3, QMF20777, DEN with RM2; OMF20785, maxilla with RMS QMF20786, DEN with LM1-3, P3; QMF20787, max- illa with LM!, P24; QMF20788, maxilla with LAI! QMF30082, DEN wath LP3, Mj-2) OMF30083, DEN with RP2.3, Mj-2; OME 30085, DEN with LIy, P4, Mi: QMF30086, DEN with Rly, P23, M1-3; QMF30087. maxilla with LP: QMFII8S, maxilla with Lp*3, QOMF30091, maxilla with LP2 3, Mi. OMP30005, maxilla with RP*, M/-?; QMP30096, maxilla wilh LP; QMF30097, maxilla with RP3; OMF30098, may- illa with RPS. Ml; QMF30099, maxilla with RP? QMF30101, maxilla with LPZ, M12; QMF30102. DEN with Lii, Py-3, M12; OMF30103, maxilla with LPS, Mi: QMF30106, DEN with Ri, Pas: QMP30(11, DEN with Lilt, P23, Mt; OMP305 12. DEN with Rly, P3; QMF301 17, DEN with RP3, Mya QMF301 18. DEN with RL. Pa. M1: QMP30114 DEN with RM2,9; QMF30120, DEN with LP3, Mp2. QMF30121, DEN with LI), P3; QMP30122, DEN with Ll), Ps; QMF30123, DEN with LP3, Mya: OMP30(24, DEN wih RP3, M 1-3; QMF30125, DEN with RP}, My; QMF30127, DEN with Li, Py QMF30128, DEN with RP3; QMF30124, R DEN: QMF30130, 30131, 1 DEN; QMF30137, DEN with RP3. Mp2: QMF30138, maxilla with LP: QMF30139, RP2; QOMP30146, 30148, 30149, 30152 LPs; OMF30147, 30150, 30154, 30155, 30179, 30182 LP: OMF30151, 30153, 30174, 30180, 30184 RPA: QMF30160, LM2; OMPF30164-30167, LM: OMEF30168, 30173, 30177 RM! QMPROL76, DEN with RP2-3, Mi-a QMF3O0TSI, 30183 RP3: OMEP30185, JOL9U RM3, QMF3VISO, RMŽ, OMP30187, Ll; OMF 30188, RI), QMP30189, RM2. Wayne's Wok Site: OMF20725, maxilla with RPS- QMF20720, maxilla with RM!**: OMF20737. maxil- lary fragment with RPS; OMF20738, DEN with RM; OMF20744, DEN wath RM 4. P3. OMF20745. DEN with LM2.3; QMF20746, DEN with RMi2; OMF 22816. maxilla wilh RP2-* MM! 4 QMF30108, DEN with RP2-3, Mp2; OMF30136, DEN with LPa, My Wayne's Wok 2 Site: QMF30100, DEN with RI 7. Ps. Mj-3; QMF30175, LP?, SYSTEM BOR C - Clef ul Ages | Site: OMF20905, R DEN, Cleft of Ages 2A Site’ OMP29772, maxilla with RP?,M!. Cleft ot Ages 4 Site: QMF20767_ RP: QMF20835, RP}: QMF20834, RP3; QMF23200. RP. SYSTEM Č . Encore Site; QMF20752, 1.M3; QMPF20753, LPY; QMF20754, LM), OME20904, DEN with RM p2, Po 3: OMF23462, DEN with RMy-2, Pa, QMF24334, DEN with LLMj2; QMF24424, DEN with LM4; QMF24426, DEN with LIL, P3. M2; OMF24552. RPS: OMF24727, DEN with Lily. Pa, Mi-2 Gag Site- OMF30078, DEN with RP3, OMEP30093, maxilla with LP) OMP3NI04 DEN with Ll), Mya. Pi; QMP30134. L DEN: OMF301 45. DEN with LP, M1- MEMOIRS OF THE QUEENSLAND MUSEUM FIG, 5. A-C, Burramys brutyi n. sp. holotype QMF30102. Left dentary with I; P2.3 M1-2 in (A) buccal, (B-B’) occlusal stereopair and (C) lingual views. D-F, Burramys bruryi paratype QMF30091, Left maxilla with P% 4 4 4 ‘ t À 4 . M "^+ in (D) buccal, (E-E’) occlusal stereopair and (F) lingual views. Scale = 2mm. QMF30137, LP?. QMF30156, LM2; QMF30157, RM!; QMF30158, RM1; QMF30161, LM3; QMF30170, RP?: QMF30171, RP3. Henk's Hollow Site: QMF30172, LP. Jim’s Jaw Site: QMF30178, DEN with RP3. Kangaroo Jaw Site: QMF30115, DEN with RP3, M1-2. Last Minute Site: QMF30105, DEN with RI}, Mj-3, P2-3; QMF30116, DEN with RP3, Mi-2; QMF30143, LP; QMF30144, RP3: QMF30145, LP? apical fragment; QMF30162, RM; QMF30163, RM3; QMF30169, DEN with RP3, M1. Main Site: QMF30109, DEN with RP3. Ringtail Site: QMF20756, RP?; QMF20757, maxilla with RM!-2, P3. Wang Site: QMF20763, maxilla with LP3; QMF20766, DEN with RM1, P3; QMF30272, RP3. AGE UNCERTAIN - Creaser’s Ramparts Site: QMF20771, LP3. NEW OLIGOCENE-MIOCENE BURRAMYS FROM RIVERSLEIGH FIG. 6. Burramyshbruryi paratype QMF301 76; occlusal stercopair of right dentary fragment with P23 and My.4. Scale = 2mm. DIAGNOSIS. Differs from B. triradiatus and B. parvus in being smaller, in having upper and lower plagiaulacoid P3 smaller and with fewer (5-6) cuspules and associated ridges and in hay- ing 2-rooted upper and lower M4, Dentary and maxilla more robust than in B. parvus, with smaller palatal vacuitics, shorter 12-P2 interval and less reduced posterior molars. P3 with larger crown and larger posterior root than that of B. wakefieldi and diverging less trom anteroposter- ior Molar row axis, P).2 double-rooted: single- rooted in B, wakefieldi. Distinguishable trom 5. wakefieldi and B. parvas hy Mı cusp morphol- ogy: protoconid more lingual in B. wakefieldi than 5. parvus ot B. brutyt; metaconid more an- terior in B. parvus than B, bruryi or B, wakefieldi. COMPARATIVE DESCRIPTION. The dentary of B. brutyi is subequal to that of B. wakefield? in size and shape. Both are more robust than that of B. purvus but slightly less so than that of 4. triradiatus. The leading edge of the ascending ramus of B. brutyi is considerably more robust and rises at a steeper angle from the horizontal axis Ol the dentary than does that of B. parvus, but not quite as steeply as that of B. triradtatus. The I2-P2 interval is shorter in B. brutyi than in B. parvus but is not as short, relative to the length of the ramus, as that of B, ¢riraeiatus, Lowerdentition. I is long, slender and procum- bent, with the tip curved upwards and slightly twisted. It is slightly less procumbent in 8. brutyi than in 8, parvus, The crown of 1, 18 basally about the same dorsoventral thickness in Æ, bruryi and ia N A B. parvus but a litlle thicker in B. trrradiatus. l) of B. bryryi thins abruptly about half way along its exposed Jength, with the anterior half of the tooth being narrower than the pasterior half. In lateral view I of B. brutyi is more curved than in the other species. lz has not been identified in B brut, B wakefieldi or B. triradiatus. In B. parvus lz is small and single-rooted, inserting into a shallow alveolus directly behind the posterior alveolar margin ofl. hs crown inclines forward to overlie I, posterobasally. In some specimens of B. brary there appears to be the remnant of a small alveo- jus in the fragile region between lj and Py, sug- gesting a small, single-rooted Iz. P, is small, 2-rooted and cap-like, the crown swelling beyond the roots in all directions, There is a minor ridge along the anteroposterior axis of the tooth, with the crown sloping away from the crest on each side towards the lingual and buccal margins respectively. In dorsal view it is almost circular in outline, being slightly wider than long. The crown does not extend as far beyond the roots posteriorly as it does in other directions. In B. parvus the crown is shorter and flatter than in B. bruryi and is also procumbent, rising slightly at its anterior end to overlie the posterior end of Iz; it is ovoid in dorsal view (slightly longer an leroposteriorly ) and its posterior end is reduced, The anterior root of P; inserts anterobuccal to ihe posterior root, The posterior alveolus ts closer to the anterior alveolus of Po than it is tò the anterioralveolus of Pi, inserting slightly lingually and anterior to the anterior alveolus of P3. The septum separating the posterior alveolus of Py and the anterior alveolus of P: frequently breaks down so that they form a single cavity. In some specimens, therefore, there may appear to be only three alveoli in the region which had heen occu- pied by the 4 roots. of Pı and Pa. Even with the septum intact, the arrangement of alveoli might suggest that the posterior alveolus of P; and the anterior alveolus of Po belonged to the same tooth. Whereas in 8. braryi the alveoli of Pi and Pz are closely but unevenly spaced, in B. parvus the 5 ulveoli of Is, Py and Po are evenly spaced and in the adult animal there is a small gap between P; and Ps (in subadull or younger ani- mals the teeth are closer together). Pa is similar in shape bul a little larger than Pi. The slight anteroposterior crest lies at an angle (lingual posteriorly) across the alveolar margins, directly above an imaginary line joining the cen- tres of the Pi alveoli. Posteriorly the crown ex- tends beyond and rises above the rool 256 MEMOIRS OF THE QUEENSLAND MUSEUM FIG. 7. Lower cheekteeth of B, bruryi in occlusal view. A-C, QMF 30102. A, LP3. B-B’, LM). C-C’, LM2. D-D’ RM3 of QMF30100. E- E’, RM4 of QMF30176 B-E stereopairs. Scale = | mm. lerminating in a small cuspule and abutting P3. Anteriorly, the crown extends slightly beyond the root. Lingually and buccally the crown swells out and falls away to a rounded point on each side. The buccal, ventral apex is slightly higher and more anteriorly located than the lingual apex, so FIG. 8. Left upper cheekteeth of B. brutyi par: type QMF30091 in, occlusal view. A, P3, B-B“, M! CIC’ M?. D- D’, M34. B-D stereopairs. Scale = Imm. that the crown is somewhat twisted. In B, parvus P2 is larger and relatively longer, with a crown that extends further beyond the roots, particularly anteriorly, giving the anterior end of the tooth a shelf-like appearance in lateral view. The crest is Jess clearly defined than in B. brutyi and approx- imately parallel to the axis of the Ih-P2 interval. NEW OLIGOCENE-MIOCENE BURRAM YS FROM RIVERSLEIGH The crown of P2 shows less lingual-buccal asym- metry than in B. bruryi. The posterior end of the crown rises higher and more steeply than in B. brutyi with a distinct hump above the posterior root of the tooth, posterior to which the crown increases only slightly in height. The Pz of B. triradiatus is similar to, but larger than, that of B. brutyi. It is wider but shorter than P2 in B. parvus and almost circular in dorsal view. Although it protrudes beyond the roots in all directions, it is flatter than in B. brutyi and B. parvus, As with B. brutyi, the buccal side is displaced ahead of the lingual side and as with B. parvus, in lateral view the crown has an anterior ‘lip’. The anteroposter- ior crest is poorly developed. A Pa (NMV P180016) assigned to B. triradiatus by Turnbull et al. (1987) is considerably larger than and dif- ferent to Pz in the Holotype. It is 1-rooted, in contrast to P2 in the Holotype, which has 2 or 3 roots. NMV P180016 could possibly be a B. triradiatus P?. P2 is not known from B. wakefieldi but appears to have been |-rooted. The plagiaulacoid crown of P3 is longer and taller in B. bruryi than B. wakefieldi, larger in B. parvus and larger again in B. triradiatus. P3 of B. brutyi has 5 or 6 dorsal cuspules and associated ridges. The anterior edge of P3 rises vertically in B. brutyi, curving back dorsally to an almost horizontal serrated crest, The anterior profile is straight in B. wakefieldi, but leans backwards slightly as it rises to an also horizontal crest. The anterior root descends from the crown more an- teriorly and buccally in B. wakefieldi than in B. brutyi. In B. triradiatus and B. parvus, the ante- rior profile of P3 curves forward then backward as it rises, giving the corrugated tooth a ‘fanned’ appearance and increasing the length of the dorsal edge. In B. triradiatus the anterior root curves forward slightly as it rises, with its convex profile continued by the crown. In B. parvus the root rises vertically to the base of the crown, then the crown expands gently forward. The P3 blade is slightly concave lingually and convex buccally. The ex- posed portion of the anterior root of P3 protrudes further beyond the jaw margin buccally in B. bruryt than in B. parvus. It is also in high relief in B. wakefieldi and B. triradiatus. In B. parvus, the posterior end of the crest has shifted lingually and backwards (relative to its position in B. brutyi). Thus the anterior angle between the long axis of the P3 crest and the molar row is greater in B. parvus than B. brutyi, as is the angle between this crest and its underlying roots. The posterior root of P3 is also smaller buccally in B. parvus than in B. brutyi and is smaller again in B. wakefieldi because the posterior end of the crest and hence the direction of the bite force in that region has shifted lingually, The anterior end of P3 is more attenuated in B. parvus than in the other species. Some specimens of B. brutyi have cracks running from the dorsal cutting edge basally and back- wards, stopping near the base of the crown. P3s of each of the other species have similar cracks. They are particularly frequent and extensive in B. triradiatus. The P3s of B. triradiatus also gener- ally show more wear on the anterior end of the dorsal cutting edge than is evident in the other species. Lower molars are bunodont in Burramys. They differ mainly in size, Mı cusp morphology and degree of reduction of M4. Some unworn molars of B. brutyi are slightly crenulate, unlike other species of Burramys, but since crenulation is rare in B. brutyi and since molars of the other fossil species are poorly known, this feature is not re- garded as diagnostic. The molar gradient is greater in B. parvus than in other species. M; is approximately the same size in B. brutyi and B. wakefieldi and is larger in B. parvus. It has two roots in each of these species. Mı is not known from B, triradiatus but judging from its alveoli was 3-rooted and relatively small, with Ma< Mi< M3< Mbp. The trigonid rises more stee- ply against P3 in B. bruryi and B. wakefieldi than in B. parvus, with the protoconid taller in com- parison to the metaconid. P3 and M; are therefore more disparate in height in B. parvus than in B. brutyi or B. wakefieldi. Posteriorly, the crown extends further beyond the roots in B. parvus than in the other species. In B. wakefieldi the entoconid is particularly tall. In all species, the Mı postmetacristid is continuous with the longitudi- nal axis of the dorsal crest of P3. In B. brutyi and B. parvus the premetacristid swings buccally to meet the postmetacristid, creating a disjunction between the P3 crest and the lingual crests of M1. The postprotocristid/premetacristid angle is more obtuse at the metaconid in B. brutyi than B. parvus because the metaconid is more posteriorly positioned in the former than the latter. The break in the P3-M, blade system is therefore, longer in B. brutyi than in B. parvus. In B. wakefieldi the protoconid is more lingually positioned so that the crests associated with the P3 and M; pro- toconid, metaconid and entoconid form an almost straight line. M? is smaller in B. brutyi than B, triradiatus or B. parvus. Ma of the latter is slightly longer and narrower than that of B. triradiatus. It is propor- tionately shorter in B. bruryi than B. parvus and 258 MEMOIRS OF THE QUEENSLAND MUSEUM FIG. 9. B. brutyi sp. nov. A-C, left P1-2 with Ip alveolus and anterobasal portion of Pa, holotype QMF30102 in (A-A’) lingual, (B-B*) occlusal and (C-C’) buccal views. D-D’, left P7? and anterior portion of M!, paratype QMF30091 in lingual view. A-D stereopairs. Scale = I mm. NEW OLIGOCENE-MIOCENE BURRAMYS FROM RIVERSLEIGH TABLE 4. Measurements of B. bruryi types. From holotype where possible. M3-4 lengths and widths, Mı-3 and M1-4 from paratype QMF301 76, All upper tooth measurements from paratype QMF30176. Ab- breviations as for Table 1. very slightly shorter than B. triradiatus. M2 is not known for B. wakefieldi. It has two roots in each species except for B. triradiatus, in which it has three. In a few (<5%) of B. brutyi specimens the anterior alveolus has, ventrally, a septum (or re- mainder thereof) subdividing it basally into 2 compartments, suggesting a root bifurcated at its tip. This condition may be intermediate between the 2- and 3-rooted conditions. In B. brutyi and B. parvus there is frequently a small cuspid halfway along the lingual margin of the crown, at the junction of the postmetacristid and the pre- entocristid. Sometimes the cuspid is not clearly differentiated from the postprotocristid. It is the same size in both species even though the tooth is larger in B. parvus. The cuspid is not evident in B. triradiatus, although there is a small dorsal protuberance on the anterior end of the pre- entocristid of NMV P158628. In each species the postprotocristid curves lingually from the pro- toconid before straightening and running approx- imately parallel to the tooth axis until interrupted by the transverse hypoconid-entoconid lophid. Postprotocristid curvature is less extreme in B. 259 brutyi than the other species. The cristid obliqua lies parallel to the tooth axis, forming a posterobuccal cingular pocket between itself and the postprotocristid. In all species the buccal cusps are bulbous. The hypoconid causes the posterobuccal corner of the tooth to extend be- yond its basically rectangular outline. The lingual cusps are slightly ahead of the buccal cusps, skewing the sides of the tooth slightly. They are more crescentic than the buccal cusps and, to- gether with their associated crests, form a blade- like structure. M3is similar to, but smaller than, M2. Cusps are lower and basins shallower, with the crown sur- face showing more wear than M2. M3 is smaller in B. brutyi than the other species, is slightly larger in B. triradiatus than B. parvus and is not known from B, wakefieldi. Interspecific compar- isons of M3 are as for M2 except that in B. brutyr and B. parvus, but not B. triradiatus, the lingual- buccal skew is slightly more pronounced than in Mo. In all species M3 is slightly shorter an- teroposteriorly than Mo. In B. triradiatus the pro- toconid and hypoconid of M3 are subequal whereas in M2 the hypoconid is larger. In M3 there is a distinct cleft dividing the rounded pro- toconid and hypoconid. M; in B. brutyi and B. wakefieldi has 2 roots, whereas in B. triradiatus it has 3 and in B. parvus 1, While most specimens of B. brutyi have 2 roots or alveoli for M4, some have 3 and a few had | root. Such variation is not evident in the B. triradiatus or B. parvus. M4 is not known for B. wakefieldi or B. triradiatus but the alveoli of B. triradiatus suggest that it was far less reduced than in B. parvus and possibly less reduced than in B. brutyi. M4 is low-crowned, with low cusps which quickly wear down. It is smallest and most degenerate in B. parvus. Upper teeth of Burramys anterior to P* have not been recognised from Riversleigh or Hamilton, so discussion of the upper dentition will be lim- ited to P?3 and M!*, Skull fragments and upper teeth of B. wakefieldi are unknown. The upper dentition of B. triradiatus is known only from isolated teeth. Maxilla. Palatal vacuities are smaller in B. brutyi than B. parvus. The anteroventral opening of the infraorbital foramen is also smaller (and less round) in B. brutyi, as are foramina in the al- isphenoid and squamosal. Known bones of the skull are more robust in B. brutyi than in B. parvus. In both species the maxilla is swollen around the P? alveolus, between the lachrymal 260) and the infraorbital foramen. This swelling is more extensive in B. parvus than B. brutyi, with P? and the anterior limit of the molar row begin- ning further forward in the living species. In ventral view, the anteromedial limit of the zygo- matic arch in 8. brutyi is level with a point mid- way between the prolocone and protoconule of M! In B. parvus it is midway between the mmelaconule and protocune of ML The upper molar gradient is steeper in B. parvis than in the other species. In &, brary the molar row rotates buecally around the muailla from front to back, to a greater degree than occurs in B. parvics. Upper dentition. P? of B, brary? is 2-rooted and simular to, although slightly larger than, P2. A weak crest runs froma small cuspule at the high- est point on the crown, which is midway along the raised posterior edge, to the anterior base of the crown. Lingually and buceally the crown slopes towards the roots. The base of the crown expands lingually over the pasterior root, extend- ing the crown oulline posterolingually. This swelling is less pronounced in B. parvas, In crown view the tooth is teardrop-shaped, being just wider than the transverse diameter of the antenor root, In B parvus, by contrast, the 2- rooted P? crown expands beyond the roots for its whole length (more so posteriorly than anteri- orly), In both species the crown is parallel to the edge ol the medially inclined palute, lorming an angle with P? and the molar row, Although P= for B. rriradiarits has not been identilied, the small, single-rooted tooth (NMY P180016), determined by Turnbull et al. (1987) to be a Po, is similar to Pes of B. brutyi and B. parvus and is interpreted here to be P?, lu B. beutvi, asin B. triradiatus and B. parvus, P3 is similar to Py, Regarding P? anteroposterior length, B. bruryi -P> interval Length of li Basal thickening of Number of roots P4.2 Arrangement of alveoli Py_> Size of P3 Size disparity between P3 roots Number of ridges P3 Curvature of P3 anterior profile 2 Concave/ convex P3 Arched dorsal edge P3 Divergence of P3 from molar row Transverse compression P3 Distinct M; talonid and trigonid Mı protoconid position Mı metaconid position Relative length lower molars Neomorphic cuspid Loph(id) developmentM>.3, M? No. roots My_3 No, roots Mg Reduction of M4 Size of maxill Antenor limit Rotation of upper molar row Inflation of lingual cusps M! Lingual displacement of M! paracone vacuities SCOHHPOHSOOPENOSSE HOODOO HHH OF OHH than in 8B. brutyi or B. rriradiatus, with B. triradiatus the shortest. This region is incomplete in the holotype of B. wakefield but appears to be about the same length as in B. bruni. This interval is relatively long in Cercartetus and phalangerids, indicating thatthis is the plesiomorphie state, 4, li length. Ti is longerin B. parvus than B. brutyi and longer again in B, ¢riradiatus (unknown in B. wakefieldi), Wis shorter in Cercartetus than Burramys and is shorter in phalangerids. A long ly 1s regarded as apomorphic, 5. Thickened base of i1, In B. brutyi and B. triracdianes Ij is thick basally (thicker in B. rriradiatus) and im- mediately begins to taper; approximately half way along the exposed portion of the tooth itthins markedly then attenuates to the tip. In B, parvus 1 tapers grid- ually without marked reduction al a particular point. In Cereartetus. Trichosurus, Spilocuseus and Phalanger l} does not change suddenly im diameter, suggesting that a basally thickened I is apomorphic, Shape of Py. Py ts not known for B. wukefieldi or B. B. brutyt B. wakefieldi B. triradiatus B. parvus 0 1 1 ? ? A ? 1 3 1 0 0 0 3 1 0 1 0 2 ? ? 0 0 1 ? ? ? ? ? covnvote CCA N VAN ENP ee WN Qe NN FRPOCORNGOPHOHORNN NH EN NH OOOHOO triradiatus. In B. brutyiit is small, rounded and similar to Pa. In B. parvus Pi is intermediate between the cap-like P2 and the slightly elongate, procumbent 12- In C. caudatus and C. lepidus P| and P2 are both button-like and upright: in C. nanus and C. concinnus Pi resembles 12 almost as much as P2, Trichosurus and Phalanger species have extensive diastemata, lacking Pi and Pz analogous to those of burramyids. Il is therefore unclear which state of Pi is more plesiomorphie and although this character may be phy- Jogeneucally significant, a satisfactory polarity assign- ment cannot be made. 6. Number of roots P| and P2. Burramys bruni and B. parvus have double-rooted P] and P2. Burramys triradiarus has a triple- rooted P2 and double- or triple- rooted Pj: the number of roots is not clear due to damage inthe available material. Py and P? each appear lo have been single-rooted in B, wakefield’. Py-P2 of Cercartetus possess sometimes one und sometimes two roots, Pj-2 of Trichosurus and Phalanger are eilher extremely reduced or ubsent. Oulgroup analysis does not resolve the polarity of this character. The normal marsupial premolar condition is two-rooted so this is NEW OLIGOCENE-MIOCENE BURRAMYS FROM RIVERSLEIGH 263 taken to be the plesiomorphic condition. Burramys wakefieldi and B, triradiatus are interpreted as having alternative derived states. 7. Arrangement of P}-2 alveoli. In B. parvus the alveoli of P1-2 are in a straight line between I2 and P3. In B. brutyi the anterior alveolus of P2 is lingual to its pos- terior alveolus and the posterior alveolus of P1 is lin- gual to its anterior alveolus. B. wakefieldi and B. triradiatus have different numbers of roots for P1-2 from B. brutyi and B. parvus, so their alveoli are not all homologous. In all species of Cercaretus the alveoli of Pj-2 lie in a straight line; this is also the case for Trichosurus, Spilocuscus and Phalanger (where the teeth occur). Linearly-arranged alveoli are therefore thought to be plesiomorphic for burramyids. 8. Size of plagiaulacoid premolar, The sectorial premo- lar of Cercartetus and phalangerids (and Mj of C. concinnus) is smaller than that of Burramys. It is there- fore assumed that an enlarged plagiaulacoid premolar is synapomorphic for Burramys and apomorphic within the genus. Although P3 of B. parvus is larger than that of B. wakefieldi or B. brutyi, log-scaled plots of P3 buccal crown surface area against jaw length (unpubl. data) suggest that P3 of B. parvus is not disproportionately large for its body size. P3 of B. triradiatus, on the other hand, departs significantly from the line of best fit for P3 size against body size, being disproportionately large. P3 of B. wakefieldi falls below the line, suggesting that it is disproportionately small, but Studentized residuals do not show its depar- ture from the line to be significant. 9. Relative sizes of anterior and posterior roots of plagiaulacoid premolar. Buccally, the posterior root of P3 is smaller, relative to its anterior root and crown, in B. wakefieldi than in other Burramys. The posterior root of P3 is smaller (relative to the anterior root and the crown) in B. parvus than in B. brutyi. The anterior root of the large P3 of B. triradiatus is massive; al- though the posterior root is comparatively small, the disparity is not as great as that in B, wakefieldi. In Cercartetus and phalangerids, the anterior and poste- rior roots of the sectorial premolar are subequal; this is thought to be the plesiomorphic condition. 10. Number of ridges on plagiaulacoid premolar, In B, brutyi and B. wakefieldi there are 5 or 6 ridges on each of the buccal and lingual faces of P3 and 5 or 6 associated dorsal cuspules. The lack of posterior and weakness of anterior cuspules in the holotype of B. wake/ieldi appears to be the result of extreme wear on the formerly serrated tooth.In B. parvus there are com- monly 7 ridges and cuspules and in B. rriradiarus, 9. Phalangerids with smaller, unridged P3s are thought to be more plesiomorphic than those with larger, ridged P3s (Flannery et al.,1987); all have fewer ridges and cuspules than Burramys. Cercartetus nanus and C. caudatus have a single sharp dorsal cusp on the secto- rial premolar and C. concinnus one main cusp on its premolariform Mı. A larger number of ridges and cuspules is regarded as more derived within Burramys and a synapomorphy of the genus. 11. Curvature of anterior profile of P3. In lateral view, P3 of B. wakefieldi and B. brutyi has a relatively straight (approximately vertical) anterior profile. In B. triradiatus and B. parvus the crown expands anteriorly to produce a ‘curved profile, The sectorial P3 of Cercartetus does not curve forward anteriorly (al- though the autapomorphic premolariform M1 of C. concinnus does). Anterior curvature may be associated with increased P3 size, with enlargement having been achieved by anterior extension of the crown. However P3s of T. caninus and T. vulpecula, which are curved, are smaller than those of Spilocuscus and Phalanger, which are less curved. Size and curvature are therefore not necessarily linked. It is possible that Miocene P3s represent primary enlargement of the tooth without the functional elaboration of other species, in which the inflated anterior edge may disperse stress, increase occlusal area, or perform some other function. A curved anterior profile is regarded as apomorphic within Burramyidae. 12. Lingual concavity/buccal convexity of P3. The P3 blade of Burramys is concave lingually and convex buccally (particularly anteriorly). The contrast between lingual and buccal curvature is least pronounced in B. wakefieldi and B. brutyi and more pronounced in B. parvus and slightly more in B. rriradiatus, As with anterior profile curvature, this feature occurs in Trichosurus but not in Cercartetus, Spilocuscus or Phalanger. It is regarded as apomorphic. 13. Arching of dorsal edge of P3. The dorsal edge of P3 is arched in B. parvus; in the other species it is straight, but in B. triradiatus there is a slight curvature at the anterior end of the blade. The sectorial teeth of Cercartetus do not have a dorsal blade edge homolo- gous with that of Burramys and so do not provide a useful comparison. The dorsal edge of P3 is straight in phalangerids and this is assumed to be the plesio- morphic condition, 14. Divergence of P3 from anteroposterior axis of molar row, In Burramys, the longitudinal axis of P3 departs from the ramus such that it forms an angle with the anteroposterior molar row axis. This angle is largest in B. wakefieldi and is larger in B. parvus and B. triradiatus than in B. brutyi. In Cercartetus the longi- tudinal axis of the lower sectorial tooth is parallel to the anteroposterior axis of the molar row and within phalangerines, a more oblique placement of P3 is re- garded as apomorphic (Flannery et al., 1987). Diver- gence of P3 from the anteroposterior axis of the molar row is asynapomorphy of Burramys; within Burramys, the plesiomorphic condition is taken to be a less diver- gent P3. 15. Transverse apical compression of P3. In anterior view, the crown of P3 of Burramys tapers from the base, attenuating dorsally then terminating apically with a serrated longitudinal median ridge. This trans- verse apical compression is least pronounced in the Miocene species and most pronounced in B. triradiatus. Crowns of the sectorial premolars of Cercartetus, Trichosurus, Spilocuscus and Phalanger are less attenuated than those of Burramys. Increased dorsal transverse compression is synapomorphic for Burramys. Laterally compressed P3s are regarded as more derived than those with thicker apices. 16. Distinction of talonid and trigonid of M1. In B. wakefieldi the talonid and trigonid of Mı are clearly demarcated in occlusal view by lingual and buccal indentations. In B. brutyi and B. parvus the talonid and trigonid are less distinct. Mı is not known for B. triradiatus. Talonids and trigonids are more distinct in Cercartetus than in Burramys, indicating the plesiomorphic state. The fused talonid and trigonid departs further from primitive tribosphenic morphol- ogy. Alternatively, the structure of M1 in B. wakefieldi could be autapomorphic, with the crests defining the talonid and trigonid functioning primarily as buttresses for the anterolingual crests which may, in this animal, have extended the function of P3. However, the former hypothesis is preferred, 17. Lingual displacement of protoconid of Mj. The protoconid of Mj is displaced further lingually in B. wakefieldi than in B. brutyi or B. parvus so that in B. wakefieldi the crests associated with P3 and the M1 protoconid, metaconid and entoconid form an almost straight line. The position of the protoconid is variable in Cercartetus and phalangerids. In the primitive tribosphenic molar, the protoconid is a buccal cusp, so lingual displacement is regarded as apomorphic. 18. Anterior displacement of metaconid of M1. The paraconid is absent in Burramys and the most anterior lingual cusp is the metaconid. In B. parvus the metaconid is more anterior than in B, wakefieldi or B. brutyi, narrowing the gap in the P3-M1 crest. In the Phalangeridae and Cercartetus position of the metaconid relative to the protoconid is variable. Out- group analysis does not resolve the polarity of this character. The metaconid of B. parvus occupies the position that in a plesiomorphic (tribosphenic) molar would have supported the paraconid, so the anteriorly displaced metaconid is regarded as apomorphic. Inclination of M] trigonid against P3. The trigonid of Mi] rises more steeply against the posterior face of P3 in B. brutyi and B. wakefieldi than in B. parvus. Neither Cercartetus nor phalangerids give a clear indication of the polarity of this character. It has developed a number of times in phalangerids and pilkipildrids and is prob- ably homoplasious, 19. Relative length of lower molars. M2-4 of Burramys differ in their lengths (relative to widths) such that B. brutyi 2 [14] 11| 0—1 [16] 13] O—1 [17] 0—1 12| O—1 15| 1—2 [18] 0—1 19| 2—1 REREEREE SF ees] FIG. 10. A phylogenetic hypothesis of intrageneric relationships of Burramys. Apomorphies listed at nodes; character numbers in boxes refer to Table 5. Character state transformations indicated by arrows. tion generated a single most parsimonious tree. The topology of this tree is identical for both algorithms. Burramys parvus and B. triadiatus form a clade to which B. wakefieldi is the plesiomorphic sister group; B. brutyi is the plesiomorphic sister group to a clade containing MEMOIRS OF THE QUEENSLAND MUSEUM all other species of Burramys. For some charac- ters, the path of transformation differs depending upon whether transformation is accelerated or delayed. There are several convergent character states in the DELTRAN tree, no convergences and more reversals in the ACCTRAN tree. When transformation is delayed the following character states arise convergently in B. brutyi and B. parvus: loph(id)s of M2-3 reduce; neomorphic cuspid appears on M2-3; and M; talonid and trigonid become less distinct. Basal thickening of I; occurs independently in B. brutyi and B. triradiatus. The relative length of the lower mo- lars decreases independently in B. brutyi and B. wakefieldi. With delayed transformation B. parvus reverses to a more plesiomorphic state of reduced robusticity and relatively long molars, and the relative size of M4 in B. triradiatus in- creases secondarily. These reversals also occur when transformation is accelerated, as do the following: in B. parvus and B. triradiatus the relative length of lower molars increases (to a greater degree in B. parvus); in B. parvus the h-P2 interval increases; in B. triradiatus loph(id)s de- velop on M2-3 and the neomorphic cuspid disap- pears from M2.3; and in B. wakefieldi the talonid and trigonid of M; are relatively distinct from one another. Although a single most parsimonious tree was generated by this analysis, another tree only one step longer placed B. wakefieldi as the plesiomorphic sister-group of the other three spe- cies, and B. brutyi as the plesiomorphic sister- group of the B. triradiatus + B. parvus clade. A bootstrap analysis using a branch and bound search with 100 repetitions, to place confidence estimates on clades (from ACCTRAN) found the node defining the B. triradiatus + B. parvus clade to be supported 84% of the time, but the node separating B. bruryi and B. wakefieldi occurred in less than 50% of repetitions. Using DELTRAN the B. triradiatus + B. parvus clade was supported 78% of the time, and the node separating the other 3 species from B. brutyi was supported by 55% of repetitions. In both cases, the node separating B. brutyi and B. wakefieldi is poorly resolved. DISCUSSION Burramys brutyi is the only species of Burramys at Riversleigh and is not known else- where. It is represented by >150 specimens from 23 Sites in Systems A, B and C; it is one of the most widely distributed (spatially and tempo- rally) marsupials at Riversleigh. Its earliest oc- NEW OLIGOCENE-MIOCENE BURRAMYS FROM RIVERSLEIGH currence at late Oligocene (Myers & Archer, 1997) White Hunter Site is of similar age to the type locality of B. wakefieldi on Mammelon Hill, Lake Palankarinna, South Australia (Woodburne et al., 1993). Metric analyses did not reveal any significant size variation between sites; variation within sites being as great as between sites. This persistence in unchanged form from the late Oligocene through much of the Miocene suggests an un- usual degree of ecological stasis for the species. Fossil Burramys in Victoria, South Australia and NW Queensland shows that small existing populations of B. parvus are remnants of a pre- viously more diverse and far more widespread lineage, now apparently in decline. This fact urges particular conservation concern for the ex- tant species, Although populations of B. parvus are apparently stable, they are threatened both by habitat disturbance and greenhouse warming, which could jeopardise their ability to survive (Geiser & Broome 1993), ACKNOWLEDGEMENTS Comparative material was made available by J. Dixon, L. Frigo, T. Rich and E. Thompson, Mu- seum of Victoria; N. Pledge, South Australia Museum; and J. Wombey, C.S.I.R.O., Gunghalin. The vital support of the following organisations is also gratefully acknowledged: the Australian Research Council (grants to M. Archer); the National Estate Grants Scheme Queensland (to A. Bartholomai and M. Archer); the Department of Environment, Sports and Ter- ritories; the Queensland National Parks and Wildlife Service; the University of New South Wales; IBM Australia Pty Ltd; ICI Australia Pty Ltd; the Australian Geographic Society; Wang Australia Pty Ltd; the Queensland Museum; the Australian Museum; Mt Isa Mines Pty Ltd; Sur- rey Beatty & Sons Pty Ltd; the Riversleigh Soci- ety Inc.; the Royal Zoological Society of New South Wales; the Linnean Society of New South Wales; and many private supporters. LITERATURE CITED ANON. 1966. A relict marsupial. Nature 212: 225. APLIN, K. & ARCHER, M. 1987. Recent advances in marsupial systematics with a new syncretic clas- sification. Pp. xv-Ixxii. In Archer, M. (ed.), Pos- sums and Opossums: Studies in Evolution. (Surrey Beatty & Sons and Royal Zoological So- ciety of New South Wales: Sydney). ARCHER, M. 1984. The Australian marsupial radia- 267 tion. Pp. 633-808 In Archer, M. & Clayton, G. (eds), Vertebrate Zoogeography and Evolution in Australasia. (Hesperian Press: Sydney). ARCHER, M., GODTHELP, H., HAND, S.J. & MEGIRIAN, D. 1989. Fossil mammals of Riversleigh, northwestern Queensland: prelimi- nary overview of biostratigraphy, correlation and environmental change. Australian Zoologist 25: 29-65. ARCHER, M., HAND, S.J. & GODTHELP, H. 1991. Riversleigh. 2nd ed. (Reed: Sydney). BAVERSTOCK, P.R., BIRRELL, J. & KRIEG, M. 1987. Albumin immunologic relationships among Australian possums: A progress report. Pp. 229- 234. In Archer, M. (ed.), Possums and Opossums: studies in evolution. (Surrey Beatty & Sons and the Royal Zoological Society of New South Wales: Sydney). BROOM, R. 1896. Report on a bone breccia deposit near the Wombeyan Caves, N.S.W.: with descrip- tions of some new species of marsupials. Proceed- ings of the Linnean Society of New South Wales 81: 48-61. CREASER, P. 1997. Oligocene-Miocene sediments of Riversleigh: the potential significance of topogra- phy. Memoirs of the Queensland Museum 41: 303-314, FLANNERY, T., ARCHER, M. & MAYNES, G. 1987. The phylogenetic relationships of living phalangerids (Phalangeroidea: Marsupialia) with a suggested new taxonomy. Pp. 477-506. In Archer, M. (ed.), Possums and Opossums: studies in evolution. (Surrey Beatty & Sons and the Royal Zoological Society of New South Wales: Sydney). FLEMING, M.R. 1985. The thermal physiology of the Mountain Pygmy-possum Burramys parvus (Marsupialia: Burramyidae). Australian Mam- malogy 8: 79-90. FLOWER, W.H. 1867. On the development and suc- cession of teeth in the Marsupialia. Philosophical Transactions of the Royal Society, London 157: 631-641. GEISER, F. & BROOME, L.S. 1993. The effect of temperature on the pattern of torpor in a marsupial hibernator. Journal of Comparative Physiology B 163: 133-137. GEISER, F., SINK, H.S., STAHL, B., MANSERGH, I.M. & BROOME, L.S. 1990. Differences in the physiological response to cold in wild and labora- tory-bred Mountain Pygmy-Possums, Burramys parvus (Marsupialia). Australian Wildlife Re- search 17: 535-539, GINGERICH, P.D. 1974. Size variability of the teeth in living mammals and the diagnosis of closely re- lated sympatric fossil species. Journal of Paleon- tology 48: 895-903. LUCKETT, P. 1993. An ontogenetic assessment in dental homologies in the therian mammals. Pp. 182-204. In Szalay, F.S., Novacek, M.J. & Mc- Kenna, M.C. (eds), Mammal phylogeny: Mesozoic differentiation, multituburculates, monotremes, early Thertins, and Marsupials’. (Springer-Verlag: New York). MANSERGH, I.. KELLY, P, & SCOTTS, D. 1987. Draft management strategy and guidelines for the conservation of the Mountain Pygmy-possum, Burramys parvus, in Victoria, Arthur Rylah Insti- tute of Environmental Research Technical Report Series No. 53. MANSERGH, |, & SCOTTS, DJ. 1989. Habital conli- nuity and social orgamsation of the Mountain Pygmy-possum restored by tunnel. Journal of Wildlife Management 53: 701-707. 1990. Aspects of the life history and breeding biol- ogy of the Mountain Pygmy-possum, Burramys parvus, (Marsupialia. Burramyidae) in alpine Victora, Australian Mammalogy 13; 179- 191. MAURER, B.A., BROWN, J.H. & RUSLER, R.D. 1992. The micro and macro in body size evolution. Evolution 46; 939-953. MUIRHEAD, J. 1994. Systematics, evolution and pal- acobiology of Recent and fossil bandicoots (Per- amelemorphia, Marsupialia). Unpublished PhD thesis, University of New South Wales, N.S.W. MYERS, T.J. & ARCHER, A. 1997. Kuterintia ngama (Marsupialia, lariidae); a revised and extended systematic analysis based on material from the late Oligocene of Riversleigh, northwestem Queens- land. Memoirs: of the Queensland Museum 41: 379-392. PLEDGE, N. 1987. A new species of Burramys Broom (Marsupialia: Burramyidae) from the middle Miocene of South Australia, Pp, 725-728, In Archer, M. (ed.). Possums and Opossums: studies in evolution. (Surrey Beatty & Sons and the Royal Zoological Society of New South Wales: Sydney), RIDE, W.D.L. 1956. The affinities of Burramys parvus Broom a fossil phalangeroid marsupial. Proceed- ings of the Zoological Society of London 127; 413-429, ROTH, V.L. 1981. Constancy in the size ratios of sympatric species. American Naturalist 118: 394- 404 SIMPSON, G.G. 1933. The ‘plagiaulacoid’ type of mammalian dentition, a study of convergence. Journal of Mammalogy 14: 97- 107- SIMPSON, G.G., ROE, A. & LEWONTIN, R.C. 1960. MEMOIRS OF THE QUEENSLAND MUSEUM Quantitative Zoology. (Hareourt, Brace & Co: New York). SPRINGER, M. & KIRSCH, J.A.W. 1989, Rates. of single-copy DNA evolution in phalangeriform marsupials. Molecular Biology and Evolution 6; 331-34], STIRRAT, S. 198]. The functional morphology of the plagiaulacoid premolar af Burramys parvis Broom. Unpublished Honours thesis, Monash University, Clayton, Victoria, TURNBULL, W.D. & LUNDELIUS, E.L. Jr 1970. The Hamilton Fauna, alate Pliocene mammalian fauna from the Grange Bum, Victoria, Australia, Fieldi- ana: Geology 19: 1-163. TURNBULL, W.D. & SCHRAM, F.R. 1972. Broom Cave Cercartetus: With observations on Pygmy Possum dental morphology, variation and tàxon- omy. Records of the Australian Museum 28: 437- 464. TURNBULL, W.D., RICH, T.H.V..& LUNDELIUS. E.L. Jr 1987, Burramyids (Marsupialia: Burramyidae) of the early Pliocene Hamilton Loval Fauna, southwestern Victoria, Pp. 729-739. In Archer, M. (ed.) ‘Possums and Opossums: stud- ies in evolution, (Surrey Beatty & Sons and the Royal Zoological Society of New South Wales: Sydney). WAKEFIELD, N.A., 1960. Recent mammal bones in the Buchan District, Victorian Naturalist 77: L64- 198. WARNEKE, R.M. 1967. Discovery of a living Burramys. Australian Mammal Society Bulletin 2: 94-95, WILEY E.O., SIEGEL-CAUSEY, D,, BROOKS, D.R. & FUNK, V.A. 199]. The compleat cladist: a rimer of phylogenetic procedures. University of ansas Museum Natural History Special Publica- tion: 19. WOODBURNE, M.O., MACFADDEN, B.J., CASE, J.A, SPRINGER, M\S,, PLEDGE, N.S., POWER, J.D., WOODBURNE, J.M. & SPRINGER, K.B. 1993. Land mammal biostratigraphy and magnetostratigraphy of the Etadunna Formation (late Oligocene) of South Australia. Journal of Vertebrate Paleontology 13; 487-515, TWO NEW BALBARINE KANGAROOS AND LOWER MOLAR EVOLUTION WITHIN THE SUBFAMILY B.N, COOKE Cooke, B.N. 1997 06 30: Two new balbarine kangaroos and lower molar evolution within the subfamily. Memoirs of the Queensland Museum 41(2):269-280. Brisbane. ISSN 0079- 8835. Lower Jaws and teeth of Nambaroo couperi sp. nov. and Wururoo dayamayi gen. et sp. nov., fossil balbarine kangaroos from the late Oligocene White Hunter Site of Riversleigh, are described. Mj trigonid cuspid homology in Hypsiprymnodon is re-interpreted such that a reduced protoconid is recognised, the anterobuccal cuspid is regarded as the protostylid and the anterolingual cuspid as the metaconid. The evolution of lophodont lower molars within Balbarinae is examined on the bases of this interpretation and information supplied by the new species. [_] Riversleigh, kangaroo, Balbarinae, Nambaroo, Wururoo, cuspid homology, lophodonty. B.N. Cooke, School of Life Science, Queensland University of Technology, GPO Box 2434, Brisbane, Queensland 4001, Australia; 18 December 1996. The Balbarinae Flannery, et al., 1982 was erected for a group of fossil macropodids in which the M; protolophid is markedly com- pressed. It assumed phylogenetic significance when Flannery (1989) proposed that balbarines were ancestral to both sthenurines and macropodines. Balbarines form a major component of the fos- sil kangaroo fauna of Riversleigh and appear to have had a wide distribution in the Oligocene- Miocene of Australia. Three genera and 8 species have so far been named, but itis more diverse than this. Balbaroo camfieldensis Flannery et al., 1982 is known from Bullock Creek and Balbaroo sp. Flannery et al. (1982) from the Kangaroo Well Local Fauna, both in the Northern Territory. Three species of Nambaroo Flannery & Rich, 1986 were described from the Tarkarooloo Local Fauna of South Australia. Unnamed balbarines have been reported by Flannery (1989) from the Kutjumarpu Local Fauna of South Australia and Woodburne et al. (1993) from the Etadunna For- mation of South Australia. Riversleigh balbarines include Balbaroo gregoriensis Flannery et al., 1982 and 3 species of Ganawamaya Cooke, 1992. The present paper includes descriptions based on lower jaws and teeth of a new species of Nambaroo and a new genus and species of balbarine. METHODS Molar homology follows Luckett (1993). Pre- molar homology follows Flower (1867). Homol- ogy of molar structures has been determined by reference to a generalised tribosphenic pattern, outlined by Szalay (1969), following Ride (1993). Terminology follows Van Valen (1966), Szalay (1969) and Butler (1990). However, the ‘anterior cingulid’ is restricted in use to that com- ponent of the macropodoid lower molar anterior cingular shelf lying lingual to the anteriorly di- rected paracristid. ‘Precingulid’ is refers to that component lying buccal to the paracristid. Van Valen (1966, 1994) used ‘precingulid’ for the anterior cingular shelf anterobuccal to the paracristid of plesiomorphic mammalian lower molars. The distinction is made here because lingual and buccal components of the macropodoid anterior cingular shelf are demon- strably of different origins. The buccal compo- nent is the more plesiomorphic since it occurs in plesiomorphic balbarines such as Nambaroo couperi sp. nov. More derived balbarines such as Wururoo gen. nov. and Balbaroo demonstrate the development of the neomorphic lingual compo- nent via lingual displacement of the paraconid and lingual extension of the paracristid. Supra- generic classification follows Aplin & Archer (1987). QMF denotes Queensland Museum fossil collection catalogue numbers. Measurements are in millimetres. SYSTEMATICS Family MACROPODIDAE Gray, 1821 Subfamily BALBARINAE Flannery, Archer & Plane, 1982 Nambaroo Flannery & Rich, 1986 Nambaroo couperi sp. nov. (Figs 1, 2, 5A; Table 1) DIAGNOSIS. Nambaroo with a hypoconulid at the posterior, buccal base of the entoconid on Mı and marked convexities along the lateral margins adjacent to the ends of the interlophid valley on all lower molars except M4. MATERIAL. Holotype QMF30401, a partial right dentary consisting of the entire horizontal ramus, most of the angular process and portion of the ascending ramus to the level of the damaged condyle. P3 and M 1-4 are preserved; from White Hunter Site, Hal’s Hill, D Site Plateau, which has been correlated (Myers & Archer, this volume) with the Ngama Local Fauna from the Tirari Desert which Woodburne et al. (1993) have shown to be late Oligocene, about 24 to 26My. ETYMOLOGY. For Patrick Couper, Queensland Mu- seum, for his assistance during the course of this re- search. DESCRIPTION. The holotype is a fragment of a right dentary consisting of the entire horizontal ramus, most of the angular process and portion of the ascending ramus to the level of the damaged condyle (Fig. 1 A, B). Dorsal edge of the diastema delineated by a ridge with matrix-filled alveolus for a very small Iz or analogous tooth at anterior end. Horizontal ramus twisted, with mesial sur- face inclined slightly dorsally below P3 and slightly ventrally below M4. Mandibular sym- physis extending as far posteriorly as the level of the anterior margin of P3. On the buccal surface the anterior mental foramen located below and slightly anterior to P3, with a much smaller pos- terior mental foramen below the hypolophid of M2. Horizontal ramus deepest below M1, with its zone of most ventral protrusion below M3. Ven- tral margin straightest below P3-M3, curving up- wards below the diastema and more steeply so posterior to M3. Buccal margin of the masseteric fossa straight so that the entrance to the masse- teric canal is ‘D’ shaped in cross section. Ventral margin of the masseteric fossa low on the ramus, well below the level of the molar row. Inferior dental canal recessed into the lingual wall of the masseteric canal but not partitioned from it. Be- cause of the confluence of the two canals, forward extent of the penetration of the masseter difficult to determine, but the gradient of anterior canal constriction suggesting insertion no further for- ward than M3. Lingual border of the angular process in the same vertical plane as the lingual margin of the horizontal ramus but the angular process extending more posteriorly than the MEMOIRS OF THE QUEENSLAND MUSEUM ramus. Large portion of the floor of the pterygoid fossa lost. Ascending ramus rising at 100° relative to the plane of the molar row. Condyle situated 9.4mm above the molar row, a transversely elongate structure, broader lingually, tapering to the buc- cal side, obliquely inclined to the plane of the ascending ramus. Dentition. Molar row straight in both occlusal and lateral view. P3 flexed slightly buccally out of alignment with molar row. Occlusal surfaces of anterior molar lophids inclined slightly buccally. Those of more central molars more or less hori- zontal. Hypolophid of Mg inclined lingually. Slight increase in molar size posteriorly. P; gracile, short and blade-like with horizontal occlusal margin. In occlusal view with an ellipti- cal outline with the occlusal crest occupying ap- proximate midline, although curving lingually posteriorly. Five cuspids, of which most posterior largest, occupying the occlusal crest, Lingual and buccal transcristids associated with each cuspid, although those on lingual surface partly obscured by wear. Anterior and posterior margins of crown delineated by vertical cristids. In occlusal view M; with a rounded anterior margin and lateral convexities low on the crown adjacent to the ends of interlophid valley. Pro- tolophid shorter than hypolophid: protoconid po- sitioned on approximate midline of tooth, Protoconid taller than metaconid, buttressed buccally by a protostylid about same height as metaconid, Paracristid running almost directly forward to anterior margin where it meets ante- rior edge of a precingulid which descends steeply to buccal margin of tooth. No anterior cingulid, trigonid basin open anterolingually. Cristid ob- liqua inclined slightly lingually, descending an- terior face of the hypoconid, turning anteriorly to cross interlophid valley and terminating at base of protostylid. Entoconid taller than hypoconid. Occlusal crest of hypolophid forming shallow ‘V’ with lingual arm steeper than buccal. Low point of hypolophid slightly lingual of midline. Preentocristid de- scending to interlophid valley floor from apex of entoconid. Wedge shaped prominence at poste- rior base of entoconid, interpreted here as a hypoconulid because its position corresponding to that occupied by hypoconulid in other marsu- pials and, as in these animals, contacted by the posthypocristid. Short, lingually displaced, diag- onal posthypocristid and hypoconulid forming posterobuccal border of small fossette in poste- TWO NEW BALBARINE KANGAROOS 271 FIG. 1. QMF30401, Holotype of Nambaroo couperi sp. nov. A, buccal view. B, lingual view. C, stereopair of occlusal view. Scales = 10mm. rior face of entoconid. Rounded, lingual border probably representing postentocristid. From base of hypoconulid a hypocingulid extending across about half width of posterior base of hypolophid. M2 larger than Mı, approximately rectangular in outline, but with lateral convexities of crown base adjacent to ends of interlophid valley. Pro- tolophid and hypolophid about equal in length, lingual cuspids taller than buccal cuspids. Occlu- sal margins of both lophids shallowly concave. Paracristid running anterolingually to anterior margin from which enamel has been lost, al- though sufficient remains to indicate that an an- terior cingulid ran between paracristid and lingually positioned premetacristid. Short, stee- ply sloping precingulid extending from anterior end of paracristid to buccal margin of the tooth. Cristid obliqua runing anterolingually across interlophid valley to join posterior base of pro- tolophid at about midline. Preentocristid short, not reaching floor of interlophid valley. Postentocristid runing vertically down posterior, lingual edge of entoconid to meet low and some- what irregular hypocingulid which runs trans- versely across half width of the posterior face of hypolophid. No posthypocristid. M; similar to M2 but metaconid and hypoconid more nearly equal in height, hypocingulid less developed and cristid obliqua interrupted adja- cent to base of protolophid. M; differing from more anterior molars in the following: protolophid longer than hypolophid; no obvious convexities of crown base adjacent to ends of the interlophid valley; some wrinkling of the enamel within trigonid basin; precingulid less obvious; rounded postmetacristid descends pos- terior face of metaconid to floor of interlophid valley; cristid obliqua interrupted before contact- 272 MEMOIRS OF THE QUEENSLAND MUSEUM TABLE 1. Dental parameters for QMF303401, Holotype of Nambaroo couperi sp. nov. Cat. No. Mo Ma w | hw | 7 ace l w | hw | QMF30401 | 5.2 | 29 | 3.9 | 52 | 26 | 33 | 5.4 a 41 | 54 | 38 | 3.6 | ing base of protolophid and ridge of enamel runs bifurcates and contacts both protostylid and pro- transversely across floor of the interlophid valley toconid. Mı trigonid morphology in N. couperi from anterior end of cristid obliqua to buccal is most similar to N. novus in that both have an margin; hypoconid taller than entoconid (which anteriorly directed paracristid, lack a paraconid is slightly damaged); no distinct postentocristid; and havea protostylid which is closely associated hypocingulid in form of rounded, transverse with the protoconid. prominence crossing posterior face of hypolopid. Wururoo gen. nov. DISCUSSION. N. couperi is similar in size to N. saltavus, N. tarrinyeri and N. novus which were TYPE SPECIES. Wururoo dayamayi sp. nov. described by Flannery & Rich (1986) from the Tarkarooloo Local Fauna of South Australia. DIAGNOSIS. Balbarines with a large, trenchant Apart from the plesiomorphic hypoconulid, N. P3 and a posterobuccally inclined enamel ridge couperiis in other aspects of talonid morphology, (the ‘protostylid crest’) descending from the apex more derived than N. saltavus. It lacks a of the protoconid of Mı. postentocristid which is present in N. saltavus, sat E : : : ETYMOLOGY. Gulf coast aboriginal wuru, a long and hasa well-developed hypocingulid crossing time ago (Breen, 1981), and roo a common Australian buccally from the hypoconulid on the posterior 4. °°. ig, ' ; base of the hypolophid, present also in N. tar- Pitninaalve: ton Mareen Paneer rinyeri and N. novus but undeveloped in N. saltavus. The Mı cristid obliqua contacts the pro- tostylid of N. couperi as it does in N. saltavus, a condition which Flannery & Rich (1986) consid- ered plesiomorphic for macropodoids. In N. novus the cristid obliqua contacts the base of the MATERIAL. Holotype QMF19820, A fragment of the protoconid and in N. tarrinyeri the cristid obliqua horizontal ramus of a right dentary, extending from the anterior end of the diastema to the posterior of the molar row and preserving P3 and M1-4. from White Hunter Site, Hal’s Hill, D Site Plateau, which has been corre- lated (Myers & Archer, 1997) with the Ngama Local Fauna from the Tirari Desert which Woodburne et al. (1993) have shown to be late Oligocene, about 24 to 26My. Wururoo dayamayi sp. nov. (Figs 3, 4, 5B; Table 2) DIAGNOSIS. As for genus. ETYMOLOGY. Waanyi daya, chop; mayi tooth. for the tall, robust plagiaulacoid premolar of the holotype. DESCRIPTION. Only small portion of anterior margin of ascending ramus present, remainder of dentary posterior to M4 lost. Horizontal ramus deepest below M3/M4, tapering gently anteriorly, ramus twisted so that mesial face inclines dorsally below P3 and slighly ventrally below M4. Low ridge running length of dorsal margin of very short diastema. Mental foramen well below and slightly anterior to P3, small posterior mental foramen below hypolophid of M2. Masseteric FIG. 2. Dimensions of QMF30401, holotype of canal elliptical in cross section, extending for- Nambaroo couperi sp. nov. A, buccal view. B, lingual ward at least as far as anterior of M2, from which view. position anterior of canal blocked by undissolved TWO NEW BALBARINE KANGAROOS matrix. A narrow canal, the inferior dental canal or branch thereof, partitioned from lingual side of masseteric canal by thin lamina of bone anterior to M4. Mandibular symphysis extending posteri- orly to below mid point of P3. Molar row concave in lateral view and straight in occlusal view. Molar size increasing from Mı to M3 but M4 a little smaller than M3. Dentition. P3 large, robust, plagiaulacoid, occlu- sal edge well above occlusal plane of molar row, long axis flexed slightly buccally out of align- ment with line of molar row. Viewed occlusally with a convex buccal margin and slightly concave lingual margin. Lingual face of crown inclined at steeper angle than is buccal. Occlusal margin on approximate midline for most of length, but slightly lingually displaced at posterior end. Six cuspids on occlusal margin, all but most posterior having associated lingual and buccal trans- cristids. Cristids descending anterior and poste- rior margins of crown from corresponding cuspids. Miı roughly rectangular in occlusal outline but with distinct lingual convexity in crown base adjacent to interlophid valley. Anterior margin abutting posterior buccal margin of P3. Pro- tolophid markedly shorter than hypolophid with protoconid set on approximate midline. Lingual cuspids positioned closer to margin and with steeper lateral walls than buccal cuspids. Both protolophid and hypolophid crests concave ante- riorly, posterior faces of both lophids vertical. Protoconid taller than metaconid. Paracristid straight but anterolingually inclined as descends from protoconid apex to a prominence positioned just posterior to anterior margin. A short but clearly defined ridge is directed lingually from apex of prominence, ending abruptly anterior to midpoint of protolophid. While the prominence may represent the paraconid, an alternative view preferred here is that this ridge represents a lin- gual extension of the paracristid, and its termina- tion the paraconid (see Fig. 7c). Line of paracristid continued by a ridge running anteri- orly from anterior prominence for very short dis- tance and at steeper angle to anterior margin. From here a continuous ridge descends ventrobuccally on anterior margin, forming bor- der of narrow precingulid which ends before reaching buccal margin. No premetacristid or postmetacristid. An enamel ridge, the ‘pro- tostylid crest’ descends posterobuccally from apex of protoconid for about half height of that cusp where it is contacted by ascending, anterior portion of cristid obliqua. Cristid obliqua de- scends anterolingually from apex of hypoconid before turning anteriorly to cross wide inter- lophid valley and ascend diagonally on posterior of protoconid. Entoconid taller than hypoconid which shows evidence of wear: enamel breached on lingual side of apex. Preentocristid descends directly an- teriorly from apex of entoconid to floor of inter- lophid valley. Posthypocristid short, extremely lingually displaced, originating at lowest point of hypolophid crest, closer to entoconid than hypoconid, descending ventrolingually to meet postentocristid just above base of entoconid. In- verted triangular fossette enclosed laterally by posthypocristid and postentocristid. Below junc- tion of postentocristid and posthypocristid is short, prominent enamel ridge, descending at dif- ferent angle to postentocristid and which may represent a reduced hypoconulid. This ridge forms lingual margin of broad, horizontal hypocingulid extending across two-thirds of base of hypolophid., Broad wear facets on posterior face of pro- tolophid extend across link between protostylid crest and cristid obliqua; similar facets on poste- rior face of hypolophid extend across posthypocristid. Facets bear fine vertical striae. M2 rectangular in occlusal outline but with slight concavity of lingual margin adjacent to end of interlophid valley. Protolophid and hypolophid subequal in length, occlusal crest of hypolophid slightly more anteriorly concave than that of protolophid. Lingual cuspids set closer to lateral margin and with steeper lateral walls than buccal cuspids. Metaconid taller than protoconid which shows evidence of wear: enamel breached lingually adjacent to apex. Paracristid directed anterolingually as descends from protoconid apex to anterior margin, Prominent pre- metacristid runs slightly buccally as descends from metaconid apex to anterior margin. Short, broad anterior cingulid enclosed between these cristids. Short precingulid buccal to paracristid. In posterior view occlusal crest of protolophid forms shallow ‘V° with low point located closer to protoconid than metaconid. Cristid obliqua forms thick enamel ridge as crosses interlophid valley, tapering somewhat anteriorly as ascends short distance on lingual side of posterior face of protoconid. No postmetacristid or preentocristid: broad interlophid valley widely open lingually. Floor of interlophid valley considerably more elevated on lingual side of cristid obliqua than on buccal side which slopes steeply towards crown MEMOIRS OF THE QUEENSLAND MUSEUM FIG. 3. QMF19820, Holotype of Wururoo dayamayi gen. et sp. nov. A, buccal view. B, lingual view. C, stereopair of occlusal view. Scales = 10mm. AR number is an informal system used in the Vertebrate Palaeontology Laboratory, University of New South Wales. base. Entoconid and hypoconid subequal in height. No posthypocristid. Thick ridge of postentocristid continuous with similarly promi- nent hypocingulid which descends slightly ven- trally as crosses posterior base of the hypolophid. M3 with longer anterior cingulid, lingual cus- pids taller than buccal cuspids, protolophid slightly longer than hypolophid and thickened ridge, representing remnant of posthypocristid, descending vertically from low point of occlusal crest on posterior face of hypolophid. Less worn than is Mo. Cristid obliqua can be seen to arise from more anterolingual position relative to apex of the hypoconid. Mg smaller than M3 with much narrower ante- rior cingulid. Lacking precingulid buccal to paracristid. Hypolophid markedly shorter than protolophid. Metaconid taller than protoconid, crest of protolophid formed chiefly by buccal crest from apex of metaconid, descending to meet lingual flank of protoconid at a point directly in line with anterior end of cristid obliqua. Cristid obliqua originates lingual to hypoconid apex and runs directly anteriorly across interlophid valley. Hypoconid and entoconid subequal in height but entoconid set somewhat anterior to hypoconid. Apex of entoconid damaged during life and bro- ken edges of enamel subsequently smoothed as a result of wear. Hypolophid crest with a narrow V-shape in posterior view with lingual arm run- ning slightly anteriorly towards damaged apex of entoconid. No preentocristid but very thick postentocristid running ventrobuccally on poste- rior face of entoconid, merging with equally prominent hypocingulid which crosses half width of hypolophid base. TWO NEW BALBARINE KANGAROOS ined ~ Ww TABLE 2. Dental parameters for QMF19820, Holotype of Wururoo dayamayi gen. et sp. nov. Cat. No. P3 | | a [mw] n QMF19820 | 8.6 | 49 | 67 | DISCUSSION. P3 is much larger and more mas- sive than that in Nambaroo or Ganawamaya Cooke, 1992, and is similar to that in undescribed Riversleigh species of Balbaroo. The large P3 in the latter balbarines is more similar in profile and size relative to molars to propleopines and hypsiprimnodontines than to macropodids. The diastema in W. dayamayi is also much shorter than in Nambaroo or Ganawamaya. The shorter diastema and markedly more robust P3 may be indicative of a greater reliance on the use of premolar shearing action in food collection and/or processing. The similarity of plagiaulacoid premolars in more derived balbarines, propleopines and hypsiprimno- dontines may be the result of convergence in species placing a similar emphasis on premolar shearing. However, premolars in these groups are similar in form to those of phalangerids and it is likely that the palgiulacoid premolar form is plesiomorphic for macropodoids. If robust, plagiaulacoid premolars are plesiomorphic, within Balbarinae the markedly more gracile pre- molars of Nambaroo and Ganawamaya would represent an apomorphy for a clade containing these two genera, The lingually displaced Mı posthypocristid seen in W. dayamayi is also present in species of Nambaroo and in Ganawamaya aediculis Cooke, 1992. Its widespread occurrence among plesiomorphic balbarines supports the view of Flannery & Rich (1986), that lingual displace- ment of the posthypocristid has played an import- ant role in hypolophid formation in Balbarinae. The connection of the cristid obliqua to the pro- FIG. 4. Dimensions of QMF1I9820, holotype of Wururoo dayamayi gen. et sp. nov., buccal view. tostylid crest resembles the connection between the cristid obliqua and the discrete protostylid of Nambaroo saltavus as noted by Flannery & Rich (1986), and suggested by them to represent the plesiomorphic state of this character in macropodids. The M; precingulid in other undescribed plesiomorphic balbarine species (pers. obs.) re- ceives the posterolingual cuspule of P3. The pre- cingulid is low in W. dayamayi, suggesting that a prominent posterolingual cuspid is to be expected on its P’. The lingual ridge associated with the Mı paraconid forms the margin of a small anterior cingulid which receives the posterior end of the P? occlusal margin, as indicated by signs of wear on the posterior face of the ridge. The broad wear facets on the posterior faces of the Mı lophids indicate broad contact with the lophs of M! while the orientation of the striae indicate greater verti- cal rather than lateral relative movement between lophs and lophids. CUSPID HOMOLOGY AND EVOLUTION OF LOWER MOLAR MORPHOLOGY IN BALBARINAE Differing views of cuspid homology and evo- lution of lower molar morphology among macropodoids are examined and alternative hypotheses proposed to determine homology of structures on lower molars of plesiomorphic balbarines from Riversleigh and to elucidate the course of molar evolution within Balbarinae. Since interpretations of cuspid homology in Hypsiprymnodon have been central to wider ar- guments pertaining to macropodoid molar evolu- tion, such interpretations of previous authors are reviewed first and a new interpretation presented which incorporates evidence provided by the new species described herein. The posterobuccal protostylid crest in W. dayamayi is very similar to that which descends from the dominant trigonid cuspid toa tiny buccal cuspid in dP3 of Hypsiprimnodon moschatus. Ride (1961) identified the dominant cuspid as the metaconid and the tiny cuspid as the protoconid, but has since modified this interpretation (Ride, 1993), considering the central trigonid cuspid posterior to the paraconid on dP3 to be a 276 4 ae FIG. 5. Posterior occlusal views of Mi, A, QMF303401, holotype of trigonid thereby becoming lat- Nambaroo couperi sp. nov. B, QMF19820, holotype of Wururoo dayamayi gen. et sp. nov.. parametaconid and the small cuspid posterobuc- cal to this to be the protoconid, with the normal relationship of protoconid, metaconid and paraconid preserved on Mj, the parametaconid having been lost in this tooth. Archer (1978) raised the possibility that the anterobuccal cusp in M; (his M2) of H. moschatus (and other macropodoids) may be ‘the homo- logue of the phascolarctid protostylid and not the protoconid’ and further that ‘the tiny cusp ob- served by Ride (1961) on Mı ( Ride’s dP4 and here dP3) of Hypsiprimnodon may be the serial homologue of this protostylid’. Archer & Flann- ery (1985) and Flannery & Rich (1986) also identified the posterobuccal cuspid of the Mı trigonid of H. moschatus as the protostylid having here an anterior cristid descending to the anterior margin, and the cuspid lingual to this, also with an anterior cristid, as the protoconid, the metaconid being displaced posterolingually. The interpretation of trigonid cuspid homology on dP3 offered by Archer (1978) is accepted here but none of the above views regarding Mı trigonid morphology in H. moschatus are upheld. For reasons explained below, the most buccal cuspid-like structure associated with the pro- tolophid is accepted as the protostylid but the lingual cuspid of the protolophid is regarded as the metaconid. As Ride (1993) noted, constraints imposed by functional interactions with premolars can alter the topography of teeth at the premolar/molar boundary, so cuspid homology of Mı in H. moschatus has been examined with reference to MEMOIRS OF THE QUEENSLAND MUSEUM wa more posterior molars. Molars posterior to M; have a buccally positioned protoconid which has a paracristid running to the anterior margin. The metaconid of these molars is lingually positioned and has an associated premetacristid. Both cristids are present on M; and bear the same relationship to each other as they do on more posterior molars. The metaconid is distinct but the protoconid, from which the paracristid originates, may be considerably reduced and is more centrally positioned, the erally compressed in a manner closely resembling that seen in balbarines. In unworn H. moschatus specimens (QMJ10233, QMJ9327, QMJ145), the paracristid originates from a slight elevation, interpreted here as the reduced pro- toconid (Fig. 6F). A posterobucccal crest links this to a larger cuspid which is thus the protostylid in the sense in which the term is used by Van Valen (1966), and Butler (1978): it occurs in the same position as the protostylid of other diprotod- ont marsupials, e.g., pseudocheirids and phascolarctids and Nambaroo, a position in which it fulfils the function ascribed to it by Butler (1978), i. e., ‘to shear against the an- terolingual surface of the paracone of the corre- sponding upper tooth’ (a function which could not be fulfilled by the ‘one or more protostylids’ indicated by Ride (1993), as occurring on the precingulid anterobuccal to the M; trigonid in the propleopines, Jackmahoneya and Propleopus). Ride (1993) indicated 3 cuspids immediately posterior to the paraconid of dP3 of H. moschatus: the large, central cuspid identified as the parametaconid, a posterolingual metaconid and a very small posterobuccal protoconid. The posterolingual cuspid is accepted here as the metaconid but since the large, central cuspid has a cristid running anteriorly from its apex toward the paraconid, the cuspid is here regarded (as it was by Archer, 1978) as the protoconid, its cristid as the paracristid and the small posterobuccal cuspid as the protostylid, homologous with that which also occurs on dP3 in undescribed Riversleigh specimens of Nambaroo (pers. obs.), the protostylid on Mı of H. moschatus and the TWO NEW BALBARINE KANGARGOS dP/3 M/1 ee | © Paraconid © Protoconid + Protostylid a Parametaconid o Metaconid Pe phases FIG. 6. Interpretations of trigonid cuspid homology in Hypsiprymnodon moschatus. A & B, Ride (1993), C, Archer (1978), D, Archer & Flannery (1985). E, F, Cooke (herein). protostylid crest on My of Wururoe and on dP3 of bulungamayine species (Cooke, 1997). The homology of the buccal cuspid on the M; protolophid of propleopines is Jess clear. Ride (1993) labelled this cuspid as the protoconid and the cuspid immediately lingual fo it as a neomorph, the parametaconid. Flannery & Archer (1985) labelled the buccal cuspid as the protostylid and the cuspid immediately lingual to it as the protoconid. While it is clear that the protostylid, as a discrete cuspid or as a protostylid crest, is common among plesiomorphic macropodoids, itis unusual in this group for it to have acquired a cristid mimicking the paracristid in linking to the anterior cingulid. If Flannery & Archer are correct, this would constitute a syn- apomorphy for propleopines as would the neomorphic parametaconid in the interpretation of Ride. The suggestion of Archer & Flannery (1985) that the buccal cuspid of the protolophid in My of propleopines and potoroines was the protostylid and that the metaconid was lost in potoroines but retained in propleopines was enlarged upon by Flannery & Rich (1986), They suggested that in potoraids the protostylid formed the buccal pra- tolophid cuspid and the protoconid the lingual cuspid, in contradistinclion to macropodids in which the protostylid is lostand the buceal cuspid is the protoconid and the lingual cuspid the metaconid (lost in potoroids other than pro- pleopines). The plesiomorphic balbarines, Nambaroe, Wururoo and Ganawamaya certainly indicate that the protostylid has been reduced und ultimately lost in balbarines, however there is less evidence to support a belief that it has been re- tained in potoroids. The Mı (Cooke, 1997) pro- tostylid of Palaeopotorous priscus Flannery & Rich, 1986, bears a similar relationship to the protoconid and cristid obliqua as does that of H. moschatus, exhibiting a condition intermediate between the distinct cuspid of the protostylid of Nainbareo and the protostylid crest of W. dayamayi. It thus may represent an early stage in the reduction of the protostylid among potoroids. In the proposals of cusp homology in macropodids and potoroids referred to above, the condition occurring in H. moschatus is crucial because it is used in both cases to represent the intermediate condition between a phalangerid an- cestorand more derived potoroids, If, as has been proposed here, the reduced cuspid from which the paracristid originates on the short Mı protolophid of H, moschatus is the protoconid, then there is no reason Lo suppose that this is not also the case in other potoroids and the sequence, Pal- aeopotorous - Hypsiprimnoden - Potarous, used by Flannery & Rich (1986) to illustrate their argument can be taken as indicating a transition involving reduction of the protostylid. The protostylid is also exhibited in varying degrees of development within phalangerids. A very similar structure to the Mı protostylid crest of W. dayamayi occurs in Phalanger inter- castellanus and joins the cristid obliqua. In Trichosurus vulpecula there i$ a straight ridge FIG. 7. Evolution of lower molar morphology in balbarines illustrated by RMy. A, hypothetical macropodoid ancestor. B, Nambaroo couperi. C, Wururoo dayamai. D, Balbaroe grade. Abbreviations: Ac=anterior cingulid: co=cristid obliqua; Ec=entoconid, He=hypocingulid; Hd=hypoconulid; Hy=hypoconid; Me=metaconid; Pa=paraconid: Pe=pre- cingulid; pec=postentocnstid, phc=posthypocrisid: Pr=pro- toconid; Prs=protostylid: pre=protosty|id crest. descending posterobuccally to the base of the protoconid where it links to the cristid obliqua. In T. caninus there is a ridge on the protoconid resembling that in T. vulpecula, but it does nat contact the cristid obliqua which independently ascends the posterior face of the protoconid, Flannery & Rich (1986) indicate a discrete pro- tostylid in Phalanger vestitus and a similar situa- tion occurs in Spilocuscus maculatus, butin some specimens of this species the protostylid may be reduced to a ridge on the posterobuccal [lank of the protoconid. Flannery & Archer (1987) sug- gest aprotostylid on Mj in Strigocuscus reidi and Trichosurus dicksoni, both from System C sites at Riversleigh. It would appear that the genetic potential for the formation of a protostylid is widespread among phalangeridans and may be considered as MEMOIRS OF THE QUEENSLAND MUSEUM plesiomorphic for the group. In pseudocheirids among Petauroidea and Nambarao among Macropodoidea the protostylid is fully developed but in most phalangerids and in Palaeopatoraus, Hypsiprimnoedon and Wurureo the poten- tial is not as fully expressed. This expres- sion has independently been suppressed among more derived macropodoids (with the possible exception of propleopines) and other phalangeridans. Wururoo is more derived than Nambaroo in terms of My morphology: the protostylid has been reduced, a neomorphic anterior cingulid has begun to form and the trigonid basin has become partly enclosed. Nambaroa and W. dayamayi indicate initial stages of a trend among balbarines towards the develop- ment of an anterior cingulid and the enclo- sure of the trigonid basin, Concomitant with this trend is a decrease in the relative importance of the precingulid which occu- pies much of the Mı anterior margin in plesiomorphic species such as N. couperi. Hypolophid formation in balbarines proceeded as outlined by Flannery & Rich (1986), involving elevation and lingual displacement of the posthypocnstid which contributes a buccal component to the hypolophid, the lingual component con- tributed by a buccally directed cristid from the entoconid. The latter cristid is not necessarily a neomorphic “entohypo- cristid’ (Ride, 1993). However, the hypoconulid on the Mj of N. couper and its apparent presence in a reduced orm on Mı of W. dayamaye indicates the addi- tional involvement of this cuspid in hypolophid formation in plesiomorphic balbarines (Fig. 7). The hypoconulid in such forms 1s contacted by a lingually displaced, diagonal posthypocristid, An inverted triangular fossette occurs on the lin- gual posterior face of the hypolaphid, bounded by the posthypocristid and the postentocristid (when present). A shallow fossette develops buccal to the posthypocrjstid and a neomorphic cingulid, the hypocingulid according to the definition of Butler (1990), develops running transversely across the posterior base of the hypolophid from the hypoconulid or, when that structure has been lost, from the ventral end of the posthypocristid, as in N. saltavus. The posthypocristid is lost in more derived species such as those of Balbarve in which the postentocristid links to a transverse, TWO NEW BALBARINE KANGAROOS posterior cingulid homologous with the hypocingulid. The modification offered above to the hypoth- esis of Flannery & Rich has been extended by Ride (1993) to macropodoids in general but I apply it only to balbarines. The bilophodont lower molar morphology of bulungamayines is derived from a bunolophodont ancestral form by different means to those outlined above (Cooke, 1997) and clearly indicates independent evolu- tion of lophodonty within this group. However, a hypoconulid positioned low on the posterior, lin- gual face of the hypolophid and associated with a posthypocristid, is the probable plesiomorphic condition for all macropodoids The basal macropodoid may well have had a bunolophod- ont lower molar morphology similar to that of Palaeopotorous or of phalangerids, but with a more distinct protostylid buccal to the protoconid and a hypoconulid contacted by the posthypocristid at the posterior base of the en- toconid. The lingual component of the balbarine hypolophid would represent a reduction of the buccal cristid of the entoconid which forms the transverse posterior lophid of bunolophodont macropodoids. The hypoconulid in plesiomorphic balbarines and its absence in any known potoroids of com- parable age suggests that balbarines diverged early from the main stem of macropodoid evolu- tion and independently and probably rapidly evolved lophodonty, the better to exploit a herbi- vore niche. The combination of plesiomorphic dental characters present in balbarines, including lateral compression of the M; trigonid and the presence of both a hypoconulid and protostylid on this tooth, contrasts markedly with the absence of such characters in bulungamayines, There has been parallel (and probably later) evolution of lophodonty within Bulungamainae and the ab- sence of plesiomorphic molar characters such as those indicated above, suggests that bulungamayines may be more likely to be ances- tral to the highly derived macropodids. ACKNOWLEDGEMENTS Research grants from the Australian Research Council and the University of New South Wales have been the primary mechanism for providing the research material examined in this study. Ad- ditional support for the Riversleigh project has come from the National Estates Program grants, the Australian Geographical Society, The Austra- lian Museum, The Riversleigh Society, ICI Pty Ltd, Century Zinc Limited, the Mt Isa Shire and private donors. I thank the Director and staff of the Queensland Museum for provision of workspace and facili- ties; Jeff Wright and Alex Cook for preparation of photographic prints; Michael Archer and David Ride for their helpful discussions; the staff and students at the University of New South Wales involved with the Riversleigh Project who have been so willing in their assistance. LITERATURE CITED APLIN, K.P. & ARCHER, M. 1987. Recent advances in marsupial systematics with a new syncretic classification, Pp xv-Ixxii. In Archer, M. (ed), Possums and opossums: studies in evolution’. (Surrey Beatty & Sons: Sydney). 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: 157-164. ARCHER, M. & FLANNERY, T.F. 1985. Revision of the extinct gigantic rat kangaroos (Potoroidae: Marsupialia), with a description of anew Miocene genus and species and a new Pleistocene species of Propleopus. Journal of Paleontology 59: 1131- 1149. ARCHER, M., GODTHELP, H., HAND, S.J. & MEGIRIAN, D. 1989. Fossil mammals of Riversleigh, northwestern Queensland: prelimi- nary overview of biostratigraphy, correlation and environmental change. Australian Zoologist 25(2): 29-65. BREEN, G. 1981. The Mayi languages of the Queens- land gulf country. Australian Institute of Aborig- inal Studies AIAS new series 29. BUTLER, P.M. 1978. Molar cusp nomenclature and homology. Pp 439-453, In Butler, P.M. & Joysey, K. (eds), Development, function and evolution of teeth’. (Academic Press: London). 1990. Early trends in the evolution of tribosphenic molars. Biological Reviews 65: 529-552. COOKE, B.N. 1992. Primitive macropodids from Riversleigh, northwestern Queensland. Al- cheringa 16: 201-217. 1997. New Miocene bulungamayine kangaroos (Marsupialia, Potoroidae) from Riversleigh, northwestern Queensland. Memoirs of the Queensland Museum 41: 281-294. FLANNERY, T.F. 1989. Phylogeny of the Macropodoidea; a study in convergence. Pp 1-46. In Grigg, G. Jarman, P & Hume, I. (eds), Kanga- roos, wallabies and rat-kangaroos. (Surrey Beatty & Sons: Sydney). FLANNERY, T. & ARCHER, M. 1987. Strigocuscus reidi and Trichosurus dicksoni, two new fossil phalangerids (Marsupialia: Phalangeridae) from the Miocene of northwestern Queensland. Pp 527- 536, In Archer, M. (ed.), Possums and opossums: 280 studies in evolution. (Surrey Beatty & Sons: Syd- ney). FLANNERY, T.F., ARCHER, M. & PLANE, M. 1982. Middle Miocene kangaroos (Macropodoidea: Marsupialia) from three localities in northern Aus- tralia, with a description of two new subfamilies. Bureau of Mineral Resources Journal of Austra- lian Geology and Geophysics 7: 287-302. FLANNERY, T.F. & RICH, T.H.V. 1986. Macropodoids from the Middle Miocene Namba Formation, South Australia, and the homology of some dental structures in kangaroos. Journal of Paleontology 60(2): 418-447. FLOWER, W.H. 1867. On the development and suc- cession of teeth in the Marsupialia. Philosophical Transactions of the Royal Society of London B157: 631-641. LUCKETT, W.P. 1993. An ontogenetic assessment of dental homologies in therian mammals. Pp 182- 204. In Szalay, F.S, Novacek, M.J. & McKenna, M.C. (eds), Mammal phylogeny. (Springer-Ver- lag: New York). MYERS, T.J. & ARCHER, M. 1997, Kuterintja ngama (Marsupialia, Ilariidae): a revised systematic analysis based on material from the late Oligocene of Riversleigh, northwestern Queensland, Aus- tralia. Memoirs of the Queensland Museum 41: 379-392. MEMOIRS OF THE QUEENSLAND MUSEUM RIDE, W.D.L. 1961. The cheek teeth of Hypsiprimnodon moschatus Ramsay 1876 (Macropodidae: Marsupialia). Journal and Pro- ceedings of the Royal Society of Western Aus- tralia 44: 53-60. 1993. Jackmahoneya gen. nov. and the genesis of the macropodiform molar. Memoirs of the Asso- ciation of Australasian Palaeontologists 15; 441- 459. SZALAY, F.S. 1969, Mixodectidae, Microsyopidae, and the Insectivore-Primate transition. Bulletin of the American Museum of Natural History 140: 193-330. VAN VALEN, L. 1966. Deltatheridia, a new order of mammals. Buletin of the American Museum of Natural History 132: 1-126. 1994, Serial homology: the crests and cusps of mam- malian teeth. Acta Palaeontologica Polonica 38, 3/4: 145-158. WOODBURNE, M.O., MACFADDEN, B.J., CASE, J,A., SPRINGER, M.S., PLEDGE, N., POWER, J.D., WOODBURNE, J.M. & SPRINGER, K.B. 1993. Land mammal biostratigraphy and mag- netostratigraphy of the Etadunna Formation (late Oligocene) of South Australia. Journal of Verte- brate Paleontology 14: 483-515. NEW MIOCENE BULUNGAMAYINE KANGAROOS (MARSUPIALIA: POTOROIDAE)FROM RIVERSLEIGH, NORTHWESTERN QUEENSLAND B. N. COOKE Cooke, B.N. 19970630: New Miocene bulungamayine kangaroos (Marsupialia: Potoroidae} from Riversleigh, northwestern Queensland. Memoirs of the Queensland Museum 41(2): 281-294, Brisbane, ISSN 0079-8835, Nowidgee matrix gen, et sp. noy. and Ganguroo bilumina gen. el sp, noy, are described from freshwater Miocene System B limestone at Riversleigh, NW Queensland. Subfamilial diagnosis of Bulungamayinae is emended. The new species indicate that lophodunty was achieved in bulungamayines by a different process from that in balbarines. Similarities in dental morphology between bulungamayines and lale Miocene macropodids suggest that Bulungamayinae is ancestral to Macropodidac. [7] Riversleigh, kangaroo, balbirines, Bulungamayinae, lophoadonty. EN. Cooke, School af Life Science, Queensland University of Technology, GPO Box 2434, Brisbane, Queensland 4001, Australias 17 December 1906, Rat-kangaroos or potoroids, in the sense of Archer & Bartholomai (1978) and Bartholomai (1978), were unknown in the pre-Pliocene fossil record of Australia until Archer (1979) described Wabularoo naughtoni as an enigmatic, lophodont kangaroo [rom the Riversleigh Local Fauna of the Carl Creek Limestone. Flannery et al, (1982) described Bulungamaya delicata from the Carl Creek Limestone and placed it and W. naughtoni in the potoroid Bulungamayinae. Gumardee paseuali, also from the Carl Creek Limestone was described in the same paper but placed in the Potoroinae, More recent additions to the record of potoroines include Wakiewakie lawsoni Woadburne, 1984 and Purtia mosateus Case, 1984, from the Neapakaldi Local Fauna of South Australia and Betrongia moyesi Flannery & Archer, 1987, from Two Trees Site at Riversleigh. Flannery & Rich (1986) described Gumardee and indeterminate potoroines from the Tarkaroploo Local Fauna of South Australia. Archer & Flannery (1985) erecied Pro- pleopinae for Ekalradete: ima, a giant rat kanga- roo from Gag Site at Riversleigh and Pleistocene and Pliocene species of Prop/eapus, Flannery & Archer (1987) described Hypsiprimnodan harthelamaii from the Gag Site at Riversleigh and Flannery & Rich (1986) reported hypsiprimnodontine material from the Tar- kurooloo Local Fauna. Palaeopotoroinae Flann- ery & Rich, 1986 accommodates Palaeopotorous priscus from the Tarkarooloo Local Fauna. The diversity of pre-Pliocene potoroids is such that only 2 of the more recently discovered spe- cies have been assigned to existing genera and 3 new subfamilies have been proposed. Of these Bulungamayinae Flannery et al, 1982, has at tracted most discussion. Woodburne (1984) and Case (1984) argued that the lophodony bulungamayines W. naughtoni and B. delicata share characteristics with plesiomorphic macropodids (their macropodines) such as Dorcepsoides fossilis Woodburne, 1967. Dorcopsis and Dorcopsulus and should he in- cluded in Macropodidae (their Macropodinae). They also argued a similar placement of Gumardee. Flannery et al. (1984) identified syn- apomorphies which they considered unjted potoroids in a monophyletic group and defended their placement of bulungamayines and G pascuali within Potoroidae on the basis of severul of these derived states. Flannery & Archer (1987a, b) demonstrated that one suggested syn- apomorphic character, squamosal-frontal contact on the lateral wall of the cranium, is not universal within the group and could no longer be consid- ered as a potoroid synapomorphy. The state of this character is unknown in pew bulungamayine material described below. The new species are similar in size and have similar premolar morphology to that of R. delicata and together with that species represent a sequence Which reveals much about the evolu- uon ol lophodonty within Bulungamayinue. METHODS Molar homology follows Luckett (1993), pre- molar homology follows Flower (1867). Dentil descriptive terminology is principally that used by Archer (1984) but with some terms adopted from Szalay (1969) and Ride (1993), In upper molars the term, ‘paracingulum’ is used to indi- cate an anterior cingulum bounded laterally by the preparacrista and preprotocrista as indicated in Szalay (1969). Ride used ‘precingulum’ tor this structure. | use ‘precingulum’ for an anterior cingulum extending lingually from the pre- protocrista, following Szalay. This structure was referred to hy Ride as the “anterolingual cingulum’. ‘Metacingulum” is used for u poste- rior cingulum bounded posterolingually by the postmetaconule erista, “paracrista’ and ‘meta- crista’ are used for the tingually directed, loph- forming cristae from the paracone and metacone, respectively. Use of the latter term in this manner is a departure from Szalay who uses “metucrista’ synonymously with ‘“postmetacrista’, “Postmetacrista’ is used here in the sense of Archer (1984). Cusp homology of upper molars is that of Tedford & Woodburne (1987), with the posterior buceal and lingual cusps designated as metacone and meltuconule respectively, and the cuspule be- tween these as the neometaconule, Suprageneric classification follows Aplin & Archer (1987). Material is housed in the Queensland Museum (QMP). Measurements are in millimetres, SYSTEMATICS Superfamily MACROPODOIDEA Gray, 1821 Family POTOROIDAE Gray, L821 Subfamily BULUNGAMAYINAE Flannery, Archer & Plane 1982, cmend. Cooke. 1997 Bulungamayines have a buccally expanded mas- seteric canal confluent over its length with the inferior dental canal, the common canal penetrat- ing deeply within the dentary below the molar row, The digastric process of the dentary is ex- panded so that the ventral margin of the dentary is convex below the molar row. T) has. enamel confined 10 the buccal surface but extensive on that surface and not confined to the ventral por- tion as it is in potoroines. Ventral and dorsal enamel flanges are present on 4). P3 is elongate with many fine transcristids and a bulbous base. A small tooth, I but which may be a small canine, is Just posterior to the dorsal margin of the I alveolus, Molar teeth may be bunolophodont or Jophodont as defined hy Flannery etal. (1984). Bulungamayines differ from hypsiprimno- dontines and propleopines by having an elongate P whose occlusal margin in lateral view is. straight or concave, rather than a plagiaulacuid P3 MEMOIRS OF THE QUEENSLAND MUSEUM with a Convex occlusal margin. They differ tren potoroines by having much more bulbous premo- lars, by having an (2 and a much more extensive arca of enamel on the buccal surface of l. They differ from palaeopotoroines by lacking a distinct protostylid on Mı. REMARKS, Type specimens ol bulungaumayines erected herein are far more complete than those of previously described species and reveal details of anatomical and dental features absent in the holotypes of Bulungamaya delicata and Waubularoe naughtoni - This additional intorma- tion forces the above subfamily revision. Nowidgee gen, nov. TYPE SPECIES, Newidgee matrix sp. nov. DIAGNOSIS. Bulungamayine with bunolophodont molars, Upper molars with large stylar cusp C extend- ing posteriorly to close the buccal end of the imerloph region, ETYMOLOGY, Waanyi (as spoken by Ivy Stinken, formerly of Riversleigh Station) New/dgee, grand- mother. Nowidgee matrix sp. nov. (Figs 1, 2; Table!) DIAGNOSIS. As for the genus, MATERIAL EXAMINED. Holotype QMF3039(), from Camel Sputum, Godthelp’s Hill, D-Site Plateau. Puratypes QMF19961, 20255, 22761, 30393, 30394, 30395 from Camel Sputum Site, QMF19937, 20069, 20080, 30391 from Wayne’s Wok Site, Hal's Hill, D-Site Plateay, Both System B sites (Archer et dl, 1989) of Miocene age. ETYMOLOGY. Latin matrix, mother of an animal, refers tu ils ancestral position, DESCRIPTION OF HOLOTYPE. Right dentiry fragment of most of the horizontal ramus to the level of M4 and part of the ascending ramus. Ih, Ps and M1-4 preserved, Ascending ramus at about 110° to occlusal plane of molar row. Masseteric canal confluent with inferior dental canal, mak- ing it difficult to assess extent of forward penc- tration of the masseter, but anterior wall of masseteric fossa extending anteriorly to about level of the M3 protolophid . At this level diame- ter of common canal not greatly exceeding that of sulcus representing inferior dental canal in posterior, lingual wall of masseteric fossa, This Suggests anterior portion of common canal occu- NEW BULUNGAMAYINE KANGAROOS pied chiefly by vessels associated with dental canal and masseter not passing, much more ante- riorly than this level, Masseteri¢ fossa buccally expunded with lat surface for attachment of su- perticial layer of masseter at anjeroventral bor- der, extending dorsally on anterobuceal margin of ascending ramus, Ventral margin of horizontal ramus gently convex with lowesi point below Ma/M3, Mental foramen just anterior to P3, be- tween root of Ty and dorsal margin of the dia- stema; much smatler posterior mental foramen ventral to protolophid of Mg. Posteroventrally melined buccinator sulcus between P3 alveolar margin and posterior mental foramen. Diastemi short, as long as P3. Damage tod) alveolas margin ohseuring [2 or its alveolus (2 alveolus in OME 19937), Mandibular symphysis extending poste- nurly almost to level of Ps posterior margin, Dentition, Molar row straight in occlusal view, P3 flexed slightly buecally out of alignment with it. In lateral view molat row concave: occlusal sur- faves of M2 and M3 lying below tine joining occlusal surfaces of Mi and Mas, Effects of wear on molar teeth progressively less obvious 10- wards. posterior of molar row. Molar size in- greases from M1-M3 but M3 larger than Ma. l; broken at anterior end, depth uniform over preserved length, rising, at approximately 20° rel- ative to dorsal margid of horizontal ramus. Enamel confined to buccal side, has both dorsal amd ventral enamel flanges, ventral being more Strongly developed. Dorsal Nange forms occlusal margin. Circular cross-section close ta alveolus becoming more elliptical anteriorly. P; blade-like, 50% longer than Mi. Occlusal outline semilunar with straight lingual margin and convex buccal margin. Occlusal crest slightly lingual to midline, flexes lingually at posterior end. Six small cuspids on occlusal margin ante- rior to longer, posterior cuspid, Transeristids as- sociated with cach of 6 minor cuspids and anterior and posterior margins of blade delineated by ver- tical cristids. Lingual surface of occlusal blade more steeply inclined than buccal. Mı almost square in occlusal outline but nar- rower anteriorly than posteriorly, Protolophid shorter than hypolophid: protoconid closer to midline than is hypoconid, Lingual cuspids taller, more sharply angular and closer to adjacent lat- cral margin than rounded, buceal counterparts which are also more worn. Lingual surfaces ver- cal, buccal surfaces more gently sloping, Pro- tolophid formed by metacristid descending buccally from metaeonid apex to lingual flank of prowweonid, Thick paracristid running antero- 283 lingually from apex of proloconid to antenoe margin, Shorl, broad anterior cingulid anterior to anterior face of protolophid, bounded buccally by paracristid, lingually by anteriorly directed pre- metucristid, Broad precingulid sloping steeply ventrally buccal to paracrisid. Sharply-delined postmetacristid curving buccally from metacanid apex, descending to narrow interlophid valley. Anteriorly oriented preentocristid separated from ventral end of posimelacristid by narrow clell- Cristid obliqua very thick, descending an- Jerolingually from apex of hypuconid to inter- lophid valley, then inclining buccally to apex of protoconid. Paracristid and eristid obliqua form continuous longitudinal ridge extending from an- terior margin to hypoconid. Hypolophid formed by buccal crest from the entoconid descending from enloconid apex and running buceally to meel lingual Nank of hypoconid. Posthypocristid descending lingually from hypoconid apex, crossing lingually posterior to buccal crest from entoconicd to posterior of entoconid below apex. Me larger and squarer than My with protolophidl and hypolophid of about equal length and pro toconid and hypoconid in alignment. Similar to My but anterior cingulid longer and broader, pre- vingulid shorter and interlophid valley broader. My worn in trigonid region but talonid rela- lively unworn, Crown very similar to Ma but moss structures more clearly defined. Cristid obliqua massive in imterlophid region, not much lower than apices of buccal cuspids, Lingual side of Interlophid valley more open with greater separa- tion of ventral ends of postmetacristid and pre entocristid. Posthypocristid sharply detined crossing posterior surlave from hypoconid apex to short, almost vertical postentocristid descend- ing from apex of entoconid, Deep, narrow trench between crest of posthypocristid and buccal cresi from entoconid anterior to it. My unworn. Hypolophid shorter than pro- tolophid. Cristid obliqua originating on an- terobuccal face of hypoconid, below apex, Preentocristid and postmetacristid separated only by narrow clett in interlophid valley. Posthypocristid crest rounded, meeting en- toconid much closer to tts apex than ou anterior molars. Buccal crest from entoconid shorter and less sharply defined. DESCRIPTION OF PARATYPES, DENTARY FRAGMENTS. Horizontal ramus nol as deep in juveniles as in adults, Posterior mental foramen varies from beneath M2/M3, (ho- 284 MEMOIRS OF THE QUEENSLAND MUSEUM FIG, 1. QMF30390, Holotype, Nowidgee matrix sp. nov. A, buccal view; B, lingual view; C, stereopair of occlusal view. Scales = 10mm. AR number is that of the Archer collection, University of New South Wales. lotype) to as far anterior as beneath hypolophid of Mı. QMF20080 preserves angular process and much of ascending ramus but lacks condyle and coronoid process. Lingual margin of angular pro- cess low, aligned with molar row, posterior mar- gin sloping ventrally towards lingual end. Pterygoid fossa triangular in dorsal view, buccal margin slightly undercutting base of ascending ramus, Mandibular foramen a narrow, vertically oriented ellipse, opening to very short posterior portion of inferior dental canal opening via mas- seteric foramen to masseteric fossa. Masseteric foramen just visible when masseteric fossa viewed from buccal side. Molar row ventrally concave. Ascending ramus at about 113° to line of a straight edge laid on molar row, LOWER DENTITION. dP? and dP3 preserved in QMF20080 and 20063, detached dP3 available for QMF30392. dP2 (Fig. 2A) short, blade-like, with bulbous base tapering anteriorly and poste- riorly. Occlusal margin straight, relatively hori- zontal, with 4 small cuspids anterior to longer, posterior cuspid overhanging posterior base of tooth. Fifth cuspid incompletely differentiated from large, posterior cuspid in QMF20080. Transcristids associated with each of 4 anterior cuspids, Posterior, buccal face of crown abutting anterior lingual face of dP3. dP3 (Fig. 2A) narrower anteriorly than posteri- orly, dominated by massive, laterally compressed protoconid, the tallest cuspid on tooth. No distinct metaconid, Paracristid anterolingually directed, descending to paraconid on anterior margin. Cristid descending steeply from paraconid apex to crown base on buccal side of antenor margin. Cristid descending posterior face of protoconid to interlophid valley. Paraconid, paracristid and protoconid form blade-like crest complementing that of dP2. Entoconid taller, more angular than hypoconid. Cristid obliqua running anterolingu- ally from apex of hypoconid, crossing interlophid valley and ascending buccal flank of protoconid. In QMF20069 and QMF30392 a short, buccally- NEW BULUNGAMAYINE KANGAROOS 285 FIG. 2. Paratypes of Nowidgee matrix sp. nov. A, stereopair of occlusal view of QMF20080, right dentary fragment with dP2-3, M1-3, (Ma). B, stereopair of occlusal view of QMF30395, right maxillary fragment with P3, Mı-4. Visible number is that of the Archer collection, University of New South Wales, directed protostylid crest joining protoconid apex to a prominence (reduced protostylid), contacted by anterior end of cristid obliqua. Sharp pre- entocristid running to interlophid valley from en- toconid apex. Buccally-directed crest from entoconid descending steeply buccally from en- toconid apex to about midline of tooth. Posthypocristid descending lingually from hypoconid apex to shorter postentocristid ascend- ing to entoconid apex. P3 in QMF19937 and 30394 resembles holo- type but with seventh minor cuspid, imperfectly differentiated from long posterior cuspid of oc- clusal crest. Seventh minor cuspid more clearly differentiated in P3’s removed from crypts in QMF20080 and QMF30392. Except for QMF19961 in which anterior molars very worn, molar teeth in paratypes less worn than those of holotype. Molar morphology among paratypes very similar to holotype, MEMOIRS OF THE QUEENSLAND MUSEUM TABLE 1. Dental parameters for type specimens of Nowidgee matrix sp. nov. Feso | | | |eolsz{sef6| | | | | | | | Tess] 2) Maxillary fragment. QMF30395 occludes extremely well with dentary fragment, QMF30394 found in close proximity. Preserves buccal surface of maxilla from diastemal region to masseteric process, including ventral margin of infraorbital foramen, suborbital shelf, alveolar process containing entire cheek tooth row and very narrow portion of palatine wing. Masseteric process of no more than a ventral prominence separated from alveolar process by short, narrow sulcus. Maxillary foramen of infraorbital canal at anterolingual corner of triangular suborbital shelf of maxilla, numerous smaller foramina within ventral margin of foramen. Infraorbital foramen dorsal to midpoint of P?. UPPER DENTITION. (Fig. 2B). Molar row slightly convex in lateral view, occlusal edge of P? aligned with buccal margin of molar row which curves slightly lingually a rt ie i Molar size increasing, from M 4 markedly smaller than M°. P? almost twice length of M!, lingual margin damaged, buccal margin convex for 2/3 length, becoming concave for remainder. Occlusal mar- gin anteroposteriorly straight, on midline of tooth. Six small cuspules on margin, succeeded by larger, posterior cuspule which has strong lingual ridge associated with its base. Fool ao laalaa] «| | | | [oaleelzefsslealso] soles zler] | |_| Fzooof 3s|27[30] 4 [ea[s2| 35] e s4 [20|2sfsefez|2z|szfaafe| | | TTT eet TATT EX Hoo za) a7 [aa T ||| Iralealsa[e| [1 [eolzal ANANTA [wean [aalav[aa| [salad] sal e [aa] 2a] a] s0]24[ 20] <0[2a] 20 [36] 0] 30 falza] 27 [so sfsl fol s[s[sfofs [al als [a[s[alo[s[uslal s [alla] reese] [LL keelel | O [aalselsr]so[sal <0] rae | _ [22|8|22 41 [31| 2] 41]32| 20 [42/31] 28 | [aslac| a1 [a1 | | ù| [sofas] so fsfs: 2s] TST] M! with straight anterior and buccal margins and convex lingual and posterior margins. Ante- rior width greater than posterior width but pro- toloph and metaloph of about equal length. Low crowned with lingual cusps more massive and more rounded than buccal counterparts. Buccal cusps closer to lateral margin of the tooth: buccal surfaces of crown almost vertical, lingual sur- faces sloping. Narrow lingual cingulum reaching from anterior, lingual base of protocone to base of metaconule. Protoloph formed by strong paracrista directed lingually from paracone apex and which meets buccal flank of protocone below apex. Preparacrista runs anteriorly from paracone apex to anterior margin and is continuous with anterior margin of paracingulum bounded later- ally by preparacrista and anterobuccally inclined preprotocrista which meets anterior margin ante- rior to junction of paracrista with protocone. Very large stylar cusp C closing interloph valley on buccal side. Postparacrista and premetacrista reaching floor of interloph valley from respective cusp apices, but not united. Postprotocrista strongly developed but worn in interloph region, contacting metaloph crest just buccal to apex of metaconule. Prominent neometaconule at about centre of metaloph with rounded crista running posteriorly for about half height of metaloph. Postmetaconule crista buccally inclined on pos- NEW BULUNGAMAYINE KANGAROOS tenor face of metaconule, crossing posterior base W metaloph as margin of strong metacingulum, contacting posteriorly directed postmetacrista at base of metacone. M- considerably wider anteriorly than posteri- urly, Occlusal outline more bluntly triangular. Crown differing from M! in; lingual cingulum continuous with precingulum extending anteri- orly across base of protocone to anterior end of preprotocrista, stylar cusp C shehtly more ante- riorly positioned on buccal flank of paracone and does not completely close buccal end of interloph valley; ncometaconule and its crista less obvious, M2 very similar to M? but lingual cingulum not as sharply defined, stylar cusp © smaller, postparacrista and premetacrista unite 10 form continuous centrocrista, M? much smaller than anterior molars, Metaloph markedly shorter than protoloph. Lingual cingulum separated from pre- cingulum, all cristae sharply defined. Metaconule lower; no ncometaconule or stylar cusp C. REMARKS, Nowidgee matrix is similar in size to Bulungamaye delicata but has bunolophodont. rather than lophodont molars. Tts bunolophadont lower molars resemble those of Partia mosaicus. but molar occlusal outline in N. matrix is more rectangular, rather than square as in P. mosaicus. P3 of N. matrix differs from thal of P. mesaicuy in having 6-7 rather than & transcristids and, while having a bulbous base, lacks the distinet lateral cingulids of P. mosaicus, Itdiffers from P. mosai- cus in having an Ti which has both ventral and dorsal enamel flanges and in having enamel which, while confined to the buccal surtace, ex- tends over that surface rather than being confined toits ventral portion as in P, mosaicus and other paturoines. Lower molars of N. matrix are similar to lower molars from the Tarkuroolow Local Fauna assigned by Flannery and Rich (1986) to Gumardee, but differ from them by being smaller in size. P3 of N. matrix has 6-7 transcristids, apparently many fewer than the P3 from the Tar- karooloo Gumardee, in which the postenor hall, the only portion recovered, has 6 transcristids, Among potoroids a dorsal enamel flange on Ti is confined to Hypsiprinneden, Potorous, bulungamayines (Flannery et al,, 1984) and Millivowi bungandiny (Flannery et al, 1992), N. matrix differs trom Hypsiprimnedon by having permanent premolars which are elongated with horizontal or concave occlusal margins rather than plagiaulacoid withconvex occlusal margins, hy failure to etun dP» after the eruption of P4 and by having less disparity between the lengths of protolophid and hypolophid on My. N. marrix differs from Potorous by having lower molars in which the buccal cuspids are positioned lingual to the adjacent lateral margin with the result that buccal crown walls are not as Sleep as in Porar- ous. N, matrix differs fram the similarly strongly bunolophodont carly Pliocene Milliyowi bun- gandit in having a strongly developed stylar cusp Con M! (absent in M, bangantij) and in lacking branches of the transeristids of Ps, The resemblance of the lı dorsal enamel lange to thatofmacropodids is Suggestive of a similarly muacropodid-like cutting action during occlusion of upper and lawer ingisors, an unusual feature in an animal with bunplophodont molars similar to those of omnivorous poloroids in which incisors perform a more forcipulate function. The posteroventrally inclined buccinator sul- cus in M. matrix was termed the ‘labial groove” by Stirton (1963) who noted it in Protemnoden and other macropodids. Woodburne (1967) re- ported a similar structure in Hadronomus puckridel, Where such a sulcus occurs among macropodids and other potoroids ib is usually closer to and parallels the alveolar margin. The reduced cuspid on the buccal flank of the large, central cuspid of the dP3 trigonid which is linked by ridges to the apex of that cuspid and to the cristid obliqua, is interpreted herein as a re- duced protostylid since i} occurs in the corre sponding position and bears the same relationship lo the cristid obliqua as do similar structures on Mi; of other species, i.e., the protostylid crest ol Wururoo dayvamayi Cooke, this volume, the dis- erete protostylid of Nambarvo saltavns Flannery & Rich, 1986, and the protostylid of Palaeaparer- ous priscus. The dominant trond cuspid fin- gually adjacent to the protostylid on dP3 of N. matrix and front which the paracristid arises is thus the protoconid and the metaconid has been lost. The loss of the metaconid of dP may be, as suggested by Ride (1993), the result of a need to supplement the shearing crest of dP2 which is shorter than the permanent premolar in this spe- cies. Apari from the discrete prorostylid rather than a protostylur ridge, the holotype tooth of P. pris» cus, designated as My (their M2) by Flannery & Rich (1986), bears strony similarities tò dP3 in paratypes of N. matrix. Undeseribed Riversleigh bulungamayines also have a protostylar ridge on dP; and a pasterobyceally inclined protolophid similar lo P. priscus. Since the latter character dees nol oceur on molar teeth ot plesiomorphic species sugh as No mereiy which have otherwise similar bunolophodont molars, it is suggested that the holotype tooth of P. priscus is dP3 rather than Mı. If this is the case, P. priscus must still be regarded as more plesiomorphic than N. matrix in view of the discrete protostylid on this tooth, but other differences in this tooth, or in molars re- ferred to this species, are here regarded as insuf- ficient to warrant the erection of anew subfamily. Subfamilial affinities of the species remain un- certain: its bunolophodont molar morphology suggests it may represent either a plesiomorphic potoroine or bulungamayine. However, the dis- crete protostylid on the holotype (dP3) is plesiomorphic and the species may prove to be basal to both these taxa. Lower molars in N. matrix are suitable to be ancestral to B. delicata. Lophids in N. matrix are clearly formed by transverse cristids extending buccally from the lingual cuspids. The posterior cingulid is enclosed by the posthypocristid which sweeps lingually posterior to the hypolophid and low on the crown before linking to the postento- cristid on the posterior of the entoconid. In B. delicata the protolophid is formed in a manner similar to that of N. matrix but joins the pro- toconid closer to its apex. The posthypocristid is more elevated on the crown, more transversely oriented and links to the entoconid much closer to the entoconid apex. The buccally oriented crest from the entoconid is reduced in length and in prominence, the posthypocristid having formed a neomorphic hypolophid. In the low-crowned upper molars of N, matrix, lophs are formed by cristae extending lingually from the buccal cusps, upper and lower molars showing reversed symmetry in this respect. Lon- gitudinal crests, notably the pre- and postproto- crista are emphasised, as they are in Gumardee pascuali, Strong longitudinal cristae characterise bunolophodont upper molars as defined by Flannery et al. (1984) who suggested that these might work in a different way to lophodont mo- lars in which transverse rather than longitudinal cutting crests are emphasised. In some undescribed plesiomorphic Riversleigh balbarines (pers. obs.) stylar cusps C and D or their stylar crests are present in the interloph region. N. marrix retains only stylar cusp C and lacks any trace of stylar cusp D. While both balbarines and bulungamayines are likely to be derived from bunolophodont ancestors, the absence of stylar cusps other than C in what is an extremely plesiomorphic bulungamayine, sug- gests that loss of other stylar cusps had already occurred in the bulungamayine ancestor which MEMOIRS OF THE QUEENSLAND MUSEUM must in this aspect at least, be more derived than that of balbarines. Ganguroo gen. nov. TYPE SPECIES. Ganguroo bilamina sp. nov. DIAGNOSIS. Bulungamayines with lower molars which are completely bilophodont, lacking any trace of a buccally-directed crest originating from the en- toconid and anterior to the hypolophid. REMARKS. Ganguroo gen. nov. differs from all potoroines, hypsiprimnodontines and pro- pleopines by having bilophodont lower molars. It differs from all macropodines and sthenurines by having a combination of: low-crowned mo- lars; finely-ridged, elongate premolars; a deeply penetrating masseteric canal confluent over its length with the inferior dental canal. It differs from all balbarines by having the elongate, finely- ridged premolars referred to above, lacking a transversely compressed trigonid on M; and in lacking a posterior cingulid on lower molars. ETYMOLOGY. Waanyi (as spoken by Ivy Stinken, formerly of Riversleigh Station) gangu, grandfather and ‘roo’ is a common Australian diminutive for kan- garoo, Ganguroo bilamina sp. nov. (Fig. 3, Table2) DIAGNOSIS. As for the genus MATERIAL EXAMINED. Holotype QMF19915 from Wayne's Wok, Hal’s Hill’ D-Site Plateau. Paratypes QMF19591, 18810, 19814, 19835, 30398, 30399 from Wayne’s Wok Site; QMF19868, 19870, 19966, 30400 from Camel Sputum Site, Godthelp’s Hill; QMF19642, 20293, 30396, 30397 trom Upper Site. Godthelp’s Hill; QMF19988 from Mike’s Me- nagerie Site, Godthelp’s Hill; QMF23777 from Bites Antennary Site,eastern part of D Site Plateau, All System B, Miocene sites (Archer et al., 1989). ETYMOLOGY. Latin lamina, blade, bilophodont lower molars. refers to the DESCRIPTION OF HOLOTYPE. Left dentary including horizontal ramus, most of angular pro- cess and part of ascending ramus. I1, P3 and M1-4 preserved. Ventral margin of horizontal ramus strongly convex, deepest below protolophid of M3, distinct digastric prominence on the ventral margin at this point. Diastema relatively short, less than 20% of length of cheek tooth row, Slender I} almost horizontal with dorsal occlusal NEW BULUNGAMAYINE KANGAROOS margin’ Well below plane of cheek tooth row, Alveolus for very small I on dorsal margin of diastema just posterjvr to margin ol I alveolus, Mental foramen close to dorsal Margin of dia» stema below anterior margin of P3, 2 very small posterior mental foramina more posieriony, | below hypolophid of Ma, the other below pr- tolaphid of M3. Very shallow sulcus for attach- ment of buccinator muscle sloping diagonally ventrally on buccal surface below My-Mz. As- cending ramus at about 105° io line of a straight odge laid across high points of occlusal surfaces of cheek tooth row, Since tooth row concave dorsally, such a line contacts posterior of P3 und hypulophid of M4. Buccal margin of masseteric fossa straight with flat area for attachment of parts of superticial layer of masseter extending an- teroventrally, Masseteri¢c canal and inferior den- tal canal confluent anterior to large masseteric foramen. Diameter of foramen and anterior con- striction of common canal suggest insertion of deep layer of masseter unlikely to have reached much more anteriorly than M3. Posterior to mas- seteri¢ foramen interior dental canal very short: masseteric foramen almost overlapped by man- dibular foramen. Lingual margin of angular pro- cess aligned with molar row. Wide. shallow basin of pterygoid fossa overhung buccally by remain- ing anterior portion of ascending ramus. Mandib- ular symphysis decreases in height posteriorty. extends to level of posterior margin of P3. Dentilion, Cheek tooth row anteropoasteriorly straight; P3 flexed slightly buccally out of align- ment, In lateral view occlusal margin of Ps above that ol molars. Molars low crowned, bilophogont- no trace of any buceally-directed crest associated with the entoconid. Molar size increases from MI to Ma; Ma is smaller than M3. l long and slender with low dorsal and ventral enamel flanges, Dorsal and ventral margins. al- mast horizontal for most of length before later converges on former al anterior end. Enamel confined to buccal surface, Cross section circular close to alveolar margin, triangular anteringly, resulting from development of rounded, longitu- dinal ridge central to lingual surface. Mi elongate, blade-like with mostly horizontal occlusal margin serrated over most of length an- terior 10 large, posterior cuspid taller than rest of tooth, Serrations formed by 6 minor cuspids, each with associated transcristids, most posterior ol these least distinct and shortest. In vcclusal view crown base lapering anteriorly and posteriorly to rounded margins. Buccal margin convex, lingual margin relatively straight. Lingual hase of crown 19 bulbous adjacent to central region of occlusal blade, forming poorly-defined, rounded lingual cingulid, Occlusal margin following midline hu posterior cuspid flexed lingually. Cristid de- scending anteriorly from apex Of most anterior cuspid to crown base and posterior margin delin- eated by similar bur shorter eristid descending [rom posterior cuspid, Mı subrectangularin occlusal ouline, narrower anteriorly than posteriorly. Lingual walls of crown subvertical, buccal side sloping more gently. Tooth worn. more wear on buccal side. Wear facet on posterior of metavonid and breaches in enamel of protacanid, hypoconid and entoconid. Metaconid taller, more angular than protoconid which has been reduced in height by wear, Sharply defined crest of protolophid de- scending buccally trom apex of metaconid to that of protoconid, Short paracyistid running anteri- orly from base of proioconid to anterior margin. Steeply descending enamel ridge on anterior mar- gin buccal to paracristid, forming margin of short precingulid. Broad anterior cingulid enclosed he- (ween paracristid and peemetacristid which runs between metaconid apex and anterior margin, Sharp postmetacristid descending posteriorly from metaconid apex to interlophid valley, meet- ing similarly well-defined preentocristid running anteriorly from apex of the entoconid, Cristid obliqua worn, running anterolingually on anterior face of hypocenid before turning anteriorly across interlophid valley, meeting posterior face of protolophid lingually adjacent to protoconicd base. Entoconid taller than hypoconid which is more worn, Crest of hypolophid crossing from posterior of hypoconid apex to meet enfoconid apex centrally, Postentocristid descending poste- rior face of entoconid but no other ornamentation of posterior Jace of hypolophid, Mo similar in vulling to Mı bur slightly larger and less worn. Meticonid taller than protocunid bin hypoconid and entoconid subequal in height. Postmeta- cristid and preentocristid not uniting in inter- lophid valley. My largest of molars, unworn, with all cuspids about equal height; lingual cuspids with more angular apices than buccal cuspids, My damaged, lacking protoconid and mostotanteriar margin. Hypoconid taller than entoconid; no pastentacrisid DESCRIPTION OF PARATYPES. Condyle pre- served in QMFI9814, 19870. 19810 and 30396, In QMF 19870 and 19814 is transversely clongare with rounded posterior margin, In QMPMI396 condyle has more circular outline, That of 290 MEMOIRS OF THE QUEENSLAND MUSEUM FIG. 3. QMF19915, Holotype of Ganguroo bilamina sp. nov. A, buccal view; B, lingual view; C, stereopair of occlusal view. Scales = 10mm. QMF 19810 slightly damaged lingually but sim- ilar to, although somewhat smaller than, that of QMF19870 and QMF19814. Differences in shape possibly age related since QMF30396 is from a subadult animal, indicated by in- completely erupted P3. Height of condyle above plane of molar row varies from 7mm in QMF 19870 and QMF30396 to 11mm in QMF19810, variation possibly being size related. Angle of ascending ramus relative to plane of molar row varies 120° (AR12517)—108° (QMF188 14). Digastric process on ventral margin of horizon- tal ramus apparently variable within species: QMF 19814 level of development comparable to that of holotype, but other paratypes show lesser or no such development. Number of posterior mental foramina also variable: most paratypes have only one such foramen, consistently located below Mz, but none present in QMF30398 while two present in QMF30400 and QMF19966. QMF19988 has number of smaller foramina ac- cessory to mandibular foramen and also has sul- cus for vessels of inferior dental canal on lingual wall of masseteric fossa. Posterior portion of inferior dental canal between masseteric foramen and mandibular foramen longer in AR12517 than in holotype and most other paratypes. QMF30400 and QMF19988 have direct opening via single foramen from pterygoid fossa into masseteric fossa with no intervening canal (the condition usual among extant macropodoids). Damage to ventral margin of horizontal ramus reveals extent of anterior insertion of masseter, in QMF19868 and QMF20293 it reaches level of M2 hypolophid, but only to level of M3 hypolophid in QMF30397. DENTITION. dP2 and dP3 in QMF19835, 23777. dP2 small, blade-like with rounded anterior and posterior margins, strongly convex buccal mar- gin and straight lingual margin. Occlusal crest serrated over much of length anterior to large posterior cuspid overhanging posterior base of crown. QMF19835 has 3 small cuspids in ser- rated region, each with associated transcristids; 4 NEW BULUNGAMAYINE KANGAROOS 291 TABLE 2. Dental parameters for type specimens of Ganguroo bilamina sp. nov. such cuspids in QMF23777. Anterior and poste- rior margins of crown delineated by vertical cristids descending from ends of occlusal crest. Occlusal crest runs slightly lingual to midline, lingual surface of crown more steeply inclined than buccal. dP3 better preserved in QMF23777 and used as basis for description below. Crown base roughly rectangular in occlusal outline, narrowing some- what anteriorly. Protolophid extremely laterally compressed, inclined posterolingually, domi- nated by tall protoconid with thick, rounded pro- tostylid crest descending its buccal flank. Metaconid cannot be distinguished from pro- toconid. Prominent paracristid (less so in QMF19835) runs anteriorly to tall paraconid (shorter in QMF19385) on anterior margin. Paraconid, paracristid and protolophid form blade-like crest complementing that of dP2. Ver- tical cristid descends from posterior margin of protolophid crest to interlophid valley and is con- tacted by anteriorly directed preentocristid in QMF19835, but not in this specimen. Hypolophid transversely oriented, concave in posterior view. Cristid obliqua runs anterolingu- ally on anterior face of hypoconid, turning ante- riorly across interlophid valley and contacting protostylid crest. No ornamentation on posterior face of hypolophid. Froets [11922 eee E ai Ka oe o |__|3.9/2.8]2.9|3.7]2.8|2.8]/3.8]2.6/27 mem aa atata T T A AA [F19868 10.921] | | | [sefat[37]s] | | | | | es[2s]2s8|38]29]29] | | | F19870) | | | | | | | | | [42|26[28[40]28]3.0[3.8]3.1[31[4.0]29]28] | | | [F196] | | | | | [59/31[39] 6 [s4[26]28[3.8]28[3.0]39]3.1]29]3930]27 | | | [Frovss| | | | | | | | | | | TJ | 1 | | J f2s[3e[28|28|36]29]25] Fos pa | | | eja fel | | [sa[aelze[a7ja7i27|sei2aja71 | |_| [Faose7| | | | | | j63fasj37[ 7| | | |35/23/25/3.9]2.7/26]3.9/2.9/28/3.9]27[2.5| [Fiea2] | | | | | [sel2s[se] 6] | | Psp | | [| | ft Tt | eee [Fsoseo| | | | | | jezj2s{se/7] | | [35[22|27[35]25/29/36|27/28/3.8/2.7|27| fracas! | [| TJesfsojas| e| [| [ss[2e|27|s6|s0|29[38)30|28|35|25|23 msa O E e e T oean T T F110] | | | | | | T | | | | T [3s[26|26/3.8/28[2.8]3.8[27]29/3.7[28] - | [F181] | | | | T Tselzte]| | | Pej2sj2sjsz[26j30] | | | TT] [F19835| | [3430|36] 4 [5728|40] 5 |29/23/23]/31[2s/27] | | T | | [ TT] | MEAN [11.3 2.0|3.8|2.8/3.4|3.5/5.9|2.9|3.8] 6 |3.5|2.4|2.8|3.6|2.6| 2.7 |37 |2.9|2.9| 38| 3.0] 2.8] 3.7|2.9] 2.6] | sd [49] 3] 6] 4] 4] 7] 3[3] 2] 7] 5] 3[ 4] 2] 4] 2] 4] sf 1] 3] 3] 4[ 3] 4] 2] P3 in most paratypes closely resembles that of holotype but QMF30399 has 7 minor cuspids rather than 6. Molar morphology very similar to that of the holotype although variable postentocristid be- tween different specimens and between different teeth in single specimens. DISCUSSION The horizontal orientation of I is similar to that in macropodines in which there is considerable ventral flexion of the rostrum, necessary to bring upper and lower incisors into occlusion and there would presumably be a corresponding flexion of the rostrum in this species. dP3 is very similar to that of N. marrix but is more derived in that the reduced protostylid of N. matrix is here further reduced to a protostylid crest. Molars in this species are more derived than in either N. matrix, B. delicata or Wabularoo naughtoni because they are lophodont. N. matrix, B. delicata and G. bilamina represent stages in an evolutionary sequence in which a bunolophodont, omnivorous ancestral form is changed to that of a lophodont herbivore (Fig. 4). Hypolophid morphology is particularly informa- tive in this respect. As discussed earlier, a neomorphic hypolophid has been developed in B. delicata by elevation of the posthypocristid on the crown and directing the posthypocristid more transversely. Hypolophid morphology in W. naughtoni closely resembles that of B. delicata. The bunolophodont origin of this morphology is indicated by the reduced buccal crest from the entoconid anterior to the new hypolophid, repre- senting the remnant of the original hypolophid crest. No trace of this crest is evident in G. bilamina, the neomorphic hypolophid crest being formed entirely by the elevated, transverse posthypocristid as indicated by the presence of a postentocristid on the posterior face of the en- toconid. Loss of the buccally-directed crest from the entoconid represents a subtle change in mor- phology between N. matrix and G. bilamina but a highly significant apomorphy. The evolutionary series represented by these bulungamayine taxa demonstrates that lophodonty evolved independently twice among Oligocene-Miocene kangaroos - once in balbarines in a process which seems to have been essentially that proposed by Flannery & Rich (1986) and once among bulungamayines using the mechanism proposed above. While Flannery (1989) suggested that balbarines were ancestral to macropodines and sthenurines, the similarity of premolar and molar morphology of derived bulungamayines such as G. bilamina to that of the later Miocene macropodids from Alcoota is greater than that of more derived balbarines such as Balbaroo in which on M; there is still consid- erable lateral compression of the protolophid and little development of the anterior cingulid. The premolar of balbarines is also much shorter than that of bulungamayines and the plesiomorphic Alcoota macropodids (Cooke, 1997). Lower molar morphology of G. bilamina has strong similarities to that of the much larger Hadronomus puckridgi Woodburne, 1967 from Alcoota which Murray (1991) regarded as a plesiomorphic sthenurine. Both species are low- crowned and bilophodont, have long anterior cingulids, have Mı protolophids which are not laterally compressed and lack posterior cingulids, although Hadronomus has a bulbous base to the hypolophid. Hadronomus also has an elongate premolar, resembling in that respect the premolar of bulungamayines, but that of Hadronomus is more coarsely serrated than that in any of the known bulungamayine species and bears well developed lingual and buccal cingula, not present in bulungamayines. However, paratypes of N, matrix show variable differentiation of minor cuspids and transcristids on P3, indicating some lability in degree of serration of the occlusal MEMOIRS OF THE QUEENSLAND MUSEUM margin of this tooth in bulungamayines. The bul- bous base of the bulungamayine P3 could serve as an adequate precursor of lateral cingula (a lingual cingulum is poorly developed on P3 of G. bilamina), The premolar of all known balbarines is in contrast a shorter, more plagiaulacoid tooth. Similarities also exist between dental morphol- ogy in G. bilamina and in Dorcopsoides fossilis, also from Alcoota. While this species was origi- nally included within Potoroidae, Bartholomai (1978) placed it in Macropodinae. Both species have elongate premolars. Lateral cingula are lacking in P3 of Dorcopsoides while a lingual cingulum is poorly developed in that of G. bilamina and there are again differences in serra- tion and transcristids between the two species. dP3 in N. matrix and B. bilamina has some similarity with that of Dorcopsoides in that the metaconid is reduced or absent in each. Woodburne (1967) also noted the ‘fused pro- toconid and metaconid’ of dP3 in Dorcopsoides and ‘a short posterolabial crest ... which turns abruptly posteriorly before descending into the transverse valley and continues posterolabially up the anterior face of the hypoconid’. This crest may be homologous with the protostylid crest which is linked to the cristid obliqua of dP3 in N. matrix and G. bilamina but which is also present on dP3 in undescribed Riversleigh balbarines ref- erable to Nambaroo and in which there is also considerable abbreviation of the protolophid (pers. obs.). Ride (1971) suggested that close proximity of the protoconid and metaconid on dP3 is plesiomorphic for macropodoids (his macropodids), and the protostylid or its reduced form of a protostylid crest in both potoroids and macropodids suggests that this character is simi- larly plesiomorphic. While lower molar morphologies in G. bilamina and Dorcopsoides are similar in many respects, they differ markedly in that Dorcopsoides has a well-developed posterior cingulid, absent in all bulungamayines but pres- ent in balbarines. Derivation of Dorcopsoides from a bulungamayine ancestor would require development of a neomorphic posterior cingulid, such development possibly indicated by the swollen hypolophid base of Hadronomus. Evolu- tion from a balbarine ancestor would require modification to both the anterior cingulid and compressed protolophid of Mj, but modification of pre-existing structures is a more usual evolu- tionary phenomenon than the development of new structures. This notwithstanding, dental morphology in bulungamayines is such that, on NEW BULUNGAMAYINE KANGAROOS 293 Henk Godthelp for encourage- ment and assistance, Anna Gil- lespie, Steph Williams and others at the University of New South Wales for preparation, FIG, 4, Development of lophodonty in bulungamayines, illustrated hy RM), A, Nowidgee matrix , B, equivalent to B. delicata. C, G. bilamina, Abbre- viations: Pr=protoconid, Me=metaconid, med=metacristiil, HY=hypoconid, Ec=entoconid, ecd=huccal crest from entoconid, phe=posthypocristid, pec=postentocristid, co=cristid obliqua. the grounds of parsimony, they, rather than balbarines, must be preferred as the group most closely ancestral lo macropodids. In the hypothesis of molar evolution within Bulungamayinae advanced herein, there is a tran- sition from a potorojd-like molar in basal species lo a macropodid-like molar in derived species. Such a transitional sequence within the group may explain the differing views of familial affin- ity of bulungamayines (Case, 1984; Woodburne, 1984; Flannery et al., 1984). At the time their respective Views were advanced, only 2 bulunga- mayine species, B, delicata and W. naughtoni. had been described, Molar morphology in both those species is intermediate in the transitional sequence and it is not surprising that both macropodid and potoroid affinities could be ar- gued on the basis of these species. If, as seems likely from the evidence provided herein. that bulungamayines are directly ances- tral to macropodids, then monophyly of Bulunga- mayinae cannot be stated with certainty, Further doubts also arise concerning monophyly of Macropodidae. ACKNOWLEDGEMENTS Research grants from the Australian Research Council and the University of New South Wales have been the primary mechanism for providing the material studied. Additional support for the Riversleigh project has come from the National Estates Program grants, the Australian Geograph- ical Sociely, The Australian Museum, The Riversleigh Society, ICI Pry Ltd, Century Zine Limited, the Mt. Isa Shire and private donors. Ithank the Director and staff of the Queensland Museum for facilities and assistance, Jeff Wright tor photographic assistance, Mike Archer and LITERATURE CITED APLIN, K.P. & ARCHER, M. 1987. Recent advances in marsupial systematics with a new syneretic Classification. Pp xv-Ixxiiin Archer, M. (ed) ‘Pos- sums and Opossums: Studies in Evolution’. (Sur- rey Beatty & Sons: Sydney). ARCHER, M. 1979. Wabularoo naughtoni gen. et sp. nov., an enigmatic kangaroo (Marsupialia) from the middle Tertiary Curl Creek Limestone of northwestern Queensland. Results of the Ray E Lemley Expeditions, part 4. Memoirs of the Queensland Museum 19: 299-307. ARCHER, M. 1984. The Australian marsupial radia- tion, Pp. 633-808 In Archer, M, & Clayton, G (eds) ‘Vertebrate Zoogeography and Evolution in Australasia’, (Hesperian Press: Perth, Western Australia). ARCHER, M. & BARTHOLOMAT, A. 1978. Tertiary mammals of Australia, a synoptic review. Al- cheringa 2: 1-19, ARCHER, M.: & FLANNERY. T.F. 1985. Revision of the extinct gigantic rat kangaroos (Potoroidag: Marsupialia), with a description of a new Miocene genus and Species and a new Pleistocene species of Propleopus, Journal of Paleontology 59, 1131- 1149. ARCHER, M., GODTHELP, H.. HAND, S.J. & MEGIRIAN, D. 1989, Fossil mammals of Riversleigh, northwestem Queensland: prelimi- nary Overview of biostratigraphy, correlation and environmental change. Australian Zoologist 25(2); 29-63, BARTHOLOMAI, A. 1978. The fossil kangaroos. Aus- tralian Mammalogy 2: 15-22. CASE, J,A. 1984, A new genus of Potoromae (Marsupialia: Macropodidae) from the Miocene Ngapakaldi Local Fauna, South Australia, and a definition of the Potoroinac. Journal of Puleontal- ogy S8(4): 1074-1086. 294 COOKE, B.N, 1992, Primitive macropodids from Riversleigh, northwestern Queensland. Al- cheringa 16; 201-217, 1997. Two new balbarine kangaroos and lower molar evolution within the subfamily. Memoirs of the Queensland Museum 41: 269-280. FLANNERY, T.F. 1989. Phylogeny of the Macropodoidea; a study in convergence, Pp 1-46. In Grigg, G., Jarman. P & Hume, |. (eds) “Kanga- roos, Wallabies and Rat-Kanguroos’. (Surrey Beatty & Sons: Sydney). FLANNERY, T. & ARCHER. M. 19872. Hypsiprimneden bartholomaii (Potoroidae: Marsupialia), a new species from the Miocene Dwamamor Local Fauna and a reassessment ot the phylogenetic position of H. moschatus, Pp 749-758. In Archer, M. (ed), Possums and opos- sums; studies in evolution, (Surrey Beatty & Sons: Sydney). 1987b. Bettongia moyesi, a new and plesiomorphic kangaroo (Marsupialia: Potoroidae) from Miocene sediments of northwestern Queensland, Pp 759 -767, In Archer. M, (ed) ‘Possums and Opossums: Studies in Evolution’. (Surrey Beatty & Sons: Sydney). FLANNERY, T.F. & RICH, T.H.V. 1986. Maeropodoids from the Middle Miocene Namba Formation, South Australia, and the homology of some dental structures in kangaroos. Joumal of Paleontology 60(2): 418-447. FLANNERY, T.F., ARCHER, M. & PLANE, M. 1982. Middle Miocene kangaroos (Macropodoidea: Marsupialia) from three localities innorthern Aus- tralia, with a description of two new subfamilies. Bureau of Mineral Resources Journal of Austra- lian Geology and Geophysics 7: 287-302. 1984. Phylogenetic relationships and a reconsidera- tion of higher level systematics within the Potoroidae (Marsupialia), Journal of Paleontol- ogy 58(4): 1087-1097, FLANNERY, T.F., RICH. T.H., TURNBULL. W.D. & LUNDELIUS, E.L., 1992, The Macropodoidea (Marsuptalia) of the early Pliocene Hamilton Local Fauna, Victoria, Australia, Fieldianea Geol. 25: 1-37, FLOWER, W.H, 1867. On the development and suc- cession of teeth in the Marsupialia. Philosophical MEMOIRS OF THE QUEENSLAND MUSEUM Transactions of the Royal Society of London B 157: 631-641. LUCKETT, W.P. 1993. An ontogenetic assessment of dental homologies in therian mammals, Pp 182- 204. In Szalay, F.S, Novacek, M.J. & McKenna, M.C. (eds) ‘Mammal Phylogeny’. (Springer-Ver- law: New York). MURRAY, P.F. 1991, The sthenurine affinity of the Late Miocene kangaroo, Hadrenomas puckridgt Woodburne (Marsupialia, Macropodidae). Al- cheringa 15:255-283. RIDE, W.D.L. 1971. On the fossil evidence of the evolution. of the Macropodidae. Australian Zool- ogist 14(1): 6-16. 1993, Jackmahoneva gen, nov. and the genesis of the macropodiform molar, Memoirs of the Asso- ciation of Australasian Palacontologists 15: 441- 459, STIRTON, R.A. 1963. A review of the macropodid genus Protenmodon, University of California Publications in the Geological Sciences 44:97- 162, SZALAY, F.S. 1969. Mixodectidae, Microsyopidae, and the Insectivore-Primate transition, Bulletin of the American Museum of Natural History 140. 193-330, TEDFORD, R.H. & WOODBURNE, M.O. 1987. The Nariidae, anew family of vombatiform marsupials from Miocene strata of South Australia and an evaluation of the homalogy of molar cusps in the Diprotodontia. Pp 401-418, In Archer, M. (ed) ‘Possums and Opossums: Studies in Evolution’. (Surrey Beatty & Sons; Sydney). WOODBURNE, M.O. 1967. The Alcoota Fauna Cen- wal Australia: an integrated pulacontological and geological study, Bureau of Mineral Resources, Geology and Geophysics Bulletin 87, 1984. Wakiewakie lawsoni, anew genus and species of Poturoinae (Marsupialia: Macropodidae) of medial Miocene age, South Australia, Journal of Paleontology 58(4); 1062-1073. WOODBURNE, M.O., MACFADDEN, B.J.. CASE, J,A,, SPRINGER, M.S., PLEDGE. N,, POWER, J.D.. WOODBURNE, J.M. & SPRINGER, K.B. 1993, Land mammal biostratigraphy and mag- netostratigraphy of the Etadunna Formation (late Oligocene) of South Ausiralia, Journal of Verte- bråte Paleontology 14; 483-515. BIOSTRATIGRAPHIC IMPLICATIONS OF FOSSIL KANGAROOS AT RIVERSLEIGH, NORTHWESTERN QUEENSLAND B.N, COOKE Cooke, B.N, 1997 06 30; Biostratigraphic implications of fossil kangaroos at Riversleigh, northwestern Queensalnd. Memoirs of the Queensland Museum 41(2): 295-302. Brisbane. ISSN 0079-8835, Kangaroos form an important component of the faunal assemblages of Riversleigh and most other Australian Cainozoic fossil deposits. Attempts to use fossil kangaroos to determine the relative ages of Riversleigh sites suggests that overall faunal composition may be & more useful guide to relative age than presence or absence of purlicular species, Marked changes in kangaroo faunal composition occur between Riversleigh System B and C sites with the apparent extinction of most balbarine species and rise to dominance by lophudunt bulungamayine species. This change correlates with climatic decline following mid-Mioccne climatic optima. Approximate age equivalence is suggested between Riversleigh System B sites and faunal zones B +C (Woodburne et al., 1993) of the Etadunna Formation, South Ausiralia, Riversleigh’s System A sites are older. The more derived, lophodonr bulungamayines of Riversleigh System C are considered potentially ancestral to plesiomorphic macropodids such as Hadyonemas from the Alcoota Formation. Kangaroos Support an age span for pre-Pliocene deposits at Riversleigh that extend from the lale Oligocene to late middle Miocene and possibly early late Miocene. [|] Kangaroos, Oligocene, Miovene, Pliocene, Riversleigh. B.N, Cooke, School of Life Science, Queensland University of Technology, GPO Bax 2434, Brishane, Queensland 400), Australia; I4 December 1996. Tedford (1967) provided the first description of kangaroos Irom Riversleigh. Since then the area has yielded many thousands of kangaroo fossils representing at least 32 new species, Archer (1979) described Wabularea nanghtont from D Site. Flannery et al. (1982) described Bulungamaya delicata (erecting the Bulungamayinae), W. naughront, Balharoo eregoriensis from Riversleigh, B. camfieldensis fromBullock Creck. Northern Territory (middle to Late Miocene) and a single specimen of Balbaroe trom Kangaroo Well. The species of Balbaroo were placed in a new macropodid Balbarinae. They also described the potoroine Gumardee pascuali and macropodid Galanarla ressellata from Riversleigh, Archer & Flannery (1985) described the Riversleigh propleopine Ekaltadeta ima, Flann- ery & Archer (1987a) described Hypsipryhin- odon bartholomait and (1987b) Bettongia moyesti, the only pre-Pliovene representatives of these genera, Cooke (1992) deserbed the balbar- ine Ganawamaya with G. acris, G. aedicilusand G. ornata, The number of Riversleigh fossil kangaroo specimens, ranging from) isolated teeth, isolated postcranial elements, maxillary and dentary frag- ments to complete skulls is almost overwhelm- ing. L have concentrated on the more complete remains available. These indicate a further 21 new species, 4 number of which are described in this volume. Praremnodon sp. has been recorded from the Pliocene Rackham’s Roost Site and Macropus agilis has been recorded from the Pleistocene Terrace Site (M. Archer pers. com.). This brings the total Riversleigh macropodoid fauna to 34, including the Potoroinae, Hypsi- prymnodontinae, Propleopinue, Balbarinac, Bulungamayinae and Macropodinae. Archer ct al. (1989) recognised more than 100 local faunas from the Riversleigh fossil area, with ages estimated to range from Oligocene-Mijocene to Holocene. (Discoveries since 1989 have raised that number to more than 150 sites with faunal assemblages). They grouped the Oligocene- Miocene sites into three ‘Systems’ designated A-C, with System A sites regarded as oldest and System C sites as youngest, Megirian (1992) treated the entire sequence of fossiliferous Limestone overlying the Thorntonia Limestone ws comprising the Carl Creek Lime- stone. He used the ‘Systems” of Archer et al. (1989) in a purely biostratigraphic sense and has later (1994) challenged the use of the System concept on the grounds that the terminology has been unsatisfactorily defined and that there is no precedent lor such usage in lithostratigraphie no mencliature, MEMOIRS OF THE QUEENSLAND MUSEUM TABLE 1. Sites of occurrence and numbers of specimens of each identified pre-Pliocene Riversleigh macropodoid species. QL=Quantum Leap; WH=White Hunter; DU=Dunsinane; COA=Cleft of Ages; BA=Bitesantennary; DT=Dirk’s Towers; OUT=Outasite; WW=Wayne’s Wok; BO=Boid; CS=Camel Sputum; IN=Inabeyance; MM=Mike’s Menagerie; UP=Upper; CR=Creaser’s Ramparts; N’sG=Neville’s Garden; LM=Last Minute; FF=Fireside Favourites; H’sH=Henk’s Hollow; TT=Two Trees; DO=Dome; J’sC=Jim’s Carousel; ENC=Encore; ?=uncertain position within the System sequence of Archer et al. (1989). jal. tessellata Gan. acris an. aediculis Gan. omata l Gan.acris | Gan. aediculis | [Gan omata | Gan sp4 | p Wur. dayamayi Wur. sp.2 w g = Ey S g le ke te ke is tela jE a lam. couperi jam. sp.7 Tu} Archer et al, (1989) Archer, Hand and Godthelp (1991) and Archer et al. (1994) interpreted the older Riversleigh local faunas, those from sites in Systems A-C, as representing rainforest commu- nities. This interpretation was made on the basis of characteristics of those faunas such as: high species diversity; presence of complex feeding guilds of small, sympatric mammals; high num- bers of obligate leaf eaters in single local faunas, indicating high tree species diversity; presence of high numbers of known rainforest taxa; high numbers of browsing marsupials but absence of grazers. Megirian (1992) disputed the hypothesis of a rainforest palaeohabitat for the older Riversleigh local faunas on the grounds of his interpretations of drier, even semi-arid environmental conditions prevailing during the genesis of the fossiliferous limestones. He suggested instead that the high species diversity evident in the local faunas re- sulted from a combination of resident rainforest communities inhabiting rainforest refugia con- fined to permanent water bodies, and the use of 7B intr.| LowC | Mid C such permanent water by animals drawn from more distant, mesically adapted communities. None of the Pre-Pliocene kangaroos from Riversleigh exhibit dental adaptation for a graz- ing habit which might be expected if they were drawn from a more arid environment. Regardless of their palaeohabitat, kangaroos have a wide- spread occurrence and are abundant among the many local faunas now known from Riversleigh. They should prove to be important in attempts to determine relative ages of those assemblages. This paper presents the results of a preliminary attempt to use kangaroo fossils to assess relative ages of sites within Riversleigh, assess the age of Riversleigh deposits relative to those of other Australian Tertiary sites which have yielded comparable kangaroo fossils, and to correlate changes apparent within the macropodoid fauna of Riversleigh with Tertiary climatic and geo- logic events, As part of that attempt, the Systems concept of Archer et al. (1989) has been used as an hypothesis which can be tested using the dis- BIOSTRATIGRAPHIC IMPLICATIONS OF FOSSIL KANGAROOS tribution of fossil kangaroo species in sites pre- viously assigned to each of systems A, B and C. INTER-SYSTEM COMPARISONS AND OTHER RELATIONSHIPS AT RIVERSLEIGH Pre-Pliocene Riversleigh sites that have yielded identified macropodoid species, together with the number of specimens of each, are shown in Table 1. Column headings in this Table follow Archer et al. (1994) who listed sites known to that date, indicating those confidently assigned to Systems A-C, provided indications of possible System affinities of some sites, and estimated ages of other sites whose faunal composition differs from those allocated to these Systems. As indicated in Table 1 and discussed further below, evidence drawn from the distribution of kangaroo fossils is not in complete agreement with ages suggested by Archer et al. (1994) for some Riversleigh sites. Site G of Flannery et al. (1982) is included within System A on this table since it is described by those authors as lying within the Carl Creek Limestone. Late Oligocene (Archer et al., 1995; Myers & Archer, this volume) System A may have been a time of ‘icehouse conditions’ with relatively low temperature, rainfall and biodiversity (Frakes et al., 1987). Five macropodoid species have now been iden- tified in System A sites. Of these, Balbaroo eregoriensis and Wabularoo naughtoni are known also from System B and Bulungamaya delicata occurs in sites in all 3 Systems. Gumardee pascuali and Galanarla tessellata are so far known only from System A. Unfortunately molar teeth preserved in the holotypes of the last two species are badly worn and/or damaged, re- ducing their usefulness for biostatigraphic analy- sis. However, lower molars of Galanarla tessellata exhibit a well-developed posterior cingulid linked to a postentocristid, a feature typ- ical of balbarines (see Cooke this volume), but not present among bulungamayines. G. tessellata is here assigned to the Balbarinae. As noted by Flannery et al. (1982), Gumardee pascuali is of comparable size to Wabularoo naughtoni, a common variable species within Riversleigh’s Systems A and B. G. pascuali is considered to fall within the range of variation observed among specimens of W. naughtoni. Nambaroo, Wururoo (Cooke, this volume) and Ganawamaya, Balbaroo sp.2, Nowidgee matrix (Cooke, this volume) and Ganguroo bilamina 297 (Cooke, this volume) all occur in System B, but not in System A or C. These taxa in newly dis- covered sites of unknown relative age may there- fore be suggestive of an age comparable to other System B sites. Occurrence of B. delicata in all three systems and of W. naughtoni and B. gregoriensis in Systems A and B suggests that caution is indicated before declaring any of the above taxa, so far found only in System B, as definitive indicators of that System. Balbaroo gregoriensis, for instance, is more derived in molar morphology than any Nambaroo and it seems likely that representatives of the latter plesiomorphic balbarine and perhaps of others, such as Wururoo and Ganawamaya, may also ultimately be found in System A. If System B sites are correctly interpreted as early Miocene in age (Archer et al., 1994, 1995), remains found in those sites may have accumulated during ‘green- house conditions’ with high temperatures, rain- fall and biodiversity (Frakes et al., 1987). The macropodoid fauna of System C is more distinctive than those of Systems A and B, con- taining representatives of 5 subfamilies: 4 species of Bulungamayinae and 1 each of Hypsiprymno- dontinae, Potoroinae, Propleopinae and Balbarinae. Of the 8 macropodoid species known from System C, Bulungamaya delicata is the only one occuring in other Systems. Of the remainder, Wan (which includes the ‘Gag Site macropodine’ of Flannery, 1989) and Ganguroo sp.2 are among the most highly derived bulungamayines and Balbaroo sp.4 is more derived in molar morphol- ogy than any other known balbarine species. Presence of these species in particular in any given site may suggest but not define an age comparable to, or perhaps younger than that of System C sites. Archer et al. (1994, 1995) sug- gested that System C is middle Miocene. If this is so, this interval was also characterised by ‘greenhouse conditions’ (Frakes et al., 1987). Overall macropodid faunal composition, rather than presence of particular species, may provide amore reliable guide in assessing relative ages of Riversleigh sites. If Galanarla tessellata is ac- cepted as a balbarine and Gumardee pascuali as a bulungamayine, these subfamilies are repre- sented in System A by 5 species, roughly equally divided among between the two: 3 bulunga- mayines versus 2 balbarines, with 7 identified specimens from each subfamily. System A de- posits may thus be characterised by roughly equal diversity and abundance of balbarine and bulungamayine species. 298 MEMOIRS OF THE QUEENSLAND MUSEUM The diversity of macropodo species is greal- ésLin absolute terms in System B. Balbarines and bulungamayines are both present, with balbarines dominating in terms of numbers of species (13 versus 4), If numbers of identified specimens are taken as a crude guide to abundance within spe- cies, bulufigamayines appear to have heen more abundant, the 3 species of lophodont bulunga- mayines beng represented by a total of 102 spec- imens, compared ta a total of 32 halbarine specimens from 12 species, System B deposits may thus be characterised by high species diver- sity of balharines with accompanying low species abundance, und by a lower species diversity of bulungamayines hyi a relatively higher abun- dance of members uf those species. Sublumilial diversity is greatest in System C, but Potoroinae, Hypsipryinniodontinae and Pro- pleopinae are cach represented by single species, Among the lophudunt species, bulungamayines have gaiped the ascendanty over balbarines in terms of numbers of species (4 versus 1) and iñ relative abundance (16 wWentlied specimens ver- sus 3), System C may thus be characterised hy domi- nance of lophodont bulungamayines, low inci- dence of balbarines and possible fypsiprymno (onnnes, propleopines and potorvines, although undescribed propleopine remains are known from sites such as Dirk's Towers which is prob- ably equivalent in age to system B. Lower absolute diversity in System A deposits may resull from cooler, drier climatic conditions, exacerbated hy the lower number of System A sites (2) yielding identified macropadoid species, compared tù the number of System B sites (10) yielding such species, Macropodeid diversity within System Ais thus likely to have been higher than that Currently known, System B deposits are suggested to have accu- mulated in pools or shullow lakes and System C deposits in deep pools, shallow pools or emergent accreting surfaces, or cave oulwashes (Archer et ul, 1994), Depositional environments are thus more comparable for Systems B and C. The prob- abilities of accumulating remains of terrestrial maniinals are also likely to be comparable for (hese systems. Differences in overall macropodoid faunal composition between these two Systems are therefore more likely to be a true reflection af conditions prevailing during the times of deposition of these Systems. There is a striking change in macropodoid fau- nal composition hetween Systems B and C With the exception of a Single, highly-derived lapho- dont, Balbaroo sp.4, browsing balburines are ap- parently absent from System C, as is Wabularene nauehtent which persists through Systems A and B. Bulungamaya delicate persists only in sites low in the sequence of System C deposits, System B omnivores, such as Nowidgee. are replaced in System C by Bettongia moyesii and Aypsiprym- nodon bartholomaii. Compared to System B, System C macropodoid assemblages are depau- perate in numbers of species and dominated by larger, derived, lophodont bulungamayines whose long premolars and gencral molar mor- phology bear strony similarities to those of plesiamorphic macropodids. The System B mucropodoid assemblages have a high species diversity, particularly among balbarines. High faunal diversity is a characteristic of ruinfneest habitats, suggested by Archer et al. (1989) to be the habitat for older Riversleigh local faunas Cin- cluding those from System C), The decline in overall magropodoid diversity evident within System C and the dominance of larger, lopho- dont hulungamayine species suggests that rainforest habitat indicated by Archer et al. (1989) may have been in decline during accumu- lation of System C. Thal this was not a sudden event is indicated by the persistence ol the pre- sumably rainforest adapted A. dèliċata into the lower levels of System C al Gag Sile, the occur- rence at this same site of Hypsiprymnodon, mod- em representatives of Which are confined to rainforestin northern Queensland, and the occur- rence there of a high diversity of possum species (Archer ct al., 1997), Archer el al. (1994, 1995) estimated the age of System B as early Miocene and System C as middle Miocene, McGowran & Li (199-4) corre- jale Lhe planktonic foraminifera record of south- ern Australia, oceanic d'O levels and sea level fluctuations and indicate generally Warmer, Wet- ler climatic conditions during the Oligocene- carly Miocene, with 3 warm and moist climate optima occurring during the Miocene. Two of these occur during the early Miocene during the Janjukian and Longtordian stages respectively, and a double-peaked optimum occurred during the early middle Miocene, corresponding with Batesfordian and Baleombian stages, 16-| Srna, Following the latest of these climatic optima there is 4 gencral and world Wide decline towards a cooler, drier climatic regime associated with a ‘rowerse greenhouse’ effect. In northern Australia the effects of this decline would have been exac- erbated by the middle Miocene uplill of the New Guinea Highlands. Archer etal. (1989) have sug- BIOSTRATIGRAPHIC IMPLICATIONS OF FOSSIL KANGAROOS gested these newly upthrust mountains could have created a ‘rain shadow’ effect across north- ern Australia which may have been one of the most important factors causing the decline of central and northern Australian rainforests. The climatic decline following the early middle Miocene climatic optimum coincides well with the early middle Miocene estimate of Archer et al. (1994, 1995) for the System B/C boundary. It is therefore likely that the decline in macropodoid diversity at the System B/C boundary is a reflec- tion of the combined effects of the geological and climatic phenomena outlined above. A number of Riversleigh sites are of uncertain stratigraphic relationship. The local biostrati- graphic implications of the macropodoid taxa which have been found in some of these are discussed below. The low stratigraphic position of White Hunter Site and its unusual faunal assemblage make it uncertain whether it belongs to System A or B. Of its 7 identified macropodid taxa found, 5 are unique to the site and, of themselves, do little to settle the question either way. Of the remainder, the plesiomorphic balbarine, Nambaroo sp.8, is found elsewhere only at Dunsinane Site whose relative age is also uncertain (but see Arena, this volume). Nowidgee matrix, is found only within System B at sites in both lower and higher levels of the sequence. The occurrence of this species and the overall composition of the macropodoid fauna of White Hunter Site — dominated by plesiomorphic balbarines with 2 plesiomorphic bulungamayines, suggests that the site is possibly a basal member of the System B sequence. The 2 species of Nambaroo found in this site are ex- tremely plesiomorphic balbarines (Cooke, this volume) and the species of Nowidgee found there are similarly plesiomorphic bulungamayines (Cooke, this volume). Occurrence of such plesiomorphic species in the one site sugests that White Hunter may be even older, perhaps belong- ing to System A (Creaser, this volume), The latter interpretation is supported by Myers & Archer (this volume) who report the occurrence at White Hunter Site of Kuterintja ngama, an ilariid con- specific with one in the Mammalon Hill Local Fauna of central Australia that is dated as 24myo (late Oligocene) by magnetostratigraphy. The occurrence of N. sp.8 at Dunsinane Site complicates rather than clarifies understanding of this already enigmatic site in which are preserved plant material, insects and fossil bone. Dunsinane occurs in an area close to the boundary of the Tertiary limestone and Precambrian quartzite. 299 The occurrence of this plesiomorphic balbarine suggests the site may be equivalent in age to White Hunter Site, but there are several reasons for caution. No other vertebrate remains support- ing this age determination have so far been iden- tified from this site. All vertebrate remains here are fragmented and poorly preserved and may well have been re-deposited after previous re- working. Bitesantennary Site and Dirk’s Towers Site are both intrusive deposits on the D Site Plateau, regarded as probably equivalent to System B assemblages (Archer et al., 1994), Only a single macropodoid, Ganguroo bilamina, has been identified from Bitesantennary Site, but this spe- cies is otherwise known only from System B. Dirk’s Towers Site has 3 macropodoids, includ- ing Bulungamaya delicata, known from Systems A, B and C. Balbaroo gregoriensis is known from both Systems A and B. However, Nambaroo sp.5, is known only from System B. On balance, macropodoid fauna of Dirk’s Towers Site supports a System B age. Archer et al. (1994) suggested that the faunal assemblage of Quantum Leap warrants its likely inclusion in System A. Only 3 macropodoid spe- cies are so far known from this site. Bulungamaya delicata is uninformative since it occurs in all 3 Systems. However, 2 species of Nambaroo are known from the site, both species otherwise known from System B sites or those likely to be of equivalent age, e.g., Dirk’s Towers and Neville’s Garden Sites. The macropodoid fauna of this site therefore suggests a closer affinity with System B rather than System A. Neville’s Garden Site is a possible cave out- wash deposit considered to be equivalent in age to System B (Archer et al., 1994). Macropodoids of this site include the ubiquitous B. delicata, W. naughtoni and Balbaroo gregoriensis, the latter two occurring in systems A and B, and 2 species of Nambaroo known from both upper and lower levels of System B. The occurrence of the latter species and the typical System B composition of the Neville’s Garden macropodoid fauna support a System B age for this site. Dome Site, Jim’s Carousel Site and Cleft of Ages Site have all been suggested to be younger than System C sites (Archer et al., 1994). Gen. Wan. sp.1 (the ‘Gag Site macropodine’) is the only macropodoid so far identified from Dome Site. It is known elsewhere only from Gag Site in System C and Encore Site of possible late Miocene age (Archer et al., 1994). Ganguroo sp.2 is the only macropodoid so far known from Jim’s 30) Carousel Site. This species is also known from Gag and last Minute Sites, low in the sequence of System C deposits, and from Henk's Hollow Site in the higher levels of thal sequence, Roth species are highly derived bulunga- mayines which first appear in System C, but there is no reason 10 suggest that (hey may not have persisted beyond System C. Their presence in Dome and Jim's Carousel Sites therefore cannot contirm or deny the younger age postulated for these sites. The single macropodoid, Wururoo sp.2, from Clefi of Ages is a plesiomorphic balbarine other- wise known only from System B. Its presence therefore suggests system B, This is in conflict with wombat teeth in this site, not known from any of the older Riversleigh siles, and the ‘gener- ally more modern’ (presumably ‘more derived’) appearance of other remains (Archer etal.. 1989), Encore Site has been suggested to be younger still, possibly late Miocene (Archer et al., 1994), Encore has 2 highly derived bulungamayine spe- cies, gen, Wan. spp. 1&2, Both species are known from System G; but there is 10 reason to suggest that these moderately robust, lophudont species might not persist beyond the age of thal system, Their presence dees nat preclude the age esti- matted for this site. MACROPODOID CORRELATES BETWEEN RIVERSLEIGH AND OTHER AUSTRALIAN TERTIARY FOSSIL AREAS OF COMPARABLE AGE Archer et al, (1989) suggested that Riversleigh System A units may fall “somewhere between the Ditjimanka and Kutjumarpu and Tarkarocloo LFs of South Australia’ a view supported by Woodburne ct al, (1993). The latter authors main- tuined that the base of the mammiul-beuring se- uence in the Etadunna, Namba and Wipajiri Formations predates that of the Riversleigh suc- cession, They noted in their (lowest) faunal zone A at Lake Palankaninna Xyeema mahoneyi, claimed to be the most plesiomorphic potoroid so fur found. Formal description of this species has yet to be published and its level of evolutionary development therefore cannot be compared with any of the plesiomorphic Riversleigh potoroids. Woodburne et al. (1993) also report 2 new species of Namibarae — spevies A and B, both occurring in faunal zone C from the Ngapakaldi Local Fauna of the Etadunna formation, with species B also present in the Neama Local Fauna in faunal zone D. Bath species are described as MEMDIRS OF THE QUEENSLAND MUSEUM more primitive than Nambaroe saltayus and N. tarrinyeri from the Tarkarooloa Local Fauna, provisionally equated with zone D of the Etadunna Formation at Lake Palankurinna. Archer et al. (1989) correlated Riversleigh Sys- tem B Local Faunas with the Tarkarooloo und Kutjumarpu Local Faunas, Woodburne et al, (1993) considered the Kutjumarpu Local Fauna of the Wipajiri Formation to represent the upper- most faunal unit in the eastern Lake Eyre Basin, and indicated a possible latest Oligocene age for this Local Fauna. Mı morphology in Nambaroe sp.8 from White Hunter Site is more plesio- morphic than that of any of the Nambarvo species in the Tarkarooloo local Fauna. It retains a hypoconulid, has a straight or only slightly curved paracristid, a short, low anterior cingulid, an under-developed precingulid and a diagonal posthypocristid on the posterior face of the hypolophid. Using the level of evolutionary de- velopment argument of Woodburne et al. (1993), the White Hunter assemblage would be older than etther the Lake Tarkarooloo or Kutjumarpu Local Faunas, possibly of equivalent age to their zone B+C, estimated by them to be between 25.5- 25.0myo. If White Hunter Site is a member of System B, System A sites at Riversieigh would therefore be older, possibly equivalent in age to zone A of the Etadunna formation, ie., > 25.5myo- Woodbume et al. (1993) reported a new genus and species of macropodid (their macropodine Gen, P sp. A.) from zone C of the Etadunna Formation This species was described as more apo- morphic in My irigonid morphology than species of Nambaroo, having a ‘reduced protostylid’. It was considered by them to be potentially ances- iral to two new species of Balbaroo Irom the Kugjumarpu Local Fauna, Given this information the new genus must be a balbarine comparable ta Wururoe (Cooke, this volume) from System B. The comparable levels of development of these genera provide further evidence for equating Sys- tem B sites with zone B+C of the Etadunna for- mation. The overall high diversity of balbarine species reported from the various zones of the Etadunna Formation and from the Kutjumarpu Local Fauna (5 species of Nambaroo, 2 of Balbaroa, 3 of the balbarine genus P, and aspe- cies of a macropodine [possibly balbirine’’] genus W) is, as has heen noted above, typical of the high balbarine diversity of System B at Riversleigh and lends further support to a late Oligocene age for Riversleigh System B, its basul BIOSTRATIGRAPHIC IMPLICATIONS OF FOSSIL KANGAROOS units being probably equivalent in age to the Ditjimanka and Ngapakaldi Local Faunas of the Etadunna Formation. At the opposite end of the time scale, derived, Jophodont bulungamayines suchas gen. Wan., Ot- curring in Riversleigh System C and possibly younger sites, have been suggested elsewhere (Cooke, this volume) to be likely antecedents of plesiomorphic macropodids such as Hadronemas puckridgi from the late Miocene Alcoota Local Fauna. Youngest Miocene sites at Riverleigh are therefore probably older than the Alcoota Local Fauna. Balbaroe sp.4 has lower molar morphol- ogy that is more derived than that of & camfteldensis from the late middle Miocene Bul- lock Creek Local Fauna. B. sp.4 occurs in the highest levels of System C which may therefore approximate the age of the Bullock Creek Local Fauna. Kangaroo fossils so tar recovered from Riversleigh thus support an age span tor pre- Pliocene deposits extending from late Oligocene ia at least late middle Miocene. For some Riversleigh sites, c.g., Dunsinane and Cleft of Ages Sites, they suggest ages older than those previously indicated hy Archer et al. (1994), ACKNOWLEDGEMENTS Research grants provided from the Australian Research Council and the University of New South Wales have been the primary mechanism for providing the research material examined in this study, Additional support for the Riversleigh project has come from the National Estates Program grants, the Australian Geographical Society, The Australian Museum, The Riversleigh Society, ICI Pty Lid, Century Zinc Limited, the Mt Isa Shire and private donors. LITERATURE CITED ARCHER, M. 1979. Wabularea nuwghtoni gen. et sp. nov an enigmatic kangaroo (Marsupialia) from the middle Tertiary Carl Creek Limestone of northwestem Queensland. Results of the Ray E. Lemley Expeditions, part 4, Memoirs of the Queensland Museum 19: 299-307, ARCHER, M. & FLANNERY, T.F. 1985, Revision of the extinct gigantic rat kangaroos (Poloroidae: Marsupialia), with adeseriptian ofa new Miocene genus und species and a new Pleistocene species of Propleopus. Journal of Paleontology 59: 1131 - 1149, ARCHER, M., GOQDTHELP, H., HAND, SJ. & MEGIRIAN, D, (989 Piessil mammals at 301 Riversleigh, nonhwestem Queensland: prelimi- tary overview of biostratigraphy, correlation and environmental change. Australian Zoologist 25(2); 29-65, ARCHER, M, HAND, S.J & GODTHELP. H. 1991 Riversleigh. (Reed: Sydney). ARCHER, M, et al, 1994. List of the principal Riversleigh local faunas and their interpreted rel- alive ages, Abstracts; The Riversleigh Sympo- sium 1994 (Supplement): 28-31, ARCHER, M., HAND, S.J & GODTHELP, H. 1995. ‘Tertiary environmental and biotic change in Aus- tralia. Pp. 77-90. In Vreba, E.S., Denton, G.H., Partridge, T.C. & Buckle, L.H: (eds), Palacocli- mate and evolution, with emphasis on human origins. (Yale University Press: New Haven), ARENA, R. 1997, The palaeontology and geology of Dunsinane Site, Riversleigh. Memoirs of the Queensland Museum 41. 171-179, COOKE, B.N. 1992, Primitive macropodids from Riversleigh, northwestern Queensland, Al- chennga 16: 201-217. 1997, New Miocene bulungamayine kangaroos (Marsupialia, Potoroidae) from Riversleigh, northwestern Queensland. Memoirs of the Queensland Muscum 41; 269-280, 1997. Two new balbarine kangaroos and lower molar cyolution within the subfamily. Memoirs of the Queensland Museum 41; 281-294, CREASER, P. 1997, Oligo-Miocene sediments of Riversleigh: the potential significance of topogra- phy. Memoirs of the Queensland Museum 41: 303-314. FLANNERY, T.F. 1989. Phylogeny of the Macropodoides: a study in convergence. Pp l-46, In Grigg, G, Jarman, P & Hume, |, (eds), Kanga- roos, wallabies and rat-kangaroos, (Surrey Beatty & Sons: Sydney). FLANNERY, T, & ARCHER, M. 1987a. Hypsiprimnodon bartholomaii (Potoroidae: Marsupialia). a new species from the Mincene Dwarnamor Local Fauna and a reassessment of the phylogenetic position of H. moschatus. Pp 749-758, In Archer, M. (ed.), Possums and opos- sums? studies in evolution, (Surrey Bealty & Sons Sydney), 1987b. Belfongia moyesi, a new and plesiomorphic kangaroo (Marsupialia: Potoroidac) fram Miocene sediments of northwestern Queensland. Pp 759 -767. In Archer, M. (ed), Possums and opossums: studies in evolution. (Surrey Beatty & Sons: Sydney). FLANNERY, T.F. & RICH. T.H.Y. 1986, Macropodoids from the middle Miocene Namba Formation, South Australia, and the homology af some dental structures tn Kangaroos, Journal of Paleontology 60(2); 4148-447. FLANNERY. T.F., ARCHER, M. & PLANE, M. 1982. Middle Miocene kangaroos (Macropodoides: Mursupiatig) from three localines im northern Aus- tralia, with a deseription of (wo new subfamilies, 302 Bureau of Mineral Resources Journal of Austra- lian Geology and Geophysics 7: 287-302. FRAKES, L.A., MCGOWRAN, B. & BOWLER, J.M., 1987. Evolution of Australian environments. Pp. 1-16. In Dyne, G.R. & Bolton, D.W. (eds), Fauna of Australia. Vol.1 A: general Articles. (Australian Government Publishing Service: Canberra). McGOWRAN, B., & LI, Q., 1994. The Miocene oscil- lation in southern Australia. Rec. S. Aust. Mus. 27(2): 197-212. MEGIRIAN, D. 1992. Interpretation of the Miocene Carl Creek Limestone, northwestern Queensland. The Beagle. 9: 219-248. 1994. Why the “Systems” terminology used at Riversleigh should be abandoned. Abstracts: The Riversleigh Symposium, 1994: 17. MEMOIRS OF THE QUEENSLAND MUSEUM MYERS, T.J. & ARCHER, M. 1997. Kuterintja ngama (Marsupialia, Ilariidae): a revised and extended systematic analysis based on material from the late Oligocene of Riversleigh, northwestern Queens- land. Memoirs of the Queensland Museum 41: 379-392. WOODBURNE, M.O. 1967. The Alcoota Fauna cen- tral Australia: an integrated palaeontological and geological study. Bureau of Mineral Resources, Geology and Geophysics Bulletin 87. WOODBURNE, M.O., MACFADDEN, B.J., CASE, J.A., SPRINGER, M.S., PLEDGE, N., POWER, J.D., WOODBURNE, J.M. & SPRINGER, K.B. 1993. Land mammal biostratigraphy and mag- netostratigraphy of the Etadunna Formation (late Oligocene) of South Australia. Journal of Verte- brate Paleontology 14:483-515. OLIGOCENE-MIOCENE SEDIMENTS OF RIVERSLEIGH: THE POTENTIAL SIGNIFICANCE OF TOPOGRAPHY PHIL CREASER Creaser, P1997 06 30, Oligocene-Miocene sediments of Riversleigh: the potential signifi- cance of topography. Memoirs of the Queensland Museum 41 (2): 303-3 14, ISSN 0079-8835. Although faunal assemblages provide the best indication of relative ages and environments of deposition in Tertiary and Quaternary sediments of the Riversleigh region, geological evidence provides additional significant information about the prehistory of the area. Data presented herein on topographic heights of sites in aréas of horizontally-bedded sediments lead to an hypothesis of cyclical sedimentation. At least 3 cycles of Oligovenc-Mincene sedimentation consist of 4 stages: 1) uplift and/or lowering of the water table; 2) erosion and development of a karst landscape; 3) subsidence and/or raising of the water table; and 4) sediment accumulation within the karst terrain and in surrounding shallow basins, Phil Creaser, 3 Paroo Place, Kaleen ACT 2617, Australia; received 3 February 1997. This paper uses data from Archer et al. (T989, 1994) and Megirian (1992), palacontology and preliminary mapping using photogrammetric base maps at 1:2,000, 1:15,000 and 1;20,000 to interpret stratigraphy and palaeogeography of Oligocene-Miocene sediments at Riversleigh. I focus on Oligocene-Miocene sediments on D Site and Gag Plateaus dated primarily through biocorrelation with magnetostratigraphically dated deposits in South Australia (Woodburne et al.. 1994), Archer et al. (1989) and Megirian (1992) fo- cussed on D Site and Gag Plateaux, particularly Godthelp’s Hill and Hal's Hill areas and the northern Gag Plateau. | concentrate herein on the nurthern D Site Plateau and the southern Gag, Plateau. I also builds on Archer et al.’s (1989) observation on the Gag Plateau that it is possible lo correlate widely separated exposures of flat- lying sediments. In areas where the sediments are faulted or areas on the margins of microbasins where the beds are dipping, correlation is limited. All sites have been plotted onto base maps lodged with the Vertebrate Palaeontology Labo- ratory, University of New South Wales, Queens- land Museum and the Queensland National Parks. and Wildlife Service. I provide new information on relative topographic heights which contributes to geological understanding of the region. l consider geographic sections of cach plateau and look at the geology of each section (Fig. 1) noting the range of sediment types, presumed age of the sediments based on palaeontological evi- dence and, Where appropriate, topographic heights. The other area considered is the "Mesus’. iso- lated erosional remnants of Tertiary sediment E of the Riversleigh/Lawn Hill road. A number of these sites appear similar in lithology and stratip- raphy tosites in the northern section of the D Site Plateau. I recognise 3 sedimentary sequences (Verdon Creek, Godthelp’s Hill and Gag Plateau). The Verdon Creek sequence is best represented in the northern section of D Site Plateau and consists uf mainly System A sites (Archer et al., 1989). The Godthelp's Hill sequence occurs in the central secuion of D Site Plateau with System B sites. The Gag Plateau sequence is best represented in the northern section of Gag Plateau and contains mainly System C sites but may also include Sys- tem A sites. Recent fieldwork has indicated a lack of uniformity and continuity of the basal sodi- ments in the northern section of Gag Plateau. D SITE PLATEAU The northern section includes Neville’s Gar- den/Burnt Olfering area and the major gully sys- tem to the south of this area with sites such as Quantum Leap, Gillespie’s Gully and MIM, It also includes sites on the eastern edge of the plateau (LSO and Dirk's. Towers) as well as the sites on the western edge (BIB). The southern boundary is at or about Syp's Siberia Site to the north of Godthelp’s Hill. The central section includes Godthelp's Hill, Hal's Hill and other sites in the valley to the south east of these hills including White Hunter, ABRS, Sticky Beak and Wayne’s Wok. The southern section includes sites south of Hal's Hill commencing with the Biggles Flies Again Site, SM and TOTE Sites and Bone Reef, Jeanette’s Amphitheatre and Chinatown Sites, 304 D SITE PLATEAU North Section Central Section VERDON CREEK SEQUENCE Sediments overlying D Site level Sediments overlying D Site level equivalents GODTHELP’S HILL SEQUENCE (System B sediments) D Site level (System A) with cave and related deposits D Site level equivalents (System A) System A sediments System A sediments (inc. White Hunter) MEMOIRS OF THE QUEENSLAND MUSEUM GAG PLATEAU North Section South Section GAG PLATEAU SEQUENCE Uplift and Erosion | - ı Sequenceofcave | System C sediments | deposits, fissure fills, i and ' ' stratigraphic/shallow | pool deposits. i 1 | There is at least one j | period of upliftand | \ erosion. ; i I | Sediments range in age i 1 from?SystemAto | ; System C i i Overlying calcarenites (Possible System A sediments) “Basal sediments. Basal sediments with White Hunter equivalent FIG. 1. Generalised Riversleigh stratigraphy based on geological observations, biocorrelations and topographic heights, NORTHERN SECTION (VERDON CREEK SEQUENCE).This sequence is best seen in the area of Bitesantennary, Burnt Offering and Neville’s Garden Sites where it consists of a basal conglomerate, overlain by arenites and calcaren- ites, up to 20m thick. A 3m homogeneous lime- stone, the D Site Limestone, overlies these sedi- ments, Cutting into and lying on the D Site Lime- stone are a series of cave deposits and possibly related tufa deposits. A further series of calcaren- ites are the highest units of this sequence. The basal conglomerate consist of a series of OLIGO-MIOCENE SEDIMENTS OF RIVERSLEIGH 175m 172m 170m 165m 162m 155m 152m Black Coffee (??Possible System C) D Site level (Damaged Digit etc) i also Bitesantennary, Neville’s Garden i Burnt Offering (BO), Upper BO, VIP, Dirk’s i Towers, Judith Horiz’is, Gillespie’s Gully : Quantum Leap, Steph’s Small Reward, MIM r 1 ?LSO A ! l ] 305 i Judy’s Jumping Joint ı (??System B) | level* Wayne’s Wok | Creaser’s Ramparts ?Sticky Beak, Bole’s ı {Bonanza IT. PA White Hunter | t 1 L] I Low Lion, Phil’s Fangs, KJO i Hiatus ' i Carbonate cemented chert conglomerate FIG. 2. Verdon Creek Sequence, System A with possible Systems B and C, ’type’ section in northern section of D Site Plateau . Heights are in metres above sea level. Dashed boxes indicate relative positions of sites from the Hal’s Hill area in the central section. massive or normally graded, matrix-supported breccias and conglomerates and clast-supported cobble and pebble conglomerates (Megirian, 1992). Archer et al. (1989) questioned whether this conglomerate is contemporaneous through- out the region, suggesting that it could represent different, non-contemporaneous cycles of local weathering. The thickness of conglomerate var- ies considerably throughout the D Site Plateau. There are several fossiliferous levels in the overlying arenites and calcarenites (Fig. 2) dom- inated by turtles, crocodiles, large birds and rare marsupials (usually diprotodontids). The lowest level at the northern end of this section includes Low Lion and Phil’s Fang Sites. A distinctive 25m wide fossiliferous horizon at the Low Lion level is at the same level as KJO Site. The level is dominated by large bone frag- ments similar in colour and preservation to fossils from Low Lion Site. Above this level at the northern extremity of the Plateau are the IT and PA Sites which yielded jaw fragments of Yalkaparidon. These two adjacent sites, which contain small terrestrial assem- blages, are the only ones known from the lower part of the sequence in this area. Although there 306 are other fossiliferous assemblages at about this level, they tend to consist of well- worn fragments of aquatic vertebrates. The next fossiliferous level includes LSO Site. Above this are sites in the major gully in the northern part of this section including Quantum Leap, MIM and Steph’s Small Reward Sites. Cooke (1997) considered the Quantum Leap Site kangaroos to be most similar to others from Sys- tem A or B assemblages; its stratigraphic position and sedimentology suggest that it is a System A assemblage. Above these sites but below the D Site Lime- stone level, is a higher more widespread fossilif- erous level which includes Burnt Offering, Upper Burnt Offering, VIP, Judith Horizontalis, Punky Brewster, Gillespie’s Gully and Dirk’s Towers Sites. These appear to be stratigraphically con- trolled and do not represent a later incised de- posit. While they are at the same topographic level and appear to be horizontally bedded, they may be of different ages. Black (1997a) suggested that Burnt Offering is a System A site. Cooke (1997) suggested it might be a System B site; the macropodid fauna indi- cates that it could be either System A or B. Black (1997a) considered that Upper Burnt Offering is a System B site but Neohelos n. sp. | is only found in System A deposits. Black (1997a) considered VIP to be a System A site on the basis of a plesiomorphic zygomaturine. As yet, there is in- sufficient data from the Judith Horizontalis, Punky Brewster and Gillespie’s Gully Sites to allocate these and no clear evidence from Dirk’s Towers Site as to whether it is System A or B. D Site Limestone is a distinctive marker bed that outcrops over much of the D Site Plateau and is characterised by a fossil assemblage of mainly large vertebrates dominated by mekosuchine crocodiles, dromornithids and diprotodontoids. However, the stratigraphy of D Site (Tedford, 1967) and of the ridge to the north of D Site, are less readily interpreted because of extensive scree slopes. Archer et al. (1989) equated the D Site Lime- stone and its fossil assemblages with their System A. However, given further research since the mid 1980s and recognition of several fossiliferous levels well below this marker bed, it is recom- mended here that System A be expanded in con- cept to include all of the lower sediments of the Verdon Creek sequence. A complex series of cave and possibly also tufa deposits have been etched into the D Site Lime- stone. This suggests a period of uplift or lowering MEMOIRS OF THE QUEENSLAND MUSEUM of the water table following deposition of the D Site Limestone, followed by karst weathering of the limestone to form caves and other sedi- ment/fossil traps, and then infilling of these microbasins. The best known cave sites are Microsite and Bitesantennary Site with the best example of a tufa deposit being Neville’s Garden Site with sediment and fossils accumulating at and beyond the entrance to a cave. However, Neville’s Garden Site may represent a strat- igraphically controlled site which can be corre- lated with Dirk’s Towers or Judith Horizontalis Sites from this section or possibly Wayne’s Wok and Creaser’s Ramparts Sites from the central section of this Plateau. Black (1997) and Cooke (1997) assign this site to System B on its diprotodontoids and macropodoids respectively. A thin series of calcarenites overlies the D Site Limestone and the cave and tufa deposits. How- ever, fossils from these sediments are not com- mon and only one site in this section, Black Coffee Site (above Gillespie’s Gully), has been sampled. Its faunal assemblage has yet to be analysed in detail. Topographically, the base of this sequence at the northern end of the Plateau is at 152m with the lowest fossiliferous level at about 155 m (Low Lion and KJO Sites), The base of the D Site Limestone is at 175m. The highest point on the D Site Plateau is at 202.7 m. Above Neville’s Gar- den Site, the highest point is 192m which would give a thickness of at least 40m. CENTRAL SECTION .The central section con- tains the discreet, richly-fossiliferous Godthelp’s Hill sequence which is separated both stratigraphically and topographically from other sediments in this section which are similar in lithology and stratigraphy to the Verdon Creek sequence. The Godthelp’s Hill sequence may be the equivalent of the cave and tufa deposits of the Verdon Creek sequence. The other sediments can probably be equated to the other Verdon Creek sequence sediments. GODTHELP’S HILL SEQUENCE. Because they are separated, possibly due to faulting (Megirian, 1992), from the main sequence itis not clear whether the tufa deposits on Godthelp’s Hill are the equivalent of the cave and related tufa deposits of the Verdon Creek sequence. Although faunal assemblages indicate a similar age, the Godthelp’s Hill sediments, which have been re- garded as System B (Archer et al., 1989), are regarded as a distinct sequence (Fig. 3). The OLIGO-MIOCENE SEDIMENTS OF RIVERSLEIGH 307 GODTHELP’S HILL SEQUENCE VERDON CREEK SEQUENCE Highest View Delightful Panorama Ten Bags Boid CAVE DEPOSITS Bitesantennary Microsite Boid Site East CAVE/POOL DEPOSITS Neville’s Garden All occur at the D Site Level Mid/High Helicopter Mike’s Potato Patch Mid Upper cs Inabeyance Mike’s Menagerie Mid/Low RSO RV Outasites Souvenir Others G Spot, DDDD, Victor’s Vacuum, Paul Willis, Dredge’s Ledge FIG. 3. Godthelp’s Hill Sequence, System B with possible System B equivalents from the Verdon Creek Sequence. 308 MEMOIRS OF THE QUEENSLAND MUSEUM 200m Jaw Junction Henk’s Hollow, Bob’s Boulders, Grimes Site 197m No Name, Arthur’s Seat, Golden Steph, Phalanger Crusty Meat Pie, Skull, Neville’s Riches, Fireside Favourites, H2rry’s Hump Flowstone Level: Bat sites: Bat Smear, Gotham City (E & N), First Drop, Sticky Wicket. Other sites at or about this level: Sue’s Diprotodontid, GOH, Bird Bone, Group, Main, Quentin’s Quarry, Courtney’s Cache, Two Trees 192m Ringtail, Melody’s Maze 188m Gag, LD’94, Jim’s Jaw, Last Minute 179m Don Camillo FIG. 4. Gag Plateau Sequence, System C, ’type’ section in northern section of Gag Plateau based at Don Camillo Site. Heights are in metres above sea level. Sites to be plotted: Archie’s Absence, Archie’s Parlour, Bernie’s Bedford, Kangaroo Jaw, Lockwood’s Link, and Bruty and the Beast. OLIGO-MIOCENE SEDIMENTS OF RIVERSLEIGH thickness of the sediments on Godthelp’s Hill has been estimated to be 7m (Archer et al., 1989) to 12m (Megirian, 1992). Detailed photogrammetry of the area indicates that the estimate of 12m is more accurate. OTHER SEDIMENTS IN THIS SECTION. These sediments are around Hal’s Hill to the south of Godthelp’s Hill. Hiatus Site, the lowest in this area, is just above Precambrian sediments on the northern side of Hal’s Hill. Although there is some doubt about its position because of fault- ing, it appears to be a base level. Black (1997a) notes that Silvabestius michaelbirti from Hiatus South Site is the most plesiomorphic zygomatur- ine known. Stratigraphic and topographic posi- tion and faunal assemblage suggest that White Hunter Site on the south side of Hal’s Hill is very low in the sequence. However, its lithology dif- fers from that of Hiatus Site and the basal sedi- ments in the northern section of the Plateau. Black (1997a) indicates that Hiatus South Site belongs to System A. Cooke (1997) considers it a System ?A/B Site because several macropodoid species from White Hunter Site are also found in other undoubted System A and B sediments. However, there are also 5 unique macropodoid species from this site. Myers & Archer (1997) indicate White Hunter Site is the only one at Riversleigh that contains ilariids. Taken together, these suggest that White Hunter either represents a distinct interval of time or, if contemporaneous, a different ecosystem. I suggest that White Hunter Site is a basal System A deposit. There are no distinctive fossiliferous levels im- mediately above either Hiatus or White Hunter Sites. The next site up section may be Sticky Beak Site which is a lower level than the D Site Lime- stone equivalent, approximately at the same level as Boles’ Bonanza Site. Black (1997a) suggests that Sticky Beak Site belongs to System A. The next level up is immediately below the D Site Limestone level equivalent. In this section, Wayne’s Wok and Creaser’s Ramparts Sites are at this level and may be correlated with Dirk’s Towers and Judith Horizontalis Sites from the northern section. Both Black (1997a) and Cooke (1997) consider Wayne’s Wok Site to be low in System B. However, this site contains a number of species that are found in Systems A and B. The age of Creaser’s Ramparts Site is also unclear and more palaeontological information is needed. Black (pers. comm.) recognised a phascolarctid from this site, that is the most primitive from the Australian Tertiary. In this section, the only D Site Limestone equivalent site is Neville’s Pancake Site (with a plesiomorphic meiolaniid turtle; Gaffney et al., 1992). Fig Tree Site, with the most plesiomorphic zygomaturine Nimbadon, Hand et al., 1993 is stratigraphically below Neville’s Pancake Site. Both sites are NE of Hal’s Hill. Above the D Site Limestone level, in the over- lying calcarenites, only Judy’s Jumping Joint site has been sampled. This site is a localised con- glomerate found on the crest of Hal's Hill and belongs to System B. However, because the rela- tionships are not clear, it is not possible to deter- mine at present whether these calcarenites represent System A or B. SOUTHERN SECTION. There has been rela- tively limited exploration in this section of the Plateau with preliminary fieldwork indicating a thin series of sediments. Sites such as Bone Reef and Jeanette’s Amphitheatre are as yet largely unassessed. They are dominated by large animals more or less of the same kind (but far more abundant) that characterise the D Site Limestone at, for example, Site D. Black (1997a) considers these two sites System A deposits. Immediately below these are fossiliferous sediments. Two other sites collected from this region are China- town, which produced a System B assemblage, and a possible cave deposit with a rich bat fauna. SM and TOTE Sites, in the northern part of this section, are dominated by large vertebrates. GAG PLATEAU The northern section includes the vast range of sites at the northern end of the Plateau including Golden Steph, GOH, LD94 and First Drop. The central section is relatively barren apart from Wang Site and this relative lack of sites is its defining feature. The southern section includes AL90, COA, Dunsinane, Dome, JC, Encore and others. The northern boundary of this rich south- ern section is at Peter the Pilot Site. NORTHERN SECTION (GAG PLATEAU SE- QUENCE). Based on acomposite section starting with Don Camillo Site at the base (near the north- ern point of this section), up through Gag Site to Jaw Junction Site at the top, together with equiv- alent sites at the appropriate levels at the northern end of this section, the Gag Plateau sequence was considered (Archer et al., 1989) to consist of fossiliferous basal sediments overlain by calcarenites and a series of tufa and ’deep water 310 pool’ deposits which, in sone cases, Contain uj- verse faunal assemblages. However, this se- quence contains a complex variety of basal sediments. At Don Camillo site, which was con- sidered the stratigraphic equivalent of the D Site Limestone, significantly different lithologies are present. At the eastern end of this section 1s a fossiliferous conglomerate and at the western end a vertical sequence of richly tossiliferous sedi- ments, Qverlying the Precambrian is a thin se- quence of arenaccous sediments overlain by fossiliferous calearenites which in turn are óver- lain by (?)weathered lateritic sediments. These are overlain by the typical’ calcarenites which in some places have rich vertebrate assemblages, However, none of these sediments are apparently continuous across the northern section of the Gag Plateau. Some of these Dasal sediments may be lateral equivalents of System A sediments in the Verdon Creek sequence. While the basal sediments vary considerably, there 1s apparently more uniformity higher in the sequence including the tufa and ‘deep water pool’ deposits of Archer et al, (1989). The tufa deposits such as Gag, Henk’s Hollow and Golden Steph Sites, are dominated by terrestrial faunas, In con- trast. the “deep water pool’ deposits, such as Crusty Meat Pie, Quentin's Quarry. Bob's Boul- ders and Ringtail Sites are dominated by aquatic faunas, Don Camillo Site is at approximately 179m with the highest point at 201m (Jaw Junction Site), giving a maximum thickness of 22m as- suming the beds are horizontal. Gag, Last Minute and LD'94 Sites are all at 188m withthe erosional break recognised by Archer et al, (1989) and Megirian (1992) at 194m, The only sites with significant bat accumulations (Gotham City, Sucky Wicket, Bat Smear and First Drop Sites) are found at this level (Fig. 4). CENTRAL SECTION. Only Wang Site is known from this section, with no clear ‘dividing line’ between the northern and southern sections. However, sediments in the southern section diller in the type and extent of their lithologies and faunas, SOUTHERN SECTION. The stratigraphy of this section is complex, Unlike the northern section of the D Site Plateau or the upper northern section of the Gag Plateau, the series and types of sedi- ments in the southern section are apparently not limited in lateral extent, often significantly differ- ent in age and do not appear to be horizontally MEMOIRS OF THE QUEENSLAND MUSEUM bedded, This makes it particularly difficult jo correlate this section with others (Fig. 5), Pul- acontological evidence [rom this section allows same correlations, The oldest recognised sediments ure at Dunsin- ane Site (and equivalents) with plant and animal fossils, These sediments are overlain by less fos- siliferous calcarenites into which are incised richly fossiliferous cave and fissure fill deposits. Dunsinane, Sue's Rocky Road, Custard Tart and Bernie’s Cooking Pot Sites may representihe oldest from the Gag Plateau, possibly System A, on the basis of correlation of mammals with White Hunter Site (Arena, 1997; Cooke, 1997). Fossils are not common in the overlying calcarenites at Anna's Horribilis, Two Gloves. Anton's Pixie, Arachnid Ridge and Don’t Ask Me Sites, all of which have yet to be studied in detail but are probably System A sites, Faunas and lithologies indicate cave deposits at Dome, AL*90, Peter the Pilot, Creaser’s Crouch and Angela's Bat Pate Sites. Fissure fill deposits such as COA and Keith's Chocky Block Sites are easily recognisable because of their lithologies, Jim's Carousel and Encore Sites could represent tufa deposits incised into a pre-existing Tertiary limestone. While the lithologies and environment of depo- sition of many of these sites are similar, 1118 clear that they represent a wide range of ages. Black (1997a) suggests that AL.9O, Jim's Carousel and Dome Sites may all be System C deposits, Black (1997a) suggests that COA Site may be a System A site, but Cooke (1997) suggests it is either System A or B. A. Gillespie (pers. comm.) con- siders itto be System B. Encore Site despite being lithologically very similar to other sites in the area, is early late Miocene, System C (Archer eb al., 1994; Black, 1997b). ENVIRONMENTS OF DEPOSITION AND PALAEQGEOGRAPHY The Tertiary limestone deposits of the Gregory River area are freshwater f{luyio-lacustrine depos- its, Archer et al, (1989) recognised a complex series of lacusirine, alluvial, travertine and cave deposits while Meginan (1992) has documented alluyial, tufa and karst facies. I agree with these views and propose that a cycle of sedimenjation/erosion that involves: 1, uplift and or lowering of the water table; 2, ero- sion and development of a karst landscape; 3, OLIGO-MIOCENE SEDIMENTS OF RIVERSLEIGH Central Section Southern Section Cave Deposits Fissure fills Peter the Pilot, Creaser’s Crouch (CC) Upper CC Nicole’s Boulders W AL’90 Captain Androgen Jolly Roger Bernards Belted Belfry Angela’s Bat Pate Bat Eerk Anne’s Bat Room Dome (N & S) Jeanctte’s Bat Stuff COA Keith’s Chocky Block Stratigraphic/Shallow Pool Deposits Arachnid Ridge* Nicole’s Boulders JC Sites Not JC8 Anna’s Homibilis* Two Gloves* Encore Angela’s Sinkhole Don’t Ask Me* Anton’s Pixie* Dunsinane, Custard Tart, Sue’s Rocky Road, Bermie’s Cooking Pot Equivalent to White Hunter Site on D Site Plateau FIG. 5. Gag Plateau Sequence, central and southern sections. The age of these sites is unclear (apart perhaps from Dunsinane Site and its equivalents) although there is evidence that there is a range of different ages represented. Deposits marked * appear to represent faunas from the calcarenites which overlie the Dunsinane Site and its equivalents. subsidence and or raising of the water table; 4, sediment accumulation within the karst terrane and surrounding basins, This 4-stage cycle oc- curred at least 3 times during Oligo-Miocene time at Riversleigh. Each new cycle may have been initiated by minor tectonic activity that may haye been responsible for changes in the hydrogeologic system. Megirian (1992) noted faulting on Godthelp’s Hill which may have been responsible for deposition of the basal conglom- erates as a debris flow, Following initial observation by B. Cooke that Tertiary limestone on the mesas not uncommonly rest directly on Precambrian quartzite, M. Archer demonstrated that there did not appear to be any- where on either the D Site Plateau or the Gag Plateau where Tertiary sediments directly over- lay the Cambrian limestones. In a number of places, however, Tertiary sediments can be found adjacent to Cambrian limestones (e.g. Microsite which is topographically situated between a high of Thorntonia Limestone to the west and D Site Limestone to the east) which suggests that karst topography of Cambrian limestones may have controlled sedimentation patterns in the Tertiary. [ suggest that the basal arenaceous sediments and the overlying calcarenites were deposited in an alluvial fan/braided stream environment. Given the oulerop paltern, it is likely that the streams flawed in a northeasterly direction. As Mevinun (1992) has noted, fossils are not com- mon ia this type of environment, However, in some pans of this environment, possibly in swamp areas or stagnant waler away from the forest, assemblages of large aquatic and terres- inal animal fossils accumulated. In other areas freshwater limestone tufa pools developed in or at the edge of the rainforest but these are rela- ively rare. The level containing rich faunas from sites such as Dirk's Towers, Creaser’s Ramparts, Judith Horizontalis and Burnt Offering Sites, is a notable example. These sites are often found in gullies today because they are more easily weath- ered than the surrounding more resistant sedi- ments, There does not appear to be any breaks in this part of the sequence and it is Suggested that ihis sedimentation continued until (? tectonic ac- tivity Or a raised water table led to the empond- ment òf larger lacustrine bodies of waterin which accumulated the sediments (e g., the D Site Lime- stone) exposed at localities such as D Site. These sediments are fuirly uniform m lithology apart from the northeastem corner of the D Site Plateau (near the Lawn Hill road and Verdon Creek) where Lhere is a high percentage of quartz grains Suguesjing 2 Precambrian source t0 the north- cast, Lurge vertebrates, including crocodiles and luriles. are common in this limestone which ts considered to represent an open lake or swamp, al some distance from the rainforest. This would explain the scarcity of terrestrial faunas espe- cully when compared to the shallow pool Lufa deposits (Archer et al., 1989). Renewed tectonic uplift and/or a lowering of the waler lable resulted in renewed development of a karsi landscape in the Tertiary as well as Cambrian limestones, This enabled formation of caves and their subsequent filling with fossilifer- ous deposits fe.g. Microsite and Bitesantennary Site), as well as deposits formed at cave entrances (e.g. Neville's Garden Site where fragments of speleothems {straws} and in situ travertine rills and small stalagmites have been found), Godthelp’s Hill sequence is regarded as a sep- arate unil, which may be equivalent in part to these cave deposits, Lithology of the sediments and the fossils from sites on Godthelp's Hill suggest Tufa deposits that accumulated over a period of lime. The uppermost series of Calcarenites in the Verdon Creek sequence suggest à relurn to the alluvial braided stream facies, possibly following erosion of the Karst landscape, These sediments MEMOIRS OF THE QUEENSLAND MUSEUM may be the laleral equivalents ol the System C sediments al the northern end of the Gag Plateau, or altematively they may be part of System A, System B faunal assemblage from the conglom- erate al Judy's Jumping Joint Site does not enable any definite conclusions to be drawn because the relationship of this Site to the surrounding sedi- ments is unclear, The basal sediments at the northern end of the Gag Plateau vary from fossiliferous conglomer- ates to arenaccous sediments and weathered lat- erilic sediments. These sediments are overlain hy calearenites that suggest an alluvial braided stream facies, and a fossil-rich tufaand deep pool (aquatic) deposits. Like similar deposits in the Verdon Creek sequence, the tufa deposits appear to have accumulated in and around shallow ler- resttial pools. There is no clear pattern evident in the distribuon of these twò types of shallow water (ula and deep water deposits, There is also à Significant erosional break in the sequence that can be recognised in the field by flowstone and travertine deposits, Palacontological evidence also indicates a significant break between the upper and lower parts of the sequence in this area and it may be no coinvidence that the only sites in this area Which contain significant bat accumu- lations are at this level, This suggests another cycle of uplift and or lowering of (he water lable, cave development, cave fill, and crosion before sedimentation recommenced. In contrast, there are apparently no similar se- quences of shallow water tufa or deep ponl de- posits at the southern end of the Gag Plateau. Isolated sites such as Jim's Carousel Site, how- ever, may be of this type. Il is nol clear if the sediments at the northern end ure equivalent to the main sequence of sediments al the southern end of the Plateau, Mammals from Dunsinane and related sites at the southern end of the Plateau, which appear to represent basal sediments, suggest correlation with White Hunter Site, a probable System A assemblage (Arena, 1996, 1997). These sedi- ments are overläin by a series of calearenites which were probably deposited in an alluvial fan/braided stream environment, Associated with the calcarenites are cave and fissure fill deposits which have been incised into the earlier Tertiary limestones indicating that a karst landscape had already been formed. The age of these cave and fissure fill deposits appears to range considerably with some (e.g,, Encore Site) being the youngest Oligo-Miocene sediments in the region. QLIGO-MIOCENE SEDIMENTS OF RIVERSLEIGH PALAEOENVIRONMENT On the basis of the fossil faunas, Archer et al. (1989, 1994) suggested that rainforest covered the region at least during the carly to middle Miocene, Archer (pers. comm., 1996) suggests that there is much less faunal evidence for tainfor- est being ubiquitous during the late Oligocene. Other researchers support this view although there seems to be evidence that this rainforest was unlike any found in Australia today and that the rainforests of New Caledonia or mid-montane Papua New Guinea may be more similar. Boles (1997) and White (1997) suggest that there were some patches of open forest. Megirian (1992) suggested the rainforest was a refugium confined lo the proximity ol perennial, spring-fed streams. He concluded that the palaeoclimate was rela- tively dry, perhaps semi-arid, despite the fact that the sediments were considered to be characteris- tic of humid alluvial fans, Archer et al. (1995) refuted this suggestion on faunal evidence. Creaser (1977) has also showed that although calclithiles, lerrigenous clastic rocks in which carbonate fragments dominate, form mainly in alluvial fans in arid/semi-arid and glacial/perigla- cial environments, they are also forming today on the Huon Terraces in Papua New Guinea where tectonism and climate both appear to influence the accumulation of the calclithites. The Huon Terraces are in a rainforest environment with 2000-2500mm of rain per annum. The rainfall has a marked peak in December to February, with the nearby mountain ranges forming a rainshadow, and a pronounced dry season at other times of the year, While there are differences between the Huon Terraces and the Gregory River region, it is possible for a species-rich rainforest lo exislin an area of tectonic activity and produce the full range of sediments evident in the Gregory River basin. Although there is no evidence for season- ality in the early to middle Miocene sediments of Riversleigh, growth rings in wood fragments (?Nethofagus sp.) from Dunsinane Site suggest seasonalily (Jane O'Brien, pers. comm.) or al least episodic changes in growth rates. Unfortu- nately, the age and relative stratigraphic position of these fragments is in doubt (Arena, 1997), They could be either late Oligocene or late Miocene, both icehouse intervals (Frakes & Macgowran, 1987) when rainforest is less likely to have characterised the region. The plants of Dunsinane Site may have grown on the edge of a forest clearing, surrounding the Dunsinane body of water. Hs ACKNOWLEDGEMENTS Many of the ideas and thoughts in this paper ure the result of the observations and fieldwork of others, especially fellow workers at Riversleigh and in particular, Michael! Archer, Henk Godthelp und Suzanne Hand, Vital support for research al Riversleigh has come from the Aus- tralian Research Grant Scheme; the National Es- tale Grants Scheme (Queensland): the University of New South Wales; the Commonwealth De- partment of Environment, Sports and Territories; the Queensland National Parks and Wildlife Ser- vice; the Commonwealth World Heritage Unit; ICI Australia Pty Lid; 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 Zine Pty Lid; Mount Isa Mines Pry Ltd; Surrey Beatty & Sons Pty Lid; the Riversleigh Society Inc.; and private supporters including Elaine Clark, Margaret Beavis, Martin Dickson, Suc & Jim Lavarack and Sue & Don Scott-Orr, Vital assistance in the field has come from many hundreds of volunteers as well as stalf and postgraduate students of the University of New South Wales. LITERATURE CITED ARCHER, M.. GODTHELP, H., HAND, SJ, & MEGIRIAN, D. 1989, Fossil mammals of Riversleigh, North Western Queensland: prelimi- nary overview of biostriligraphy, correlation and environmental change. Australian Zoologist 25; 29-65. ARCHER, M., MAND, § I, & GODTHELP, H. 1994. Riversleagh. Second Edition, (Reed: Sydney) ARCHER, M.. HAND, S.J, & GODTHELP, H. 1995. Tertiary environmental and biotic change in Aus- tralia. Pp. 77-90, In Vrba, E.S., Denton, G-H, Partridge. T.C. & Burekle, L.H. (eds), Paleocti- mite and evolution, with emphasis on hunian Origins, (Yale University Press: New Haven). ARENA, R. 1996. Dunsinane Site: the case of ihe Tertiary time capsules. Riversleigh Notes 30; 4-6. ARENA, R. 1997, The palacontology and geology of Dunsinane Site, Memoirs of the Queensland Mu- seum 41: 171-179. BLACK, K. 1997a. Diversity and biostratigraphy of the Diprotodontoidea of Riversleigh, northwestem Queensland. Memoirs of the Queensland Museum 4l: 187-192. BLACK, K. 1997b. A new Species of Palorchestidie {Marsupialia) from the [a middle to early late Miocene Encore Local Fauna, Riversleigh, north- Western Queensland. Memoirs of the Queensland Museum 41: 181-185 314 BOLES, W. 1997. Riversleigh birds as pal- aeoenvironmental indicators. Memoirs of the Queensland Museum 41: 241-246. COOKE, B.N. 1997. Biostratigraphic implications of Riversleigh fossil kangaroos. Memoirs of the Queensland Museum 41: 295-302. CREASER, P.H. 1977. Lithogenesis and diagnostic features of recent and ancient terrigenous lime- stones (calclithites). Unpubl. M.Sc.Thesis, ANU. FRAKES, L.A., MCGOWRAN, B. & BOWLER, J.M. 1987. Evolution of Australian environments. Pp. 1-16. In Dyne, G.R. & Walton, D.W. (eds), Fauna of Australia, Vol. 1A, General Articles. (Austra- lian Government Publishing Service: Canberra). GAFFNEY, E.S., ARCHER, M. & WHITE, A. 1992. Warkalania, a new meiolaniid turtle from the Tertiary Riversleigh deposits of Queensland. The Beagle 9: 35-47. HAND, S.J., ARCHER, M., GODTHELP, H., RICH, T.H. & PLEDGE, N.S. 1993. Nimbadon, a new genus and three new species of Tertiary zygomaturines (Marsupialia: Diprotodontidae) from northern Australia, with a reassessment of Neohelos. Memoirs of the Queensland Museum 33: 193-210. MEGIRIAN, D. 1992. Interpretation of the Miocene MEMOIRS OF THE QUEENSLAND MUSEUM Carl Creek Limestone, northwestern Queensland. The Beagle 9 : 219-248. MEGIRIAN, D. 1994. Approaches to marsupial biochronology in Australia and New Guinea. Al- cheringa 18: 259-274. MYERS, T. & ARCHER, M. 1997. Kuterintja ngama (Marsupialia, Ilariidae): a revised systematic anal- ysis based on material from the late Oligocene of Riversleigh, northwestern Queensland. Memoirs of the Queensland Museum 41: 379-392. TEDFORD, R.H. 1967 Fossil mammals from the Carl Creek Limestone, northwestern Queensland. Bul- letin of the Bureau of Mineral Resources, Geology and Geophysics, Australia 92: 217-236 WHITE, A., 1997. Cainozoic turtle assemblages from Riversleigh, northwestern Queensland. Memoirs of the Queenland Museum 41: 413-421. WOODBURNE, M.O., MCFADDEN, B.J., CASE, J.A., SPRINGER, M.S., PLEDGE, N.S., POWER, J.D., WOODBURNE, J.M. & SPRINGER, K.B. 1994. Land mammal biostratigraphy and magnetostratigraphy of the Etadunna Formation (Late Oligocene) of South Australia. Journal of Vertebrate Paleontology 13: 483-515. PALORCHESTES AZAEL (MAMMALIA, PALORCHESTIDAE) FROM THE LATE PLEISTOCENE TERRACE SITE LOCAL FAUNA, RIVERSLEIGH, NORTHWESTERN QUEENSLAND A.C. DAVIS AND M. ARCHER Davis, A.C & Archer, M.. 199706 30. Palorchestes azael (Mammalia, Palorchestidae) from the late Pleistocene Terrace Site Local Fauna, Riversleigh, northwestern Queensland. Mèm- oirs of the Queensland Museum 41(2); 315-320, Brisbane. ISSN 0079-8835. A maxilla of Palorchestes azael is described from gravel deposits at Terrace Site, Riversleigh Station, A radiocarbon date of 23,900 +4 100-2700 years BP is reported trom the fossiliferous unit at Terrace Site supporiing previous interpretations of a lite Pleistocene age for the Terrace Site Local Fauna. Angela Davis, Geology Department, Ausiralian National University, Canberra; ACT 0200, Australia (Present address: Western Australian Musettin, Francis Street, Perth, Western Australia 6000, Australia) ; Michael Archer, School of Biological Science, University of New South Wales, Sydney, New South Wales 2052, Ausiralia; received 4 December 1996. Terrace Site occurs inunconsolidated fluviatile sediments on the eastern bank of the Gregory River 5 km downstream from the Lawn Hill road crossing on Riversleigh Station, NW Queens- land. These deposits were interpreted as Pleistocene gravels resting on Tertiary and Cam- brian limestones (Archer et al., 1989, 1994). The Terrace Site Local Fauna is listed in Archer et al. (1994). Material is deposited in the Australian Museum (AMF), the Natural History Museum, London (BM), Museum of Victoria(NMVP), Queensland Museum (QMF), South Australian Museum (SAM), Department of Geology, James Cook University (P). Molar number follows Luckett (1993); premolar number follows Flower (1867): molar crown morphology follows Archer (1984), STRATIGRAPHY AND AGE Terrace Site has a 3m high cross-section through horizontal and lenticular beds in an up- wardly fining sequence. The basal sediments are poorly sorted, light grey sands and gravels with abundant mussel shell fragments and most of the vertebrales including QMF30882. The section grades upwards into finer sands, silts and clays with finer shell fragments, Charcoal particles up to 5mm occur throughout in small lenses or iso- lated fragments. The charcoal occurring in lenses, maxilla and most other specimens being un- abraded and the recovery of articulated material, Suggests that al least part of the fauna and associ- ated charcoal is a primary accumulation. Based on the mammal fauna the age was inter- preted as possibly late Pleistocene (Archer et al.,1994), Charcoal from the basal bone-rich layer that contained QMF30882 gave a conventional radiocarbon date of 23,900 +4100 -2700 BP (ANU-7620). Although the standard errors are high, the range within two standard errors con- firms the Late Pleistocene age. SYSTEMATICS Order DIPROTODONTIA Owen, 1866 Suborder VOMBATIFORMES. Woodburne, 1984 Family PALORCHESTIDAE Tate, 1948 sens. Archer & Bartholomai, 1978 Palorchestes Owen, 1873 TYPE SPECIES, Palorchestes azael Owen, 1573, Palorchestes azael Owen, 1873 (Fig. 1; Table 1) MATERIAL. QMF30882, a left maxillary fragment with sear complete M?-3, the fragmented alveolus of P3, and a portion of the palate and jugal. DESCRIPTION. Upper molars high-crowned, bilophodont, trapezoidal in occlusal view, wiih the protoloph wider than the metaloph, Lophs broad at their bases, narrow toward the apices, slightly crescentic, Broad anterior and posterior cingula on each tooth, extending around the tooth forming narrow lingual and buccal cingula. Molar enamel crenulated and rough on anterior, posterior and inter-loph surfaces, but smooth on lateral surfaces, 316 MEMOIRS OF THE QUEENSLAND MUSEUM FIG. 1. Palorchestes azael Owen, 1873, QMF30882 from Terrace Site, maxillary fragment with M!-M3, in occlusal view. a.c.=anterior cingulum; p.c.=posterior cingulum; l.c.=lingual cingulum; f.1.=forelink; m.1.=mid- link; h.l.=hindlink; l.a.l.=lingual accessory link; pr=protocone; pa=paracone; me=metacone; mcl=metaconule; prl=protoloph; mel=metaloph; p.s.=palatal sinus. PALORCHESTES AZAEL FROM RIVERSLEIGH Table 1. Dimensions (mm) of upper cheek teeth of QMF30R82 and comparative samples ot P. azael (BM46316, AMP452, P186593, AMF7272, QMF7074, QMF772), P. parvus (from Woods 1958), P. paine? (from Woodburne, 1967) and P. selestiae (from Mackness, 1995). The dimensions of the holotype of P. azael equal the minimum values of the observed ranges.e = estimate, H=holotype. P, parvis P. pamei P; selesniite MEAN | RANGE] (H) 13 8-144 - 16.5-18.2 13.6- Lhd 13.2-15.5 18.0-19,4 18.1-23.1 P? missing. Partly preserved alveolus suggest- ing P? as subtriangular, about 17 mm long. M! nearly complete, missing only a small part of the enamel of the buccal side of the protoloph, with 2 high forelinks joining the protoloph to a broad anterior cingulum. Lingual forelink offset diagonally, joining the anterior cingulum from the protocane at the midline of the tooth. Deep pits between, and on both sides of, the forelinks. A high double-midlink between the Jophs. On the lingual side of the double-midlink is an accessory link forming a deep lingual pit. Lingual accessory link high and well-developed on M!. Posterior cingulum compressed by molar crowding from M?; an irregular folded posterior face of the metiulaph formed from a thickened crest (hindlink) descending posteriorly from the metaconule, M? Jarger than M!, missing the anterofingual fuce of the protoloph, with asingle weak forelink, and midlink a single (ef. double in M’) high link following the midline of the tooth. Lingual acces- sory link V-shaped in lateral view, differing from the straighter link of M! and slightly lower, An- terior and posterior cingula well-developed; an- terior cingulum pressed into M! because of molar crowding. Hindlink present, M? similar in size and morphology to M3 dif- ferences in Jength being attributed to molar crowding, Forelink, midlink and hindlink all sin- gle, Lingual accessory midlink more reduced than in M?, Base of the jugal projecting perpendicular to ihe maxilla, its anterior edge originating at the M!-M? contact before sloping gradually posteri- 19,7-26.0| 3 | 15.0-16.3| - 12.6-14.1 | : orly, with its posterior edge perpendicular to the maxilla at the M*-M? junction, A few fragments of the orbital surface preserved on the upper part of the specimen, A minute palatal sinus level with the posterior edge of P? alveolus 12 mm from the lingual edge of the tooth row, Only the left side of the palate preserved, extending fram the toath row to near the midpalatal suture. COMPARISON. QMF30882 compares well with Woods (1958) diagnosis for P. azael and its molar dimensions (Table 1) lie within the ranges for P, azael and outside those of other species of Palorchestes. When compared with a cast of the holotype (BM46316) and description by Owen (1873, pl. $2 fig. 1), the Riversleigh specimen is larger in all dimensions but is otherwise similar, The holotype is the smallest of the comparative sample of P. azael examined. QMF30882 is most similar 1o AMF452, from Wellington Caves (Dun, 1893, pl, 16), in which the only notable difference is the lingual acces- sory midlink being only slightly developed in M!, less so in Mand absent in M3. QMF30882 shows a similar gradieal of development from well-de- veloped anteriorly to simplest posteriorly. QMF772 (Woods, 1958, fig. 1) also shows it lingual accessory midlink gradient. and is inter- mediale in development between the other two specimens. SAMP31370 and 31371 (Pledge. 199], fig. 4) also show the lingual midlink, P186593 from the Wyandotte LF (McNamara. 1990) does not show the midlink at all, and itis unclear in the worn holotype RM46310 and AMP?7272. The extreme expression of thts cher- 318 MEMOIRS OF THE QUEENSLAND MUSEUM @ Sites with A azae/ D Sites with P. spp. other than F azae/ 7 a a 4, Alcoota f Cuddy a (2) 1 i rings 2. Wellington Coven @,B)3 3 Wee Jasper Caves (f) 1 14. Grong Grong (a) 1, 15. Buchan Caves (a £) 4 16, Gipp's Land, Jambo River (@) 1 17. Sorrento (a) 1 18. Werribee (a) Mi 19, Hamilton Grange Burn (7% 1 20. near Keilor (@) 1 2h Teaia a4 ing Creek (%) 4 E lt Dein a B)2 a4 Goulden's Hole Cave (6) 2 25. Henschke Fossil Cave (?a) 2 26. Victoria Cave (@ 1 27. Thebarton (4) 1 28. Toolapinna Fauna; Warburton River (al) 1 29. Guramujker g 1 30. Scotchtown Cave (%) 4 31. Mowbray Swamp (%2) 1 32. Pulbeena Swamp ( uncertain locality (a,P,%) 26 specimens KEY TO SPECIES SYMBOLS P. azael, Pct, azae/ E pavus, P çf. parvus at sslestiag P sp., P sp, nov, P. sp. indet. FIG. 2. Distribution of Palorchestes in Australia. The minimum number of individuals ateach site (number after the species symbol in the site listing) includes published and unpublished specimens. acter in QMF30882 is within the intraspecific range of variation. The large variation in molar size for P. azael (Table 1), coupled with intraspecific variation support Woods’ (1958) view that it is a highly variable species. Molar morphology grades along the tooth row from complex anterior molars with additional links to simple diprotodontid-like teeth posteriorly. Sample sizes are too small to estimate coefficients of variation for molar dimensions, or identify sexual dimorphism. The Riversleigh specimen is larger than P. parvus from the early Pliocene Chinchilla Sand (Woods, 1958) (Table 1). Its molars are relatively broader resulting in squarer molars in occlusal view. The lack of a double hindlink on the M! as in QMF789 (Darling Downs;Woods, 1958, fig.4) and NMYP48987 (Buchan Caves) also dis- tinguishes the two species. The Riversleigh specimen differs from P. selestiae, from the early Pliocene Bluff Downs Local Fauna (Mackness, 1995), in being about 25% smaller and in the lingual forelink being in contact with the anterior cingulum rather than terminating in the cingular basin. Itdiffers from P. painei, from the late Miocene at Alcoota (Woodburne, 1967) in being signifi- cantly larger (Table 1), with more complex molar morphology including higher crowns, haying a double forelink on M1, and in having a single midlink on M2. DISCUSSION. Over 80 Plio-Pleistocene Pal- orchestes specimens have been registered with museums from at least 32 open sites and cave deposits throughout Australia (Fig.2). P. azael, the most widely distributed species, is present in 21 of the 32 sites, Few sites with P. azae/ have radiocarbon dates available. Published dates range from PALORCHESTES AZAEL FROM RIVERSLEIGH 19.8007 390 BP at Spring Creck (Flannery & Gott, 1984) to 54,200 +11,000 -4,500 BP at Pulbeena Swamp (Banks et al., 1976) and 30,400 +750 -700 BP at Wyandotte (McNamara, 1990). Specimens at Wellington Caves, Naracoorte Caves (Wells et al., 1984) and the Warburton River (Toolapinna Fauna; Tedford et al, 1992) could be considerably older. The presence of 2 species of Palorchestes al Cement Mills (Bartholomai, 1977), Wellington Caves. Buchan Caves, and Strathdownie (Flann- ery & Archer, 1985) has been interpreted by some to mean that they are mixed Pliocene and Pleistocene assemblages. Alternatively P. parvus may have extended into the Pleistocene as sug- gested by Bartholomai (1977) or Pleistocene P. parvus may be incorrectly assigned. Although widespread, P. azael specimens are Ture at any one site, generally being represented by only one or two individuals and no other fossils (at [east 8 known sites) or a single speci- men of P. azael with associated faunas (at least $ sites), The low density may be an artefact of preservation, but could indicate that P. azael was a solitary animal (Flannery & Archer, 1985), At Terrace Site P. azael is associated with the large to medium-sized herbivores ina fauna dom- inated by riverine turtles, crocodiles, water rats and fish (Archer etal., 1994), The palaeoenviron- ment is interpreted to have been similar to thal which characterises the area today (Archer ct al., 1994), The other northern Queensland assem- blage, the Wyandotte Local Fauna (McNamara. 1990), contains a similar faunal assemblage. More southern faunas are dominated by large browsing and grazing mammals (¢.g. Diprotodon optarum and diverse Macropus, Proremnodon and Sthenurus spp.) in open sclerophyll forest (Bartholomim, 1977), Eucalyptus woodland (Banks ey al,, 1976), low heath (Flannery & Gon, 1984) and wees & shrubland (Dodson et al., 1993). P. azael evidently tolerated a wide range of climate conditions and habitat types. ACKNOWLEDGEMENTS We acknowledge support from: Australian Re- search Grant Scheme, The University of NSW, the National Estate Grants Scheme (Queensland), the World Heritage Unit in Canberra, the Depart- ment of Environment, Sports and Territories, the Queensland National Parks and Wildlife Service (puricularly Paul Sheehy), the Waanyi People and the Carpentaria Land Council, ICI Australia, the Australian Geographic Society, the Queens- $19 land Museum, the Australian Museum, Century Zine (particularly Doug Fishburn), MIM, the Mount Isa City Council, Surrey Beatty & Sons, the Riversleigh Society; private supporters in- cluding Elaine Clark, Sue & Jim Lavarack, Sue & Don Scott-Orr, Margaret Beavis and Manin Dickson: research colleagues notably Henk Godthelp, Suzanne Hand, Alan Bartholomai, Phil Creaser, Peter Murray, David Ride, Sue Solo. mon, Arthur White, Anna Gillespie, Virginia O’ Donoghue, Cathy Nock, Syp Praseuthsouk and Stephan Williams; and postgraduate students working on Riversléigh fossil materials wha have generously shared their understanding including Bernie Cooke, Jeanette Muirhead. Paul Wills and Anita Van der Meer, Particular thanks are owed to Sue and Jim Lavarack who spent many years slogging lo and from Terrace Site as leaders of the “Terracists’. LITERATURE CITED ARCHER, M. 1984. The Australian marsupial radis- tion. Pp, 633- 808. In Archer , M. & Clayton, G (eds), Vertebrate zoogeography and evolution in Australasia, (Hesperian Press;Perth), ARCHER, M.. GODTHELP, 1L, HAND, SI. & MEGIRIAN, D. 1989. Fossil mammals of Riversleigh, northwestem Queensland. prelimi- nary overview of biostratigraphy, correlation and environmental change, The Australian Zoologist 25: 29-65. 1994, Riversleigh, 2nd ed. (Reed:Sydney). BANKS, M.R.. COLHOUN. E.A. & VAN DE GEER, G. 1976, Late Quaternary Pulorchestes azael (Mammalia, Diprotodontidae) from northwestern Tasmania. Alcheringa L: 159-166. BARTHOLOMAL A. 1977. The lossil vertebrate fauna froin Pleistocene deposits at Cement Mills, Gore, southeastern Queensland. Memoirs of the Queensland Museum 18: 41-51. DODSON. J.. FULLAGAR, Ra FURBY, J.. JONES, R. & PROSSER, I. 1993. Humans and megafauna in a fate Pleistocene environment from Cuddie Springs, north westem New.South Wales. Archae- ology in Oceania 28; 94-99. DUN, W.A, 1893, On palatal remains of Palerchesies aael, Owen, from the Wellington Caves hone deposit. Records of the Geological Survey of New South Wales 3: 12-124, FLANNERY, T.F. & ARCHER, M. 1985. Palerchesies Owen, 1874, Large and small palorchestics, Pp 234-239, In Rich, P.V. & Van Tets, G. (eds). Kadimakara extinct vertebrates of Australia, (Pi oneer Design Studio: Lilydale, Victoria). FLANNERY, T.F, & GOTT, B, 1954. The Spring Creek Locality, southwestem Vietoria.a late sur- viving megafaunal assemblage, The Australian Zoologist 21; 385-422. 320 FLOWER, W.H. 1867. On the development and suc- cession of teeth in the Marsupialia. Philosophical Transactions of the Royal Society of London 157: 631-641. LUCKETT, W.P. 1993. An ontogenetic assessment of dental homologies in therian mammals, Pp. 182- 204. In Szalay, F.S., Novacek, M.J. & McKenna, M.C. (eds). Mammal phylogeny: Mesozoic differ- entiation, multituberculates, monotremes, early therians, and marsupials. (Springer-Verlag, New York). MACKNESS, B. 1995. Palorchestes selestiae, a new species of palorchestid marsupial from the early Pliocene Bluff Downs Local Fauna, northeastern Queensland. Memoirs of the Queensland Museum 38: 603-609. MCNAMARA, G.C. 1990. The Wyandotte Local Fauna: a new, dated, Pleistocene vertebrate fauna from northern Queensland. Memoirs of the Queensland Museum 28: 285-297. OWEN, R. 1873. On the fossil mammals of Australia part IX. Family Macropodidae; Genera Macropus, Pachysiagon, Leptosiagon, Pro- coptodon and Palorchestes. Philosophical Trans- MEMOIRS OF THE QUEENSLAND MUSEUM actions of the Royal Society of London 164: 783- 803. PLEDGE, N.S. 1991. Occurrences of Palorchestes spe- cies (Marsupialia: Palorchestidae) in South Aus- tralia. Records of the South Australian Museum 25: 161-174. TEDFORD, R.H., WELLS, R.T. & BARGHOORN, S.F. 1992. Tirari Formation and contained faunas, Pliocene of the Lake Eyre Basin, South Australia. The Beagle, Records of the Northern Territory Museum of Arts and Sciences 7: 173-194. WELLS, R.T., MORIARTY, K. & WILLIAMS, D.L.G. 1984. The fossil vertebrate deposits of Victoria Fossil Cave, Naracoorte: an introduction to the geology and fauna. The Australian Zoolo- gist 21; 305-333 WOODBURNE, M.O. 1967. The Alcoota Fauna, Cen- tral Australia, Bulletin of the Bureau of Mineral Resources Geology and Geophysics Australia 87: 1-187. WOODS, J.T. 1958. The extinct marsupial genus Pal- orchestes Owen. Memoirs of the Queensland Mu- seum 13: 177-193, PRISCILEO ROSKELLYAE SP. NOV. (THYLACOLEONIDAE, MARSUPIALIA) FROM THE OLIGOCENE-MIOCENE OF RIVERSLEIGH, NORTHWESTERN QUEENSLAND ANNA GILLESPIE Gillespie, A., 1997 06 30. Priscileo roskellyae sp. nov. (Thylacoleonidae, Marsupialia) from the Oligocene-Miocene of Riversleigh, northwestern Queensland. Memoirs of the Queens- land Museum 41(2): 321-327. Brisbane. ISSN 0079-8835. Upper dentition of the marsupial lion Priscileo roskellyae sp. nov. from the Upper Site Local Fauna of Riversleigh provides the first detailed information about upper dentition of the genus. The upper adult dental formula is 11-3, C1, P1-3, M1-4. This species is smaller than P. pitikantensis and is the most plesiomorphic thylacoleonid. Relative to Wakaleo, P3 is less bowed and molars are square and have a metaconule at their posterior margin. LIThylacoleonidae, Wakaleoninae, Priscileo, Riversleigh, Oligocene, Miocene. Anna Gillespie, School of Biological Science, University of New South Wales, New South Wales 2052, Australia; received 15 February 1997. The marsupial lion genus Priscileo Rauscher, 1987 contains only P. pitikantensis Rauscher, 1987, from the late Oligocene Ngapakaldi Local Fauna, S. Aust which is known only from a max- illary fragment, a few teeth, and a number of post cranial elements. Additional Priscileo material has been recovered from the Oligocene-Miocene of Riversleigh, northwestern Queensland. This material includes a near complete skull from the Upper Site Local Fauna representing Priscileo roskellyae sp. nov. Dental terminology for molars and the last pre- molar follows Luckett (1993) who has shown that a molariform dP3 (M1 of Archer, 1978) in mar- supials is replaced by P3. Accordingly, the re- maining molars represent M1-4. However, homology of the other premolars follows Flower (1867). Material is housed in the Commonwealth Palaeontological Collection, Bureau of Mineral Resources, Canberra (CPC), Northern Territory Museum (NTM), Queensland Museum (QMF), South Australian Museum (SAMP), Museum of Palaeontology, University of California, Berke- ley (UCMPB). SYSTEMATICS Superorder MARSUPIALIA Illiger, 1811 Order DIPROTODONTIA Owen, 1866 Family THYLACOLEONIDAE Gill, 1872 Priscileo Rauscher, 1987 TYPE SPECIES. Priscileo pitikantensis Rauscher, 1987. DIAGNOSIS. Small; P3 length usually less than 12mm; M1 and M2 relatively square in basal outline, with a posterolingual metaconule; ante- rior root of the zygomatic arch projecting an- terolaterally dorsal to M2-3. Priscileo roskellyae sp. nov. (Figs 1-3) MATERIAL. Holotype QMF23453, a skull with left and right P3, M1-2, alveoli for left and right I2-3, C1, P1-2, M3-4, and partial alveoli for the left and right I1 from early Miocene Upper Site, Godthelp Hill, Riversleigh. ETYMOLOGY. For the former Australian Minister of Arts, Sport, the Environment, Tourism and Territories, the Hon. Ros Kelly, who provided significant support for the Riversleigh Project. DIAGNOSIS (by comparison with the type and only other species). Smaller; P3 approximately 2/3 as long; metaconule on M1 and M2; alveolus of P1 closer to the P2 alveolus than to the canine alveolus; lingual root on M1 smaller, not intrud- ing as far medially into the palate. DESCRIPTION. Upper dentition. Formula 11-3, C1, P1-3, M1-4. Alveoli for I] large and incom- plete. Alveolus for I2 smallest of incisor alveoli. Alveolus for C1 ovoid, larger than alveolus for 13, smaller than that for I1. Canine alveolus closer to I3 alveolus than to P1 alveolus. Two small alveoli for two single-rooted premolars between the canine and P3. C1 and P1 alveoli separated by an approximately 3 mm, P2 alveolus close behind that for P1, abutting the anterior base of P3. P3. As in Wakaleo but 25-66% smaller. Posterior portion only slightly broader than the anterior MEMOIRS OF THE QUEENSLAND MUSEUM FIG. 1. Priscileo roskellyae sp. nov., holotype, QMF23453, partial skull, from Upper Site, Godthelp Hill, Riversleigh. Ventral view stereopair, showing upper dentition. portion (unlike Wakaleo where the posterior is much broader). Longitudinal blade slightly in- wardly-curved in contrast to the distinctive in- wardly-curved blade of most Wakaleo. Relatively uniform width with gently curved lon- gitudinal blade giving P3 rectangular shape. Lon- gitudinal blade running between 2 major cusps. In buccal view, the shearing blade of P3 W- shaped, the longitudinal blade forming the rise in the middle, and the anterior and posterior blades ascending from the major cusps at each end. The anterior cusp 1s slightly higher than the posterior (as in W. alcootaensis). In W. vanderleueri the cusps are approximately equal in height. Three vertical blades, one anterior, one lingual, and one buccal, ascending from the anterior cusp to the base of the crown. Anterior blade curving lin- gually as it ascends, bending slightly posteriorly before merging with the base of the crown. Lin- gual curvature of blade producing a small vertical lip towards the base of the blade. Similar lip in some Riversleigh Wakaleo. Posterior to this blade lingual face of the crown curving concavely forming an anterolingual basin. Lingual blade curving slightly anteriorly as it ascends and merges with the base of the crown. Many T. carnifex and T. crassidentatus also with lingual blade in contrast to W. alcootaensis and W. vanderleueriin which it is absent. Posterior to the lingual blade the lingual flank of P3 formiing a sharp depression extending from the base to the crown. Lingual flank following the curve of the posterior root, curving convexly to the posterior border. Buccal blade of the anterior cusp ascend- ing in a slight posterior direction, merging with the crown midway up the tooth. Short buccal blade also ascending from the posterior cusp, merging with the crown midway up the tooth, not as prominent as the anterior buccal blade. Similar anterior and posterior buccal blades in Wakaleo. Posterior blade running posterolaterally from the posterior cusp to the posterior margin of the tooth, an extension of the longitudinal blade, contiguous with the preparacrista of M1, in P3 of Wakaleo but absent in Thylacoleo. Short oblique blade at termination of posterior blade at the posterior margin of P3, ascending anterolaterally on the buccal flank, merging midway with the base of PRISCILEO ROSKELLYAE SP. NOV. FROM RIVERSLEIGH TABLE |. Dimensions of upper cheek teeth of species of Priscileo and Wakaleo. Data from Clemens & Plane (1974), Archer & Rich (1982), Rauscher (1987) and Murray & Megirian (1990). Measurements for P. pitikantensis (except P3) were taken from froma caste of UCMP88448. a=alveolus measurement. Length (mm) | P |m | m2 | m3 | ma [Pmi P. roskellyae 8.2 | 5.8 | 43 QMF23453 65a | 53 | 3.7a| 3.5a| 1.69 33a | 29a | 1.41 . pitikantensis UCMP88448 |11.0a W. alcootaensis NTMPI |233ļ147|72| 2 | - | 158 W. vanderleueri CPC26604 | 17.9 1.59 Riversleigh Wakaleo QMF24680 |12.9a| 8.9a | 6.3 | 4.9a | 4.3a | 1.44 126| 96 | 60 |52| - | 131 QMF23446 | omr23443 |122| 94 | 60 | 45 |35a| 130 | the crown, lacking in Thylacoleo and W. al- cootaensis. W. vanderleueri with a small blade in a similar position, differing by commencing a short distance before the end of the posterior blade, more vertically oriented. On the buccal flank of P3, a broad valley running between the anterior and posterior buccal blades, much broader, in Wakaleo. Molars. Left and right M1 and M2 relatively unworn. Alveoli for M3 and M4 indicating 3 roots for each; 2 equal anterior roots, with slightly larger posterior one. In Wakaleo vanderleueri anterior roots of M3 larger. Molar gradient steep, similar to P. pitikantensis and Wakaleo. M1 and M2 square, unlike the triangular molars in Wakaleo. Molar morphology similar to Wakaleo. M1 wider anteriorly than posteriorly. Paracone highest cusp. Small blade ascending anteriorly from the paracone to the anterior edge, joining stylar cusp B, contiguous with blade ascending from the posterior cusp of P3. Postparacrista run- ning posteriorly, meeting the ascending pre- metacrista, forming a notch midway along the straight centrocrista. Metacone well-developed. Postmetacrista ascending posteriorly from the metacone to the posterior margin of M1. Buttress- ing the steep lingual face of the paracone a short crescent-shaped preparaconulecrista straighten- ing medially, terminating at a paraconule. Short postparaconulecrista running posteriorly into the trigon basin. W. vanderleueri and Riversleigh Wakaleo with much straighter and longer blade ascending lingually from the paracone. Pre- protocrista arising from the anterior edge of the trigon basin medial to the paraconule, running posteromedially to the protocone, From the pro- tocone a postprotocrista running posteriorly to a metaconule. Postmetaconulecrista curving posterolaterally from the metaconule, ascending to merge with the posterior margin of the postmetacrista. Short crescent-shaped ridge as- cending lingual face of metacone, terminating midway between metacone and metaconule. M1 of W. vanderleueri and Riversleigh Wakaleo with similar ridge. Blades joining the 4 major cusps forming margins of a deep, square trigon basin. Within the basin, fine enamel crenulations radiate outwards. Similar crenulated trigon basins occur in Wakaleo and T. crassidentatus. Buccal flank of M1, especially anteriorly, swollen resulting in a broad base for the stylar shelf. A stylar basin running from the posterobuccal face of the paracone to the posterobuccal margin of the metacone, becoming shallower posteriorly. Lat- eral wall of the basin lined with small vertical ridges. M2. M2 smaller than M1, lacking the distinctive broad stylar shelf and basin, Paracone highest cusp, more lateral than in M1. Short, semicircular preparacrista running anteromedially from the paracone. Small ridge buttressing lingual base of the paracone. Medial to this ridge a preprotocrista arising, running posteromedially to the pro- tocone. Protocone prominent, more anteriorly placed than in M1. Protocone lingually bulbous as in P. pitikantensis and Wakaleo. A narrow anterior shelf running from the base of the pre- paracrista to the protocone. Postprotocrista run- ning posterobuccally to a gently-rounded, posterior metaconule, producing squaring of the lingual outline. No metaconule on M2 of P. pitikantensis but could be lost by damage to rear portion of the tooth. Short postmetaconulecrista running posterolaterally, merging with the poste- rior margin of the tooth. Metacone rounded and more posteriorly situated than in M1, with lingual blade running medially, connecting with a small blade running laterally from the metaconule, forming anterior border of a small, oval, basin at the posterior. Postparacrista running posterolaterally to the buccal margin of the stylar shelf, converging with anterolaterally orientated premetacrista. Deep, square, trigon basin be- tween the major cusps with fine enamel crenula- MEMOIRS OF THE QUEENSLAND MUSEUM FIG. 2. Priscileo raskellyae sp. nav., holotype, QMF23453, measurements (mm) of left tooth row. P? on left, M? and M-. tions concentrated on the lateral and posterior walls. Crenulated basins in M2 of P. pirikantensis and Wakaleo, On the buccal margin of M2 a small, elongate, stylar basin parallel to the posiparacrista, terminating midway between the paracone and metacone. Similar basin in Riversleigh Wakaleo, difficult to discern in the heavily worn M2 of W, vanderleueri. COMPARISON. P. roskellyae differs from all species of Waka/eo in: being smaller; the P3 of P. roskellyae 1s approximately 1/3 length of P3 of W. alcootaensis, 1/2 the length of P3 of W. vanderleuveri, and 2/3 the length of P3s of Riversleigh Wakaleo; having a lingual crest as- cend from the anterior cusp of P3; having the shearing blade of P3 straighter; having the poste- rior root of P3 only slightly enlarged; having the posterior half of P3 relatively square in basal outline; having M1 and M2 relatively square in basal outline, P. roykellyae differs from Thy- lacaleo in: being smaller; P3 being approxi- mately 1/3 length of P3 of T. hilli and 1/6 length of P3 of T, carnifex, having M3+4, having M1+2 relatively square in basal outline. DISCUSSION The diagnosis of Priscileo is amended to include P. roskellyae. Rauscher (1987) distinguished Priscileo from other thylacoleonids by M4, a P3/M1 length ratio less than 1.70, and M2 with a crenulate, anteropostenorly broad trigon basin. Some of the new Riversleigh Wakaleo specimens have these features. Two Wakaleo specimens have an M4 (Table 1), and most have a relatively broad crenulated basin on M2. All species of Wakaleo have a P3/M1 ratio of Jess than 1.70. The value for P. roskellyae (1,41) falls midway within this range. Rauscher (1987) distinguished Priscileo from Wakaleo by the loss of P1 or 2 and M4 in the latter, Although some new Riversleigh Wakaleo specimens have these teeth, the plesiomorphic features of P. roskellyae exhibits (significantly smaller size, square molar shape, metaconule and relatively straight cutling blade on P3) require generic distinction, P. pitikantensis and P. roskellyae share generic features of dental dimen- sions, shape of M2 and position of the anterior base of the zygomatic arch. Specific distinction of P. roskellyae is based on the size difference between these two species,P. pitikantensis being 33% larger. INTRAFAMILIAL RELATIONSHIPS Rauscher (1987) found no synapomorphies unit- ing Priscileo and Wakaleo, or uniting Priscileo and Thylacoleo. Analysis of Priscilea and Thy- lacoleo included comparison of postcranial ma- terial and for a number of character-states, Thylacoleo exhibited the plesiomorphic condi- tion while Priscileo was derived. Rauscher con- PRISCILEO ROSKELLYAE SP. NOV. FROM RIVERSLEIGH 325 5mm FIG. 3. Priscileo roskellyae sp. nov. holotype, QMF23453, left cheek dentition. Ib=longitudinal blade; absan- terior blade: pb=posterior blade; ac=anterior cusp; pe=posterior cusp; abb=anterior buccal blade; linb=lingual blade; pbb=posterior buccal blade; alb=anterior lingual basin; pa=paracone; pacl=paraconule: prpac=pre- paracrista; popac=postparacrista; prpacie=preparaconulecrista; popacle=postparaconulecrista; pr=protocone: mel=melaconule;, pamelc=postmetaconulecrista; me=metacone; pome=postmetacrista; prmc=premetacrista: stB=stylar cusp B; trb=trigone basin. cluded that Thylacoleo's primitive features were secondarily derived and not a retained plesiomorphic condition. Because Wakaleo and Thylacoleo share the synapomorphy of loss of M4, Rauscher (1987) considered them to be sister groups. Priscileo was regarded as the sister group of a Wakaleo/Thylacoleo clade. Rauscher (1987) suggested that loss of the metaconule could be a diagnostic feature for the Thylacoleonidae based on the tritubercular upper molars of Wakaleo and Priscileo and the second- arily quadritubercular M1 of Thylacoleo, How- ever, MI and M2 of P. roskellyae have a metaconule and are basically square in outline, The metaconule on these molars, especially on M2, is posteriorly positioned and its seeming absence from M2 of P. pitikantensis may be a result of damage which is evident at the rear margin of this tooth. Consequently, loss of the metaconule can no longer be considered a syn- apomorphy for the family. Murray et al. (1987) placed Wakalea and Thy- facoleo in separate subfamilies: the Waka- leoninae includes Wakaleo, and the Thylacoleoninae which includes Thylacoleo and possibly Priscileao, Wakaleonmes were regarded to differ from thylacoleonines in absence of P| and formation of a tympanic wing composed of alisphenoid and squamosal contributions. Fea- lures distinguishing thylacoleonines from wakaleonines. include P1, squamosal contribu- tion to the tympanic wing and frontal-squamosal contact on the lateral cranial wall, The new thy- lacoleonid specimens from Riversleigh indicate that, in terms of dental morphology, Priscileo exhibils no features that prevent it from being ancestral to Wakalea and Thylacoleo. The dental features of Priscileo, including small P3 and molar size, square (nearly bunodont) molar shape, metaconule, and full premolar and molar complement are almost certainly plesiomorphic features within the family. However, these same features clarify some questions about relationships of the family with the Order Diprotodontia. It was once commonly believed that thylacoleonids evolved from a phalangerid-like diprotodontian which had quadritubercular upper molars including a hyper- ue [S koa trophied metaconule (Krellt, 1872; Broom, 1898: Bensley, 1903; Ride, 1964, Archer, 1976). Archer & Rich (1982) hypothesised that the fritubercular shape of the molars of W. aleooraen- sis were secondarily derived Irom an ancestral quadrituberculur shape through suppression of the metaconule. It has been suggested that the triangular molars of Wakaleo are plesiomorphic for the family (Murray etal., 1987). The primitive dental features of P. raskellyae, especially the metaconule and square molar shape, provide sup- port for Archer & Rich (1982). Priscilea and Wakaleo have been collected from Riversleigh’s System B sites indicating overlapping of early Miocene thylacoleonid lin- cages. Temporal overlapping of species of Fhy- lacoleo (T. crassidentatus and T, hilli) also occurred in the early Pliocene (Archer & Daw- son, 1982). In each case there was a distinct size dilference in the lineages involved. In terms of P3 length, System B specimens of Wakalee are ap- proximately 1.5 times larger than P. roskellyae. Similarly, P3 of 7. crassidentatus is twice (he length of thartooth in T. Hilli (Pledge, 1977), Itis possible that size differences of this magnitude among sympatric thylacoleonids were an import- ant factor in reducing competition. Morphologi- cal studies of the limbs of P, pitikantensts suggest it was arboreal (Rauscher, 1987). Wakaleo, being larger and No doubt heavier, may have been more terrestrial. Murray & Megirian (1990) also inti- mated a terrestrial existence for Wakaleo based on the wrist joint and heavily-worn dentition which they consider may indicale a scavenging mode of life. ACKNOWLEDGEMENTS I thank Michael Archer and Henk Godthelp for constructive comments and Stephan Williams, Karen Black and Jenni Brammall for assistance with photography. Support for research at Riversleigh has come from the Australian Re- search Grant Scheme; the National Estate Grants Seheme (Queensland); the University of New South Wales; the Commonwealth Department of Environment. Sports. and Territories; the Queensland National Parks and Wildlife Service; the Commonwealth Heritage Unit, ICT Australia; the Australian Geographie Society; the Queens- land Museum; the Australian Museum: the Royal Zoological Sociely of NSWsthe Linnean Society of NSW; Century Zinc; Mount Isa Mines; Surrey Beatty & Sons; the Riversleigh Society: and pri- vale supporters including Elaine Clark, Margaret MEMOIRS OF THE QUEENSLAND MUSEUM Beavis, Martin Dickson, Sue & Jim Lavarack and Sue & Don Scott-Orr. Invaluable assistance in the field has come from numerous volunteers, stalf and postgraduate studénis of the University of New South Wales. LITERATURECITED ARCHER, M. 1976. Phascolarctid origins and the po- tential of the selenodont molar in the evolution of diprotodont marsupials, Memoirs of the Queens- land Museum 17: 367-371. 1978. The nature of the molar-premolar boundary in marsupials and a reinterpretation of the homol- ogy of marsupial cheekteeth, Memoirs of the Queensland Museum 18: 157-164, ARCHER, M. & DAWSON, L, 19823, Revision of marsupial lions of the genus Thy/acolee Gervais (Thylacoleonidae, Marsupialia) and thy- lacoleomd evoluGon in the late Cainozoic, Pp. 477-494, In Archer, M. (ed.), Curnivorous marsus plals. (Royal Zoological Society of New South Wales: Sydney). ARCHER, M., HAND, J.H. & GODTHELP. H. 1991. Riversleigh. (Reed Books: Sydney). 1994, Patterns inthe history of Australia’s mammals and inferences about palacohabitats. Pp, 80-103, In Hill, R. (ed), History of Australian vegetation, (Cambridge University Press; Cambridge), 1995, Tertiary environmental and biotic change in Australia. Pp 77-90. In Vrba, E.S., Denton, G-H., Partridge, T.C. & Burckle, L.H, (eds), Paleoeli- mate and evolution, with emphasis on human origins, (Yale University Press: New Haven). ARCHER, M. & RICH. T.H. 1982. Results of the Ray E. Lemley Expeditions. Wakaleo alceottensis n. sp. (Thylacoleonidae: Marsupialia), a new marsu- pial lion from the Miocene of the Northem Teri- tory. with a consideration of the early radiation of the family, Pp, 495-502. In Archer, M. (ed.), Car- nivorous marsupials, (Royal Zoological Society of New South Wales: Sydney). BENSLEY, B.A, 1903 On the evolution of the Austri- tian Marsupialia; with remarks on relationships of the marsupials in general. Transactions of the Linnaean Society of London (Zool) Ser, 29; 83- 217 BROOM, R. 1898. On the affinities and habits of Thy- lacoleo . Proceedings of the Linnean Society uf New South Wales 22: 57-74. CLEMENS, W.A. & PLANE, M, 1974, Mid-Tertiary Thylacoleonidae (Marsupialia, Mammalia), Jour- nal of Paleontology 48; 652-60. FLOWER, W.H, 1867, On the development and sut- cession. of teeth in the Marsupialia. Philosophical Transactions of the Royal Society of London 157: 631-641. GITTLEMAN, JL. 1989. Carnivore group living: comparative trends. Pp. 183-207. In Gittleman, PRISCILEO ROSKELLYAE SP. NOV. FROM RIVERSLEIGH J.L. (ed.), Carnivore behaviour, ecology and evo- lution. (Cornell University Press: Ithaca). GUGGISBERG, C.A.W. 1975, Wild cats of the world. (David & Charles: London). KREFFT, G. 1872. A Cuverian principle in palaeonto- logy, tested by evidences of an extinct leonine marsupial (Thylacoleo carnifex) by Professor Owen etc, Reviewed by Krefft, Annals and Mag- azine of Natural History 4:169-182. LUCKETT, W.P, 1993. An ontogenetic assessment of dental homologies in therian mammals. Pp. 182- 204. In Szalay, F., Novacek, M.J. & McKenna, M.C. (eds), Mammalian phylogeny. Mesozoic differentiation, multituberculates, monotremes, early therians, and marsupials. (Springer-Verlag: New York). MURRAY, P., WELLS, R. & PLANE, M. 1987. The cranium of the Miocene thylacoleonid, Wakaleo vanderleueri ; click go the shears-a fresh bite at thylacoleonid systematics. Pp. 433-466. In Archer, M. (ed.), Possums and opossums: studies in evolution. (Surrey Beatty & Sons and the Royal Zoological Society of New South Wales: Sydney). 327 MURRAY, P. & MEGIRIAN, D. 1990. Further obser- vations on the morphology of Wakaleo vanderleueri (Marsupialia: Thylacoleonidae) from the mid-Miocene Camfield Beds, Northem Territory. The Beagle 7: 91-102. PLEDGE, N. 1977. A new species of Thylacoleo (Marsupialia, Thylacoleonidae) with notes on the occurrence and the distribution of Thy- lacoleonidae in South Australia. Records of the South Australian Museum 17: 277-283. RAUSCHER, B. 1987. Priscileo pitikantensis, a new genus and species of thylacoleonid marsupial (Marsupialia: Thylacoleonidae) from the Miocene Etadunna Formation, South Australia. Pp. 423- 432, In Archer, M. (ed.), Possums and opossums: studies in evolution. (Surrey Beatty & Sons and the Royal Zoological Society of New South Wales: Sydney). RIDE, W.D.L. 1964. A review of Australian fossil marsupials. Journal and Proceedings of the Royal Society of Western Australia 47:97-131. ZYZOM YS RACKHAMI SP. NOV. (RODENTIA, MURIDAE) A ROCKRAT FROM PLIOCENE RACKHAM'S ROOST SITE, RIVERSLEIGH, NORTHWESTERN QUEENSLAND H. GODTHELP Godthelp, H., 1997:06:30. Zyzomys rackhami sp. nov. (Rodentia, Muridae) a rockrat from Pliocene Rackham’s Roost Site, Riversleigh, northwestern Queensland. Memoirs of the Queensland Museum 41(2): 329-333. Brisbane. ISSN 0079-8835. Zyzomys rackhami sp. nov. from the Pliocene Rackham’s Roost Site, Riversleigh, northwest- em Queensland is the first fossil member of this genus and only the second Tertiary murid described from Australia. It is known from many hundreds of dental fragments recovered as a part of an ancient megadermatid roosting cave. It appears to be the most plesiomorphic member of the genus and is part of a diverse suite of extinct murids from this site. M Rodentia, Muridae, Zyzomys, Pliocene, Riversleigh. H. Godthelp, School of Biological Science, University of New South Wales, New South Wales 2052, Australia; 4 November 1996, The murid genus Zyzomys contains 5 living spe- cies which are restricted lo tropical Australia. Zyzomys has been placed in the tribe Conilurini (Baverstock, 1984) and is an ‘Old Endemic’ sensu Ride (1970) or an ‘Older Immigrant’ of Tate (1951). Along with Mesembriomys and Còn- turus, Zyzomys comprises a group of taxa unified hy a number of cranio-dental characters, phallic morphology (Lidicker, 1987) and chromosomes (Baverstock et al., 1981). Dental nomenclature follows Musser (1981), Measurements are in millimetres. Unless other- wise Stated material is housed in the Queensland Museum (QMF). SYSTEMATICS Order RODENTIA Bowdich, 1821 Suborder MYOMORPHIA Brandt, 1855 Infraorder MYODONTA Schaub, 1958 Superfamily MUROIDEA Miller & Gidley, 1918 Family MURIDAE Gray, 1821 Subfamily MURINAE Gray, 1821 Zyzomys Thomas, 1909 Zyzomys rackhami sp. nov. (Figs 1-2, Table 1) MATERIAL. Holotype QMF108 18, partial left maxil- lary with MI-3, ETYMOLOGY. For Alan Rackham, the discoverer of the Rackham’'s Roost Site. Paratype QMF10819, left maxillary fragment with M* and zygomatic plate, Other material QMF23365, partial left maxillary with M12: QMF10821, partial right maxillary with M!-3 QMF23325, partial right maxillary M!-3; QMFIO819, left M}; QMF24001, partial left maxillary with M12; QMEF24004, partial left maxillary with M!: QMF24002, right M!; QMF24003, right M}. All from Rackham's Roost Site (19°02'09" $, 138°41°60" E) at Riyersleigh, NW Queensland. The site represents the remnants of an ancient cave which has largely eroded away leaving the indurated floor sediments exposed. The sediments of the floor contain myriad bones and teeth, mostly fragmented, which are interpreted to be megadermatid (Macroderma and Megaderma spp.) prey remains (Godthelp, 1988; Hand, 1994). Its age is Pliocene on the basis ol a macropodid similar to Protemnodon snewini Bartholamai (1978) which species occurs in the early to middle Pliocene Bluff Downs Local Fauna (Archer & Wade, 1976; Bartholamai, 1978). DIAGNOSIS. Zyzomys rackhami differs from other species of the genus by the following com- bination of characters; Relatively well-developed series of cusps (T3,6,9) particularly T3,relatively small proportions of the lingual series of cusps (1,3,7), reduced molar gradient, frequent T1bis; tooth row short and narrow. DESCRIPTION. Small to medium-sized dental arcade arcuate and concave lingually with jhe internal edge of M? as the most lingual point of the tooth row. M! longer than M2. M? small and relatively reduced. All cusps and cusp complexes with marked posteriorly inclined slant except in M? with cusps nearly vertical. Molar overlap minimal. M!, Relatively long and narrow often with u prominent anterior cingulum which forms a semi- 330 EHT= 20.0 KY WD= 26 mn 2.0mm -———_—_—__—_ FIG, 1. QMF10818, SEM (stereo pair). circle around the T2,3 complex in the holotype. Anterior cingulum with a series of small but apparently occlusally functional accessory cusps randomly positioned. T1 elliptical with its long axis obliquely inclined to the axis of the T2,3 complex, directed to the rear of the tooth and positioned posterior to the base line of the T2,3 complex. T1 joined to the T2,3 complex only in extreme stages of wear and in some specimens via a variably sized and shaped Tlbis. Tlbis MEMOIRS OF THE QUEENSLAND MUSEUM structure in the holotype small, becoming more prominent with increased wear. T2 large and arcuate anteriorly with a straight posterior edge, more than half the width of the ante- rior portion of the tooth. T2 joined to T3 at its buccal edge. T3 small and circular, with its posterior third be- hind the posterior edge of T2, almost entirely incor- porated into the T2 com- plex with wear. T4 elliptical with its long axis inclined obliquely to the axis of the T5,6 complex and directed posteriorly. T4 occlusal surface approxi- mating T1 in size and shape if unworn, larger in worn specimens, joined to T5 in early stages of wear, nearly wholly incorporated in older individuals. TS large, with a subangular arcuate anterior edge, with posteri- orly concave posterior edge, with a smaller occlu- sal surface area than T2. T6 strongly attached to T5 even in early stages of wear, with junction of these two cusps marked by acleft on the occlusal surface that continues anteriorly and ventrally as a furrow in the enamel indicating that the distinction between T5 and T6 is never lost. T6 circular, smaller than T3, approxi- Zyzomys rackhami sp. nov., Rackham’s Roost Site, holotype, mately half of T6 behind the posterior edge of T5. T7 larger than T4, elliptical, with its long axis inclined obliquely to the axis of the T8,9 complex but in reverse to the angles of T1 and T4 and as such is directed forwards. T7 and T4 in very close proximity, forming a single complex after moderate to extreme wear. T8 large, with a nearly circular occlusal surface. T9 barely discernible, incorporated into the T8,9 complex at the onset of wear. Slight furrow in the anterior surface of the T8,9 complex marking the anterolingual edge of T9 (would remain even ZYZOMYS RACKHAM SP. NOV. FROM RIVERSLEIGH FIG. 2. Zyzomys rackhami sp. nova paratype, QMF10821, SEM showing zygomatic plate. with moderate wear). Posterior cingulum (z) ab- sent. M?, T1 large, tear-shaped, with a long axis obliquely inclined to the anterior edge of the tooth and directed posteriorly. T2 and T3 absent. T4 of moderate size, elliptical, posterior to the domi- nant TS. T4 joined to T5 in very early stages of wear, TS subtriangular in occlusal outline, with its anterior most edge of the enamel boundary contacting the posterior edge of M! in extreme wear. T6 small, circular, joined to T5 with mod- erate wear. As in the T6 structure of M!, the connection associated with a well-defined fur- row, this cusp always retaining ils identity. T7 an elliptical cusp of moderate size, with its axis running almost parallel to the main axis of the tooth, not observed to merge with the T8,9 com- plex in M? even with extreme wear. TS large, nearly circular, with distal edge extending well beyond the distal edge of T7, giving the posterior region of the tooth an arcuate shape. T9 lost, with a remnant of the furrow that marked the position of the lingual side of the cusp. Mì. T1 small, almost circular. T2 and T3 ab- sent. T4 small, tear-shaped, joined to TS after A little wear, with an axis directed across the width of the tooth, TS small, subtriangular, displaced toward the buccal edge of the tooth, T6 absent, with only a poorly defined remnant of the lingual furrow forming a weak buccal cingulum at the anterior edge. T7 moderate in size, shaped as a bisected semi-circle. T8 a mirror image of T7, with the 2 cusps joined in early wear. T9 absent. M! with 4 roots, 3 well-developed, fourth small and probably reduced; anterior root large, directed forwards, exposed occlus- ally; medio-lingual root long, narrow, posi- tioned under TI and the anterior edge of T4; medio-buccal root a remnant of a more sub- stantial root, with an alveolus for this root on all specimens examined even though there is not always a root. Posterior root large, nearly as wide as the tooth at its origin, directed towards the buccal edge of the max- ilary. M7 with 3 roots; anterolinqual and anterobuccal roots of equal size, together below the anterior margin of the tooth; pos- terior root wide, running obliquely with re- spect to the anterior roots, with lingual extremity its most posterior point. M? with 3 roots of equal size, forming a triangle with a toot at each apex. As in M? there are 2 anterior roots (buccal and lingual). Occlusal surface of the tooth row concave, with the highest points being the anterior third of M! and the M*. Anterior palatal vacuity extending distally as far the anterior edge of MÍ. Attachment node for the origin of the superficial masseter large and well defined. Anterior edge of the node alligned with the anterior edge of the zygomatic plate and just behind the maxil- lary/premaxillary suture, Zygomatic with ante- rior edge rising vertically and straight, of moderate width, “with one nutrient foramen dis- tally near its base. COMPARISON. Although the nearest species in size Z. argurus, Z. rackhami is more similar to Z. pedunculatus because of the relatively cuspidate nature of the teeth, the development of the buccal series of CUSPS, and the relatively unreduced ap- pearance of M3. These similarities are probably simplesiomorphies. A nutrient foramen is found anterior to the tooth row in all species with the exception of Z. rackhami and Z. maini, in which the foramen is on the zygomatic plate. Unlike Z, maini however Z. rackhami has a longitudinal palatal crest as in all other species of Zyzomys. Zyzumys rackhami differs from Pseudomys, 352 TABLE | Measurements (mm) of Zyzamys rackiiawn sp. nov. L=length; W=width. Mt? ! a2 SPEC. NO. L L QMF10818 53 QMF10821 56 QMF23325 53 QMF10819 QMF30065 QMF30063 MF24503 — 4.2 QMF10810 - | OMF30268 | 2. omr30265 | - | 43 | 26 | 15 | (i a a a eee O nD Mastacomys, Leporillus, Notomys, Leggadina, Rattus, Melamys and Uromys in having a well- developed T7. Z rackhamiis distinguished Irom Pagonomys by the lack of a posterior cingulum (z) and by smaller and fess pronounced buccal cusps (T3,6,9), Z, rackhami differs from Hydromysand Neromys in having 3 upper molars. Z. rackhami dilfers from Conilurus in having a well-developed T3 and T1 isolated fram and dis- lal to the T2,3 complex. Z. rackhami is removed from Mesembriomys by retaining a T9 on MÌ, close proximity of T7 to the T&.9 complex, re- duced molar overlap and less cuspidate nalure of the molars. DISCUSSION. Zyzomys rackhari is the most abundant {mainly isolated molars) rodent in the Rackham’s Roost deposit but only a few maxil- lary fragments are known. Skulls apparently break up more readily than other murids from the deposit. Lower jaws of rodents are uncommon in the deposit and not well preserved. No elements of the lower dentition of Z. rackhami have been identified. There is apparently some taphonomic process limiting the number of dentaries pre- served. This might be caused by some aspect of the feeding behaviour of Macroderma gigas, the presumed predator, or by some physical process of sorting within the original cave system. In the latter case it is possible that we might find a MEMOIRS OF THE QUEENSLAND MUSEUM concentration of these elements elsewhere in Ihe deposit. Lower jaws and teeth of other mammal groups{ i.e., marsupials, bats) found in the deposit appear ta be as abundant as their upper counter- parts. Z. rackhami is the only species of Zyzomys recovered from the Rackham’s Roost Local Fauna and is represented by hundreds of speci- mens. This is im contrast to modern and Pleistocene faunas, in which there are usually at least 2 species present and sometimes, as in the Nouralangie Rock area, N.T., 3 (Kitchener (1989) reported only 2 bul the author has sighted 2 specimens of Z. woodwardi, CM7200 and CM7200, from the area.) Areas on the east coast where only Z. argurus is now found appear to have lost a second species, Z. woodwardi, only recently, as is the case in the Chillagoe area. Z pedunculatus occurs in surficial cave deposits at Cape Range, Westem Australia along with Z. argurus which is still extant locally, In the Riversleigh area Z. argurus has been trapped and Z. weadwardi has heen recovered from Recent owl pellet deposits along the Gregory River. Both Z. argurus and Z, woodwardi have been recov- ered from Mucroderma gigas prey remains in Carrington’s Cave on Riversleigh Station. The deposits in Cartington's Cave are as yet undated bul appear to range from Recent to Pleistocene. The presence of both species in these deposits indicates thal M. gigas is capable of taking the relatively larger Z, woodwardi as prey, Pliocene M. gigas from Rackham's Roost were not differ- ent in size (Hand, 1994) from the modern popu- lation and prey size would seem to be an unlikely explanation for the absence of a second Zyzomys in the Rackham's Roost deposit, ACKNOWLEDGEMENTS Work at Riversleigh has been supported by the Australian Research Council, the Department of the Environment, Sport and Tourism, National Estate Programme Grants (Queensland), the Aus- tralian Geographic Society, ICI, the Queensland Museum and the University of NSW. Access to comparative material was kindly provided by S. Van Dyck, J. Calaby, T. Flannery, L. Gibson, D. Kitchener and P, Jenkins, SEM photographs were taken by J, Muirhead at the University of NSW. Tam grateful to A, Gillespie and S. Williams who have done much of the preparation of the mate- tial. Particular thanks are due to the Rackham family for their unwavering enthusiasm and as- sistance throughout the past years. 1 must thank ZYZOMYS RACKHAMI SP. NOV. FROM RIVERSLEIGH my colleagues M. Archer and S. Hand for their support and advice, and B. Turnbull and A. Baynes for constructive criticism of earlier drafts of this manuscript. LITERATURE CITED ARCHER, M. & WADE, M. 1976. Results of the Ray Lemley Expeditions, part 1. The Allingham For- mation and a new Pliocene vertebrate fauna from northern Queensland. Memoirs of the Queensland Museum 17: 379-397, ARCHER, M., GODTHELP, H., HAND, S.J. & MEGIRIAN, D. 1989, Perliminary overview of mammalian diversity, biostratigraphy, correlation and environmental change evidenced by the fossil deposits of Riversleigh, northwestern Queens- land. The Australian Zoologist 25: 29-65. ARCHER, M., HAND, 8.J. & GODTHELP, H. 1991. Riversleigh. (Reed: Sydney). BARTHOLOMAI, A. 1978, The Macropodidae (Marsupialia) from the Allingham Formation, northern Queensland; results of the Ray E. Lemley expedition, Part 2. Memoirs of the Queensland Museum 18: 127-143. BAVERSTOCK, P.R. 1984. Australia’s living rodents: a restrained explosion. Pp 905-913, In Archer, M. & Clayton, G. (eds), Vertebrate zoogeography and evolution in Australasia, (Hesperian Press:Perth) BAVERSTOCK, P.R., WATTS, C.H.S., ADAMS, M. & COLE, S.R. 1981. Genetical relationships 333 among Australian rodents (Muridae). Australian Journal of Zoology 29: 289-303. GODTHELP, H. 1988, Riversleigh scene 4: Rackham’s Roost — the beginnings of the modern world. Pp. 81-83. In Hand, S.J. & Archer, M. (eds), The antipodean ark. (Angus & Robertson: Sydney). 1993. Zyzomys rackhami, a new species of murid from the Pliocene Rackhams Roost deposit, northwestem Queensland. Abstracts, CAVEPS, Adelaide. 1994. The three R’s of Riversleigh: the Rackhams Roost rodents (Placentalia: Rodentia). Abstracts Riversleigh Symposium, UNSW, Sydney, 14. HAND, S.J. 1995. First record of the genus Megaderma Geoffroy (Microchiroptera: Megadermatidae) from Australia. Palaeovertebrata 24: 48-66. KITCHENER, D.J. 1989. Taxonomic apraisal of Zyzomys (Rodentia, Muridae) with descriptions of two new species from the Northern Territory , Australia. Records of the West Australian Mu- seum 14: 331-373. MUSSER, G. 1981. The giant rat of Flores and its relatives east of Bomeo and Bali. Bulletin of the American Museum of Natural History 169: 71- 175. RIDE, W.D.L. 1970. A guide to the native mammals of Australia. (Oxford University Press: Melbourne), TATE, G.H.H. 1951. The rodents of Australia and New Guinea. Results of the Archbold Expeditions. No. 65. Bulletin of the American Museum of Natural History 97: 189-424, NEW MIOCENE LEAF-NOSED BATS (MICROCHIROPTERA: HIPPOSIDERIDAE) FROM RIVERSLEIGH. NORTHWESTERN QUEENSLAND SUZANNE HAND Hand, S.J. 1997 06 30: New Miocene leaf-nosed bats (Microchiroptera: Hipposideridae) from Riversleigh, northwestern Queensland. Memoirs of the Queensland Museum 41(2): 335-349, Brisbane. ISSN 0079-8835. Two new Australian Tertiary hipposiderids are described on the basis of skull and dental material recovered from Bitesantennary Sile. a Miocene cave-fill on the Site D Plateau, Riversleigh, northwestern Queensland, The new species are closely related to Hipposideras (Brachipposideras) nooraleebus Sigé, Hand & Archer from Riversleigh’s Microsite, and the living northern Australian Rhinonicteris aurantius (Gray), One species is referred to Rhinonicteris, the other tentatively referred to Brachipposideros, The subgenus Brachipposideros Sigé is raised to generic rank to better reflect relationships of its species. C Miovene, Riversleigh, hipposiderids, leaf-nosed bats. Suzanne J. Hand, School of Biological Science, University of New South Wales, New South Wales 2052, Australia: received 4 December 1996, Ritesanteninary Site, in early Miocene (Archer ct al., 1989; Creaser, 1997) freshwater limestone on the NE edge of the Site D Plateau at Riversleigh (Hand et al., 1989; Archer et al., 1989,1994) covers approximately 150m? and contains thousands of bat skulls, limb bones and snails. Almost all are complete, suggesting fossilisation at or very near the point of accumu- lation. This deposit is interpreted as a cave-fill (Hand et al.. 1989) and contains at least 11 microchiropteran species - 10 hipposiderids and a megadermatid, At least 4 of the Bitesantennary hipposiderids are known from many hundreds of partial and complete skulls. Two of the hip- posiderids, which are morphologically similar to Microsite’s Brachipposideros nooraleebus Sigé etal., 1982, are described and their phylogenetic relationships and palaeoecology are discussed. Skull terminology follows Hand (1993, 1995); dental terminology follows Sigé et al. (1982), Stratigraphic nomenclature for the Riversleigh region follows Archer et al, (1989, 1994; Creaser this volume). The prefix QMF refers to speci- mens held in the fossil collections of the Queens- Jand Museum, Brisbane. SYSTEMATICS Suborder MICROCHIROPTERA Dobson, 1875 Superfamily RHINOLOPHOIDEA Weber, 928 192 Family HIPPOSIDERIDAE Miller, 1907 Rhinonicteris Gray, 1847 Rhinonicteris tedfordi sp. nov. (Figs 1-2, Table 1) MATERIAL, Holotype QMF22910, partial skull with RM, LM. Paratypes QMF22911, partial skull with RP -M7 god tM” , QMF22912, maxillary fragment with RC’-M", QM F22840, rostrum with LC -M7; types from early Miocene (System B) Bitesantennary Site, Other material: Bitesantennary Site: QMF22831, QMP22841, QMF22842, QMF22845, QMF22854, QMF22859, QMF22865, QMF22871, QMF22890), QMF22891, QMF22893, QMF22909. White Hunter Site (System A); QMF22921, QMF22922. RV Site (System B): QMF22930, QMF22931, QMF22932, QMF22913. Upper Site (System B): QMF2291/4. White Hunter, RV and Upper Sites are about 2km SSW of the type locality. ETYMOLOGY. For Richard Tedford, American Mu- scum of Natural History who described the first Terti- ary mammals from Riversleigh in 1967. ASSOCIATED FAUNA AND TAPHONOM Y. The cave-fill (Hand et al., 1989) at the type local- ity contains thousands of well preserved, almost complete bat skulls and limb bones, suggesting fossilisation at or near the point of accumulation, Contact between the fill and older cave wall have been identified. The deposit’s many freshwater snails suggest that the depositional area was open to light and under water for some period during its history. A travertine floor, including a large stalagmite, has been found at the base of the deposit. MEMOIRS OF THE QUEENSLAND MUSEUM FIG. 1. A-C, Rhinonicteris tedfordi sp. nov., QMF22910, holotype, from Bitesantennary Site, Riversleigh, northwestern Queensland. A, dorsal view; B, lateral view; C, ventral view. D-F, Rhinonicteris aurantius, AR 15400, Klondyke Queens Mine, Marble Bar, Western Australia. D, dorsal view; E, lateral view; F, ventral view Scale indicates 5 mm. Bitesantennary Site contains R. tedfordi, ?B. watsoni and at least 8 other hipposiderids and a megadermatid with rarer frogs, lizards, a boid, a stork, swift, peramelids, a dasyurid and a bulungamayine macropodid (Archer et al., 1994). In the complex lacustrine White Hunter, Upper and RV deposits the vertebrate faunas are much more diverse, with the Upper Local Fauna (Archer et al. 1994) one of Riversleigh’s richest. DESCRIPTION, In comparison to Miocene B. nooraleebus Sigé et al., 1982 and Recent Rhinonicteris aurantius (Gray, 1845). Skull 10-20% smaller than R. aurantius and approximately same size as B. nooraleebus (braincase may be slightly longer in R. tedfordi). Proportions similar to B. nooraleebus: rostrum wide and long with respect to braincase, approx- imately 2/3 braincase length, 2/3 maximum (mas- toid) width and twice interorbital width. Zygomatic width greater than mastoid width. Maximum height of the skull dorsal to the glenoid process as in R, aurantius. In dorsal view, poste- NEW MIOCENE LEAF-NOSED BATS, RIVERSLEIGH 337 FIG. 2. Rhinonicteris tedfordi sp. nov., QMF22912, paratype, maxillary fragment with C!-M3, from Bitesantenn- ary Site, Riversleigh, northwestern Queensland. A, oblique-occlusal view; B-B’, occlusal view, stereopairs. Scale indicates 1 mm. rior margin of the skull quadrate rather than rounded as in R. aurantius and B. nooraleebus. Rostrum distinctly lower than the braincase, more so than in B. nooraleebus but less than in R. aurantius. Rostral inflations much more promi- nent than in B. nooraleebus and R. aurantius , mainly because of the very marked groove lead- ing to a deep frontal depression delimited sharply by well-developed supraorbital ridges. R. au- rantius with inflations better developed, with very little development of supraobital ridges, with frontal depression and groove between ros- tral inflations more limited in depth and extent. Infraorbital foramen wholly above M? as in B. nooraleebus, but unlike R. aurantius (above M> 3), larger and more rounded than in B. nooralee- bus, smaller and slightly more elongated than in R. aurantius. Bar of bone forming its dorsal mar- gin (anteorbital bar; e.g. Hill 1963) straighter and wider anterodorsally than in R. aurantius (being roughly the same thickness throughout), (In R. aurantius this bone curved, about 3 times as wide posteroventrally as anterodorsally.), more curved than in B. nooraleebus, in which it is roughly the same thickness throughout and very straight. Zy- goma (as in B. nooraleebus and R. aurantius) with an enlarged jugal projection occupying much of its length, as tall as the level of the lower insertion of the anteorbital bar, with slightly con- vex posterior margin, with its anterior edge posterodorsally directed (rather than vertically). Sagittal crest well-developed (but see QMF22871), much better developed than in B. nooraleebus and different to R. aurantius, with maximal height anterior to the middle of the braincase level with the posterior zygomatic roots, not terminating as abruptly nor in a for- wardly curving projection as in R. aurantius, extending further anteriorly onto the moderately constricted interorbital region, not joining the supraorbital ridges as distinctly as in B. nooralee- bus, extending posteriorly to the lambdoidal crest, rather than attenuating in the interparietal region as in R. aurantius. 338 MEMOIRS OF THE QUEENSLAND MUSEUM TABLE 1. Skull and dental measurements (mm) of type material. H=holotype; P=paratype; two measurements in parentheses in a column indicate (left) and (right), respectively. Premaxillae not known but make a V-shaped junction (often stepped) with the maxillae rather than a rounded V-shape as in R. aurantius and B. nooraleebus. Palate shorter, with posterior mar- gin extending to the level of the metacone of M? (rather than the anterior face of M3), marked by a short postpalatal spine, as in R. aurantius. Mesopterygoid fossa narrow anteriorly, necking in before broadening posteriorly, more similar to R. aurantius than B. nooraleebus in which it is broad and rounded anteriorly and of uniform width throughout its length. Lacrimal foramen much larger than in R. au- rantius and larger than in B. nooraleebus. Lateroventral fossa broader than in R. aurantius and similar in width to B. nooraleebus. Postpala- tal and sphenopalatine foramina much larger than in R. aurantius or B. nooraleebus (QM F19039 but not 19040), closely paired, more distant in R. aurantius and well separated in B. nooraleebus. | Rhinonicteris tedfordi ?Brachipposideros watsoni QMF 22910 | QMF22911 | QMF22912 | QMF22915 | QMF22828 | QMF22916 (H) (P) (P) (H) (P) (P) | Greatest skull length (dorsal) 15.0 15.4 14.4 Rostral length 5.0 5.5 4.6 24.2 Braincase length 10.0 9.9 9.6 10.1 Rostral width (at lacrimal) 5.3 5.2 4.5 Min. interorbital width 2.5 2.4 19 E Zygomatic width 8.5 8.8 7.5 Mastoid width (8.9) | 75 8.3 Rostral height 43 44 4.0 4.6 Braincase height (max.) 7.9 T2 6.6 73 Palate length 1.6 1.6 1.7 L5 Palatal width (base of M?) 27 3.0 | Interperiotic distance C1-M3 5.6 P4-M3 5.0 (4.1) (4.1) 43 MI1-M3 3.5 (3.3) (3.2) 3.5 cL 1.4 c! w Ll PİL 0.9 1.1 (0.9) (1.0) 0.9 L1 PW 1.1 1.3 (1.0) (0.9) 1.3 1.2 | M!L 1.4 1.5 (1.3) (1.3) 13 1.4 M! W 1.4 1.4 anan 1.5 13 MCL (1.3) (1.3) 13 13 (1.2) (1.3) 13 13 M? Ww (1.4) (1.4) 1.5 14 | 0302) 1.5 1.4 MBL 10 | 09 (0.8) (0.9) 0.9 M? W 13 | 14 (1.1) (1.2) L5 Anterior diploic, ethmoidal and cranio-orbital fo- ramina fused, larger than in B. nooraleebus, not fused and large in R.aurantius, separated from the optic foramen by a thick bar (rather than broader plate) of bone. Like R. aurantius, palate pierced by many foramina, none especially distinctive. Sphenorbital bridge relatively broad, not greatly constricted posteriorly, with sphenorbital fissure well-exposed. Hammular process very similar to R. aurantius, with a conspicuous wing projecting backwards making up at least half its length, with a laterally directed flange of variable length (long in QMF22859) posterior to the ham- mular process, as in R. aurantius and B. nooralee- bus. Sphenorbital fissure shorter and broader than in R. aurantius; optic foramen more lateral than in R. aurantius, with the orbitosphenoid splint separating them directed medially rather than posteromedially as in R. aurantius and B. nooraleebus. NEW MIOCENE LEAF-NOSED BATS, RIY ERSLEIGH Basisphenoid shallow, Basioccipital width be- tween the periotics as in R, aurantins (perhaps slightly narrower), narrower than in 8, nooraleg- bus. Postglenoid fossa (temporal emissary fora- men) larger than in R. aurantius and B. newraleebus, postglenoid process also slightly bener developed than in R. anrantius, and much better than in B, neeraleebus. Foramen ovale very large; a bar of bone separating the foranien ovale from a ?posteriorly opening fenestra in B. nearaleebus is absent in R, tedfordi and R. au- rantins as is the fenestra. The lambdoidal cresi is better developed than in B, nooraleebus, and in this way more similar to R, aurantius (although in the laller this varies intraspecitically e.g. AR 15400 and M8416), Unlike R, aurantius, it is continuous across the occipilals in K. tedfordi cather than attenuating at the ?nuchal point. Fora- men magnum more dorsally oriented than in ZB. novraleebus and R. aurantius with indentation of its dorsal margin in A. aurantius lacking in K. tedfordi and B. nooraleebus. Periotic, its orientation and its attachment to surrounding busicranial elements similar to that in R. aurantius and B. neoraleebus Upper tecth approximately the same size inthe 3 species, those of R. aurantius more hypsidont. Upper incisors unknown. C! similar tu that in 8. noeraleebus in width, length and posterior sec- ondary cusp, bul with shallower anterolingual cingulum, removing ils squared appearance (but set QM F22845).C! wider and longer in the tooth row than in X aurantius. P extruded such that C! and P? are in close contact, almost touching (e.g QM F22845), being closer than in A, nooraleebus (although this varies) and at least as close as in R. auranmins. P? narrower than in R, auranting (es- pecially anteriorly), the lingual cingulum deeper than in R. aurantius and similar to B. novralee- bus, and the anterolingual cingular cusp better developed than in R. aurantius. M! with 4 roots, with heet similar to R. qurantins and broader than in B. neoraleebus, with a very strong dihedral crest and thickened posterolingual cingulum. Lingual notch incipient, well-developed in QMF22840. M? with 4 roots, evenly spaced, as in R. aurantius, B. nooraleebus with 3. Its heel much weaker than in R. aurantius, similar to B. nooraleebus bul with the postprotocrista reaching the base of the metacone and with a slight ridge (rather than crest) issuing from its end point (or more anteriorly in worn specimens) and extend- ing to the slightly thickened posterslingual cin- gulum. M? similar in the 3 taxa. 339 COMPARISON. This species differs from the Recent R. aurantius in its smaller size, relatively shorter braincase (especially in the postglenoid region), flattened rostral inflations, deep groove between inflations, strong supraorbital ridges, C! with less pronounced posieriort accessory cusp, relatively narrow with greater anterobuccal cx- tension and M? hee] much less expanded, From Riversleigh’s Brachipposideras nooraleebus Sigé et al., 1982 itdiffers most con- spicuously in its relatively shorter palate with posterior medial spine, its lang, slim mesopterygoid fossa, well-developed sagittal and lambdoid crests, more inflated nasals, C! without deep anterolingual cingulum, M?! with broader heel and M? with four roots. It differs from B, omeni Sigé, 1995, B. sp. cf. B. branssutensis or ‘Form X* from St Victor La Coste (Sigé et al., 1982), B. sp. cl. B. branssaten- sis from La Colombiére (Sigé et al., 1982) and B, aguilari Legendre. 1982 in M? having 4 roots. It differs from A, collongensis (Depéret, 1892) and B. dechaseauxi Sigé. 1968 in the heel of M'? not being posterobuccally extended and M? invari- ably having 4 roots, ILuiffers from B, branssaten- siy (Hugueney, 1965) in its posterolingual development of the heel of M!+ and generally less Conspicuous lingual notch separating pru- tocone from heel in M7. Brachipposideros Sigé, 1968 ?Brachipposideros watsoni sp. nov. (Figs 3-4, Table 1) ETYMOLOGY, For Neil Watson in recognition of his long association with the University of NSW. MATERIAL. Holotype QMF22915, skull with LPZ- M? and RP?-M+. Paratypes. QME22828, skull wilh LC T-M? andRP2-M3, QMF229 16, maxillary fragment with LC!-M?, Other material QMF22824,QMF22826, QMF22833, QMF22846, QMF22857, QMF22840, QMP22861. QMF22862, QMF22870, QMF22844, QMF22896. QMF22898, QMF22900, QMF22904, QMF22907. All material from the early Miocene Bilesantennary Site (discussed above). DESCRIPTION. ?Brachipposideres watsoni 1s described in Comparison with the Miocene R. tedfordi sp.nov., B. nooraleebus Sigé et al., 1982 and Recent R, aurantius (Gray, 1845). Skull approximately 10% shorter and narrower than R. tedfordi, 20-30% shorter than KR, cr- radius, With braincase length similar to B. neoraleebus, with similar overall proportions to B. neeraleebus and R tedferdi. rather than K. 340 MEMOIRS OF THE QUEENSLAND MUSEUM FIG, 3. ?Brachipposideros watsoni sp. nov., QMF22915, holotype, from Bitesantennary Site, Riversleigh, northwestern Queensland. A, dorsal view; B, lateral view; C-C’, ventral view, stereopairs. Scale indicates 5 mm. nooraleebus, with rounded posterior margin. Sagittal crest lower anteriorly than in R. tedfordi and R. aurantius but probably slightly taller than in B. nooraleebus, As in B. nooraleebus, maximal height of braincase more pos- terior than in R. tedfordi and R. au- rantius, being posterior to the glenoid; sagittal crest remaining tall anteriorly onto the postorbital region (unlike R. aurantius), joining the supraorbital ridges fairly distinctly (in QMF22828 supraorbital ridges almost develop wings or flattened plates like an incipi- ent frontal shield), of variable posterior extent (in QMF22915 attenuating in the parietal region, as in B. nooraleebus and R. aurantius, butin QMF22828 and QMF22843 extending to the lambdoidal crest as in R. tedfordi. Zy- gomatic width greater than mastoid width as in R. tedfordi and unlike R- aurantius. Rostrum lower than braimcase (not as low as in R, aurantius and R. tedfardi). Rostral inflations similar in proportion to R. tedfordi and R. aurantius, more distinct than in B. naoraleebus, less dis- tinct than in R. tedfordi and R. au- rantius. Trough between the inflations less pronounced than in R, tedfordi but more than in R. aurantius and slightly more than in B. nooraleebus. Frontal depression shallower than in R. tedfordi but deeper than in R. aurantius and B. nooraleebus, with an unpaired medial frontal foramen. Nasal opening dorso- ventrally compressed in anterior view compared to that in R. tedfordi and R. aurantius, bony nasal septum much longer than in R. tedfordi and similar to R. aurantius; opening of the vomer sinus round as in R. aurantius rather than slit-like as in R. tedfordi. Infraorbital foramen dorsal to M* as in R. aurantius, rather than M? as in R. tedfordi and B. nooraleebus, more elongate than in R. tedfordi and R. au- rantius. Anteorbital barslim and gener- ally the same width throughout, as in B. nooraleebus, sometimes with a flange or wing, often slightly curved as in R. aurantius (latter much longer in the postglenoid tedfordi and R. aurantius. Zygomatic arch with region), with lambdoidal crest generally weaker very enlarged jugal projection (QMF22857) ex- than in R. tedfordi, more like R. aurantius and B. tending upwards to at least the level of the upper NEW MIOCENE LEAF-NOSED BATS, RIVERSLEIGH 341 FIG. 4. ?Brachipposideros watsoni sp. nov., QMF22916, paratype, maxillary fragment with C!-M2, from Bitesantennary Site, Riversleigh, northwestern Queensland. A, oblique-occlusal view; B-B’, occlusal view, stereopairs. Scale indicates 1 mm. insertion of the anteorbital bar, directed slightly posterodorsally, with a rounded but narrow apex, and slightly convex posterior margin. Premaxillae unknown, with a V-shaped junc- tion to the maxillae as in R. aurantius, B. nooraleebus and R. tedfordi. Palate extendin posteriorly to the level of the anterior face of M> as in B. nooraleebus and R. aurantius, rather than the M? metacone as in R. tedfordi. Bony medial palate spine absent, unlike R. aurantius and R. tedfordi (variable in B. nooraleebus). Mesopterygoid fossa more like that in B. nooraleebus than in R. tedfordi or R. aurantius, being broad and rounded anteriorly and uniform in width throughout its length. Lateroventral fossa narrower than in R. ted- Jordi, broader than in R. aurantius and similar in width to that in B. nooraleebus. Lacrimal and postpalatal foramina similar to those in B. nooraleebus and smaller than in R. tedfordi. Lac- rimal larger than in R. aurantius; postpalatal fo- ramen and sphenopalatine similar in size to R. aurantius (but proportionately larger), approxi- mately equidistant from each other and the three interorbital foramina (cranio-orbital, ethmoidal and frontal diploic), closely paired with the inter- orbital foramina more distant in R. tedfordi, with intermediate condition in R. aurantius, with the sphenopalatine not ‘confluent’ (i.e. 2 small fo- ramina (QMF19038, QMF19039), and the 3 ap- proximately equidistant) in B. nooraleebus. Orbitosphenoid splint separating the optic fora- men from the sphenorbital fissure, directed posteromedially like in B. nooraleebus and R. aurantius, rather than medially as in R. tedfordi. Sphenorbital bridge slightly more constricted posteriorly (posterior to pterygoid processes) than in R. tedfordi and R. aurantius. Pterygoid wings directed dorsally rather than posteriorly, resulting in shorter wings than in R. tedfordi and slightly shorter than in R. aurantius (proportion- ately). Postglenoid fossa slightly smaller than in R. tedfordi, but slightly bigger than in R. au- rantius. Postglenoid process similar to R. au- rantius and R. tedfordi and better developed than in B. nooraleebus. Foramen ovale similar to the other taxa; a posteriorly directed fossa relatively smaller than in R. tedfordi without bar. Inter- 3p periðtie distance similar to that in R. tedfordi, Periotic morphology and orientation and attach- ment to the basicranium similar to other 3 taxa. Foramen magnum similar to that in K- sedfordi, directed more ventrally, us in R. aveantios and B. nooraleehus, Teeth smaller than in R. alrantius and R. tedfordi, approximately same size as in B. nooraleebus. Upper incisors unknown. C! pro- portionately shorter (ih the tooth row) than in R. tedfordi and probably B. nooraleebus, more sim- ilar to R, aurantius, C! cingulum not developed as in R. tedfordi and B. nooraleebus, more like in R aurantius: anterolingual cingulum follow- ing the tooth outline rather than thickening in the anterolingual corner, Posterior secondary cusp similar to that in R. redfordi but perhaps taller (in buccal view, 1/3 to 1/2 C! length rather than 1/3 in B. noaraleebus and R, tedfordi), P* small and buceally extruded; C! and P? generally not in contact (but see QM F22907), generally closer, hut not in contact in R. tedfordi and R. aurantius, P’ narrower than in R, aurantius and B, nooralee- bus, most similar to R, tedfordi. M! has 4 roots, with heel longer than in B. nooraleebus, more sitnilar to R. tedfordi and R. aurantius. M? with 3 roots, like B. nooraleebus and unlike R. tedfordi and R, aurantius, with heel more expanded than in R. tedfordi, similar to B. nooraleebus, much less expanded than in R. aurantius. (Buccal and lingual lengths similar in 7B. watsoni and R. tedfordi, buccal length greater than lingual length in B. novraleebus,) M'* crest and cingular devel- opment and M? similar in the 4 taxa, COMPARISON, It differs from R, tedfordi in its slighty smaller size, shorter mesopterygoid fossa, less anteriorly inflated brajncase, more elongate infraorbital foramen, lack of postpalatal spine and M2 with three roots. It differs from the Recent R. aurantius in its. smaller size, much less anteriorly inflated brain- case and pronounced sagittal crest, relatively shorter braincase (especially in postglenoid re- gion), flattened rostral inflations, deeper groove between inflauions, strong supraorbital ridges. less pronounced accessory cusp on C!, P? larger and less extruded from the toothrow, P* relatively narrow with greater anterobuccal extension and M? heel much less expanded and having three TOOLS. lt differs from B. nooraleebus in its C! lingual cingulum being uniformly shallow, its narrower and shorter P*, more expanded M? heel, sharp rise in braincase height above glenoid, position of 2 MEMOIRS OF THE QUEENSLAND MUSEUM infraorhital foramen, deep frontal depression and more pronounced supraorbilal crests, Itdiffers from Brachipposideros branssatensts, B. collongensis and B. dechaseauxt in M? invari- ably having three roots. It differs from B. dmtane in its larger size and better developed heel in M*. Itdiffers trom "Form X’ in its more expanded heel in M°. it differs from B. sp. cl. B. branssatensis in its posterolingual development of the heels of M! and pronounced crests on the posterior flank of the protocone. It differs from 4 agrilariin M! having four roors. COMPARISONS OF THE NEW HIPPOSIDERIDS WITH RELATED TAXA These new species are similar in skull and dental morphology to northem Australia’s living Rhinonicreris aurantius and Méicrosite’s Brachipposideros nooraleebus in proportions of the skull, broad rostrum, subparallel tooth rows, palate and zygomatic arch, crested premaxillae, basicranial, periotic and otic morphology, pro- nounced accessory cusp on C! and little reduced upper and lower M3s, Sigé et al. (1982) recognised R. aurantius as i probable descendant of the Australian Brachipposideros \incage. Brachippusideros is known from the Tertiary of Europe, Arabia and Australia (Sigé, 1968; Sigéetal., 1982; Legendre. 1982; Ziegler, 1993; Sigé et al.. 1995). The new Riversleigh species can be compared with Euro- pean and North African taxa only on their upper dentition because: 1) skull material has not been described for non-Australian taxa and 2) dentar- ies cannat be positively referred tọ the Riversleigh taxa. A combination of dental characters is shared by Brachipposideros and the new Australian taxa: small size, P? between C! and P* near or on buccal margin of tooth row, C! with secondary cusp, P? slender with respect to other teeth, M? with four roots (loss in same), M? with three roots (ad- vanced forms have four), heel of M! separated from protocone by a notch (loss secondary) and forming a posteriorly directed lobe. M? heel rel- atively weakly developed, primitively, postprotocrista has prominent anterior portion and only incipiently developed posterior part Brachipposideros noaraleebus shares with Euro- pean Brachippesideros a small lower canine, low coronoid process and similar shape of ascending ramus (Sigé ct al. 1982). The 3 Australian Miocene species differ from the early Oligocene B. omani (Sigé ct al.. 1995) NEW MIOCENE LEAD-NOSED BATS, RIVERSLEIGH 143 in their larger size, more recurved C! with better developed secondary posterior cusp, and M? pro- tucone with weaker dihedral crest. Additionally, V tedfordi dilfers from B. omani in its 4-rooted MP. Compared with &. sp, cf. B. branssatensis or “Form X’ (Sigé et al, 1982) of the French laté Oligocene (Chattian), the Australian species have P? smaller and further extruded from the lonthvow, P* with better developed anterolingual cingular Cusp, P? wider with respect to M! (clos- esto PB. warsoni), M? heels more posterolingu- ally developed and posteriorly directed, M* heel more expanded with dihedral crest and posterolingual cingulum stronger in Australian laxa. Mè size is similar. The early Miocene (Lower Aquitanian) French species R. branssatensis (Hugueney. 1965) has quite different M!? heel development from Aus- tralian species, with heel expansion occurring al ihe posterolingual corner but directed buceally, und having a pronounced lingual notch, a variable characteristic in Australian taxa. The M? heel is hetier developed than in Australian forms but the dihedral crest is more pronounced in Australian lak as.is (probably) the posterolingual cingulum. C! is similar to that in ?B. watsoni and R. tedfordi, in which the lingual cingulum is aniform and follows the curvature of the tooth, and hence unlike A. nooraleebus. P> position and size are similar but in Australian forms P? is generally staller and more extruded. The infraorbital fora- men occurs dorsal to M? asin R, redfordi and B. nooraleebus, M'* heel expansion in the Australian taxa is more similar to that found in the French early Miocene (Lower Aquitanian) #. sp. ef, B, branssatensis from La Colombière, in direction of expansion and strong crest on the posterior Nank of the pratocone, C! is smaller in size and the lingual cingulum uniform and even in depth, bul with similar thickening in its anterolingual corner as in R, redfordi. The posterior margin of P? is very curved, the anterior margin narrower and the anterobuccal extension greater than in #B. warvont and similar to B, neoraleebus, M! and M? ?variably, has 4 roots, The French early to early middle Miocene (Upper Aquitanian) B. dechaseuuxi Sige, 1968 is larger than the Australian species.. The posterior flank on the M'? protocone is simply rounded with the dihedral crest poorly developed, the heel is directed posteriorly to posterobuccally like B, hranssatensis, Mi? width is very similas to that of M4, variably developed lingual noleh senaral- ing prolocone and heel, P* is narrow with respect w M!-3, possibly smaller than in 7B. watsoni, its anterobuccal extension much greater than in Aus- lrahan taxa F? is outside the toothrow, but is probably similar to Australian taxa in size and position. C! has a uniform lingual cingulum as in ?B. watsoni and R. tedfordi. M!* heel development in the French early to early middle Miocene (Burdigalian) B, agnileari Legendre, 1982 is sharper than the Australian species hut the direction of expansion and crest on the protocone are similar. The pusterolingual heel cingulum is nat well-developed, The erest is continuous with the posterior lingual cingulum in B aguilari. in M'* the ectoluph is different: the buccal edge 15 angular rather than rounded as in the Australian taxa. P? appears to be relatively large and C! gracile with a uniformly deep lingual cingulum like 7B. warsont, The type species, B. collongensis (Depéret. 1892), from the French early middle Miocene (Upper Burdigalian) is similar in size to Æ. warsoni and B. nooraleehus bul P? is less ex- truded from the toothruw, M!? heels posterobuccally developed like B, branssateriaty and B. dechaseauxi and M? heel belter developed but with weaker dihedral crest especially in M? whose protocone flank is rounded, P? is relatively wide with respect to M5? as in the Australian species, PHYLOGENETIC RELATIONSHIPS On denial characters, the new Australian spe- vies are more similar to cach other and to B. roordeebus than to non-Australian taxa, Sigé et al. (L982, figs &-9) found that that compared ta European species, the dental structure of B. novraleebus was more advanced than Aquitanian forms and as advanced as Burdigalian species. The Chattian “Form X' was considered close to the hase af the European radiation, with B. branssetensis close to the group that gave rise bo the &. callongensis and B. dechaseaunt lineages and £. sp. cf. B. branssatensis Closer to B, aguilar and B. nooraleebus, Apomorpbies shared hy B. sp. ef. B. branssatensis, Baguilart and B. nooraleebus included heel of M+ separated from the protocone by a slight lingual notch and heels developed postervlingually and directed posteri arly. Brachipposideros aguilari and B. nooralee- bus share further reduction of P? so that Cand P are close and sometimes in contact, PA relatively larger and M!* protecone with pronounced dih dral crest MEMOIRS OF THE QUEENSLAND MUSEUM 344 Jo\seouy Joenus snydojoulyy Siue snydojouryy snyAydebew snydojouiyy snjepidsnou) snosijesy yyy sdojaop Heansad snoop Ipsojpa} suajoAuouiyy snqueine suejoAuouiyy snqeejeloou sosapisoddiyoeig snaissed ‘sdoueeuy ISWIEIIM BYbia/SIBAnY dey sourysouex smeuso sdoyjuy eesojAejai6bew ‘H SNJBIEDIED ‘H suoads `H seq ‘H SNAINY ‘H JOYE snua H snojebew snowsapuís 418 `H ended `H ewaƏpEIF H sidi ‘H med `H 184. "H SMJENEJ `H BAIPEŅUE] `H sisuenbiznog snydojourysopnesg JUOSIƏWWOI “YH Jae 'H IÁoJƏND esoydoyAydoeejeg sdopÃə `H suapu Byes IUOWƏS ‘H snyAydousioa ‘H JƏfıspJeuseq 'H IUOJSE}IOM ‘H snunsnw sosapisoddiH <= Joseouy Joenus snydojouryy sjuyye snydojouryy sniAydebaw snydojouyy snepidsnou) snosijesy Iyi sdojaog yeajued snoaojg Hey sourysouex snaisied sdoueeuy ISWEIIM BYyBblajssaaiy Ips0jpa} suajoAuouiyy snijueine suejsAuouryy sngesjesoou sosapisoddiyoeig snojebaw siowsapuss suoeds ‘H SNUIAIB9 ‘H Jaqi H SNAN “H 184E9 `H 1818 `H snjeuso sdoyjuy erojAejaibbew ‘H SMEJEJJEI `H ended ‘4 ewapeIp 'H Sidin H med 'H 1811 "H SMEAEJ `H BAIPEŅUE] `H sisuenbiznog snydojoursopnesg MUOSJAUIWIOD “YH Jawe 'H tAauenb esoydoyAydoeejeg SCOjOA9 ‘H suepu) eyjasy IUOW8S 'H snAydouoa `H Jabispseuseq `H IUOISEJJOM ‘H snujosnu sosapisoddiH PA Joyseouy Jojyejnuis snydojouryy snAydeGew snydojouryy siuyje snydojouryy juowes "YH snjepidsnou) snosijesy 14u sdojaog yearued Soaog sna!ssəd sdouaeuy snqəəjesoou sosepisodaiyoe1g snnuesne suejoAuouiyy Ip104p9} Sua PÁuouNY ISWEIIM BYBIa/SIBAY wey soursouex sable `H JOYED“H SMAIN} H 191A ended ‘H med ‘H suoeds ‘H sdin ‘H sısuanıznoq snydojouiyIopnaSd smeuwo sdoyjuy BWAPEID | BAIpeyUe] | SMENE] | eesojAejeib6ew | SMEJE | SNUB * 4aqnu `) 4818 `H .Syoeĝəw snowsapuís IUOSIƏWWOI `H sdopA2 H snyAydouAso2 ‘H shosenb esoydojAydoaejed Susp} easy Jabispseuseq `H snupsnw ‘H 1uO}se|IOM sosapisoddiH rirrrrrī =] 0.25; 87.5% resolution); B, Strict consensus of 8 PAUP trees, some 0.23; 82.5% resolution). C, Hennig86 Nelson consensus, unordered characters. See from analyses conducted on 40 taxa and 59 dental, cranial and skeletal characters: A, Strict consensus of 4 PAUP trees, all unordered characters (CI ordered characters (CI FIG. 5. Phylogenetic hypotheses of hipposiderid relationships presented by Hand & Kirsch (in press) resulting Hand & Kirsch (in press) for characters and character states. NEW MIOCENE LEAF-NOSED BATS, RIVERSLEIGH TABLE 2. Distribution of character-states used in a phylogenetic analysis of relationships among Brachipposideros and Rhinonicteris species, and re- lated taxa, based on dental characters only. O=inter- preted plesiomorphic condition, 1-3=apomorphic states, ?=missing data or character does not apply. Taxon Character states 2127 ???? 01 ?11 100? 0011 01 011 1000 0011 01 011 1011 0101 11 011 1012 1122 01 011 10100011 11 011 1002 0022 01 011 1001 0022 01 011 1012 0022 11 011 1012 0122 12 011 Hipposideros ater 0201 ?022 01 111 Anthops ornatus 0100 ?021 ?0 100 Ancestor 0000 0000 00 000 Characters: 1: Height of ascending ramus of dentary: 0=tall, 1=low 2: C! accessory cusp: O=present, 1=poorly developed, l=absent 3: P? extrusion: 0= extruded but still separating C! and P4, 1=C! and P4 in contact or nearly so 4: P* width wrt other cheekteeth: O=narrow, 1=me- dium, 2=wide 5: M! no. of roots: 0=4, 1=3 6: M! heel development/length: 0=moderate, 1=strong 7: M! heel direction: 0=none, 1=posterobuccal, 2=posterolingual 8: M! lingual notch: O=absent, 1=inconspicuous, 2=conspicuous 9: M! dihedral crest: O=absent, 1=weak/medium, 2=strong 10:M? number of roots: 0=3, 1=4 11:M* heel length/development: Q=none, 1=slight, 2=great 12:M* heel direction: O=none, 1=posterobuccal, 2=posterolingual 13:M? dihedral crest: O=absent, 1=weak/medium, 2=strong B. omani B. brassatensis B. sp. cf. B. branssatensis B. collongensis B. aguilari B. dechaseauxi B. nooraleebus ?B. watsoni R.. tedfordi R. aurantius The new Bitesantennary species also share these apparent apomorphies and are assigned to that clade. Although tedfordi shares with the B. branssatensis, B. collongensis and B. dechaseauxi lineages a fourth root on M? it does not share the distinctive posterobuccally ex- panded heels on M!~. In this case a four-rooted M? is interpreted to be homoplastic; it occurs also in R. aurantius. A phylogenetic analysis of the interrelation- ships of 12 hipposiderid species including 10 345 species of Brachipposideros and an hypothetical ancestor, based only only dental features (13 characters) (Table 2) and using the clustering program PAUP 3.1.1 (Swofford 1993), was un- able to resolve relationships within the group (percent resolution of trees 18.2%). Tree resolu- tion did not improve when the most poorly known species, B. omani , was removed, nor if character states were ordered. However, majority rule trees (50%) did show the European B. branssatensis, B. dechaseauxi and B. sp. cf. B. branssatensis (Form X) clustering in 67% of trees, as did B. collongenis, B. aguilari, R. tedfordi and R. au- rantius. Hand & Kisch (in press) found in their phylogenetic analyses of 37 hipposiderids that dental features (20 characters) were not sufficient to interpret relationships among genera and spe- cies groups of the Hipposideridae. They found that resolution of trees was less than 33% when dental features only were used, compared with 87.5% resolution with a combined data set of cranial, dental and skeletal characters. Brachipposideros Sigé, 1968 was erected as a subgenus of Hipposideros Gray, 1831. However, probable patristic relationships between Brachipposideros and Rhinonicteris, indicate that the evolutionary relationships of these taxa are not adequately reflected by current taxonomy. Hand & Kirsch (in press) found Hipposideros to be almost certainly paraphyletic, as did Bogdanowicz & Owen (in press). Hugueney (1965), Sigé (1968) and Legendre (1982) all sug- gested that Hipposideros was paraphyletic. Hand & Kirsch (in press) also found that B. nooraleebus was more closely related to Rhinonicteris (aurantius and tedfordi), and pos- sibly other Australian Miocene hipposiderids than to Hipposideros (Fig. SA-C). Because that analysis was based on cranial as well as dental characters, European Brachipposideros taxa could not be included, and precise relationships between non-Australian and Australian Brachipposoideros species remain unclear. Relationships between Brachipposideros, Rhinonicteris and other Australian Miocene hipposiderids were not completely resolved in the analyses by Hand & Kirsch (in press), However, in all trees nooraleebus occurred as the plesiomorphic sister-species to a clade consisting of, or containing, aurantius and tedfordi. In some trees, aurantius and tedfordi formed part of a broader group of Australian Miocene hipposiderids including Xenorhinos and Riversleigh (Fig. 5B-C). When watsoni was included in PAUP analyses 346 E C È Anthops ornatus Xenorhinos halli Riversleigha williamsi Triaenops persicus Rhinonicteris aurantius Rhinonicteris tedfordi Brachipposideros nooraleebus ?Brachipposideros watsoni Cloeotis percivali Coelops frithi Aselliscus tricuspidatus Rhinolophus megaphyllus Rhinolophus affinis Rhinolophus simulator Ancestor Rhinonicteris aurantius Rhinonicteris tedfordi Brachipposideros nooraleebus ?Brachipposideros watsoni MEMOIRS OF THE QUEENSLAND MUSEUM A Anthops ornatus Xenorhinos halli Riversleigha williamsi Triaenops persicus Rhinonicteris aurantius Rhinonicteris tedfordi Brachipposideros nooraleebus ?Brachipposideros watsoni Cloeotis percivali Coelops frithi Aselliscus tricuspidatus Rhinolophus megaphyllus Rhinolophus affinis Rhinolophus simulator Ancestor rh 100 Rhinonicteris aurantius D 67 Rhinonicteris tedfordi Brachipposideros nooraleebus ?Brachipposideros watsoni FIG. 6. Phylogenetic hypotheses of relationships of 40 hipposiderids plus ?Brachipposideros watsoni resulting from PAUP analyses conducted on 59 characters (Hand & Kirsch, in press). A, Strict consensus of 44 trees (C1=0.24; 55% percent resolution), all unordered; B, 50% majority rule tree of 6A; C, D, % support (majority rule) for clustering of Rhinonicteris and Brachipposideros taxa, trees based on unordered and ordered characters respectively. of the same taxa and characters used by Hand & Kirsch (in press), resolution of relationships be- tween hipposiderid taxa fell (from over 82% to less than 60% in all analyses). Relationships among crown groups (i.e. Hipposideros, Asellia, Palaeophyllophora and Pseudorhinolophus) re- mained unchanged from those shown in Fig. 5 (indicated by broken line in Fig. 6A-B), but res- olution at the base of the trees (e.g., among Brachipposideros, Rhinonicteris, Coelops and Cloeotis) decreased markedly (cf. Figs 5A and 6A). Majority rule trees (50%) clustered species of Rhinonicteris and Brachipposideros (e.g., Fig. 6B), but with little consensus on relationships between watsoni , nooraleebus and an aurantius- tedfordi clade (Fig. 6C-D). On the basis of all analyses (Hand & Kirsch in press and herein), watsoni and tedfordi are as- signed to a clade also containing B. nooraleebus and R. aurantius. However, the interrelationships between these taxa is not as clear. Skull morphol- ogy of nooraleebus (e.g., its poorly developed sagittal crest, shallow frontal depression and poorly inflated nasals) is less derived than that of watsoni, but its dentition (e.g., large P*) would exclude it from being a structural ancestor to watsoni. Here, watsoni has been tentatively as- signed to Brachipposideros, and tedfordi to Rhinonicteris. Brachipposideros nooraleebus is known from fragmentary material with key fea- tures of the sphenorbital bridge area lacking (Hand, 1993, fig.1). However, both nooraleebus and watson lack a number of apparent apomorph- ies shared by tedfordi and aurantius, including an anteriorly vaulted braincase and low but conspic- uously inflated rostrum, a round infraorbital fora- men bordered by a curved anteorbital bar, a postpalatal spine, a narrow, scalloped mesopterygoid fossa, a poorly (posteriorly) con- stricted sphenorbital bridge with long (an- NEW MIOCENE LEAF-NOSED BATS, RIVERSLEIGH 347 difficult to determine on dental morphol- ogy alone. Until further information be- comes available, all non-Australian and the least derived Australian taxa (i.e., those lacking obvious synapomorphies for Rhinonicteris) are referred to Brachipposideros as perhaps the simplest, if not entirely accurate, reflection of the group’s evolutionary relationships. ?Brachipposideros sp. (Fig. 7), a maxil- lary fragment of a Miocene hipposiderid from Riversleigh’s Upper Site, preserves M! and M? which are strikingly similar to those of the B. collongensis and B. branssatensis lineages, particularly in their posterobuccally-directed heel develop- ment which is quite unlike any other known Riversleigh hipposiderid. The Bitesantennary Site is a Miocene FIG. 7. ?Brachipposideros sp., QMF22917, maxillary fragment cave-fill in which ?B. watsoni and R. with M!-2, from Upper Site, Riversleigh, northwestern Queens- tedfordi occur with at least 8 other land. A-A’, stereopairs, oblique-occlusal view. Scale indicates hipposiderids, 5 of which are yet to be | mm. teroposteriorly) pterygoid wings, and M? with four (rather than three) roots. Fewer apomorphies appear to be shared between watsoni and nooraleebus, but potentially include the posterior extension of the supraobital crest and elongated infraorbital foramen. Recent R. aurantius can be distinguished from the Miocene R. tedfordi by its larger size, rela- tively longer braincase (especially in postglenoid region), little or no groove between inflations, weaker supraorbital ridges, more expanded heel on M?, more pronounced accessory cusp on C!, P* relatively wide with little anterobuccal exten- sion, and P? small and further extruded. In Riversleigh’s Pliocene Rackham’ s Roost deposit, an early population of R. aurantius occurs syn- topically with other as yet undescribed Rhinonicteris and/or Brachipposideros species, and today R. aurantius is still found in the general area. DISCUSSION I raise Brachipposideros Sigé, 1968 from sub- generic to generic level. Tentatively, it would include non-Australian species B. branssatensis, B. collongensis, B. dechaseauxi , B. omani. B. aguilari and B. sp. cf. B. branssatensis) as well the Australian Miocene species, nooraleebus and watsoni. Although this may be a paraphyletic group, evidence is conflicting and relationships between Australian and non-Australian taxa are described. Five of the 10 Bitesantennary hipposiderids, including ?B. watsoni and R. tedfordi, are well represented, each by tens or hundreds of complete skulls; the other 5 hipposiderids, and a megadermatid (cf. Macroderma godthelpi), are represented by fewer, more fragmentary specimens. The gener- ally very fine preservation of the remains (often with periotics in situ) suggests that fossilisation occurred quickly with little transport, probably in still water rather than guano (in which biodegra- dation would be expected). Few juvenile bats are among among the thousands represented, sug- gesting that this cave (or part thereof) was not used as a maternity roost. By analogy with modern bat communities, the high diversity of hipposiderids in the Bitesantennary deposit suggests warm, humid conditions in the cave, and probably outside it. In Europe, appearance of Brachipposideros in the fossil record coincides with a period of steadily increasing temperature and their disappearance probably correlates with the climatic deteriora- tion across Europe in the later Pliocene (Aguilar et al., in press). In Australia, 6 hipposiderids are restricted to northern tropical areas. Rhinonicteris aurantius roosts in very warm, humid caves in colonies of 20 to several thousand individuals from NW Queensland to NW WA. It emerges at dusk to feed, mostly on moths but also on beetles, shield-bugs, parasitic wasps, ants, chafers and weevils (Jolly & Hand, 1995). Although R. au- rantius and the Miocene Rhinonicteris tedfordi 348 are closely related and similar in many skull features, the extinct species lacks the forward- projecting development of the sagittal crest that characterises R. aurantius, and it is unclear whether or not they could be described as ecolog- ical vicars. Hipposiderid bats promise to be useful biostratigraphic indicators in the limestones at Riversleigh. They are the most common bats in Riversleigh’s Miocene deposits, the best pre- served, and, with megadermatids, currently the best understood in terms of their phylogenetic relationships as well as their morphological vari- ability (Sigé et al., 1982; Hand, 1993, 1995, 1997). Brachipposideros nooraleebus is known only from Microsite and ?B. watsoni only from Bitesantennary Site. Rhinonicteris tedfordi, how- ever, is known from Bitesantennary Site in the Verdon Creek Sequence, RV and Upper Sites on Godthelp’s Hill, and White Hunter Site on Hal’s Hill. None of the species described herein has been recorded from System C sites (Archer et al., 1989, 1994: Creaser, this volume), but close rel- atives (?descendants) occur at sites such as Gotham City and Dome North Sites suggesting that lineages may be identified within the Riversleigh limestone sequence. ACKNOWLEDGEMENTS Work at Riversleigh has been supported by the Australian Research Council, the Department of the Environment, Sport and Territories, National Estate Programme Grants (Queensland), Queens- land National Parks and Wildlife Service, the Australian Geographic Society, the Linnean So- ciety of New South Wales, ICI, the Queensland Museum and the University of NSW. I thank Michael Archer, Henk Godthelp and Bernard Sigé for their help and encouragement and two referees for constructive criticism of this paper. Skull photographs are by Ross Arnett and Robyn Murphy, University of NSW. LITERATURE CITED AGUILAR, J.-J., LEGENDRE, S., MICHAUX, J., & MONTUIRE, S. (in press). Pliocene mammals and climatic reconstruction in the western Medi- terranean area. American Association of Strati- graphic Palynologists, Contribution Series. ARCHER, M., GODTHELP, H., HAND, SJ. & MEGIRIAN, D. 1989. Fossil mammals of Riversleigh, northwestern Queensland: prelimi- nary overview of biostratigraphy, correlation and MEMOIRS OF THE QUEENSLAND MUSEUM environmental change. The Australian Zoologist 25: 35-69. ARCHER, M., HAND, S.J. & GODTHELP, H. 1994. Riversleigh. The story of the animals of Australia’s ancient inland rainforests. 2nd Ed. (Reed: Sydney). BOGDANOWICZ, W., & OWEN, R.D. (in press). In the Minotaur’s labyrinth: a phylogeny for the Hipposideridae. In Kunz, T. &Racey, P. (eds), Proceedings of the 10th International Bat Re- search Conference, Boston. (Smithsonian Institu- tion: Washington). CREASER, P. 1997. Oligocene-Miocene sediments of Riversleigh: the potential significance of topogra- phy. Memoirs of the Queensland Museum 41: 303-314. HAND, S.J. 1993. First skull of a species of Hipposideros (Brachipposideros) (Micro- chiroptera: Hipposideridae), from Australian Miocene sediments. Memoirs of the Queensland Museum 31: 179-192. 1995. Hipposideros bernardsigei, a new hipposiderid (Microchiroptera) from the Miocene of Australia and a reconsideration of the monophyly of related species groups. Münchner Geowissenschaftliche Abhandlungen. 1997. Miophyllorhina riversleighensis gen. et sp. nov., a new Miocene leaf-nosed bat (Microchiroptera: Hipposideridae) from Queens- land. Memoirs of the Queensland Museum 41:351-354. HAND, S.J., ARCHER, M. & GODTHELP, H. 1989. A fossil bat-rich, Oligo-Miocene cave-fill from Riversleigh Station, northwestern Queensland. Conference on Australasian Vertebrate Evolution, Palaeontology and Systematics, Sydney, March 1989, Abstracts: 7. HAND, S.J., & KIRSCH, J.A.W. (in press). A southern orgin for the Hipposideridae (Microchiroptera)? Evidence from the Australian fossil record. In Kunz, T. & Racey, P. (eds) Proceedings of the 10th International Bat Research Conference, Boston’. (Smithsonian Institution: Washington). HUGUENEY, M. 1965. Les chiroptères du Stampien supérieur de Coderet-Branssat. Documents du Laboratoire géologique de la Faculté des Sciences de Lyon 9: 97-127. JOLLY, S. & HAND, S.J. 1995. Orange Leafnosed-bat Rhinonicteris aurantius. Pp. 464-465. In Strahan, R. (ed.), The complete book of Australian mam- mals. 2nd Ed. (Reed: Sydney) LEGENDRE, S. 1982. Hipposideridae (Mammalia: Chiroptera) from the Mediterranean Middle and Late Neogene and evolution of the genera Hipposideros and Asellia. Journal of Vertebrate „Paleontology 2: 386-399. SIGE, B. 1968. Les chiroptères du Miocene inférieur de Bouzigues, l. Etude systématique. Pal- _aeovertebrata 1: 65-133. SIGE, B., HAND, S.J. & ARCHER, M. 1982 An Aus- tralian Miocene Brachipposideros (Mammalia, NEW MIOCENE LEAF-NOSED BATS, RIVERSLEIGH 349 Chiroptera) related to Miocene representatives sis using parsimony, Version 3.1. (Computer pro- „from France. Palaeovertebrata 12: 149-71. gram distributed by the Illinois Natural History SIGE, B., THOMAS, H., SEN, S., GHEERBRANDT, Survey, Champaign, Illinois). E., ROGER, J. & AL- SULAIMANI, Z. 1994.Les ZIEGLER, R. 1993. Die Chiroptera (Mammalia) aus chiroptéres de Taqah (Oligocéne inférieur, Sul- dem Untermiozin von Wintershof-West bei tanat d’Oman). Premier inventaire systématique. Eichstätt (Bayern). Mitteilungen der Bayerischen Miinchner Geowissenschaftliche Abhandlungen Staatssammlung fiir Paläontologie und A, 26: 35-48. Historische Geologie 33: 119-154. SWOFFORD, D.L. 1993. PAUP: Phylogenetic analy- MIOPHYLLORHINA RIVERSLEIGHENSIS GEN. ET SP. NOV., A MIOCENE LEAF- NOSED BAT (MICROCHIROPTERA: HIPPOSIDERIDAE) FROM RIVERSLEIGH, QUEENSLAND SUZANNE HAND Hand, S.J. 1997 06 30: Miophyllorhina riversleighensis gen. et sp. nov.. a Miocene leaf-nosed bat (Microchiroptera: Hipposideridae) from Riversleigh, Queensland. Memoirs of the Queensland Museum 41(2): 351-354. Brisbane. ISSN 0079-8835. A new Australian Tertiary hipposiderid is described on the basis of a maxillary fragment from RV Site, on Godthelp Hill, Riversleigh, northwestern Queensland. Miophyllorhina riversleighensis gen. et sp. nov. is distinguished from all other hipposiderids in its loss of P*, retention of a large M°, and P* longer than wide with well-developed anterolingual cingular cusp. lts phylogenetic relationships to other hipposiderids are obscure, a Miocene, Riversleigh, Australia, hipposiderid, leaf-nosed bat Suzanne J, Hand, School of Biological Science, University of New South Wales, New South Wales 2052, Australia; received 4 December 1996, Tertiary deposits in the Riversleigh World Heri- tage Fossil Site, Lawn Hill National Park, NW Queensland are rich in Old World bats of the Hipposideridae (Sigé et al., 1982; Hand, 1993, 1995, 1997). They comprise the majority of bats in all Riversleigh Oligocene-Miocene sites, with as many as 8 Species occurring synlopically in deposits such as Upper Site (Archer et al., 1994). Many species are known from complete or nearly complete skulls as well as disarticulated but com- plete postcranial material. Several hipposiderid genera or subgenera are represenied and others await description. A fragment of a maxilla from RV Site (Archer el al, 1989, 1994; Creaser, 1997) represents a new hipposiderid genus distinguished by a unique combination of features. It has not been identified from adjacent, possibly contemporane- ous deposits, such as the better-sampled RSO Site (Creaser, 1997), Other vertebrates from RY Site are generally tragmentary and of small to me- dium-sized animals, They include a skink, madtsoiid, small crocodile, chelid, peramelid, pseudocheirid and the phascolarctid Nimiokoala greystanesi (Black & Archer, 1997). Other bats from the site include Rhinonicteris tedfordi (Hand, 1997) which also occurs at adjacent siles, a vespertilionid possibly Leveonee and a molossid. Dental terminology follows Sigé et al, (1982). Specimens held in the fossil collections of the Queensland Museum (QMB), Brisbane. SYSTEMATICS Suborder MICROCHIROPTERA Dobson, 1875 Superfamily RHINOLOPHOIDEA Weber, 1928 Family HIPPOSIDERIDAE Miller, 1907 Miophyllorhina gen. nov. TYPE SPECIES. Miophyllorhina riversleighensis sp. nov. ETYMOLOGY. Greek phvila, leaf and rhina, nose; Mia- refers to the interpreted Miocene age. DIAGNOSIS. P? lost; P* longer than wide, with deep lingual cingulum and well-developed an- terolingual cingular cusp; large MÌ, as wide as M? and with premetacrista 3/4 paracrista length. Miophyllorhina riversleighensis sp. nov. (Fig. 1) MATERIAL. Holot pe QMF30566, a left maxilla fragment with P4, MZ-ẹ from Early Miocene tulaat RV Sile in System B on Godihelp’s Hill, Riversleigh (Archer et al., 1989, 1994; Meginan, 1992: Creaser, 1997). RV Site is perhaps slightly younger than the South Australian Kutjamarpu Local Fauna (Woodburne et al, 1985), ETYMOLOGY. For the Riversleigh World Heritage Fossil Site. DIAGNOSIS. As for genus. DESCRIPTION, Teeth worn but not broken. Al- veolus for C! indicating this tooth wider and longer than P+. P? lost, with no sign of an alveolus for this tooth either within or extruded (lingually or buccally) from the toothrow. P* longer than MEMOIRS OF THE QUEENSLAND MUSEUM FIG. 1. Miophyllorhina riversleighensis gen. et sp. nov, from RV Site, Godthelp’s Hill, QMF30566, holotype. A, antero-occlusal view showing lack of alveolus for P?, B-B’, stereopair, oblique occlusal view. Scale = 1mm. wide, narrower particularly anteriorly than M3 and as long. Lingual cingulum deep, well-devel- oped; anterolingual cingular cusp well-devel- oped. M! and M2 with 4 evenly-spaced roots. M? with a posteriorly-directed, small but conspicu- ous heel. Protofossa probably open posteriorly but with wear the postprotocrista reaching the base of the metacone. Postprotocrista with a slight ridge or crest issuing from what was prob- ably its end point (more anteriorly in this worn specimen) and emende to the thickened posterolingual cingulum. M? not greatly reduced, as wide as M2, with premetacrista 3/4 paracrista length. MEASUREMENTS(mm). Holotype QMF 30566:- P4_M3 L=3.77; C! (alveolus)- M3=4,71, M2-M?=1.77; P4 L=0.88; P#l=0.81; M? L=1.18; M2 1=1.42; M3 L=0.82: M3 1=1.40, COMPARISONS. In lacking a P?, this species differs from all other Riversleigh hipposiderids, namely, Brachipposideros, — Rhinonicteris, Xenorhinos, Riversleigha and Hipposideros bernardsigei from the Oligocene-Miocene Neville’s Garden Site (Hand, 1995) and H. sp. from the Pliocene Rackham’s Roost Site (H. bi- color group). The lack of P? characterises living and extinct Asellia, Cloeotis and H. (Syn- desmotis), and P? is very reduced or lacking in some members of the H. cyclops group (including the only fossil taxon H. bernardsigei). It is rarely absent in other extant species of Hipposideros en possibly H. sabanus). However, in these cases: 1) M? is also very] reduced (i.e., Asellia and Hipposideros), or 2) P+ is large and anteriorly very wide, with the anterolingual cingular cusp reduced, absent or located near the buccal margin of the tooth (i.e., Cloeotis percivali and some of the H. cyclops group, including H. semoni and H. stenotis); or 3) M? is reduced and P4 is wider than long (i.e., Syndesmotis megalotis). P? is retained in all other hipposiderids, i.e., Tertiary Pal- aeophyllophora, Pseudorhinolophus and Vaylat- sia, and extant Coelops, Paracoelops, Triaenops, Anthops and Aselliscus. It is also retained in the Rhinolophidae, the immediate sister-group of the Hipposideridae. DISCUSSION. On available material it is not possible to determine the relationships of this species to other members of this family. Hand & Kirsch (in press) found that dental features alone are not sufficient to interpret relationships within the Hipposideridae. In its dentition M. riversleighensis exhibits a mixture of what appear to be plesiomorphic and apomorphic features. For example, the loss of P? NEW LEAF-NOSED BAT FROM RIVERSLEIGH 333 is probably a derived feature for hipposiderids, independently acquired in a number of separate lineages, The anteriorly-narrow P* with promi- nent anterolingual cingular cusp is more difficult to interpret, but is possibly plesiomorphic among hipposiderids (Hand, 1995; Hand & Kirsch, in press). The large M? on the other hand is probably plesiomorphic among hippusiderids (6 g., Hipposideros, Cloeotis percivali and the Brachippasideros-Rhinonicteris group) but may he secondarily denved in other groups (e.g. the H. cyclops group; Hand, 1995). All hipposiderids known in Riversleigh’s Oligo-Miocene and Pliocene sediments. retain a P% as do most living Australian hipposiderids (A. auramius, H. ater, H. cervinus, H. semoni and H. diadema), Only living H. stenotis of NW Aus- tralia, a highly specialised member of the H. cyclops group (Hand. 1995), lacks a P? M. riversleighensis may he related to the Brac- hipposideros-Rhinonicteris group, sharing asini- ilar P? and large M3. M. riversleighensis, in lacking a P*, may repre- sent an aberrant Brachippoasideras, However, no other species of the Brachipposideros- Khinenicteris group shows this abnormality de- spite hundreds of specimens being available. The other very distinctive bat taxa in RV Site (a vespertilionid and a molossid) lend weight to the urgumentthat Miophyllorhinais also a distinctive but poorly represented taxon. Hand & Kirsch (in press) suggested a close relationship between Brachipposiderns and Cleeotis, early autapomorphically specialised branches of the hipposiderid radiation. Hill (1982) grouped Rhinonicteris, Cloeotis and Triaenops, Koopman (1994) referred them lo a separate subtrihe, the Rhinonycterina, Perhaps Miophyllorhina is part of this larger group of relatively plesiamorphic hipposiderids, Cloeotis shares with Miophyllorhina a very large M? and læk of P% but its P* is autapomorphically quite distinct and its M? heel very poorly developed. Alternatively, M. riversleighensis could be a distant relative of H. hernardsigei of the H. cy- clops group, interpreted by Hill (1963), Flannery & Colgan (1993) and Hand & Kirsch (in press) as derived hipposiderids. However, although it shares with Mtophyllorhina a similar M*, it re tains a reduced P? and its P* is derived, ACKNOWLEDGEMENTS Work at Riversleigh has been supported by the Australian Research Council, the Department of the Environment, Sport and Territories, National Estate Grants Programme (Qld), Queensland Na- ional Parks and Wildlife Service, the Australian Geographic Society, the Linnean Sogiety of NSW, ICL the Queensland Museum and the Uni- versity of NSW. This study would not have been possible without the support and encouragement of Michael Archer and Henk Godthelp. The fol- lowing people kindly provided access to compar- ative specimens in their institutions: B. Engesser, H. Felten, T. Flannery, W. Fuchs, L. Gibson, J. E. Hill, M. Hugueney, P. Jenkins, D. Kitchener, K- Koopman, P. Mein, R. Rachl, B, Sigé, N. B. Simmons, G. Storch, and S, Van Dyck, 1 àm grateful to Coral Gilkeson lor access to an SEM at Macquarie University. LITERATURE CITED ARCHER. M.. GOBDTHELP, H.. HAND, S.L & MEGIRIAN, D, 1989. Fossil mammals of Riversleigh, nochwestem Queensland: prelime- nary overview of biostratigraphy, correlation and environmental change. The Australian Zoolowst 25: 35-69. ARCHER, M., HAND, 8.J. & GODTHELP, H. 1994 Riversleigh. The story of the animals of Australia’s ancient inland rainforests. 2nd Tx, (Reed: Sydney) BLACK, K. & ARCHER, M. 1997. Nimiokoala gen. növ. (Marsupialia: Phascolarctidac) from Riversleigh, northwestern Queensland, with a n=- vision of Litokoula. Memoirs of the Queensland Museum 41; 209-228. CREASER, P, 1997, Oligocene-Miocene sediments of Riversleigh: the potential significance of topogra- phy. Memoirs of the Queensland Museum 41 303-314. FLANNERY, T.F. & COLGAN, DJ, 1993, A new species and two new subspecies of Hipposideros (Chiroptera) from western Papua New Guinea. Records of the Australian Museum 45: 43-57. HAND, S.J. 1993, First skull of a species of Hipposideros (Brachipposideros) (Microchiroptera: Hipposideridae), Irom Austta- lian Miocene sediments. Memoirs of the Queens- land Museum 31: 179-192 1995, Hipposideros bernardsigei, a new hipposiderid (Microchiroptera) from the Miocene of Australia and a reconsideration of the monophyly of related species groups. Miinchiver Geowissenschaftliche Abhandlungen. 1997, New Miocene leal-nosed hals (Microchiroptera: Hipposideridae) from Riversleigh, northwestern Queensland. Memoirs of the Queensland Museum 41; 335-349, HAND, S.J. & KIRSCH, J.A.W. in press. A southern origin for the Hipposideridae (Microchiroptera)? Evidence from the Australian fossil record. In Kunz, T. & Racey, P. (eds) Proceedings of the 10th International Bat Research Conference, Boston. (Smithsonian Institution: Washington). HILL, J.E. 1963. A revision of the genus Hipposideros. Bulletin of the British Museum (Natural History), Zoology 11: 1-129, 1982. A review of the leaf-nosed bats Rhinonycteris, Cloeotis and Triaenops (Chiroptera: Hipposideridae). Bonner zoologishe Beiträge 33:165-86. KOOPMAN, K.F. 1994, Chiroptera: systematics. Handbook of Zoology, VIII, 60, Mammalia: 1- 217. MEGIRIAN, D. 1992. Interpretation of the Miocene MEMOIRS OF THE QUEENSLAND MUSEUM Carl Creek Limestone, northwestern Queensland. The Beagle, Records of the Northern Territory „Museum of Arts and Sciences 9: 219-248. SIGE, B., HAND, S.J. & ARCHER, M. 1982. An Australian Miocene Brachipposideros (Mammalia, Chiroptera) related to Miocene rep- resentatives from France. Palaeovertebrata 12: 149-171. WOODBURNE, M.O., TEDFORD, R.H., ARCHER, M., TURNBULL, W.D., PLANE, M.D. & LUNDELIUS, E.L. 1985. Biochronology of the continental mammal record of Australia and New Guinea. Special Publications, South Australian Department of Mines and Energy 5: 347-363. THE FIRST FOSSIL PYGOPOD (SQUAMATA, GEKKOTA). AND A REVIEW OF MANDIBULAR VARIATION IN LIVING SPECIES MARK N. HUTCHINSON Hutchinson, M.N. 1997 06 30: The first fossil pygopod (Squamata, Gekkota), and a review of mandibular variation in living species. Memoirs of the Queensland Museum 41(2): 355-366. Brisbane. ISSN 0079-8835. The snake-like Australian pygopod lizards (Pygopodidae in traditional taxonomies) show considerable variation in mandibular and dental structure, correlated with dietary specialisa- tion in several genera. Following a review of this variation, a fully toothed Miocene dentary from Riversleigh, northwestern Queensland, is identified as a pygopod, (he First in the fossil record, Pygopus hortulanus sp, nov. is specifically distinguishable from living Pygapus by tooth morphology and proportions of the synyphysial region of the dentary. lizards. osteology, fossils, Miocene. | Pygapods, Mark N. Hutchinson, South Australian Muséum, North Terrace, Adelaide, South Australia 5000, Australia; 4 December 1996. One of the most distinctively Australian squa- male radiations is a group of snake-like, virtually limbless lizards, variously known as Map-footed lizards or snake-lizards (Bustard, 1970), These have no external trace of forelimbs while hind limbs are reduced to fin-like flaps on each side of the vent (Kluge, 1974; Shea, 1993). There are 35 species in 8 genera (Greer, 1989; Shea, 1991; Cogger, 1992). All are restricted la Australia and New Guinea, Flap-footed lizards have long been regarded as forming a distinct family, the Pygopodidae, closely related to the Gekkonidae (Underwood, 1954, 1957), It has been suggested that the sister group of Pygopodidae is not all Gekkonidae, but only the Australian Diplodactylinae, or some part of that subfamily (Kluge, 1987; King & Mengden, 1990). Acceptance of this phyloge- nene hypothesis (dissenting views exist; see Estes et al., 1988) would mean changes in taxon- omy, with the diplodactylines becoming a sub- family of the Pygopodidae, as proposed by Kluge (1987) or the pygopodids becoming a subfamily of the Gekkonidae (Bauer, 1990). Pending a con- census yiew I use Underwood's (1957) contrac- tion of their traditional family name, ‘pygapods’, as an informal collective term. Pygopods show considerable ecological and morphological diversity. Aprasia, Pletholax and Ophidiocephalus, exhibit fossorial adaptations and behaviour (Kluge, 1974; Ehmann. 1981; Shea & Peterson, 1993). However, Pletholax gracilis and at least some species of Aprasia are regularly active on the surface by day (Shea & Peterson, 1993; pers. obs.). Species in the largest genus, Delma, and that regarded by Kluge (1974; 1976) as most generally primitive, Pygepus, as well as Lialis, Aclys and Paradelma, ure surlace- dwellers (Wilson & Knowles, 1988; Greer 1989). While most pygopodids appear to be active for- agers feeding on invertebrates, the two species of Lialis are ambush predators of scincid lizards (Patchell & Shine, 1986a, by Murray et al., 1991). The mandibular and dental anatomy of pygopods is varied, and at least partly correlated with the ecological diversity just mentioned, The one overview of pygopod osteology attempted (Stephenson, 1962) gave scant attention to the mandible, Kluge (1974, 1976) employed some mandibular characters in his analysis of the clade, and Rieppel (1984) noted some correlations be- iween anatomy and miniaturisation in Aprasia and Pletholax when compared to Pygopus, Parker (1956) noted sexual dimorphism in Aprasia, in which only males have premaxillary teeth. Australia’s fossil lizard fauna 1s poorly known (Estes, 1983a, b; Covacevich et al., 1990; Hutch- inson, 1992); no fossil pygopods have been iden- tified. Fossils are essential for establishing a minimum estimate of the time 4 taxon has inhab- ited an area and of the time since the taxon first evolved. If pygopads are the sister group of diplodactylines, especially the Diplodactylini (Kluge, 1987), then itis likely that their differen- tiation oecurred in Australia and should be re- corded in the Australian fossil record. This paper is the first report of a fossilised pygopad. MATERIALS AND METHODS Dried skeletons (21 species in 8 known gencra- Appendix ) Were examined to assess interspecific 356 MEMOIRS OF THE QUEENSLAND MUSEUM FIG. 1. Labial (A) and lingual (B) views of the right mandibular ramus of a pygopodid lizard, Delma fraseri (SAMAR2291 1, Coomalbidgup, W.A.) showing major features, Abbreviations: adfo=adductor (=Meckelian) fossa; ani/=anterior mylohyoid foramen; asf=anterior surangular foramen; c=coronoid; cp=dorsal process of coranoid; cpd=compound bone - fused prearticular, surangular and articular; ctf=foramen for chorda tympani: d=dentary; iaf=inferioral veolar foramen; prp=‘prearticular process’ (medial articular process of the surangular); psf=posterior surangular foramina: rap=retroarticular process; s=splenial. Scale=5mm, variation in the sutural relationships of the màn- dibular bones, positions of foramina, tooth num- ber and tooth morphology. With the exception of Pygopus, this sample did not permit study of intraspecific variation. Unless. otherwise speci- fied, the descriptions refer to the anatomy as seen in intact mandibles. Outgroups used to infer de- rived character states are diplodactylines as the sister group of pygopods, gekkonines the sister of these two and other scleroglossans as the most distant outgroup (Estes et al., 1988). PYGOPOD MANDIBLE (Fig. 1). The dentary 1s the largest bone of the pygopod mandible. It consists of the tooth-bearing body of the bone and a relatively long, posteriorly di- rected angular process that covers. much of the labial and ventral surface of the mandible. Pygopods share with other gekkonoids and mem- bers of several other families the complete oblit- eration by dentary overgrowth of the groove for Meckel’s cartilage, but differ, at least from all diplodactylines examined, in that the angular pro- cess extends on the labial surface of the mandible to well behind the coronoid (the dentary also extends posteriorly to a marked degree in the gekkonine Paroedura, Kluge, pers. comm.). The dentary of diplodactyline and gekkonine geckoes is generally more slender and incurved than that of pygopods and the angular process terminates at about the level of the coronoid. 3-8 mental foramina open along the labial surface of the dentary, the series generally extending posteri- orly to 1/2-3/4 the length of the tooth row. There is NO posterior extension of the bony internal septum separating the Meckelian cartilage from the inferior alveolar nerve; bony separation is limited to the immediate vicinity of the mental foramina. Estes et al. (1988) described this char- acter (their number 56) in terms which emphasised that a vertically-oriented, posteriorly extended intramandibular septum is well-devel- oped in anguimorphs, but failed to note that it occurs to almost the same extent in lygosomine skinks (Shea & Hutchinson, 1992, fig. 3). Adult geckoes typically have large numbers of small dentary teeth (Bauer & Russell, 1990; Kluge & Nussbaum, 1995); in a sample of 9 diplodactyline, 1 sphaerodactyline and 14 gekkonine genera, dentary counts ranged trom 25 (Phelsuma madagascariensis) to 62 (Saltuarius salebrosus), with a mean of 40,2 (Edmund. 1969; pers. obs.), In adult Delma and Aclys tooth num- ber is very much like that of diplodactyline and gekkonine geckoes, typically in the range 25-35. Reductions to 24 or fewer, or increases to 50 or more are therefore likely to be apomorphic PYGDPOD MANDIBLE (Kluge, 1976). Diplodaciylines and gekkonines usually have slender, upright teeth with acute crowns bearing a pair of pointed apical cusps separated by a groove (Sumida & Murphy, 1987)- Teeth with this morphology occur in some pygopods (Delma and Aclys), and are probably plesiomorphic for the group. Within pygopods there is marked intergeneric variation, including robust, upright or slightly recurved teeth, less rabust but more strongly recurved teeth, or very small teeth with compressed, sharp-edged crowns. All retain an apical groove, but the bitus- pid structure is largely lost, the labial cusp enlarg- ing to become the tooth apex, while the lingual cusp all but disappears. In Lialis tooth crowns are so compressed that the apical groove is faint and only discernible in unworn teeth, The coronoid consists o! a laterally compressed dorsal process, an anteriorly-directed dentary process and a posterior process. The dentary pro- cess is bifurcated and clasps the dentary bone behind the end of the tooth row both labially and lingually, On the lingual face of the mandible, the anterior exiremily of the dentary process termi- nates posterior to, coextensively with, or anterior to the front ot the splenial, the latter 2 character stales being apomorphic with respect to other gekkonoids and other lizards. The posterior pro- cess of the coronoid extends to the anterior ex- tremity of the Meckelian fossa. The form of the dorsal process varies from tall and fin-hke in several genera to very low in Lialis (Kluge, 1976). A well-developed dorsal process is the rule in geckoes and other lizards and is likely to be plesiomorphic for pygopods. The splenial is reduced in all pygapods com- pared to the development seen in diplodactylines ànd other geckoes, usually failing to extend ante- riorly beyond the level of the distal two or three teeth. Tn most pygopods the splensal is further reduced in length or depth, The splenial is com- pletely absent in Aprasia (pers, obs,; Parker, 1956; ‘very slight’, Stephenson, 1962). The splenial, when present, completely surrounds (as in diplodactylines) the inferior alveolar foramen and bears a notch for the anterior mylohyoid forumen on its ventral suture with the dentary. Like the majority of geckoes (Kluge, 1987), pyzopods lack a distinct angular, The splenial in Delma and Pygopus has a posteriorly extending process that separates the dentary and prearticu- lar, as would an angular, suggesting thal the an- gular has been lost via fusion with the splemial, The surangularis fused labially and pasteriorly with the fused prearticular-articular in aduh pygopods, but is completely distinet in juveniles (Delma molleri, Lialis burtanis and Pygopur lepidopodys), In aduhs of must genera a suture persists on the lingual face of the mandible run nmg antenorly from the Mecketiam lussa; in in- lact mandibles this suture may be concealed by the coronoid, 3 foramina are usually present on the labial surface of the surangular. The anteriorly directed opening of the antenor surangular lura- men lies on the labial surface on oF just posterior to the point of intersection of the sutures between the coronoid, dentary and surangular. 2 other foramina usually lie towards the postervlabial region of the surangular, but there is imergenenc variation, In some pygapods there may he only a single foramen, as in diplodactylines, but in most there are 2, The 2 openings may be close together, or moderatcly separated, and the more pasterodorsal of the 2 may itsell be subdivided, Most lizards have only a single posterior stir- angular foramen, and so the additional (more anterior) foramen must be identified. Kluge (1967) and Grismer (1988) designated the more ventral of the posterior foramina as the posterio mylohyoid and the more dorsal as the posteriur surangular foramen. My survey of gekkoneid mandibular variation suggests that this interpre- taton is incorrect. The posterior mylohyoid nerve innervates. the throat musculature (Camp, 1923; Poglayen-Neu- wall 1954) and typically exits through a medially or ventromedially direcied foramen in the angular bone. While the angular is absent in mosi gekkonoids (Kluge, 1987) some of those exam- ined (the gekkonines Gekko, Phelsuma and Phy- Hodactylus [=Christinus)) retain a small foramen on the dentary-splenial suture in the expected topographic position of the posterior mylohyoid foramen. Other pekkonines and other gekkanvids examined, including all pygopods, lack a fora- men in this position, The foramen identified by Kluge (1967) and Grismer (1988) as the posterior mylohyoid opens on the labial surface of the mandible and runs through to open lingually into the Meckelian fossa, This foramen thus follows the course of the posterior surangular foramen of other lizards, and it 100 transmits a branch of the mandibular nerve to the adductor (pterygoideus) musculature (pers. obs,). The foramen and nerve are remote from the throat musculature which, hy definition, the pos- terior mylohyoid supplies. lt therefore seems to me more probable that the additional foramen on the labial surface of the surangular in pygopods and other zekkonoids represents a duplication of 358 MEMOIRS OF THE QUEENSLAND MUSEUM FIG, 2, Labial views of right mandibular ramus of the 8 pygopodid genera. A, Delma inornata (SAMAR22408, no data); B, Aclys concinna (SAMAR38060, Badgingarra National Park, W.A.); C, Pygopus nigriceps (SAMAR21029, no data); D, Paradelma orientalis (QMJ30250, Cracow, Qld); E, Ophidiocephalus taeniatus (SAMAR28365, Abminga, S.A.); F, Pletholax gracilis (SAMAR38061, Jandakot, W.A.); G, Aprasia striolata (SAMAR35569, Mylor, S.A.); H, Lialis burtonis (NMVD15399, Warby Ranges, Vic.). Scales=5mm. PYGOPOD MANDIBLE FIG. 3. Lingual views of same specimens in Fig. 2, drawn to same scales. 359 the posterior surangular foramen, not a displaced posterior mylohyoid, The fused prearticular-articular together with the surangular constitutes the posterior 1/3 of the pygopod mandible. As in diplodactylines and other gekkonoids the articulating facet is orented to face posterodorsally rather than dorsally, re- sulting in a marked ‘step’ down from the level of the upper edge of the adductor fossa 1o the level of the retroarticular process, The latter structure is variable intergencrically in its shape (spoon- shaped to rod-like) and the degree of medial or ventral inflection (Kluge, 1976), A foramen for ihe n. chorda tympani opens on the dorsolingual aspect of the base of the retroarticular process. The posterior region of the Meckelian fossa, which contains the internal opening for the pos- terior surangular foramen, may be demarcated from the anterior region. MANDIBULAR VARIATION IN LIVING PYGOPODS DELMA. This genus has the most generally ‘gecko-like’ mandibles. However, there is signif- icant variation within the genus in proportions, tooth crown shape, size of splemal and other features. Deima fraseri (Fig. 1) and D. inornata (Figs 2A, 3A) show some of this variation, with the latter species tending to retain more plesiomorphie features than the former, The pygopod synapamorphy of extensive posterior extent of the dentary on the labial surface is present in all Delma , bul the teeth are numerous, with unmodified (e.g, D impar, D, inornata, D. nasuta or slightly expanded crowns (e.g. D. fraseri, D. miella). Whether this is inter- or in- traspecific variation will require a nore extensive survey, Tooth crowns retain the bicuspid mor- phology of diplodactyline geckoes, The dentary is moderately slender and the splenial little re- duced, although it usually fails to extend anteri- orly as far as the poSteriurmos| looth (D. inornata specimens showed intraspecific variation, the splenial failing to reach the tooth row in NMVD15448, reaching the second-last tooth in SAMAR35570 and extending as far as the sixth- last in SAMAR22408, Fig. 2A). The posterior surangular foramina are moderately to narrowly separated (variable bhoth inter- and intraspecifi- cally). The relatively unspectalised dentition is associated with a generalised arthropad dict (Shine & Patchell, 19864; Coulson, 1990). ACLYS (Figs 2B, 3B}. In general the mandible af MEMOIRS OF THE QUEENSLAND MUSEUM this genus is an elongate version of Delmas. Elongation of the jaw oceurred by lengthening of the region between the tip of the coronoid process of the dentary and the posterior end of the tooth row, with hypertrophy of the lingual ramus of the anterior process of the coronoid reducing the exposure of the splenial and widely separating it from the tooth row. Height of the dorsal process of the coronoid is. reduced relative to mos! Delma and the medial articular process (Fig. 1) of the surangular is elevated, both trends foreshadow- ing the extensive coronoid flattening and sur- angular elevation of Lialiy, The derived features of the jaw of Aciys are all seen, although not to the same degree. insome Delma, especially those with more elongate skulls such as D. burleri and D. nasuta. PYGOPUS (Figs 2C, 3C, 4B, 4D). The mandibles of the two species placed in this genus are very similar to one another, and probably indistin- guishable, The form of the mandible is apomorphic with respect to thal of Delma in being shorter and deeper, The dentition is also apomorphic, the teeth being fewer (<25), much more robusi, and with sharp, tapering, recurved crowns, The tooth crowns retain a pronounced apical groove, but the typical gekkonoid bicuspid structure is reduced, with the labial cusp being the principle tooth apex while the lower, lingual cusp is little more than the acule-angled inner margin of the apical groove. The mandible is more plesiomorphic than most Delma in that the splen- ial extends forward to underly the posteriormost teeth. The posterior surangular foramina are mod- erately to widely separated (intraspecitically yanable), The genus is characterised by a pro- nounced mesio-distal decrease in tooth size, the mesial teeth (second to fifth) being 30-40% taller than the mid-dentary teeth (tenth ta twelfth), The enlarged teeth at the front of the jaw are some- what procumbent and are supported by u deep symphysial region. Pianka (1986) described P. nigriceps as a scorpion specialist, and Patchell & Shine (1986a) found that the major prey of P. lepidopedus were mygalomorph and lycosid spi ders, Possibly the relatively powerful front teeth are adaptations for rapidly disabling such poten- ually dangerous prey. PARADELMA (Figs 2D, 3D). This monotypic genus, like Pygopus, has a reduced tooth number (21) compared with Delma but the teeth are dis- tinctive, being more slender than in Pygapus and having recurved crowns, The tooth apices are like PYGOPOD MANDIBLE those of Pygopus in that the apical groove is present but the lingual cusp is barely developed. Compared with Pygapus, the jaw is less robust and more bowed. The dict is unknown. OPHIDIQCEPHALUS (Figs 2E, 3E). The jaw of this small fossorial form is relatively robust. short und deep, similar in proportions to that of Pygopus, but is apomorphic in several characters. The splenial is greatly shortened and shifted pos- leriorly compared with Pygapus. The single jaw examined is distinctive in that the lingual ramus of the anterior process of the coronoid is reduced, exposing the bone beneath, Based on the position of the prearticular-surangular suture exposed below the dorsal process of the coronoid, this anteriorly exposed bone is the surangular. The posterior surangular foramina are narrowly sepa- tated, The teeth are similar in to those of Pygopus, but are fewer in number (13-15 in adults versus 17-24 in hatchling to adult Pygapus) and have moderately recurved crowns, Recorded prey in- dicates a relatively generalised arthropod diet fEhmann, 1981). PLETHOLAX (FigS 2F, 3F). The jaw is long and slender, but with a well-developed dorsal process of the coronoid. Tooth counts are below 20 (Kluge, 1976), with the mesial teeth relatively robust, pointed and slightly recurved while the more distal teeth are markedly reduced in size, The splenial is reduced to a narrow splint but still encloses the inferior alveolar foramen and forms the dorsal margin of the anterior mylohyoid fora- men. There is a single posterior surangular fora- men. Shea & Peterson (1993) found that most guts of this species contained little chitinous ma- terial but often included short cut lengths of grass, consistent with digestion from the bodies of in- sect prey, They suggested the most likely diet was poorly-sclerotised, readily digested prey such as termites. Ehmann (1993) suggested that Pletholax is a nectar feeder. The weak jaws and small, widely spaced teeth suggest it would be unlikely to deal effectively with the tough exo- skeletons of typical invertebrate prey. APRASIA (Figs 2G, 3G), The mandible has only 3 elements; dentary, coronoid and a compound bone representing the remainder; the splenial is absent, whether through loss or fusion is unclear. The interior alveolar foramen lies on the suture between the anterior extremity of the compound bone and the dentary, The posterior surangular foramen is single. The dersal process of the 361 coronoid is reduced and the retroarticular process is abbreviated. The teeth are very greatly reduced in number, restricted to a pateh of 3-4 relatively robust, pointed, recurved teeth situated close to (but not on) the symphysis. The dict is restricted to the eggs, larvae and pupae of ants (Webb & Shine, 1994). The mandibular and dental anat- omy of Aprasiais thus convergent to some extent on that of ant-ealing scolecophidian snakes, which also have weak, almost edentulous jaws save lor a few teeth in the upper (Typhlopidae) or lower (Leptotyphlopidae) mandible, LIALIS (Figs 2H, 3H), This genus has a highly derived mandibular morphology. The dentary is greatly allenuated, with many (lo over 60) small, recurved, sharply pointed teeth having ligamen: jous basal attachments (L burtonis, Patchell & Shine, 1986b; L jicaré, G. Shea, pers. comm), Other peculiarities include the greatly reduced height of the coronoid and a peculiar fan-like medial articular process of the surangular ascend- ing well above the level of the articular region, and higher than the coronoid, Functionally this braces the jaw against the quadrate ramus of the pterygoid, preventing lateral displacement of the jaw when the mouth is open. Along with these apomorphies, Lialts retains a plesiomorphic, rel- auvely large. antenorly extending splenial. The posterior surangular foramen is single or double. The wide gape and specialised dentition are re- lated to the obligate Jizard-eating (especially skink-eating) habits of this genus (Patchell & Shine, 1986h). SYSTEMATICS Infraorder GEKKOTA Cuvier, 1817 Family PYGOPODIDAE Pygopus Merrem, 1870 TYPE SPECIES. Bipes lepidapedus Lacepede, 1804 Pygopus hortulanus sp. noy. (Figs ) ETYMOLOGY. Latin hortulanus, of or belonging to a garden; alluding to the Neville’s Garden Site. MATERIAL. Holotype QMF16875 (Fig. 4), a right dentary preserving the complete tooth row and sym- physial region but minus the angular process from early Miocene, System B (Archer et al., 1994), Neville's Garden Site on D Site Plateau at Riversleigh, NW Queensland; the site is interpreted as representing an accumulation in a pool close toa cave entrance 362 MEMOIRS OF THE QUEENSLAND MUSEUM FIG. 4. Comparison of the dentaries of Pygopus hortulanus SP, NOV. (A and C, QMF16785) and P. lepidopodus (B and D; SAMAR38928). A-B, labial views; C-D, lingual views. Note larger mesial (anteriormost) teeth and deeper symphysis in P. lepidopodus. Scale=Smm. DESCRIPTION, Right dentary, complete and undamaged except for the absence of the angular process posterior to the level of the end of the tooth row. Dental arcade in occlusal view straight from distal tooth anteriorly to about seventh tooth (counting mesiodistally), then curving gently me- sially. Lingual rim of dental sulcus distinct, rounded in cross-section. Groove for Meckel's cartilage completely obliterated by overgrowth of dentary. Intramandibular septum not extended posteriorly. In labial view, acute-angled splenial notch extending to level of penultimate (twenti- eth) tooth, and superficial facet for anterior exten- sion of splenial extending as far as seventeenth tooth. Wedge-shaped facet for anteroventral ex- tremity of coronoid incised into posterior margin between dorsal edge of splenial notch and last tooth. Labial surface with 5 mental foramina, last level with fourteenth tooth. Shallow fossa on labial surface tapering anteriorly from posterodorsal margin of dentary. Total length 6.0mm; depth at level of last tooth 1.4mm (tooth excluded). Tooth loci 21. Counting from mesial to distal, all but first (broken), ninth and fifteenth (empty) with intact teeth. Twentieth tooth lost subsequent lo preparation of Fig. 4. Length of tooth row 5.2mm. Teeth robust, 2.5-3 times as high as wide, crowns tapering rapidly to sharp points. Well-de- fined groove traversing each tooth crown, setting off weak second apical point, lower and lingual to main tooth apex. Tooth crowns slightly in- curved. Teeth reducing gradually in size from PYGOPOD MANDIBLE mesial to distal, but anteriormost teeth only slightly larger than mid-dentary teeth, REMARKS. Identification of the fossil as a gekkonoid is based on apomorphic features: 1) obliteration of the groove for Meckel’s cartilage hy overgrowth of the dentary, 2) apex of the splenial notch level with the posterior end of the tooth row, with a narrow tongue of the splenial extending forward over the lingual face of the dentary below the tooth row, and 3) mesial (i.e, anteriormost) teeth largest (rather than mid-den- lary teeth). Many Australian skinks (Scincidae) also have the Meckelian groove obliterated by the dentary, but in these the splenial notch extends well for- ward under the tooth row (the tiny Noefoscinenus, tooth row <3mm, is an exception), and scincid dentanes with a closed Meckel's groove all have a vertically oriented, posteriorly extended in- trumandibulur septum. Skinks, unlike gekkonoids (Bauer, 1990; Bauer & Russell, 1990), have the mid- dentary teeth larger than the mesial teeth. Scincid dentaries also have a pro- nounced coronoid process on the labial surface (Estes, 1983); although this part of the specimen is incomplete, existing outlines indicate that very little of this region is lost and that there was scarcely any development of this process, The only other lizard families in which dentary oblit- eration of the Meckelian groove occurs are the Xantusiidae (all), many Gymnophthalmidae and many iguanians (Presch, 1980; Estes et al., 1988; Etheridge & de Queiroz, 1988; MacLean, 1974). These can be excluded on the basis of tooth crown shape (McDowell & Bogert, 1954, Sumida & Murphy, 1987; MacLean, 1974; Etheridge & de Queiroz, 1988), and having mid-dentary teeth Jarger than mesial teeth. Among gekkonoids, only pygopods have the stoul, Straight dentary shape, low dentary tooth counts and teeth of the robust form seen in P. hortulanus, Pygopus hortulanus shows two seemingly de- rived character states within pygopads, the rela- tively short, deep dentary and the fewer robust teeth. These relate it not only to Pygopus but also Ophidiocephalus. The fossil is plesiomorphic with respect to Ophidiocephalus in straighter sooth crowns, more teeth and greater anterior extent of the splenial. ft is plesiomourphic with respect to Pygopus perhaps in the less enlarged mesial teeth and (?correlated) shallower sym- physial region. Alternatively the more even tooth row of P. hortulanus could be interpreted as 363 autapomorphic, because enlarged anterior teeth are common in pygopods. The placement of the species in Pygopus rather than Ophidiocephalus reflects the fewer specialisations shared with Op- hidiocephalus than with Pygopus. The cladistic analysis of intergeneric relation- ships reported by Kluge (1974, 1976) not only concluded that Pygopus is the most primitive extant genus but also placed Pygopus as a pride group at the base of the pygopod radiation and sugpested that Paradelma orientalis is more closely related to Pygopus nigriceps than the latter is to Py. lepidopadus. The placement of the fossil in Pygopus could therefore be taken to mean only that the fossil is. a plesiomorphic py go- pod. However, the short, deep dentary and robust teeth of Pygopus are probably apomorphic in pygopods, and I therefore maintain the concept of this genus employed by Cogger (1985), Greer (1989) and Shea (1993). This early Miocene pygopod is consistent with Kluge’s (1987) suggestion thal pygopods evolved on the Australian continent subsequent to a Late Cretaceous vicariant event isolating the ancestral diplodactyline-pygapod stock, This find, with its apomorphic teeth, is not a generalised ancestral pygopod, implying the ori- gins of the group must be older than the Miocene, Archer et al. (1989; 1995) suggested that the Miocene environment of Riversleigh was primar- ily tropical closed forest, an assessment sup- ported by many genera in Systems B and C whose living representatives are restricted to closed for- estenvironments, Megirian (1992) suggested that sedimentological evidence argued fora more arid climate with any rainforest limited to water courses. Creaser (this volume) reports that sedi- ment patterns regarded by Meginan as being confined to arid depositional environments occur today in mid-montane New Guinea. No modem pygopod inbabits rainforest, al though Lialis and some Delma (Shea, 1987) in- habit vine scrubs and cucalypt forest on the margins of rainforest. The présence of P. hortulanus in System B could be interpreted to indicate either that there were drier, open patches nearby, or thal P. hortulanus was a rainforest- dweller with no living analogue, ACKNOWLEDGEMENTS I thank Mike Archer for enabling me to work on the Riversleigh lizurd fossils. G.M. Shea pre- pared the dentary of Delma tritella. A.G. Kluge and G. M. Shea commented on the fossil. For 364 pecnenos to prepare skeletal material from col- ections in their care and for the loan or exchange of specimens, I thank J, Covacevich and P, Couper (Queensland Museum), K. Aplin and L. A. Smith (Western Australian Museum) J, Cov- entry (Museum of Victoria) and R. Sadlier (Aus- tralian Museum), Fieldwork at Riversleigh is supported by the Australian Research Grants Scheme, Australian Department of the Arts, Sport, the Environment, Tourism and Territories, National Estate Programme Grants Scheme, Uni- versity of NSW; Wang Computers, Australian Geographie Society, Mount Isa Mines, Queens- land Muscum, Australian Museum, Royal Zoo- logical Society of NSW, Linnean Society of NSW, Anset/Wridgways, Mr Isa Shire Council, Riversleigh Society and the Friends of Riversleigh, LITERATURE CITED ARCHER, M., GODTHELP, H. HAND, S, J & MEGIRIAN, D. 1989. Fossil mammals of Riversleagh, northwestern Queensland: a prelimi- nary overview of biostratigraphy, correlation and taal change. Australian Zoologist 25: 29-65. ARCHER, M., HAND, 5. 1. & GODTHELP, H. 1994. Riversleigh. 2nd ed. (Reed: Sydney). 1995, Tertiary environments and biotic change in Australia. Pp. 77-90. In Vrba, E.8., Denton, G.H., Purindge, T.C, & Burckle, LH, Paleoclimate and evolution, with emphasis on human origins. (Yale University Press; New Haven). BAUER, A. M, 1990, Phylogenetic systematics and biogeography of the Carphodactylini (Reptilia: Gekkonidae), Bonner Zoulugisches Monographien 30: 1-217, BAUER, A.M, & RUSSELL, A.P. 1990, Dentitional diversity in Rhacodactylus (Reptilia: Gekkonidae). Memoirs of the Queensland Mu- seum 29: 31 1-321. COGGER, H,G, 1992. Reptiles and amphibians of Aus- tralia. 5th ed. (Reed: Sydney). COULSON, G. 1990. Conservation biology of the striped begless lizard, (Delma impar) an initial investigation. Arthur Rylah Institute for Environ- mental Research Technical Report 106: 1-40. COVACEVICH, J., COUPER, P, MOLNAR, R.E. WITTEN, G. & YOUNG, W. 1990, Miocene dragons from Riversleigh: new data on the history of the Family Agamidae (Reptilia: Squamata) in Australia. Memoirs of the Queensland Museum 29: 339-360. EDMUND, G, 1949, Dentition, Pp, 117-200, In Gans, C., Belluirs, A.d’ A. & Parsons, T.S. Biology of the Reptilia, vol, 1, Morphology A, (Academic Press; London). EHMANN, H. 1981. The natural history and conserva- MEMOIRS OF THE QUEENSLAND MUSEUM tion of the bronzeback (Ophidiocephalus taenians Lucas & Frost) (Lacertilia, Pygopodidae). Pp, 7-13, In Banks, C. & Martin, A.A, Proceedings of the Melbourne Herpetologi- cal Symposium. (Zoological Board of Victoria: Melbourne). 1993. Encyclopacdia of Australian animals: reptiles, (Australian Museum with Angus & Robertson: Sydney). ESTES, R. 1983a. Encyclopedia of palecherpetology,, vol, 10A, Sauria terrestria, Amphisbaenia. (Gus- tav Fischer Verlag: Stuttgart). 1983b. The fossil record and early distribution of lizards, Pp, 365-398, In Rodin, A.G.J. & Miyata, K. Advances in herpetology and evolutionary biology. (Museum of Comparative Zoology. Harvard: Cambridge). ESTES, R.. DE QUEIROZ, K, & GAUTHIER, J. 1988. Phylogenetic relationships within Squamata, Pp. 119-281. In Estes, R. & Pregill, G. (eds), Phylo- genetic relationships of the lizard families. (Stan- ford University Press:Stanford). ETHERIDGE, R, & DE QUEIROZ, K. 1988. A phy- jogeny of Iguanidae. Pp. 283-368. In Estes, R. & Pregill, G. (eds), Phylogenetic relationships of the lizard families. (Stanford University Press: Stan- ford), GREER, A.E. 1989. The biology and evolution of Aus- tralian lizards, (Surrey Beatty & Sons and the Royal Zoological Society of NSW: Sydney). GRISMER, L. L. 1988. Phylogeny, taxonomy, classifi- cation and biogeography of eublephurid geckos. Pp. 369-469. In Estes, R. & Pregill, G. (eds). Phylogenetic relationships of the lizard families. (Stanford University Press; Stanford). HUTCHINSON, M.N. 1992. Origins of the Australian selncid lizards: a preliminary report on the skinks of Riversleigh. The Beagle, Occasional Papers of the Northern Territory Museum of Arts and Sci- ences 9; 61-70. KING, M. & MENGDEN, G. 1990, Chromosomal evolution in the Diplodactylinae (Gekkonidae: Reptilia), If. Chromosomal variability between New Caledonian species, Australian Joumal of Zoology 38: 219-226, KLUGE, A. G. 1967, Higher taxonomic categories of gekkonid lizards and their evolution. Bulletin of the American Museum of Natural History 135: 1-60, 1974. A taxonomic revision of the lizard family Pygopodidae. Miscellaneous Publications of the Museum of Zoology, University of Michigan 147; 1-221. 1976. Phylogenctic relationships in the lizard family Pygopodidac: an evaluation of theory, methods and data, Miscellaneous Publications of the Mu» seum of Zoology, University of Michigan 152: 1-72. 1987. Cladistic relationships in the Gekkonvidea (Squamata, Sauria). Miscellancous Publications PYGOPOD MANDIBLE of the Museum of Zoology, University of Migh- igan 173: 1-54. KLUGE, A.G, & NUSSBAUM, R.A. 1995. A review ol African- Madagascan gekkonid lizard phylog- eny and biogeography (Squamata). Miscellaneous Publications of the Museum of Zoology, Univer- sity of Michigan 183; 1-20, MACLEAN, W, P. 1974, Feeding and locomotor mech- anisms of teiid lizards: functional morphology and evolution, Papeis Avulsos de Zoologia, Sau Paulo 27: 179-213. MCDOWELL, S. B. & BOGERT, C. M. 1954, The systematic position of Lanthanotus and the affin- ities of the anguinomorphan lizards, Bulletin of the American Museum ol Natura! History 105: 1-142. MURRAY, B.A, BRADSHAW, S.D. & EDWARD, D.H. 1991, Feeding behavior and the occurrence of caudal luring in Burton's pygopodid Lialis burronis (Sauria: Pygopodidae). Copeia 1991; 509-516. MEGIRIAN, D. 1992, Interpretation of the Miocene Carl Creek Limestone, northwestem Queensland. The Beagle, Occasional Papers of the Northern Territory Museum of Arts and Sciences 9: 219- 248. PARKER, H.W. 1956. The lizard genus Aprasia, Bul- letin of the British Museum (Natural History), Zoology 3: 363-385. PATCHELL, F.C. & SHINE, R. 1986a, Food habits and reproductive biology of the Australian legless liz- ards (Pygopodidae). Copeia 1986: 30-32, PATCHELL, F.C. & SHINE, R, 1986b, Hinged teeth for hard bodied prey: a case of convergent evolu- tion between snakes and legless lizards. Journal ot Zoulogy, London 208; 269-275. PIANKA, E.R. 1986, Ecology and natural history of desert lizards, (Princeton University Press; Princeton), POGLAYEN-NEUWALL, 1. 1954. Die Kiclermuskulatur der Bidechsen und ihre Innerva- lion, Zemsehrilt für wissenschaliche Zoologie 158: 79-] 32. PRESCH, W. 1980, Evolutionary history of the South American nierotwiid lizards (Teiidae: Gym- nophthalminae), Copeia 1980: 36-56. RIEPPEL, O. 1984. Miniaturization of the lizard skull: its functional and evolutionary implications. Pp 503-520. In Ferguson, M.WJ.(ed.), The structure, development and evolution of reptiles, Sympo- sium of the Zoological Society of London 52: - SHEA, G.M. 1987, Two new species of Delma (Lacertilia: Pygopodidae) from northeastern Queensland and a note on the satus of the genus Aclys, Proceedings of the Linnean Society of N.S.W. 109: 203-212, 199]. Revisionary notes on the genus Delma (5quamata: Pygopodidae) in South Australia und the Northern ‘Territory. Records af the South Australian Museum 25; 71-90. 1993, Family Pygopodidae. Pp 254-249, In Glasby, jicari C.J.. Ross, G.J.B. & Beesley, P.L. Fauna ol Aus- tralia. Vol 2A, Amphibia.and Reptilia. (Austra- lian Government Publishing Service: Canberra), SHEA, G.M, & HUTCHINSON, M.N, 1992. A new species of Tiligua Jrom the early Mivcene ol Riversleigh, Queensland. Memoirs of the Queens- land Museum 32; 303-310. SHEA, G.M. & PETERSON, M. 1993, Notes on the biology of the genus Pletholax Cope (Squamalia: Pygopodidae), Records of the Western Australian Muscum 16; 419-425. STEPHENSON, N.G. 1962. The comparative morphol- ogy of the head skeleton, girdles and hind limbs in the Pygopodidae. Zoological Joumal of the Linnean Society 44: 627-644. SUMIDA, 5.5, & MURPHY, R.W, 1987, Form and function of the tooth crown structure in gekkonid lizards (Reptilia, Squamata, Gekkonidae). Cana- dian Journal of Zoology 65: 2886-2892. TYLER, MJ., HAND, S.J, & WARD, V.J, 1990. Anal- ysis of the frequency of Lechriadus intergerivus Tyler (Anura; Leptodactylidae) in Oliga-Miocene local faunas of Riverslelgh Station, Guisehstand. Proceedings of the Linnean Society of N.S, W. 112° 105-109. UNDERWOOD, G, 1954, On the classilication and evolution of geckos Proceedings of the Zoologi- cal Society of London 134: 169-492. 1957, On lizards of the family Pygopodidae. Journal of Morphology 100: 207-268. WEBB, JK, & SHINE, R. 1994. Feeding habits and reproductive biology of the Australian pygopodid lizards of the genus Aprasia. Coperta 1994; 390- 398, WILSON, S.K, & KNOWLES, D.G. 1988. Australia’s rēptiles, A photographic reterence tothe terrestrial reptiles of Australia, (Collins:Sydney). APPENDIX Pygopodid skeletal specimens exammed: Actys concinna SAMA R38060; Aprasia inauriti SAMA R14275; A. pseudopulchella SAMA R406A,; A. striolata SAMA R35569, R41825: Aprasia sp. SAMA unregistered alizarin speci- men; Delma australis SAMA R15958; D, butleri SAMA R14913, RI6843A: D. fraseri SAMA R22911; D. impar NMV D15446; D. inernata SAMA R222408, R35570, NMV D15448; D. mitella AMS R65264 (partial dentary only); D. molleri SAMA R22540, R35572. D, nasute SAMA R22517: D. plebeia QM JSB9 1: D. tincta SAMA R]5189A; Lialis burronis SAMA KI5882. R4U031, QM J47481, NMY D15399; L. SAMA RIl44l; Ophidiocephalses 366 MEMOIRS OF THE QUEENSLAND MUSEUM taeniatus SAMA R28365 (mandible only); Paradelma orientalis QM J30250 (mandible only); Pletholax gracilis SAMA R38061; Pygopus lepidopodus SAMA R19604 (mandible only), R35571, R38924-28; P. nigriceps SAMA R1250, R21029. TWO NEW EARLY MIOCENE THYLACINES FROM RIVERSLEIGH, NORTHWESTERN QUEENSLAND JEANETTE MUIRHEAD Muirhead, J. 1997 06 30: Two new early Miocene thylacines from Riversleigh, northwestem eealen Memoirs of the Queensland Musewn 41(2); 367-377. Brisbane, ISSN 0079- Thylacines, Wabulacinus ridei gen. et sp. nov.and Ngamalacinus timmulvaneyi gen. el sp. nov., are described from the early Miocene of Riversleigh, northwestern Queensland. Both show camivorous adaptation intermediate between that of the plesiomorphic Nimbacinus dicksoni and derived Thylacinus. The family concept is revised to include these new taxa. All known thylacinid genera occur in late Oligocene to middle Miocene Riversleigh faunas and some may have overlapped in time followed by a decline in family diversity since the Miocene. [C] Fhylacine, marsupial, carnivore, Miocene, Riversleigh, Queensland. J. Muirhead, School of Biological Sciences, University af New South Wales NSW 2052 Australia; received 25 June 1995. The Thylacinidae consists of three species of Thylacinus (T. cynocephalus Harns, 1808, T- pres Woodburne, 1967 and T, macknessi uirhead, 1992) and the monotypic Nimbacinus dicksoni Muirhead & Archer, 1990 from the late Oligocene to middle Miocene of Queensland and the Northem Teritory (Muirhead & Archer, 1990). It is the oldest and most primitive thy- lacinid, more closely resembling dasyurids in many plesiomorphic features. Thylacinus potens from the late Miocene Alcoota Local Fauna (Woodburne, 1967) is considered (Archer, 1982) the sister species of modem T, cynocephalus and is almost as specialised. Thylacinus macknessi, from early to middle Miocene Riversleigh faunas, is also a highly specialised thylacine. Because it retains some plesiomorphic features, it is consid- ered to be the sister species to the T, patens -T. cynocephalus clade (Muirhead, 1992}. Two new early Miocene thylacinids from Riversleigh are described here. In many features they provide a continuum in morphological change fom the plesiomorphic dentition of N, dicksoni to that of specialised Thylacinus. Dental nomenclature fol- lows Flower (1869) and Luckett (1993) where the adult dentition includes P1-3 and M1-4. Taxo- nomic nomenclature follows Muirhead & Archer (1990). Material is housed in the Queensland Museum (QMF) or Northern Territory Museum. SYSTEMATICS Order DASYUROMORPHIA (Gill, 1872) Superfamily DASYUROIDEA (Goldfuss, 1820) Family THYLACINIDAE (Bonaparte, 1838) Wabulacinus gen, nov, TYPE SPECIES. Wabulacimus ridet pen. et sp. nov. ETYMOLOGY. Wanyii Wabula, long ago: Greck kynos, dog. Masculine. DIAGNOSIS. Infraorbital foramen surrounded whally by the maxillary and positioned low and anterior to M!; centrocrista and preparacrista parallel, forming contin- uous straight line on MI. entoconid absent (on Mak hypoconulid enlarged (on M3). COMPARISON, Wabulacinus differ from N. dicksoni by larger size, lack of stylar cusps B and D on M!, lack of stylar cusp B on M? and the minute size of St D on this tooth, the straight or almost straight centrocrista on Mt and M2, ante- rior cingulum of M! has no notch for placement of preceding premolar, the anterior roatof M! lies directly under the cingulum, the anterior width of the upper molar crowns are less than that of the buccal lengths, wider angle of crests at the paracone and metacone, extreme reduction of the talonid basin and protocone, particularly on M! with concurrent loss of metaconules on this tooth, extreme reduction in size of the metaconid, ab- sence of entoconid, reduced talonid basin by the more lingual position of the hypoconid and tack of diasiemata between P1 and P2, Species of Wabulacinus differ from all species of Thylacinus in the extreme reduction of the talon and protocone on Mt, the more parallel alignment of the preparacrista with the cen- trocnistaon M', a small metaconid (at least on the M3), less elongate snout by lack of diastemata between the premolars as well as between Pı and the canine. Wabulacinus ridei is similar in molar 368 size to T. macknessi, but lacks an anterior cingulum on M!. Wabulacinus ridei sp. nov. (Fig. 1) ETYMOLOGY. For David Ride for his long-term commitment to Australian vertebrate palaeonto- logy. MATERIAL. Holotype. QMF16851, right maxillary frag- ment containing M!-2 (Fig. 1 A-C). Paratype. QMF16852 left dentary fragment with broken M3 (Fig. 1D- F) from early Miocene (Sys- tem B) Camel Sputum Site, Godthelp Hill, Riversleigh. DIAGNOSIS. As for genus. DESCRIPTION. Maxilla partly preserved. Infraorbital foramen enclosed within the body of maxilla, above the pos- terior alveolus of P3, Buccal crown of M! length exceeds anterior width. Metacone largest cusp fol- lowed by paracone, protocone and St E. No other cusps. Postmetacrista longest crest, curving buccally at the poste- rior end. Preparacrista orien- tated almost parallel to the tooth row, terminating at the anterior tip of the crown. Pre- metacrista and postparacrista connecting as a straight cen- trocrista which parallels the preparacrista. Lacking pre- protocrista, postprotocrista, protoconule, metaconule, sty- lar shelf or stylar cusps anterior to St E. Buccal flank of crown forming continuous slope from paracone and metacone to low- est buccal edge of the crown. Protocone small. M? similar to M! except : St E minute. Stylar shelf region high, of many tiny indistinct cusps and crests, especially on the more posterior half of the crown. Postmetacrista longest crest on the crown, followed in declining length by preparacrista, postprotocrista, preprotocrista, postparacrista MEMOIRS OF THE QUEENSLAND MUSEUM FIG. 1. Wabulacinus ridei, A= QMF16851 ( M! and M2 2) lingual view. B= QMF16851 (M! and M?) buccal view with infraorbital foramen arrowed. C and C! = QMF16851 stereo occlusal views. D = QMF16852 (P2 and M3) buccal view. E = QMF16852 (M3) lingual view showing small metaconid (arrowed). F and F’ = QMF16852 stereo occlusal view. and premetacrista. Postparacrista and pre- metacrista forming a wide angled centrocrista. Postmetacrista leaving metacone almost parallel to the premetacrista, curving buccally. Pre- paracrista straight, connecting to the postpara- TWO NEW EARLY MIOCENE THYLACINES FROM RIVERSLEIGH crista at approximately 90° and oblique jo the tooth row, No $1 B present. Trigon basin wider than on M!, Lingual flank of trigon basin *V’- shaped with a distinct ridge running vertically down its centre. The preprotocrista and postprotocrista prominent with a minute pro- toconule and metaconule. Ectoflexus on the buc- cal surface of this tooth slightly developed. Anterior cingulum terminating anterior to base of the paracone, lacking a notch. Mental foramen under the anterior root of Pa. Alveoli for Pia, Mj-2 and the anterior root of My, M3 only molar present. Symphysis beginning ad- jucent to the anterior root of P3. No diastemata between alveoli, All alveoli pairs orientated par- allel to the tooth row except P; oblique, indicating some crowding of Pi against the canine. Alveoli size indicating relative lengths of Pa> Po> Py, Ma=Mz> Mj, M; with cusps in decreasing height paraconid, metaconid, hypoconulid, hypoconid. All cusps prominent except minute metaconid; entoconid and related crests absent. Protocristid longest crest, followed (in decreasing length) by the posthypocristid and cristid obliqua. Remains of the metacristid connect to the small metaconid. Small talonid basin open on the lingual side. Hypoconid almost medial to the trigonid basin. Posthypocristid and crisud obliqua orentated oblique to the dentary, mecting at the hypoconid al right angles, Antenor cingulum continuing buecally past the anterobuccal corner of tooth, with wide notch. Posterior cingulum poorly de- veloped, a small bulge in the enamel, Ngamalacinus gen, nov. TYPE SPECIES. Ngamalacinus tinmulvaneyi et sp. mov. ETYMOLOGY. Wanyii Ngamala, died out; Greek kvitos, dog, Masculine. DIAGNOSIS. Moderately specialised among thylacinids in the reduced conules, reduced stylar shelf, anteroposteriorly elongated molars, Re- taining small St B and D, metaconid, entoconid und hypoconulid. COMPARISON, Neamalacinus differs. from N, dicksoni in ils larger size, reduced metaconules and protoconules, reduction of St D particularly on M2, Ngamalacinus differs trom W. ridei and Thy- favinus in its: smaller size; narrower angle of crests al the paracone, metacone and protocone; 309 narrower angle of centrocrista; less reduced sty lar shell with retention of prominent St B, St D and stylar shelf crests on M! and M7; less reduced talon basin, particularly on M!; less anteroposter- ior elongation of the molars and associated crest lengths; larger talonids: and larger metaconid (larger than the paraconid) and with a distinct metacrisuid. Ngamalacitus Yurther differs from W- ridet in the more posterior position of the infraorbital foramen, presence of an entoconid on the lower molars and smaller hypoconulid. Ngamalacinus timmulyaneyi sp. nov. (Figs 2, 3) ETYMOLOGY. For Tim Mulvaney, a longtime sup- porter of research at Riversleigh: MATERIAL. Holotype QMF 16853 right dentary with Mi-4 (Fig, 2) from early Miocene (System B) Inahey- ance Site, Godthelp Hill, Riversleigh. Paratypes QMF30300 left maxillary with P2-M4 (Fig. 3A-C), from early Miocene (System B) Camel Sputum Site, Godthelp Hill. Referred specimen QMF 16855, right M? (Fig, 319), from the type locality, DIAGNOSIS. As for genus. DESCRIPTION. All articulating surfaces of dentary broken. Coronoid process rising from the ramus at approximately 120°. All four molars ani the posterior alveolus of P? present. No diaste- mata between these teeth. Degree of eruption of Mg indicating a juvenile. Protoconid of Mı tallest cusp, followed (in de- creasing height) by metaconid, paraconid, hypoconid, hypoconulid and entoconid, All cusps distinct, with crests. Paracristid longest crest on crown followed (in decreasing length) by posthypocristid, metacristid, cristid obliqua, pre-, postentocristid, Anterior cingulum with a very small notch. Paracristid almost straight with a very wide angle connecting the paraconid and pratoconid. Talonid basin entirely enclosed by crests, large and deeply concave to central point. Hypoconid more buccally positioned than pro- tocomid. Cristid obliqua continuing up the poste- rior wall of the protoconid. Posterior cingulum distinct, uniform in thickness to the base of the crown, with a slight notch formed between it and the hypoconulid Ma same as Mi except: Metaconid mit A large, distinct. taller than the paraconid. All crisiids higher and distinct, Anterior cingulum broader and the notch more distinct. Angle al crests on the protoconid approximately [OU-LIW. 370 Paracristid and metacristid longer. Metacristid straight; paracrisud changing orientation at the valley between the paraconid and protoconid. M3 same as M2 except: the anterior half of the crown thicker than the posterior because of the more lingual position of the paraconid and metaconid. Paracristid and metacristid elongated, Hypoconulid and entoconid slightly more to pos- terior, with entoconid slightly smaller than on Ma, Posthypocristid bending posteriorly to con- nect to the posteriorly positioned hypoconulid, Paracristid proportionally longer than on Ma. Ma same as Ma except: Talonid basin reduced, well defined and enclosed by crests. Entoconid minute; hypoconid small; hypoconulid highest cusp on talonid. Small posterior cingulum pres- ent. No obvious sutural boundaries on the maxilla excepta posterior suture that may have connected to either the jugal or the lachrymal. Maxilla indi- cating that the canine was large, its root extending deeply into the maxilla, Infraorbital foramen above M2. The region immediately posterior to the infraorbital foramen damaged but a depres- sion in the maxilla here and sutural boundaries of the jugal indicate that the jugal 1s likely to have contacted the external opening of the infraorbital canal. Maxilla with large extension projecting back towards and contributing to the zygomatic arch. No maxillary palatal vacuities in the region of the premolars, Small diastemata between the upper premolars. P? triangular in lateral view with both an anterior and posterior cusp, with a crest from the major central cusp to the posterior cusp and a less well defined crest anteriorly to the smaller anterior cusp, with posterior region wider than the ante- rior, with ridges extending along both sides (lin- gual and buccal) of the pasterior cusp. P? larger than P* and similar except for: anterior and pos- terior cusps relatively larger, anterior cusp with ridges off the lingual and buccal sides, posterior crest from the major cusp more prominent but not straight, posterior half of tooth relatively wider with enlarged crests bordering the posterolingual and posterobuccal edges of the crown, with an additional posterobuccal cusp, M! damaged, with anterior cingulum, a large St D larger than the distinct St B, a stylar crest running posteriorly from St D to the metastylar comer, talon broad with a possible protoconule, postmetacrista long and straight, crests at the paracone at approximately 90°, preparacrista connecting to St B, almost perpendicular to the tooth row. MEMOIRS OF THE QUEENSLAND MUSEUM FIG. 2. Ngamalacinus timmulvaneyi lower dentition. A = QMF16853 (dentary with M1-5) lingual view. B = QMF16853 buccal view. C and C! = QMF16853 M? same as M! with: in occlusal view posterolingual dimension longest followed by buccal length and anterior width, Anterior cingu- lum not notched in QMF16855 but is in QMF30300, cingulum terminating at the anterior face of the base of the paracone without connect- ing to the talon basin. No posterior cingulum. Metacone highest cusp on the crown, followed (in TWO NEW EARLY MIOCENE THYLACINES FROM RIVERSLEIGH 371 decreasing height) by: paracone, St B, metastylar cusp(s) and protocone. Postmetacrista longest crest on the crown, followed by the preparacrista, premetacrista, postprotocrista, pre- protocrista and postparacrista. All crests relatively straight. Enamel sur- face slightly raised about the pro- toconule. Metaconule not present as a distinct cusp. Slightly raised postprotocrista connecting the pro- tocone to the metacone base where a sharp crest runs up the lingual surface of the metacone. A less distinct ridge running down the lingual side of the paracone and protocone. Slight ectoflexus at the buccal side of the crown due to bulging of enamel around St B. St E a raised part of the stylar crest. Between St E and B are minute cusps on QMF16855 but St D is more distinct on QMF30300. One crest connecting the metastylar cusp(s) to the posterolingual corner of the crown. Talon basin large with a broad, flat base. Preprotocrista and postprotocrista relatively low. Cen- trocrista at approximately 100°. M? same as M? except: Ectoflexus stronger and all stylar cusps reduced to cuspules. Stylar crest not continuous along the lingual edge of the crown. St B largest stylar cusp. Anterior cingu- lum with less distinct notch than in M? of QMF30300. Preparacrista and postmetacrista longer; paracone rela- tively larger but smaller than metacone. Paracone more lingually lo- cated. Centrocrista at approximately 90°; postparacrista strongly curved. Talon narrower. Protoconule and metaconule with ridges connecting to the lingual face of the paracone and metacone respectively. THYLACINID PHYLOGENY Ngamalacinus timmulvaneyi and W. ridei do not share any apomorphies that are not also found in Thylacinus. FIG. 3. Ngamalacinus timmulvaneyi wees dentition. A = QMF30300 maxillary fragment with P?-M? and showing infraorbi- tal foramen (arrowed). B and B!=QMF30300 stereo occlusal view of P? and molars. C = QMF30300 lingual view. D and D! QMF16855 (M2 ) stereo occlusal views. These, therefore, cannot be considered to repre- Plesiomorphic N. dicksoni and apomorphic Thy- sent members of the same genus. lacinus but do not form an independent dichot- Wabulacinus ridei and Ng. timmulvaneyi have omy (Fig. 4). Neither species can be placed in a combinations of features that place them between known genus because: neither shares any 372 TABLE 1. Characters and states among thylacines. L Infraorbital foramen: 0. not bound by jugal. 1. bound by jugal, 2 + iain 1, angled. 2. straight (as indicated by t "3 Preparacrstton M", 1, angled almost perpendicular wo the woth row axis. 2. oer aae te state 1, 3. straight. 4. Angle of crests at paracone and metacone. |. wider than on plesiomorphic dasyurids, 2, further widened. 5. Entoconid. 1. small. 2. minute. a. either absent or pompony positioned and combined with the hypeconulid. 6. Hypoconulid size. 0. large. 1, reduced. 2. minute, 7. Stylar shelf size. 1, crests and cusps present but reduced compared to plesi hic dasyurids. 2. re- duction in size of some cusps and crests,3, further loss of cusps and cresis (mostly absenton M ). 4, complete loss on crests, only a single small cusp present on the posterior of the crown. - vAmýriorcinguhini, 0, completeon M”. 1 incomplete i rana and protoconule, 0. present and large. T. present and reduced. 2. further reduced or absent. 10. Metaconid size. 1. reduced compared to plesiomorphic dasyurids. 2. small. 3. absent but reten- sion of crest arrangement in postenor molars, 4, com- plete absence of cusp and associated crests. 11. Talonid basin size, 0. untreduced. 1, reduced by lingual placement of hypocondd. 2. further reduction. 12. Talon size. 1. reduction of talon width compared to plesiomorphic dasyurids with associated lengthening. 2, loss of metaconid and further widening of the crests 13. Diastemata and size of M4. 0. no diastemata in premòla region, M4shorter ihan M3, 1. diastemataand i4 equal in length to M3. 2. diastemata and Ma is longer than M3. apomorphy with N. dicksoni that is not also shared with Thylacinus, to include cither in Thy- lacinus would expand it beyond any other dasyuromorphian genus. Wabulacinus ridei shows character conflict in the plesiomorphic nature of the infraorbital foramen which is more plesiomorphic than in N. dicksont and Ng. timmulyaneyi, This character may have under- gone reversal in W. ridet, The single most parsimonious tree of thylacinid relationships was found using an Exhaustive Search PAUP 3.1 (Swofford, 1993) with 13 or- dered characters (Tables 1 & 2) using pleisomorphic dasyurids as the outgroup. Each taxon represents the sister species of all thyla- cines immediately to its right. In general, the MEMOIRS OF THE QUEENSLAND MUSEUM TABLE 2. Character state distribution among thyla- cimes. (a = either 0 or autapomorphic combination of entoconid and hypoconulid, ? = unknown slate). 00000 00000 000 t1111 11001 010 41111 12011010 Dasyurids Nunbacinus dicksoni 0232a 03112 12? 22222 24023 221 12222 24024 222 12222 24124 222 oar specialised carnivores are located on the right. Wabulacinus ridei and Ng. tinued vaneyi arc more plesiomorphic than Thylacinus in the larger size of the metaconid {small on W. ridet and much larger on Ng. timmulvaneyi) and the lack of ex- pansion of the premaxillary region. Both species (and particularly W. ride) are more specialised than N. dicksord in the reduction of the stylar shelf and the metaconule and protoconule, talon basin and degree of ectoflexus on MÊ., oepa tsa RIDEI. oe ve are more a c than in N. dicksoni are syn- on with Thylacinus are: the straight cen- trocrista; the widened angle of the preparacrista relative to the postparacnsta, particularly on the M! where this crest is parallel with the an- teropostenor dimension of the tooth; an increase in the size of the angle formed by crests at the paracone and metacone, thereby increasing over- all tooth length; further reduction in size of the stylar cusps than that seen in either Ng, yi and N. dicksoni; reduction in size of the entoconid; reduction of the metaconid to a minute cusp; reduction in size of the talonid basin by the more lingual position of the hypoconid; and reduction in size of the talon basin. Wabulacinus ridei exhibits some au- tapomorphies not seen in any other thylacinid, same of which are considered specialisations be- yond that of T. cynocephalus. The preparacrista on M! of W, ride! is parallel with the tooth row and the centrocnsta. The pre tae on N. dicksoni and Ng. timmulvaneyi show the plesiomorphic state similar to most dasyurids in which itlies almost perpendicular to the tooth row and forms almost a 90° angle with respect to the postparacrista. The morphocline otherwise shown in thylacines from N. dicksoni through to T. cynocephalus is a widening of the angle at which these crests meet (Fig, 5). This elongates the tooth in an anteroposterior direction and pro- TWO NEW EARLY MIOCENE THYLACINES FROM RIVERSLEIGH Pad C plesiomorphic state E@ Ist intermediate state W 2st intermediate state W 3rd intermediate state BB most apomorphic state Uncertainty, see text FIG. 4. Cladogram of thylacines showing character state changes. Cladogram is the single most parsimo- nious tree of 32 steps (CI = 0.906, HI = 0,094, RI = 0.917, RC = 0.831). Striped box = unknown state of either plesiomorphic or highly derived. For characters and their distribution see Tables 1 & 2. duces. an enlarged longitudinal blade formed from the postmetacrista, centrocrista and pre- paracrista, Only on M! of W. ridei does the pre- paracrista lie parallel to the tooth row, a condition more derived than that in any other thylacine. The talon basin on the M! of W. rides is also more derived in its degree of reduction than that of T. 373 macknesst but is similar to the condition in T. cynocephalus. The anterior cingulum on M! of W. ridei is reduced compared to that of N. dicksoni (it is unknown in Ng. timmulvaneyi). In W. ridei it is incomplete while in N. dicksoni it continues lin- gually to join the talon basin. This feature is more plesiomorphic than in T, cynocephalus where the cingulum is lost, but more specialised than in T. macknessi where a complete cingulum is re- tained. In addition, the anterior portion of M! of W. ridei is unique in that the anterior root lies much further forward under the crown than in other thylacines. Another trend in thylacines is for the entoconid to become reduced. Only in W. ridei is this cusp completely lost. Wabulacinus ridei is autapomorphic within the family in having an enlarged hypoconulid. In other thylacines the hypoconulid shows reduction (e.g., in N. dicksoni, T. cynocephalus) and may also move posteriorly (e.g.,in T. macknessi), This enlarged cusp in W. ridei may be compensate for loss of the entoconid, or alternatively, it may represent a combination of the hypoconulid and a more posteriorly placed entoconid, A feature previously used to distinguish thyla- cines from dasyurids is the posterior position of the infraorbital foramen posteriorly delimited by the jugal (Muirhead & Archer, 1990). Itis known in T, cynocephalus, T. potens, Ng. timmulvaneyi (Fig: 3A) and N. dicksoni. Wabulacinus ridei has the infraorbital foramen anterior to M’ and well distant from the jugal (Fig. 1B). This position is similar to dasyurids in which it most frequently occurs above M'/M2 (e.g,, in Dasyurus and An- techinus). The anterior position of this foramen in these dasyurids indicates that posterior place- ment near the jugal in most thylacines is apomorphic (Archer, 1976). The anterior place- ment of the jugal in W. ridei is therefore plesiomorphic relative to other thylacinids. Wabulacinus ridei is plesiomorphic in many respects to Thylacinus excluding it from Thy- lacinus. W. ridei has a number of features unique among thylacinids placing it outside the range of variation within Thylacinus. NGAMALACINUS TIMMULVANEY!. This spe- cies. Shares with W. ridei and Thylacinus the apomorphic reduction in the stylar shelf com- pared to N. dicksoni (Fig. 4), This includes reduc- tion in size of St D of M2, On QMF16855, St D is further reduced and replaced by a number of minute cusps that border the stylar shelf On other 374 molars, size nf the stylar shelf is comparable to that in N. dicksoni. The protoconules and metaconules of W. ridei are apomorphically reduced compared to those of N. dicksoni. The talon basin is also slightly more reduced than that of N, dickvoni. This species further differs from N. cicksont iw the less ex- treme ectoflexus, an apomorphic feature. These specialisations of Ng, timmulvaneyi compared to N. dicksoni are less marked than the degree of specialisation of these same features in W. ridei and Thylacinus. Ngamalacinus timmulvaneyt and N, dicksoni share several plesiomorphies and, in terms of overall similarity, Ng. tinmalvaneyi is much closer 10 N. dicksoni than to any other thylavinid (Fig. 4). These two species do mot share any apomorphy act also found in other thyla- eines. PALAEDECOLOGY OF RIVERSLEIGH THYLACINIDS Thylacinids described from the Riversleigh as- semblages are N, diekwani, Ne- timmulvaneyi, W. ridegi and T. macknessi. This diversity raises qués- tions about niche diversification. Allhough only one thylacine appears to haye been present at any one time in late Miocene (Alcoota; T, potens), Pliocene (Awe & Chinchilla, T. cynocephalus) and Quaternary (many assemblages, T- cyn- ocephalus) local faunas of Australia and New Guinea (Archer, 1982; Dawson, 1982), prior to the late Miocene, available resources enabled the ‘thylacine niche” to be more finely divided. Part of the explanation may be found in the apparent absence from the Riversleigh local faunas of any large dasyurids as specialised far carnivory as the late Cainozoic species of Glancoden, Sarce- philns and Dasyurus. Presence of large carnivo- rous dasyurids appears inversely correlated with thylacinid diversity, The subsequent rise of these dasyurines may, therefore, have accompanied late Miocene decline in thylacinid diversity. Although there is a greater diversity of thyla- cines in the Oligo-Miocene Riversleigh deposits than later, a wider range of large carnivores was also present in these Riversleigh local faunas. For example in single local faunas, [here were often 3 crocudilians (P. Willis, pers, comm), at least 2 large snakes (madisoiids and pythonids; J- Scanlon, pers. comm.), 2 Jineages of thy- lacaleamnds (Wekales and 4 genus similar to Priscileo: Archer et al., 1989), a possibly carniv- uwus kangaroo (Archer & Flannery, 1983; Wroe & Archer, 1995; Wroc, 1996) and an unknown MEMOIRS OF THE QUEENSLAND MUSEUM number Of raptorial birds (Bales, pers. comin; Archer et al., 1994). Hence it is probable that the relutively high diversity of Riversleigh thylacines reflects an overall higher biotic diversily in the rainforests of the Riversleigh region. Camel Sputum is the only Riversleigh site to have produced more than one thylacine: Ng. timmulvaneyi and W. ridei, These taxa are very similar in size. The maxilla of Ne. timmulvaneyi recovered from Camel Sputum Site differs from the maxilla of W. ridei in the position of the infraorbital foramen (in Ng. timmulvaneyt it typ- ically lies above M? and was probably bounded by the jugal while in W. ride/ it lies anterior ta M!) and the more plesiomorphic structure of the molars in Ng. timmulvaney/, These differences cannot be accounted for by intraspecific variation and the specimens clearly represent two dilferent species. It iš not clear how many of Riversleigh's thyla- cines co-existed. While 2 are presentin theCamel Sputum assemblage, the more generalised N. dicksoni may have been present throughout the Oligo-Miocene (Systems A to C; Muirhead & Archer, 1990). Tylacinus macknessi in Systems B and C (Muirhead, 1992) suggests that by the early to middle Miocene, all 4 genera co-existed al Riversleigh, By late Miocene Alcoota time, one lineage is known: T. potens. In late Cainozoic deposits from other areas of Australia (i.e. cave assemblages in eastern, southern and western Australia; Ride, 1964; Archer, 1974, 1982; Dawson, 1982), thylacinid remains are common. Sites where thylacinid re- mains are abundant (e,g., Thylacine Hole on the Western Australian Nullarbor; Lowry, 1972) may be interpreted to represent lairs or traps where carnivores were preferentially attracted, perhaps by the presence of other animals. 1n the Riversleigh deposits, most of which appear to have accumulated in shallow pools within rainforest environments (Archer et al., 1989; Archer et âl., 1994), thylacinid remains are rela- tively rare and therefore may more fairly repre- sent natural frequencies, THYLACINID DIAGNOSIS AND MORPHOLOGICAL TRENDS Thylacinids differ from dasyurids and other polyprouxtony marsupials hy having, in combina- tion, the following features, The premetacrista and postparacnsia join as a centrocrista, The angle fonned by these crests is straight or almost straight in ceclusal view in at Teast M? of the TWO NEW EARLY MIOCENE THYLACINES FROM RIVERSLEIGH > | PSs X © | SK el IDS ISS 375 ICO © + J ICAZA FIG. 5. Upper and lower dentitions of all known thylacine genera showing crest orientation compared to a dasyurid. A=Dasyurus. B=Nimbacinus dicksoni (P8695-92) C=Ngamalacinus timmulvaneyi. D=Wabulacinus ridei. E!=Thylacinus macknessi E=Thylacinus eynacephalus. Upper dentitions include P? where known. Scales =0,5mm. upper dentition. The cristid obliqua continues up the posterior flank of the protoconid from the talonid region rather than terminating at the base of the protoconid. This functions in elongating this crest and becomes more prominent as the metaconid is reduced (e.g., in Thylacinus). The stylar cusps are reduced. This occurs most prom- inently on M? but also occurs to varying degrees on more anterior molars. The size of the metaconid is reduced on all lower molars. This reduction is correlated with the more posterior placement of the metaconid relative to the pro- taconid, functioning in widening the angle of crests at the protoconid and enlarging the trigonid basin. Reduction of the metaconid is also found to progress in degree from the more reduced condition on anterior molars to posterior molars (Muirhead & Gillespie, 1995). The size of the talonid basin is reduced because of the more lingual position of the hypoconid, This cusp oc- cupies much of the surface of the talonid basin such that no flat surfaces occur on the basin floor. Structural morphoclines of the family (appar- ent in more specialised forms) include the follow- ing. There is an increase in the angles formed by crests of the paracone and metacone, increasing the length of the postmetacrista, More antero- posterior orientation of the preparacrista. The loss of extreme ectoflexus particularly in M? (related to the overall elongation of the teeth). A reduction in size of the protoconule and metaconule as well as the entire talon basin and reduction in size of the stylar shelf. AIL of these features of the upper dentition increase the anteroposterior length of the molars with the entire tooth row acting as a system of anteroposteriorly orientated blades (Fig. 5). These are typical specialisations in mam- malian carnivores. In the lower molars, the trends are for complete loss of the metaconid and opening of the trigonid 376 basin. Here. like the upper molars, the lower molar crests become orientated anteroposteriorly (Fig. 5). The paracristid becomes the anterior crest with the elongated cristid obliqua function- ing as the posterior crest (the postprotocrista) (Muirhead and Gillespie, 1995), The lingual side of the Lalonid is also reduced through reduction of the entoconid. Only in Thylacinus 1s the snout elongated by both diastemita between the canine and premo- lurs and clongation of M4 (such that it is longer than preceding molars). Extreme posterior place- ment Of the infraorbital foramen and partial en- closure by the jugal is also a possible synapomorphy of Thylacinus related to snout elongation. All thylacinids plesiomorphically retain the paraconid on Mj, remnants of posterior and ante- rior cingula on the lower molars and posterior increase in size from Pi to Ps. Variation among thylacinids that does not ap- pear to follow these “carnivorous trends” includes position of the infraorbital foramen which, plesiomorphically and unlike all other known (hylacinids, is more anteriorly positioned in W. ridei above P’, ACKNOWLEDGEMENTS The study was undertaken with support from the Queen Elizabeth It Silver Jubilee Trust For Young Australians und the Australian Common- wealth Department of Employment, Education and Trainimg. Material on which this study is based was due to support from: the Australian Research Grant Scheme; the University of New South Wales; the Natonal Estate Grants Scheme, Queensland; the Department of Arts, Sport, the Environment, Tourism and Territories; Wang Australia; IC] Australia; the Queensland Museum; the Austra- lian Museum; the University of New South Wales; the Australian Geographic Society; MIM; Ansett Wridgways; and Surrey Beatty and Sons. I am grateful for the advice and suggestions of referees Ken Aplin and Mike Archer and for the assistance piven by the Australian Museum (in particular Linda Gibson), the Northern Territory Museum (in particular Dr Peter Murray), Robyn Murphy, Anna Gillespie, Henk Godthelp (Uni- versity of New South Wales). MEMOIRS OF THE QUEENSLAND MUSEUM LITERATURE CITED ARCHER, M. 1974, New Information about the Qua- ternary distribution of the thylacine (Marsupialia, Thylacinidae) jn Australia. Royal Society of Westem Australia 57: 43-50. 1976. The dasyuric dentition and its relationship to that of didelphids, thylacinids, borhyaenids (Marsupicarnivora) and perametids (Peramelina: Marsupialia). Australian Journal of Zoology, Supplementary Series 39; 1-34. 1982. A review of Miocene thylacinids (Thy- lacinidae, Marsupialia), the phylogenetic posi- tion of the Thylacinidae and the problems ot apriorisms in character analysis. Pp 445-476. In Archer, M. (ed,), Carnivorous marsupials, (Royal Zoological Society of New South Wales; Syd- ney). ARCHER. M. & FLANNERY, T.F. 1985, Revision of the extinct gigantic rat Kangaroos (Potoroidae: Marsupialia), With description of a new Miocene genus and species, and a new Pleistocene species ol Propleepus, Journal of Paleontology 59: 1331- (349, ARCHER, M., HAND, S.J. & GODTHELP. H. 1994. Riversleigh. Second edition. (Reed Books: Syd- ney). ARCHER, M., GODTHELP, H., HAND, S.J. & MEGIRIAN, D. 1989, Fossil mammals of Riversleigh norhwestem Queensland; prelimi- nary overview of biostratigraphy, correlation and environmental change. The Australian Zoologist 25; 29-65, ARCHER, M., HAND, S.J. & GODTHELP, H, 1995, Teniary environmental and biotic change in Aus- traha, Pp 77-90, In Vrba, E.S., Denton, G,H., Partiridge, T. C. & Burekle, L.H, (eds), Paleocli- mate and evolution, with emphasis on human origins (Yale University Press: New Haven). BENSLEY, B.A. 1903, On the evolution of the Austra- lian Marsupialia with remarks on the relationships of marsupials in general. Transactions of the Linnean Society, London (Zoology) 9: 83-217. DAWSON, L. 1982. Taxonomie status of fossil thyla- cines (Thylacinus, Thylacinidae. Marsupialia) from late Quatemary deposits in eastern Australia. Pp 527-536, In Archer, M (el), Carnivorous marsupials, (Royal Zoological Society of New South Wales: Sydney). FLOWER, W.H, 1269, Remarks on the homologies and notation of the teeth in the Marsupialia Journal of Anatomy and Physiology 3: 262-278. HENNIG, W. 1965. Phylogenetic systematics. Annual Review of Entomology 10: 97-116. LOWRY, JWJ. 1977. The taxonomic status of small fossil thylacines (Marsupialia, Thylacinidae) from Westem Australia. Journal and Proceedings ofthe Royal Society of Westem Australia 55; 19-29, LUCKETT, W.P. 1993, An ontogenetic assessnient of dental homologies in therian mammals. Pp 182- 204. In Szalay, F.S., Novacek, M.J. & McKenna, TWO NEW EARLY MIOCENE THYLACINES FROM RIVERSLEIGH M.C. (eds), Mammal phylogeny. (Springer-Ver- lag, New York). MUIRHEAD, J. 1992. A specialised thylacinid, Thy- lacinus macknessi, (Marsupialia: Thylacinidae) from Miocene deposits of Riversleigh, northwest- em Queensland. Australian Mammalogy 15: 67- 76. MUIRHEAD, J. & ARCHER, M. 1990. Nimbacinus dicksoni, a plesiomorphic thylacinid (Marsupialia: Thylacinidae) from Tertiary depos- its of Queensland and the Northern Territory. Memoirs of the Queensland Museum 28: 203-21. MUIRHEAD, J. & GILLESPIE, A. K. 1995, Additional parts of the type specimen of Thylacinus macknessi (Marsupialia: Thylacinidae) from Miocene deposits of Riversleigh, Northwestern Queensland, Australian Mammalogy 18: 55-60, RIDE, W.D.L. 1964. A review of Australian fossil marsupials. Proceedings of the Royal Society of Western Australia 47; 97-131, SWOFFORD, D.L. 1993. PAUP (phylogenetic analysis using parsimony). Documentation for Version 3.1. (Illinois Natural History Survey: Campaign). WOODBURNE, M.O, 1967. The Alcoota Fauna, Cen- tral Australia: an integrated palaeontological and geological study. Bulletin, Bureau of Mineral Re- sorah Geology and Geophysics, Australia 87: 1-187. WROE, S. 1996, An investigation of phylogeny in the giant extinct tat kangaroo Ekaltadeta (Pro- pleopinae, Potoroidae, Marsupialia).Journal of Paleontolagy 70: 681-690. WROE, S. & ARCHER, M. 1995. Extraordinary ` diphyodonty-related change in dental function for a tooth of the extinct marsupial Ekaltadeta ima (Propleopinae, Hypsiprymnodontidae). Archives of Oral Biology 40; 597-603. 377 APPENDIX All measurements are actual distance between cusps except those with ‘(horiz)’ for which mea- surements were made from a horizontal plane above the cusps (occlusal view). Wabulacinus ridei dentition (mm) es SELIM OMF16852 M Meta-proto |420 | 5.01 | -| Anterior/width 5. t co meta-proto (horiz z 2. hypo-hypoconulid oa aj |e = al JS g | i} | | | gamalacinus timmulvaneyi upper dentition (mm) cai OMF 16855 MŽ M? | Para-meta | | [3.64] 3.66 [3:34] 3.03 | Metaproto | | [286| 3.36 [3.65] 5.00 | Proro-para | | [3.77] 5.14 [5.02] 5.52 | | Anterior/width __|2.25|4.26|4.90] 7.68 |7.20| 8.25 | | Buccalength —_|5.56]7.92[8.11] 9.10 [8.76] 8.17 | |Posterolingu/uppers| | _ [8.15] 10.29]9.30]10.27/ para-meta (horiz | [330| 3.42 [3.31] 2.80 | | meta-proio(horiz) | | [2.28] 3.05 [2.84] 3.04 | | proto-para(horiz) | | [3.34] 2 37 para-mela (hon | 3.98 | | meta-proto (horiz) | - | 251 | 264 | 2.43 |_proto-para (horiz) | - | 3.88 | 3.89 | 4.03 | ypo-hypoco | hypeno | 076 | 108 | 113 | 0.90 | KUTERINTJA NGAMA (MARSUPIALA, ILARIIDAE): A REVISED SYSTEMATIC ANALYSIS BASED ON MATERIAL FROM THE LATE OLIGOCENE OF RIVERSLEIGH, NORTHWESTERN QUEENSLAND T.J. MYERS AND M. ARCHER Myers, T.J. & Archer, M, 1997:06:30. Kulerintja ngama (Marsupialia, Uariidae); a revised systematic analysis based on material from the late Oligocene of Riversleigh, northwestem Queensland. Memoirs of the Queensland Museum 41(2); 379-392. Brishane. ISSN 0079- 8835. The Riversleigh ilariids come from the late Oligocene White Hunter Site and are Kulerintja ngama Pledge, 1987, Molar cusp morphologies are compared with those of other ilariids and vombatiforms and several morphoclines identified. The range of variation is similar to that in Phascolarctos cinereus, Cladistic analysis suggests several hypotheses about intrafamilial relationships: 1) Au. ngama is an ilariid; 2) Kovbor is not an ilariid; and 3) ilariids form a monophyletic clade with the wynyardiids, although the relationships of these taxa Lo other yombatomorphians are not resolved. Clariidae, Oligocene, Vombatiformes, White Hunter Sile, Riversleigh. T.J. Myers & M. Archer, School of Biological Sciences, University af New South Wales, Sydney, New South Wales 2052, Australia; received 10 November 1996, llartids are extinct marsupials discovered in late Oligocene deposits of central Australia. Haria includes I. illumidens and 1. lawsoni (Tedford & Woodburne, 1987); Kuterintia contains Ku. ngama (Pledge, 1987). There is also some contro- versy concerning the placement of Koobor within the Phascolarctidae because Pledge (1987) sug- gested that Ku. ngama may have been ancestral tọ Ko, jimbarratti, making the latter a potential ilariid, Tedford & Woodburne (1987) found sim- iJarities in upper dentition between J. i/lumidens and Keobor, namely; a paraconule on M!, no paraconule or neometaconule on M? or M$, but considered them symplesiomorphic, concluding that Koobor shared more synapomorphies with phascolarctids than with ilariids. Wereview Kuterintja ngama based on material from the White Hunter Local Fauna at Riversleigh, NW Queensland. White Hunter Site on Hal’s Hill, on the D-site Plateau (Archer et al., 1994; Creaser, 1997) was questionably assigned to early Miocene System B (Archer ct al., 1989, 1994) but the fauna now suggests late Oligocene System A. A tentative correlation is made of White Hunter Local Fauna with the Ngama Local Fauna of the Etadunna Formation at Lake Pal- ankarinna, South Australia. Pledge (1987) observed that Kuterintja ngama differs from /laria in being smaller, having larger cusps, pre- and postcristae on the stylar cusps, postprotocrista and premetaconulecrista sepa- rated by a crevice, and an anterior cingulum di- vided by a stronger preprotocrista. Similarities to F illumidens include a selenodont structure and well-developed buccal stylar cusps. Pledge (1987) described the holotype (SAM P24539) of Ku. ngama as a LM*. However, material from Riversleigh suggests that the holotype is a LM*. SYSTEMATICS Material is deposited in the South Australian Museum (SAMP), and the Queensland Museum (QMF), Homology of molars and the dP3 follows Luckett (1993). Homology of the other premolars follows Flower (1867). Cusp homology follows Archer (1984), Tedford & Woodburne (1987) and Pledge (1987). Order DIPROTODONTIA Owen, 1866 Suborder VOMA tra Woodburne, 1984 Infraorder VOMBATOMORPHIA Aplin & Archer, 1987 Family ILARIDAE Tedford & Woodburne, 1987 Kuterintja Pledge, 1987 TYPE SPECIES. Kuterinija ngama Pledge, 1987 from late Oligocene Etadunna Formation at Lake Pal- ankarinna, northern South Australia, DIAGNOSIS. Relative to Ilaria: Small, lack- ing transverse linking crests on the cheek teeth. h with low, almost horizontal inclination, dorsally flattened, transversely compressed and with an- terior portion inflected. P3 subrectangular, with | large anterior cuspid and two smaller posterior cuspids only slightly 380 transverse valley conga) cbapds entoconid bypoconid anten@l emaila FIG, 1. Kurerintja ngama, QMF20810, 23306, left dentary (P3 - M4). separated, one in a posterolingual position, other in a posteromedial position longitudinally aligned with the anterior cuspid, Mj suboyate, with anterior cingulum medially inflected and less developed, with lingual faces of the buccal cuspids near vertical, with less developed pre- protocristid and posthypocristid, with pre- protocristid and posthypocristid terminating in line with the ‘central’ cuspids, with small lingual basin on the hypolophid; ‘central’ cuspid on the protolophid in transverse alignment with the lin- gual and buccal cuspids on M)-M3; metaconid separated slightly from the ‘central’ cuspid of the protolophid; M2 with ‘central’ cuspid on the pro- MEMOIRS OF THE QUEENSLAND MUSEUM tolophid and hypolophid of similar widths and more closely linked, with ‘central’ cuspid on the hypolophid not linked posteriorly to the en- toconid: M2 and My with preprotocristid and posthypoeristid not extending as far lingually; Ma with lingual basins less developed, with ‘central’ cuspid on the hypolophid greatly reduced, with posterior cingulum relatively small, with postprotocnstid and prehypocristid not blocking transverse valley; M4 with compressed posterior, with ‘central’ cuspid not distinguishable on the hypolophid, with the postprotocristid and pre- hypocristid poorly developed (Fig. 1). P? subovate, much wider both anteriorly and posteriorly compared to the P3, with narrow an- terior portion, with large cusps, tri- cusped, lack- ing the posterobuccal cusp, with cusps subequal in height, with twinned central cusps separated by a larger trough, with a larger crevice separating posteromedial and posterolingual cusps, with an- terolingual cingulum, with well-developed rib running from the apex of posleramedial cusp to the posterabuecal edge of the posterior cingulum, M! with nearly vertical buccal surfaces on cusps. M! with stylar cusp C almost as large as the paracone, with stylar cusp D as large as the metacone, with buccal border slanting sharply posterobuceally, with posterior cingulum around convex structure, with all cusps subequal in height; M* with stylar cusp C relatively small, with the cristae forming the borders of the buccal basin on the paracone strongly developed, with the postparacrista separated from stylar cusp C, with stylar cusp E greatly reduced, with pre- prolocrista strongly developed and dividing the anterior cingulum; M? and M? with lingual cusps transversely aligned with the buccal cusps; M=4 with the anterior portion of the tooth larger than the posterior; M? with stylarcusps B and C equiv- alent in height to the paracone, with the buccal basin on the paracone enclosed at its buccal mar- gin, with stylar cusp D larger, M3 and M? without stylar cusp E; M? with the lingual half of the transverse valley inflected less towards the posterolingual corner, with stylar cusp C vari- able, with metaconule variable in position, and thus the lingual basin variable in size; (Fig. 4). COMPARISON: Kurerintja ngama differs from phascolarctids in lacking a paraconule and neametaconule, having longer molars, simpler selenes, separation of buccal selenes, better de- veloped stylar cusps, a strongly developed trans- verse valley. poorly developed postprotocrista and premetaconulecrista, a protocone that is more KUTERINTJA NGAMA FROM RIVERSLEIGH 38] 10mm 10mm FIG. 2. Kulerintjangama. A, QMF30057, RM. oeclu- sal view, stereo pair. B, QMF31299, RM*4, occlusal view, stereo pair, C, QMF31301, RI- Mo, buccal view. compressed relative to the metaconule on M!- M3, significant separation of stylar cusps C and D, no protostylid, a lingually convex metaconid, a protoconid that is larger than the metaconid, lingual cusps thal are not compressed towards each other, larger crown height, a well-develaped posterolingual cusp on P?, no posterobuccal cus- pid on P3, central cuspid, having dposterolingual cuspid on P3, a non-bladed P3 or P?, no longitu- dinal valley, and a bulbous PS, Ku. ngäma is distinguished from Koabar by its larger stylar cusps, higher crown, larger molars, and continuous crest between protocone and aan (Pledge, 1987). father di ferences in- clude: 1) more conical stylar cusps, 2) lower selene singles on the buccal basins of ihe upper molars; 3} Keober lacks a lingual basin on the iransverse valley; 4) Koobor has a poorly devel- oped anterior cingulum; 5) the absence of a pos- terior depression on the metaconule, as exists on most ilariid molars; 6) Keeber has molars which are slightly compressed lingually; 7) a much wider and longer longitudinal valley exists in Koober; 8) more poorly developed postproto- crista and premetaconulecrista; 9) a protocone that is compressed longitudinally relative to the metaconule on M!-M?; 10) no paraconule on M!; 11) Kovbor lacks the posterolingual cusp on P3; and 12) Koobar has an elongated, rather than bulbous, P?. Kuterintja ngama Pledge, 1987 (Figs 1-5, 7) MATERIAL. Holotype SAMP24539, LM3, presumed to be a left Mt by Pledge (1987) from the saddle between Mammalon Hill ‘and main escarpment, NW corner of Luke Palankarinna, 100km N of Marree, South Australia in the late Oligocene (Woodbume et al., 1993) Ngama Local Fauna within the Etadunna Formation. Other material, QMF31302, aright dentary fragment containing Pa, Mi and M2; QMF23306, QMF20810), a lefi déntary with all cheek teeth and the alveoli for 11; QMF31301, anterior portion of a juvenile nghi dentary, with 11, dP3, and M 1. P3 is removed from its crypt, and M2 has only part of the protoconid remaining; QMF17527, RM3 with roots missing; QMF31300 RM4 with the anterior portion of the irigonid missing; QMF30057, RM!; QMF23203, LM! with a broken anterior cingulum; QMF30058, RM? QMF31299, right maxillary fragmente containing M4: and QMF24604, right maxillary fragment with M3 and M4, and alveoli for M! and M2; QMF30332, partial right maxilla, with partial palate, anterior zygomatic arch, P3, MŽ and the alveoli for M!, All except type from late Oligocene White Hunter Site, Riversleigh, NW Queensland; previously regarded as possibly Sys- tem A or carly System B (Archer et al., 1989; Archer et al. 1994), tate Oligocene or early Miocene. This species Suggests comparable age to the South Austra- lian type locality. DIAGNOSIS. As for genus. DESCRIPTION. Dentary, Deepest below the posterior half of M3. In lateral aspect alveolus lor lı inclined slightly on its ventral side, horizontal on dorsal side. Mental foramen at the posterior end of this alveolus in the dorsoventral midline, and just anterior to P3, only foramen on the den- tary (break dorsoventrally from the junction of M2 and M3 may obscure others). I. Lower first incisor projecting horizontally from the dentary, curving lingually at its anterior (distal) extremity, subeylindrical, transversely compressed, With dorsal surface transversely flat- 382 FIG. 3. Kuterintjangama. A, QMF20810, left dentary, occlusal view, stereo pair. B, QMF23306, buccal view. tened, with enamel from the buccal to ventral surfaces. dP3. Same as P3 except in size, tricuspid, sub- triangular. Anterior cuspid tallest, with widest base. Smaller cuspid posterolingually from anterior cuspid. Third cuspid posterobuccally from ante- rior cuspid and equivalent in size to the posterolingual cuspid. All cuspids closely linked, conical, with wide bases. P3 (Fig. 1). Transversely compressed, tricuspid. Large, subovate, conical cuspid anteriorly larger, taller and more broadly based than the twin cus- pids posteriorly. More buccal of these an- teroposteriorly aligned with the large, anterior MEMOIRS OF THE QUEENSLAND MUSEUM cuspid. Posterolingual cuspid taller than its worn buccal counterpart. Thin, low cristid running from the an- terolingual corner of the apex of the anterior cuspid, anterolingually to the base of the cuspid, then turning posterolingually and running further down towards the root, then turning posterobucc- ally up the cuspid and terminating about half way up the height of the cuspid, in line with the posterior side of the apex of the large cuspid. Anterior cuspid located over the posterior portion of the anterior root; posterior cuspids located directly over the posterior root. A minor crevice on the anterolingual corner of TABLE 1. Measurements (mm) of dentition of Kuterintjangama. Le=length; Mw=maximum width; Ha=height of anterior cuspid; Aw=anterior width; Pw = posterior width; Hp=height of paracone; Hpr=height of protocone; Hm=height of metacone; Hml=height of metaconule; Hprd= height of pro- toconid; Hmd=height of metaconid; He=height of entoconid; Hh=height of hypoconid. Italicised num- bers indicate dimension may have been lessened by wear. Ee ee A 30057 130057RM'|9.7| - | - [8.7|7.7]5.4] 46 |5.2| 47 | Sata Ls esal sa las [Sa sen | - | - [s.9/8.8[47 5.4 [5.1 5.1 | eee 9.5 23 ? LARIN 3.1 [31299 RM? [8.4] 1.6 ss 4.4 E 4.1| 5.1 zarm |e | 8.1|7.0/4.5| 5.6 [4.4] 4.8 cosas ral [sb sal ea fsa) 3 is ss 35/40] 51 = as [8.3 |8.0|42] 4.8 [4.2] 4.2 | ea CE 31301 RP; |6 sL 31301 RM: | 10 31302 ERAT 31301 [31301 RM2 |9.8 | EANES 31301 Rwa 10 31302 RP; |6.6| z StS Bat eer ee 5.4 Afef [2331m foo] - |- [ealeolaal as [a7] sa | [23306 melo] - |- |7.4[75|49[ 44 [45| 5. | rasostveleal - | traf leat 49 fsa] so] [23306 Lva [92| - |- [6.8 [s.7]4.1] 42 [3.7] 327 | KUTERINTJA NGAMA FROM RIVERSLEIGH the tooth. A deeper crevice dividing the tooth into sub-equal halves, with the large anterior cuspid on the anterior side and the twinned posterior cuspids on the posterior side, blocked half way along by a crest linking the anterior cuspid to the posterobuccal cuspid. A shallower crevice be- tween the posterior cuspids blocked by a minor crest running from the apices of these cuspids. Small cristid running posteroventrally from the posterobuccal corner of the apex of the posterobuccal cuspid, turning posterolingually, joining a wider posterior cingulum. Posterior cin- gulum curving anterolingually before joining the base of the posterolingual cuspid. Lower Molars. Subrectangular. M1-3 subequal; Mg smaller. Crown heights decreasing from M1- 4. Tooth row curving posterolingually (Fig. 1). M1: ‘Central’ cuspids on the protolophid and hypolophid are neomorphs (Tedford & Woodburne, 1987). 6-cuspid; anterior portion narrower than posterior. Trigonid triangular; an- terolingual border inclined posterolingually; an- terobuccal border of trigonid inclined posterobuccally; both these inflections originat- ing from an anteromedial position of the anterior cingulum, at termination of preprotocristid. Talonid wider than trigonid. Protoconid over the posterior portion of the anterior root; posterior cuspids aligned over the central portion of the posterior root. Preprotocristid (or paracristid) rel- atively wide, generally low, with pocket between the buccal margin, the anterior cingulum and the anterior face of the protoconid, with smaller and less well defined pocket between the anterior cingulum, the lingual margin of the pre- protocristid and the anterior surfaces of the ‘central’ cuspid and metaconid. ‘Central’ cuspid of protolophid with apex slightly anterior to the protoconid and metaconid. Anterior positioning of ‘central’ cuspid or neomorph more exagger- ated on the hypolophid. Both ‘central’ cuspids of similar height, lower than main cuspids. ‘Central’ cuspid on protolophid forming a lingual basin with the metaconid, not totally enclosed, with small openings anteriorly and posteriorly. Sim- ilar, small basin formed between the ‘central’ cuspid of the hypolophid and the entoconid, with comparable openings to its counterpart on the protolophid, with anterior opening much smaller. A deep crevice dividing ‘central’ cuspids from the main buccal cuspids, continuous an- teroposteriorly, shallower in the central part of the tooth. Transverse valley interrupting this lon- 383 lingual basins metaconule protocone metacone stylar cusps B, C, D, E Cc paracone D FIG. 4. Kuterintja ngama. A-B, QMF30058, RM?. A, occlusal view; B, buccal view. C-D, QMF30057, RM!. C, occlusal view; D, buccal view. gitudinal crevice wide, blocked at its buccal ex- tremity by a small, posterobuccally slanting cin- gulum linking the base of the protoconid to the base of the hypoconid. Thin crevice in the trans- verse valley preventing symmetrical postproto- cristid and prehypocristid (cristid obliqua) and postmetacristid and preentocristid from linking. Metaconid and entoconid with apices steeply in- clined, rather than conical, with lingual surface of each much taller than the buccal. Entoconid higher than metaconid higher than ‘central’ cus- pids, with slight gradient descending from lingual to buccal. A thin posterior cingulum and a small pocket in the posterolingual corner of the tooth; pocket bordered by the lingual end of the poste- rior cingulum, with 2 crests from the postero- lingual and posterobuccal sides of the apex of the entoconid, respectively. In QMF31301 protoconid and hypoconid with lingual surfaces slightly more vertically orien- 384 FIG. 5. Kuterintja ngama. 4, SAM P24539, holotype, LM. B, OMF24604, RM*. C, QMF31299, RMF. tated, possibly due to less wear than observed in QMF23306. Mz, Like Mı except: anterior end resting upon the posterior cingulum of Mj); trigonid subrectangu- lar rather than triangular, due to the anterior cin- gulum being more transversely linear, anterior MEMOIRS OF THE QUEENSLAND MUSEUM pockets formed with the anterior cingulum smaller, trigonid and talonid equal in transverse width. *Central’ cuspid on the protolophid in direct transverse alignment with the protoconid and metaconid; ‘central’ cuspid on the hypo- lophid more to anterior; hypoconid slightly more posterior. Crevice between the linked lingual cus- pids and the buccal cuspid shallower (possibly due lo wear on protoconid and hypoconid). Cus- pid height gradient from lingual to buccal much steeper (possibly due to wear). Pocket at the buccal end of the transverse valley larger, Cristid obliqua (or prehypocristid) and hypocristid (posthypocristid) more developed. Lingual pocket on the protolophid less well defined, with the openings between the metaconid and ‘central’ cuspid larger. Small posterolingual pocket bor- dered by postentocrislid, hypocristid and a small posterior cingulum, with most of the Jatler hidden by M3. M3. Same as M2 except: anterior cingulum rounded. Crown height reduced; with height gra- dient, Metaconid and ‘central’ cuspid not closely linked, separate entities with a crevice between the two cuspids. Crevice of variable depth. Central’ cuspid on the protolophid larger, high- lighting an increase in size from Mj to M3. Crev- ice between ‘central’ and buccal cuspids shallower, decreasing in depth down the tooth row, Despite damage to the talonid, ‘central’ cuspid on the hypolophid much reduced. En- toconid sub-equal in height to the ‘central’ cuspid on the hypolophid, transversely compressed, Pos- terior cingulum and posterobuccal basin much shorter, Juvenile M3 with an anterolingual basin bigger than in Mı or Ma, paracristid terminating in lon- gitudinal alignment with the buceal side of the metaconid. Mg, Shortest and narrowest molar, with lowest crown, rounded subrectangular, with a very rounded anterior cingulum, Same as M3 except: protolophid and hypolophid slanting more an- terolingually, due to the buccal cuspids being posterior to the lingual cuspids. ‘Central’ cuspid on the protolophid not linked to the metaconid; crevice between these 2 cuspids deeper. ‘Central’ cuspid on the hypolophid greatly reduced, more so than in M3, further to posterior, Posterior cingulum short, extending to the medial line of the tooth. Posterolingual basin greatly reduced. Transverse valley closed lin- gually and buccally, Lingual end of the transverse KUTERINTJA NGAMA FROM RIVERSLEIGH valley curving posterolingually; buccal end curv- ing posterobuccally. Cristid obliqua and hypocristid relatively short. All cuspids subequal in height, with the metaconid slightly larger than the entoconid = protoconid and hypoconid. P. Subovate, tricusped, transversely wide. Ante- nor portion narrower than posterior. Cusps 3, large, subequal in height, Anterior and postero- medial cusps longitudinally aligned, separated by a shallow trough. A large crevice separating the posteromedial and posterolingual cusps. With very small anterolingual cingulum and larger posterior cingulum. A thin rib running from the apex of the posteromedial cusp to the posterobuc- cal edge of the posterior cingulum, Upper molars. Stylar cusps well-developed; gen- tral selenodont cusp pattern; high crowned, with a general gradient towards the lingual side, with 4 major cusps (paracone, protocone, metucone and metaconule), with a stylar shelf consisting of stylar cusps B,C,D and E. M! (Fig, 4), Buccal cusps of RM! positioned more posteriorly than in other molars, with posterobuccal slant, wider posteriorly than ante- riorly, giving an anterolingual slant to the buccal border. Stylar cusp B smaller and further anterior than in other molars, Stylar cusp C as large as that on Mł, anterior to the postpuracrista; crista not forming part of the posterior face of the stylar cusp. Stylar cusp D largest cusp, subequal in height to the metacone, larger than in any other molar. Stylarcusp E more developed than in other molars, larger than stylar cusp B. A minor cuspule on the anterior of stylar cusp D, buccal to the termination of the postmetacrista, larger than sty lar cusp B, but slightly smaller than stylar cusp E. All stylar cusps subconical to triangular, except posterobuccally-aligned ridge, stylar cusp C. Buccal margin wider than lingual; blocking crests in the transverse valley absent (some minor par- tial blockages buccally); anterior cingulum curv- ing posterobuccally at its buccal extremity; preparacrista orientated less transversely than in other upper molars: minor depression on the pos- terior face of the metaconule less developed than in M2; buccal basins on the paracone and metacone poorly developed compared jo other molars; posterior cingulum thinner than in other molars, M? (Fig. 4). Square. Cusp sizes: paracone >metucone> protocone = metacunule, Stylar cusp height: C>D>B>E. Stylar cusp B connected to the paracone by a preparucrista, and stylar cusp C via a postparacrista. Stylar cusp D connecting to the metacone by a premetacrista, and stylar cusp E connected lo the metacone by a postmetacrista, Buccal basin deep, formed be- tween stylar cusps B and C and the paracone. The homologous basin on the metacone less distinct, enclosed less tightly, slanting steeply posterobuccally towards the reduced stylar cusp E. Basin on the metacone deepest anterolingual to stylar cusp E, Large transverse valley dividing this tooth in half, containing the paracone (and associated stylar cusps) and protocone anteriorly, and the metacone and metaconule posteriorly, partially blocked buccally by an incomplete crest linking premetacrista and postprotocrista, blocked centrally by a small crest linking postprotocrista and premetaconule erista, stopped at its lingual extremity by avery low crest linking the lingual sides of protocone and metaconule (lingual cingulum), Buccal faces of protocone and metaconule steeply inclined, al- most to the point of being vertical; anterior cin- gulum well-developed. running buccally from the protoerista to the anterior side of stylar cusp B, and lingually from the anterolingual corner af the base of the protocone to the protocrista; pos- terior cingulum smaller than anterior cingulum, with the former extending from stylar cusp E to join the postmetaconulecrista, small depression on the lingual side of jhe postmetaconulecrista and medial posterior hase of the metaconule (per- haps remnant of the lingual portion of the poste: rior cingulum); all cusps over the mid-line of the roots; stylar cusps triangular, rather than round or conical; buccal cusps with very round apices, Mì (Fig. 5). Same as M2 except: crown lower, 4 major and stylar cusps retaining same relative heights; stylar cusp E further reduced, virtually non-existent; posterior cingulum less defined; small] pocket on the posterior side of the metaconule on M? absent; lingual cusps closer to the anterior side of their respective roots, Stylar cusp C more to posterior than in M?, with postparacrista forming part of this stylar cusp; buccal basin on the paracone of M? larger than in M2. triangular, with wider buccal edge, Trans- verse valley partially blocked buccully by a crest linking stylar cusps C and D, but not by a crest linking the premetaconulecrisia and postpara- erista, With central and lingual blocking crests, Crest linking stylar cusp D and the metacone larger and more uniform; stylar cusp E more 386 TABLE 2. Characters and character states used in the ilariid intrafamilial phylogenetic analysis. Characters States | 1_|Stylar cusp development _|0=poor; 1=well 2, | Transverse valley on Q=absent; 1=moderate; lower molars 2=well-developed 0=none; |=poor; 2=moderate; 3=well- developed 3 | Transverse linkages between cuspids P 4 |Post protocrista and pre i 0 =strongly developed; metaconulecrista l=poorly developed Protocone compressed 5 |longitudinally relafive,to Q=absent; |=present metaconule (on M -M ) O=no significant _ separation; l=significant separationby large trough O=well developed; 1=poorly developed; 2=absent. 0 = absent; 1= weak;2 = strongly developed O=present; |=absent Separation of stylar cusps 6 |e Ind D ne 7 |Paraconule on M' 8 |Paraconid on M, 9 | Protostylid 10 | Metaconid 0=conical; 1=lingually convex crest 0 = subequal ; l=protoconid larger than metaconid Relative heights of the 11 ; anterior cuspids Overall tooth size 0 =small; 1=large O=compressed together; l=not compresse: 0=low; 1=moderate; 2=high O=absent; 1= slight cusp;2 = moderate;3 = well-developed O=dorsoventrally flattened; 1=caniniform and conical; 2=dorsally flattened and distally inflected Lingual cusps Crown height 15 | Posterolingual cusp on P 16 I, (unordered) 17 | Posterobuccal cuspid on P, O=absent; |=present 18 | ‘Central’ cuspid 19 Fosterolingual cuspid on lingual closure of 20 | transverse valley by a = [cingulum (on upper molars} 21 | Bladed P; O=absent; 1=present O=absent; |=present O=cingulum absent; 1=incipient cingulum (in form of cuspules); 2=cingulum present O=present, 1=absent O=strongly bladed; 1=weakly bladed; 2=absent Posterobuccal cusp on P O=absent; 1=present Longitudinal valley (i.e. |0=well-developed; 24 | distance between lingual |l=moderately developed; & buccal cusps / ids) 2=absent O=bulbous; |=elon: 22 |Bladed P? ate MEMOIRS OF THE QUEENSLAND MUSEUM anteriorly positioned; buccal basin on the metacone narrower, slanting more anterobucc- ally. M4. Sub-triangular, posteriorly compressed. Same as M? except: crown height very small, with the protocone>paracone=metacone> metaconule. Stylar cusp B>D; stylar cusp C non- existent; stylar cusp Eextremely reduced or miss- ing. Crevice between the paracone and protocone transversely wider. Buccal surface of the metaconule and protocone far less vertically in- clined. Anterobuccal basin larger; buccal basin on the metacone absent; buccal basin on the paracone very shallow, slanting posterobuccally. Transverse valley not blocked buccally, curving posterobuccally rather than being transverse, with lingual end enclosed slightly, by a low crest (i.e. the crest does not continue to the base of the protocone). Anterior cingulum very small. Poste- rior root slants posteriorly rather than vertically. REMARKS. Comparing LM! QMF23203 to RM! QMF30057: buccal half of the anterior cin- gulum transversely shorter; stylar cusps B and E less developed; cuspule on the anterior face of stylar cusp D absent. M? of QMF24604 exhibits variation compared to the M? of QMF31299 as follows: 1) stylar cusp D is larger; 2) the distance between stylar cusps D and E is greater and therefore a bigger buccal basin is found on the metacone; and 3) the medial lingual basin is divided into two sub-basins at its lingual margin by a very small transverse crest. QMF24604 highlights the variability in M4, as follows: posterior half not as compressed as in M4 of QMF31299, and therefore has a longer and wider posterior cingulum. Stylar cusp E is much reduced. Therefore the buccal basin on the metacone is also present and it is as deep as the basin on the paracone. Stylar cusp D is also more defined and larger than in M* of QMF31299. The medial lingual basin is smaller. The anterior cin- gulum extends further lingually to the base of the protocone. The transverse valley is blocked in two places rather than one. It is blocked buccally by a crest linking stylar cusps C and D, and is partially blocked by a small crest linking postparacrista and premetaconulecrista. The crest partially blocking the lingual extremity in QMF31299 is not present. Large stylar cusp C is not present in QMF31299. A well developed and enclosed buccal basin on the paracone is absent in QMF31299, KUTERINTJA NGAMA FROM RIVERSLEIGH TABLE 3. Ilariid intrafamilial data matrix as used by PAUP.? = fossil material missing or status uncertain; a = 1&2; — c=0&2 [1 [2[3]4]s][e[7|[s]9frofi1[i2]i3] 14] 15]16]17]18]19|20]21 [22]23 | 24]25 | totolitotolotolelololololololi]?lilololelololololo] folololololojololo|ijo| An EXEREREA FAFA FRERER EEE Kuterintja ngama [2 fale fi] pnpnnpanonnnonan PHYLOGENETIC SYSTEMATICS Twentyfive dental characters with up to 4 states each (Table 2) were used to develop the data matrix following character argumentation and polarisation (Table 3) for the intrafamilial analy- sis of ilariids. Three outgroups used to determine polarities are: 1) the modern Koala, Phascolarctos cinereus; 2) Madakoala; and 3) wynyardiids. The former is the most derived member of a primitive outgroup because the Phascolarctidae is the stem taxon from which the vombatomorphian radiation diverged (Marshall et al., 1990; Aplin & Archer, 1987). Madakoala devisi and Madakoala wellsi are employed be- cause of the primitive position of Madakoala within the phascolarctid radiation (Woodburne et al., 1987). Primitive and derived phascolarctids were used to determine the relationships of spe- cies of Koobor. Wynyardiids include Namilamadeta snideri and Muramura sp. and are a closer sister group of the ilariids than the phascolarctids, therefore providing a basis for polarising character states within the Vombato- morphia. Character optimisation is performed after the character analysis has been completed and the most parsimonious trees found. The two op- timisation algorithms used by PAUP are ACCTRAN and DELTRAN. ACCTRAN accel- erates the evolutionary transformation of charac- ters so that changes occur ai the earliest possible stage on the optimal tree. As far as homoplasy is concerned, this algorithm has the effect of favour- ing reversal of character states over conver- gences. DELTRAN delays transformation of characters so that changes occur as far up the optimal tree as possible. This has the effect of favouring convergences over character reversals (Wiley et al., 1991). DELTRAN analyses are favoured here because of the large amount of missing character data in the matrices. RESULTS. the a tree (Fig. 6) has 50 steps; a consistency index (CI) of 0.800; a homoplasy index (HI) of 0.260; a retention index (RI) of 0.778; and a rescaled consistency index (RC) of 0.622. Notably the ingroup (Koobor notabalis, I. lawsoni, I. illumidens and Kuterintja ngama) did not form a monophyletic clade. Ko. notabalis is sister taxon to the Wynyardiidae, Ku. ngama, I. illumidens and I, lawsoni clade. Madakoala and Phascolarctos cinereus formed a basal monophy- letic clade. Bootstrap analysis for the most parsimonious tree had the clade excluding phascolarctids and Koobor supported 99% of the time. The ilariid clades, excluding and including Ku. ngama, oc- curred 78% and 95% of the time respectively. Removal of the wynyardiids as an outgroup had no effect on the topology in the optimal tree. A bootstrap analysis on data excluding the wynyardiids found the clade containing /. illumidens and I. lawsoni to be supported 62% of the time, slightly lower than in the previous anal- ysis. While the clade including all 3 ilariid species was supported on all occasions. Another method of testing support for the opti- mal tree is to examine the frequency and topology of the ‘next best’ trees (Simmons, 1993). PAUP evaluated 945 trees and found one optimal tree of 50 steps. Two trees of length 51 were observed as well as one tree of 52 steps. Neither of the trees of 51 steps in length are considered here as the phyletic relationships presented by each do not represent the phascolarctids as a monophyletic clade. In both cases Koobor is intermediate be- tween Madakoala spp. and Phascolarctos cinereus. DISCUSSION Classification of Ku. ngama as an ilariid was tentative (Pledge, 1987) and controversy sur- rounded placement of Koobor. Comparison of the 388 Riversleigh ilariid with species of Haria and Kuterintja ngama, confirms that the Riversleigh animal is indistinguishable from the latter. Dental variation in Phascolarctos cinereus, one of Ku. ngama’ s closest living relatives, suggests that: 1) variation in Riversleigh fossil material is in the range for vombatiform species, and represents only one taxon; and 2) the Riversleigh species is Ku. ngama. DISCUSSION OF THE PHYLOGENETIC ANALYSIS. Kuterintja ngama as the sister taxon of a clade containing Zaria illumidens and Ilaria lawsoni (Fig. 6), and not united with wynyardiids or Koobor, reinforces classification of this animal as an ilariid. Synapomorphies used by DELTRAN to unite ilariids include: 1) a well-de- veloped transverse valley; 2) poorly-developed postparacrista and premetaconulecrista; 3) pro- tocone longitudinally compressed relative to the metaconule on M!-M3; 4) large crown height; 5) moderately well- -developed posterolingual cusp on P3; 6) closure of the transverse valley on upper molars bya lingual cingulum; 7) a non-bladed and bulbous P3; 8) a non- bladed P3 and 9) a ‘central’ cuspid on the protolophid and hypo- lophid of lower molars. These synapomorphies only apply to the Ilariidae relative to the other taxa used in this analysis, and may prove to be symplesiomorphies when all other vombato- morphian taxa are included. Some of these syn- apomorphies refer to the upper dentition, which is unknown for Zaria lawsoni. However, the close similarities between the lower dentition of both Zaria species suggests that these syn- apomorphies will be generically significant when upper dentition for /. Jawsoni is found. Syn- apomorphies used by the same algorithm to unite species of //aria include: 1) moderately well-de- veloped transverse linkages between cuspids; 2) weak paraconid; 3) large tooth size; and 4) trans- versely compressed, caniniform lower first inci- sors. In constructing the most parsimonious tree, 9 characters were found to exhibit some degree of homoplasy. According to DELTRAN moder- ately well-developed transverse linkages be- tween cuspids is due to convergence between primitive wynyardiids and species of /laria, with Kuterintja ngama with plesiomorphic poorly de- veloped linkages. Conversely, ACCTRAN sug- gests that moderately well- developed transverse linkages were already a feature of the common wynyardiid/ilariid ancestor, possibly before Koobor diverged from the vombatomorphian lin- MEMOIRS OF THE QUEENSLAND MUSEUM Madakoala spp Phascolarctos cinereus Koobor spp Wynyardiidae Kuterintja ngama Ilaria illumidens Ilaria lawsoni FIG. 6. Relationships of the Ilaritdae and Koobor. DELTRAN Synapomorphies; A-weak longitudinal valley; B-well-developed stylar cusps; moderately developed transverse valley; separation, of stylar cusps C and D; weak paraconule on M!; no pro- tostylid; protoconid>metaconid; no compression of lingual portion of tooth; moderate crown height; dor- sally flattened and distally inflected I); no longitudi- nal valley; C- well-developed transverse valley; poorly developed postprotocrista and premeta- conulecrista; protocone longitudinally compressed relative to metaconule on M!- M2; large crown; mod- erately developed posterolingual cusp on P’; ’central’ cuspid; posterolingual cuspid on P3; lingual closure of transverse valley on uppers by a cingulum; non- bladed P3 and P?; D-moderately well developed trans- verse linkages; weak paraconid on Mj; large teeth; caniniform and conical l. ACCTRAN Syn- apomorphies: A-moderately developed transverse valley; moderately developed transverse linkages; no protostylid; protoconid metaconid; dorsally flattened and inflected 11; no posterobuccal cuspid on P3; mod- erately developed longitudinal valley; B-well-devel- oped stylar cusps; separation of stylar cusps C and D; no paraconule; uncompressed lingual portion of tooth; moderate crown height; moderately developed posterolingual cusp on P3; no longitudinal valley; C- well- -developed transverse valley; poorly developed postprotocrista and premetaconulecrista; protocone longitudinally compressed relative to metaconule on M! - M3; metaconid a lingually convex crest; high crowns; ’central’ cuspid; posterolingual cuspid on P3; transverse valley closed lingually by cingulum; non- bladed P3 and P3; bulbous P?; D-poorly developed paraconule; weak ‘paraconid on Mj; large teeth; can- iniform, conical I}; posterobuccal cuspid on P3; and Pe KUTERINTJA NGAMA FROM RIVERSLEIGH cage. Poorly-developed transverse linkages in Ku ngama would therefore be the result of a reversal to the phascolarctid slate. The ACCTRAN model appears preferable. although discovery of a lower dentition for Koobor would help resolve its classification. The poorly-developed paraconule on M! of Ha- ria illumidens (Tedford & Woadburne, 1987), is either a plesiomorphy dating from some time alter divergence of Koobor (DELTRAN), or the result of a reversal (ACCTRAN). The former hypothesis implies that loss of the paraconule is convergent between wynyardiids and Ku. ngama, while the latter, and possibly more parsimonious, hypothesis suggests that the paraconule was al- ready lost from the vombatomorphian lineage before the wynyardiids and ilariids diverged. The paraconid on M, is another homoplasic character, For both algorithms character transfor- mation suggests that a well-developed paraconid is convergent between J, il/umidens and Madakoala. A poorly developed paraconid is deemed to be convergent between species of //a- riq and primitive Madakoalu, with absence of a paraconid being the plesiomorphic phascolaretid character state, However, a more likely solution is: 1) that a well-developed paraconid is the plesiomorphic condition; 2) that absence of a well-developed paraconid in P. cinereus is a de- rived condition; 3) that the paraconid was grad- ually reduced or lost before or after Koobor diverged; 4) that the paraconid in species of Maria represents a reversal to the plesiomorphic state; and 5) that loss of a paraconid is convergent between P. cinereus, wynyardiids and possibly Koobor. Knowing whether there was ar was ool a paraconid in Koobor would help clarify this situation. Both algorithms suggest that 2 protostylid on Mı of Haria illumidens represents a reversal to the plesiomorphic phascolarctid condition. The only discrepancy between the two character (ransformation pathways is the point at which the protostylid was lost. DELTRAN delays loss of the protostylid until after the divergence of Koobar, while ACCTRAN muintains lhat lass occurred before the divergence. An identica character transformation occurs for ‘relative heights of the anterior cuspids’ (character 11), such that cuspids which are subequal in height represent a reversal to the plesjomorphic condi- tion for £ ilhwnidens, Possessing a protoconid larger than the metaconid is therefore a syn- apomorphy uniting wynyardads. 4 lawson Kigeringia ngama and possibly Roolwrn A 389 posterobuccal cuspid on Pa of species of Hariays deemed to be a reversal to the plesiomorphic phascolarctid condition by ACCTRAN, while DELTRAN suggests that absence of this strug- ture is convergent between wynyardiids and Kurerintja ngama. Again. ACCTRAN seems to be the most parsimonious, implying thii the posterobuccal cuspid was lost before wynyardiids and ilariids. and possibly Koobor, diverged. For lingual closure of the transverse valley on upper molars (character 20) the pathway for char- acler transformation is unclear due primarily to the variable nature of this structure in Madakoala. However, the suggested transformation sequence is: 1) a partial cingulum, in the form of 2 cusputes on the anterolingual and posterolingual bases of the metaconule and protocone respectively, was present in the ancestral koala; 2) the two cuspules eventually joined, convergently forming the de- rived lingual cingulum in P; cinereus, some spet- imens of Madakoala and ilariids; and 3) other Madakoala and wynyardiids developed in the opposite direction, convergently losing the cuspules altogether. The two cuspules occur in Koabor notahalis but not in Koober jimbarati (Archer, 1977), perhaps suggesting that the latter is more derived than the former, Alternatively, a lingual cingulum may be a plesiomorphic phascolaretid condition. implying that the cuspules in Ko, efabealis are an apomorphic ves- tige. In this case, absence of a lingual cingulum would be a mure derived condition, convergent between some specimens of Madakoala, Ka, jimbarani and wynyardiids. The final character transformation found to contain some degree ol homoplasy involved a cusp on the posterobyceal margin of P? (character 23). The sequence of change suggested by the algorithms is that absence of such a cusp is the plesiomorphic condition, and that species of Madakeala, Haria tlumidens and possibly /. lawsoni conyvergently share aden ved posterohuc- cal cusp. This transformation sequence seens unlikely because Madakoala are overall more plesiomurphic phascolarctids (Woodburne el al., 1987} and a posterabuccal cusp on P* is more likely la be the plesiomorphic condition. If so, loss of this structure is a synapomorphy for Koobor, wynyurdiids and Ke agama, and is com- Vergent on the condition in Phascolarctos cinereus. Absence of the cusp ip this contest is yet another potential character stale separating Kooher [rom phascolarctids. According tò Simmons (1993) the expected 390 value for the consistency index of a tree with 7 taxa is: CI = 0.90 - 0.022 (7) + 0.000213 (7)? = 0.736 (3 s.f). The observed CI for the optimal intrafamilial tree is 0.800, 0.064 from the expected value. The observed CI value implies slightly less homo- plasy for the intrafamilial analysis than would be expected for seven taxa. Similarly the retention index (RI) and the rescaled consistency index (RC) are reasonably large, emphasising the low degree of homoplasy and potential for homoplasy respectively. The optimal tree which includes all outgroup taxa is reasonably well-supported by bootstrap analysis and by the lack of significantly different trees, of plausible topology, within a few steps of the most parsimonious. The low bootstrap result for the /laria clade is almost totally due to the amount of missing data for /. lawsoni. This study supports the notion that Ku. ngama is an ilariid and forms a monophyletic clade with Zaria. CLASSIFICATION OF KOOBOR Pledge (1987) discussed the possibility that Kuterintja ngama is more closely related to Koobor than Ilaria. The lower dentition and upper molars in addition to M? demonstrates that Ku. ngama is an ilariid. One of the few similari- ties between Koobor and Ku ngama is the smooth rounding of the lingual faces of the lingual cusps, a character state previously thought to unite the taxa phyletically (Pledge, 1987). However, smooth and rounded lingual faces on lingual cusps are also a feature of wynyardiids, and to a lesser extent Madakoala, suggesting that it is plesiomorphic, The ambiguity of this character state also increases the possibility of homoplasy. Pledge (1987) hypothesised that Ku. ngama may be ancestral to Koobor. Our study does not sup- port this view. Koobor notabalis appears to be the primitive sister-group of wynyardiids plus ilariids. We have no clear support, however, for Koobor being in the Phascolarctidae. This may be indirect sup- port for the suggestion that Koobor represents a distinct family of vombatiform marsupials. DELTRAN found only one synapomorphy po- tentially uniting Koobor with the wynyardiids and ilariids: the less well-developed longitudinal valley on the molars. ACCTRAN found 7 syn- apomorphies fora Koobor, wynyardiid and ilariid clade. This should not be taken at face value, however, as 6 of these character states refer to the lower dentition which is unknown for Koobor. 10 MEMOIRS OF THE QUEENSLAND MUSEUM synapomorphies were found by DELTRAN to unite wynyardiids and ilariids to the exclusion of Koobor (Fig. 6). BIOCORRELATION OF RIVERSLEIGH AND THE ETADUNNA FORMATION Ku. ngama occurs in the White Hunter Local Fauna at Riversleigh and in the Ngama Local Fauna in the upper Etadunna Formation. Maria lawsoni occurs in the Ditjimanka Local Fauna in the lower Etadunna Formation. /laria illumidens occurs in the Pinpa Local Fauna of the Namba Formation, at Lake Pinpa. Woodburne et al. (1993) suggested at least 6 magnetic reversals within the Etadunna se- quences, correlated them with a biostratigraphic zonation and the MPTS (Fig. 7) and suggested 24-28 Ma for the base of the Etadunna Formation. Woodburne et al. (1993) correlated Zone D with the Ngama and Tarkarooloo Local Faunas. Correlation of magnetic polarity and biostratigraphic zones places zone D in lower magnetozone R3, which in turn correlates with Chron 7n.1r of the MPTS, or 24.7 - 25.0 Ma. Ku. ngama therefore, correlates White Hunter Site with the Ngama Local Fauna at 24.7 - 25.0 Ma providing: |) that ilariid material in White Hunter Site has not been reworked from older deposits (which, given the lack of evidence for weathering or transport, does not appear likely); and 2) that the apparently short temporal range of Ku. ngama in the Etadunna Formation is the full range of this species. ACKNOWLEDGEMENTS Vital support for research at Riversleigh has come from the Australian Research Grant Scheme, the National Estate Grants Scheme (Queensland), the University of New South Wales, the Commonwealth Department of Envi- ronment, Sports and Territories, the Queensland National Parks and Wildlife Service, the Com- monwealth World Heritage Unit, ICI Australia, the Australian Geographic Society, the Queens- land Museum, the Australian Museum, the Royal Zoological Society of New South Wales, the Linnean Society of New South Wales, Century Zinc, Mount Isa Mines, Surrey Beatty & Sons, the Riversleigh Society, and private supporters in- cluding Elaine Clark, Margaret Beavis, Martin Dickson, Sue & Jim Lavarack and Sue & Don Scott-Orr. Vital assistance in the field has come KUTERINTJA NGAMA FROM RIVERSLEIGH MPTS MAGNETOZONE 24.2 Ma ————? 6 Cr R4 ART MS in N3 a goa T 7n.lr R3 25.0Ma 7n.2n N2 Tr R2 25.5 Ma m i Tar R1 25.7 Ma ——————? 391 Etadunna Etadunna Namba Riversleigh Mammal Formation Formation Assemblage Zone Assemblage Assemblage E "Treasure/ Lungfish" , a? Tark; l White D Hunter Site ? C+B "Wynyardiid" ? FIG. 7. Geochronology and biocorrelation of the Etadunna and Namba Formations, Lake Palankarinna and Lake Pinpa, S.A. and lower Riversleigh faunas, northwestern Queensland. (Modified from Woodburne et al., 1993, figs 2 and 15), N = Normal magnetic polarity; R = Reversed magnetic polarity; MPTS = magnetic polarity time scale. from many hundreds of volunteers as well as staff and postgraduate students of the University of New South Wales. Skilled preparation of most of the Riversleigh material has been carried out by Anna Gillespie. TJM acknowledges the assis- tance of Prof. Alberto Albani, Karen Black, Jenni Brammall, Henk Godthelp, Steve Salisbury, Anne Musser, Mary Knowles and his family. LITERATURE CITED APLIN, K.P. & ARCHER, M. 1987. Recent advances in marsupial systematics with a new syncretic classification. p.xv-Ixxii. In Archer, M. (ed.), Pos- sums and Opossums: studies in evolution. (Surrey Beatty & Sons and the Royal Zoological Society of New South Wales, Sydney). ARCHER, M. 1977. Koobor notabalis (De Vis), an unusual koala from the Pliocene Chinchilla Sand. Memoirs of the Queensland Museum 18: 31-35. ARCHER, M. 1984. The Australian marsupial radia- tion. Pp. 633-808. In Archer, M. & Clayton, G. (eds), Vertebrate zoogeography and evolution in Australasia. (Hesperian Press: Perth). ARCHER, M., GODTHELP, H., HAND, S.J. & MEGIRIAN, D. 1989. Fossil mammals of Riversleigh, northwestern Queensland: prelimi- nary overview of biostratigraphy, correlation and environmental change. Australian Zoologist 25: 29-65. ARCHER, M., HAND, S.J. & GODTHELP, H. 1994. Riversleigh. 2nd ed. (Reed: Sydney). CREASER, P. 1997. Oligocene-Miocene sediments of Riversleigh: the potential significance of topogra- phy. Memoirs of the Queensland Museum 41(2): 303-314. FLOWER, W.H. 1867. On the development and suc- cession of teeth in the Marsupialia. Philosophical Transactions of the Royal Society, London 157: 631-641. LUCKETT, W.P. 1993. An ontogenetic assessment of dental homologies in therian mammals. Pp 182- 392 204. In Szalay, F.S., Novacek, M.J, & McKenna, M.C. (eds), Mammal phylogeny: Mesozoic differ- entiation, multituberculates, monotremes, early therians and marsupials. (Springer-Verlag: New York). MARSHALL, L.G., CASE, J.A. & WOODBURNE, M.O. 1990. Phylogenetic relationships of the fam- ilies of marsupials. Pp. 433-505. In Genoways, H.H. (ed.), Current mammalogy, vol. 2. (Plenum Press: New York), PLEDGE, N.S. 1987. A new genus and species of primitive vombatoid marsupial from the medial Miocene Ngama Local Fauna of South Australia. Pp.419-422. In Archer, M. (ed.), Possums and Opossums: studies in evolution. (Surrey Beatty & Sons and the Royal Zoological Society of New South Wales: Sydney). SIMMONS, N.B. 1993. The importance of methods: archontan phylogeny and cladistic analysis of morphological data. Pp. 1-51. In MacPhee, D.E. (ed.), Primates and their relatives in phylogenetic perspective. (Plenum Press: New York), TEDFORD, R.H. & WOODBURNE, M.O. 1987. The Ilariidae, a new family of vombatiform marsupials from Miocene strata of South Australia and an evaluation of the homology of molar cusps in the MEMOIRS OF THE QUEENSLAND MUSEUM Diprotodontia. Pp. 401-418. In Archer, M. (ed.), Possums and Opossums: studies in evolution. (Surrey Beatty & Sons and the Royal Zoological Society of New South Wales: Sydney). WILEY, E.O., BROOKS, D.R., SIEGEL-CAMSEY, D, & FRANKS, V-A. 1991. The compleat cladist: a primer of phylogenetic procedures. University of Kansas Museum of Natural History Special Publication 19. WOODBURNE, M.O., TEDFORD, R.H., ARCHER, M. & PLEDGE, N.S. 1987. Madakoala, a new genus and two species of Miocene koalas (Marsupialia: Phascolarctidae) from South Aus- tralia, and a new species of Perikoala, Pp. 293- 317. In Archer, M. (ed.), Possums and Opossums: studies in evolution. (Surrey Beatty & Sons and the Royal Zoological Society of New South Wales: Sydney). WOODBURNE, M.O., MACFADDEN, B.J., CASE, J.A., SPRINGER, M.S., PLEDGE, N.S., POWER, J.D., WOODBURNE, J.M. & SPRINGER, K.B. 1993. Land mammal biostratigraphy and magnetostratigraphy of the Etadunna Formation (late Oligocene) of South Australia. Journal of Vertebrate Paleontology 13: 483-515. NANOWANA GEN. NOV. , SMALL MADTSOIID SNAKES FROM THE MIOCENE OF RIVERSLEIGH: SYMPATRIC SPECIES WITH DIVERGENTLY SPECIALISED DENTITION JOHN D SCANLON Scanlon, J,D, 1997 06 30: Nanewana gen. nov, small madisoiid snakeS from the Miocene of Riversleigh: sympatric species with diVergently specialised dentition. Memoirs of the Queensland Museum 41(2): 393-412. Brisbane. ISSN 0079-8835. Two sinall early Miocene madtsotid snakes from Riversleigh, NW Queensland are described as Nanowana godrthelpi gen. et sp. nov. and N, schrenki gen. et sp. nov. Jaw elements of the former are depressed, lack ankylosed teeth, and have alveoli of nearly uniform size, these features are interpreted as signs of a coadapted character complex (‘arthrodonty’) where the teeth are attached to the jaws by a fibrous hinge. This condition is associated with a diet of hard-scaled scincid lizards. The latter species retains ankylosis, and has strongly enlarged teeth on the anterior dentary and middle maxilla indicating a distiner method of subduing prey, but extant analogues are also predominantly scincivorous. Departure in each species from the nearly homodont, ankylosed condition in other madtsoiids is interpreted as adapta- uon tò a diet of scincid lizards. These divergent, but functionally parallel specislisauions are likely to be independently derived from the ancestral condition, John D, Scanlon, School of Biological Sciences, University of New South Wales, New South Wales 2052, Australia (email: johns @ geko.netan); received 7 February 1997, Maudtsoiid snakes in Tertiary faunal assemblages of Riversleigh (Scanlon 1992, 1993, 1995, 1996) have been referred to Yurlunggur Scanlon, 1992 and Wonambi Smith, 1976, Other Riversleigh madtsoiids cannot be included in previously known genera. Two small species, estimated. to reach Im long, are represented by upper and lower jaw elements from System B (Archer et al., 1989, 1994) on Godthelp Hill. Some are associ- ated with vertebrae, but the two species cannot be distinguished unambiguously on vertebral char- acters. I include them in a single genus which possibly unnatural treatment allows generic iden- ulication of isolated vertebrae from other siles, This paper provides descriptions of the two species including some ontogenetic stages. While analysis of phylogeny of madtsolids awaits de- tailed comparisons with other primitive snakes, some functional and evolutionary points are nored by analogy with extant forms. MATERIALS AND METHODS Material is housed in the Queensland Muscum (QMF), Australian Museum (AMF), Northen Territory Museum of Arts and Sciences (NTMP), Museum of Victoria (NMVP), and South Austra- lian Museum (SAMP). ...(SMNR) specimens €x- amined in Paris by courtesy of J.-C. Rage, Teeth or alveoli are numbered beginning from the anterior on Complete jaw elements; on frag- ments where the tooth row is or may be incom- plete anteriorly the numbers are spelled out in words, In illusirating cranial bones, views of the same specimen are usually arranged parallel to each other, in lateral, dorsal, medial, ventral us- pects. Figures of vertebrae have left lateral, ante- rior, posterior, dorsal, and ventral views of cath elementin 4 vertical row. If more than one verte- bra are shown in an illustration, they ate arranged (tòr) in order from anterior to posterior, SYSTEMATICS Family MADTSOIIDAE Hoffstetter, 1961 Nanowana gen. nov. TYPE SPECIES. Nanowana godthelpi sp. nov. OTHER SPECIES. Nanowana sehrenki sp. nov. ETYMOLOGY. Greck nanos, a dwarf and Warlbiri (Tanami Desert, central NT) Wana, Rainbow Serpent of Aboriginal mythology, DIAGNOSIS, Small, upto 1.5m long; neural spine low to moderately high, not extending close to anterior edge of zygosphene; zygosphene shal- low, with anterodorsal edge straight, slightly con- vex or concave in dorsal view; subcentral ridges well-defined, straight or slightly concave or con- vex in ventral view; haemal keel relatively nar- row, with ‘paired hypapophyses’ in posterior trunk defined laterally, but not projecting vën- 394 MEMOIRS OF THE QUEENSLAND MUSEUM C FIG. I. Nanowana godthelpi sp. nov., QMF31379, holotype, upper jaw bones (right (A) and left(B) maxillae, palatines (C,D) and pterygoids (E,F) of a single individual) in palatal view, CS Site. Scale bar=5mm. dorsal process with steep an- terior edge; dentary with at least 2 mental foramina. COMPARISON. This genus is distinguished from all madtsoiids other than Al- amitophis by the zygo- sphene in dorsal view frequently (but not always) having a convex anterior margin; the convexity is broad rather than a distinct median tubercle as in Al- amitophis. It is distinguis- hed from = Yurlunggur, Wonambi, Rionegrophis, Gigantophis and Madtsoia by being smaller. Its neural spines are lower, at corre- sponding positions in the trunk, than in Madtsoia, Rionegrophis, Wonambi and Alamirophis, but higher than Patagoniophis or Giganto- phis. Itis distinguished from all genera except Patagoni- ophis by the less steeply converging subcentral ridges (relatively more elon- gate centrum in ventral view). Maxillae resemble Madtsoia sp. (SMNR 2879, Itaborai) and are distin- guished from Wonambi and Yurlunggur by prefrontal process having a steep ante- rior edge; distinguished from each of these by devel- opment of the septomaxill- ary process (condition unknown in other madtsoiids). DISCUSSION. Vertebrae can be distinguished from other madtsoiids, but their common features (including small size) may be sym- plesiomorphic; the concept trally or separated by a median concavity; neural of Nanowana containing these 2 species can be arch in posterior trunk depressed, its lateral por- described as a ‘marriage of convenience’. The tions strongly concave dorsad. Anterior tip of phylogenetic relationships of these with other maxilla with medial expansion (septomaxillary madtsoiids remain unknown, but they are treated process) contributing to floor of narial chamber; as a unit because their vertebrae (which provide NEW MIOCENE MADTSOIID SNAKE FROM RIVERSLEIGH 395 the only taxonomically use- ful material in most depos- its) are unable to be distinguished in many cases. In a number of aspects of the vertebrae, including size, Nanowana is comparable to Patagoniophis sp. cf. P. parvus from the early Eocene Tingamarra Local Fauna (Scanlon 1993), dif- ferences include the higher neural spine (in adults), nar- rower haemal keel in the posterior trunk, frequently convex anterior edge of the zygosphene, and dorsolat- eral concavities of the poste- rior neural arch, It differs from Alamitophis, which also occurs in the Australian Eocene (Scanlon, 1993): the anterior edge of the zygosphene, when convex, is broadly so rather than forming a distinct promi- nence; paradiapophyses do not project anteriad; zyg- apophyses are more steeply inclined at equivalent posi- tions within the column. The lower neural spine, broader zygosphene, and features of the haemal keel or hypapophyses differentiate Nanowana from Wonambi Smith, (1976) (Wonambi is known from Riversleigh, much smaller than W. nar- acoortensis but larger than Nanowana; Scanlon, 1996), The only other known Australian madtsoiid is Yurlunggur, at least 2 spe- cies of which occur at Eea pE ti tae ie FIG. 2. Nanowana godthelpi sp. nov. QMF31379, holotype, upper jaw bones 4 spg . &, NaNOwaNE i - : s Pia Aria reak (left maxilla (A-C), left palatine (D-F) and right pterygoid (G-I)) in lateral, Scanlon 1992). That genus dorsal and medial views, CS Site. Scale=5mm. exceeded 5m and thus in- cluded only ‘giant’ snakes, though not as largeas lengths from under 1m to over 7m), and need not Gigantophis garstini or Madtsoia bai, However, be considered an essential part of the diagnosis. size is rather variable in many snake genera (e.g. The vertebrae of small and large forms are rather the pythonid Morelia, sensu Underwood & Stim- Similar except in features which may be directly son, 1990, includes species with maximum related to size (neural spine height is variable 396 within Yurlunggur, and is proportionally similar to Nanowana in some), but Nanowana differs from Yurlunggur in the shape of the zygosphene, and the haemal keel of posterior trunk vertebrae being narrower and lacking a median concavity. Comparisons with non-Australian forms do not suggest any links closer than that with Yurlun- ggur, and will not be pursued here. The rib-heads of Nanowana have not been considered in detail, but appear to be similar in shape to those of Yurlunggur and Wonambi (Scanlon, 1993). Nanowana godthelpi sp. nov, (Figs 1-8, Table 1) ETYMOLOGY. For Henk Godthelp, University of New South Wales, in recognition of his contributions to Australian palaeontology. MATERIAL. Holotype QMF31379, associated ele- ments of a single individual comprising partial to com- plete maxillae, palatines and pterygoids of both sides. Paratypes QMF 31383, 31384 associated dentaries and compounds of a single individual; dentaries QMF20892, 23052, 23053, 23054, 23056); maxillae QMF31380, 31382, 31386, 31387; palatine QMF31381; pterygoids QMF23058, 31393. All types from early Miocene (System B) Came! Sputum Site, Godthelp Hill. Other material: Camel Sputum Site, trunk vertebrae QMF19741. Upper Site, dentary QMF31389; palatine QMF23066; maxilla fragment QMF31390; pterygoids QMF23067, 31385; series of cloacal vertebrae. Mike’s Menagerie Site, anterior fragmentof pterygoid QMF19742, Creaser's Ramparts Site, dentary QMF23076. DIAGNOSIS. Palatine lateral process about as long as two alveoli (nearest to 4th and 5th), ven- tral concavity of process with obtuse angle ac- commodating posterolateral angle of palatine process of maxilla. Maxilla with 23 tooth posi- tions, palatine 11, pterygoid 9, dentary 16. Teeth not ankylosed to alveoli; maxillary alveoli vary only slightly in size, dentary alveoli largest in centre of tooth row (4-8 or 5-8). Posterior part of maxilla strongly depressed. Dentary tooth row curved in dorsal view, Two or 3 mental foramina, all anterior to the 7th alveolus. DESCRIPTION OF HOLOTYPE. Upper dentig- erous elements in a single block (without verte- brae or other elements) are complete on one or both sides, missing bilaterally only the posterior (quadrate) processes of the pterygoids (Figs 1,2). Maxillae long and flat posteriorly, supporting a high lizard-like prefrontal process anteriorly; pal- atines with ‘alethinophidian’ features; pterygoids MEMOIRS OF THE QUEENSLAND MUSEUM with prominent, also lizard-like, ectopterygoid processes. Proportions of jaws indicating a rela- tively long postorbital skull and moderately short, rounded snout. Palatine: Left more complete than right, both well-preserved. Eleven alveoli forming a sigmoid tooth row, convex laterad anterior to an inflection and lateral concavity (slight, but definite and angular) between 7th and 8th. Dorsolateral crest arising above 3rd alveolus, bifurcating above 4th to form anterior edges of maxillary and choanal processes. Maxillary process with an oblique an- terior edge (near 45° from sagittal plane), longi- tudinal lateral edge and transverse posterior crest on its ventral face, level with the Sth alveolus on the left palatine (4th-5th on right side); process not perforated or notched for the maxillary nerve. Anterior edge of the choanal process smoothly concave anteriad for its full width, reaching be- tween level of 4th and Sth alveoli; then curving strongly anteroventrally, extending to front of 2nd alveolus. Vertical anteromedial part of the choanal process bilobed anteriorly, a dorsal lobe curved mediad, the other laterad (forming articu- lations with the parasphenoid and vomer); third, posterolaterally pointed, lobe on the ventral edge deflected laterad, contributing (along with the vomer and ectochoanal cartilage, presumably) to the floor of the choanal passage. Lamina of choa- nal process strongly arched anteriorly, flatter pos- teriorly, and ventrally deflected part of lamina reducing in depth posteriorly. Posteromedial cor- ner of process level with rear of 9th alveolus, posterior margin sinuous so that posterior process not sharply demarcated (as in some specimens); margin concave medially, convex posteriorly. Posterior extremities of choanal process and tooth row extending back level with each other, both with lateral margin parallel to tooth row, and separated by a distinct triangular notch extending forward to middle of 11th alveolus (thus, poste- rior edge W-shaped); on dorsal face this notch continued as a tapering trough extending to rear of 9th; ventrally a step-like groove running from the apex of the notch anteromediad to between 9th and 10th, with a shallow trough posterior and partly medial to the groove. Small foramen dorsomedially on the dentigerous process, just below the ridge continuous with the anterior edge of the choanal plate; a large foramen medial to the 8th alveolus, piercing the plate and emerging dorsally as a posteriorly widening foramen be- tween 8th and 9th; another small foramen an- teromedial to 10th alveolus. Dorsomedially on the anterior dentigerous process with tip of a tooth NEW MIOCENE MADTSOIID SNAKE FROM RIVERSLEIGH 397 emerging from the bone (this is the only tooth associ- ated with jaws of this spe- cies). Right and left palatines al- most identical; spacing of al- veoli slightly different on different sides; alveoli 2-5 in the right shifted posteriorly, relative to the lefi (alveoli 1 and 2 on the left, 5 and 6 on the right, confluent). Lateral (maxillary) process with small but distinct angular concavity marking the lon- gitudinal (lateral) and oblique (anterolateral) sec- tions of the margin. Pterygoid. Nine alveoli (complete row), anterior tip (length of approximately 1,5 alveoli) edentulous, Tooth row curving medially poste- riorly, following inner edge of bone; ventral face nar- rowing to a point anterior to tooth row, pointinterlocking with posterior notch of pala- tine. Dorsal surface forming a longitudinal trough, with foramen above Ist alveolus (opening anteriad), lateral to a dorsomedial ridge. Lateral margin smoothly convex, diverging gradually from tooth row; anterior edge of ectoplerygoid process di- verging at about 120° from this margin, level with 7th alveolus, Process nearly as wide as rest of bone at this point, about as long as wide; its anterior and lateral edges at 90° in dorsal or ventral view, lateral margin inclined strongly posteroventrally, with posterior extremity produced as a knob-like ex- tension, and posterior edge strongly concave. No part of the ectopterygoid facet ex- posed dorsally. Concave posterior surface of the pro- FIG. 3, Nanowana godthelpi sp. nov. QMF31383, 31384, paratype lower jaws. A, B, left dentary in lateral and medial view (upper posterior process broken and slightly displaced). C, D, E, right dentary in medial, dorsal, and lateral view. Scale=Smm, cess continuous with the ventrolateral face of the medially by a narrow extension of the ventral posterior lamina (quadrate process), bounded (occlusal) surface. Quadrate process broken off 398 MEMOIRS OF THE QUEENSLAND MUSEUM TABLE 1. Measurements (mm) of jaws of Namowana godthelpi sp. nov. C1, C2, etc.=single individuals; L=left, R=right. Alveoli were selected as landmarks for some measurements because they could be identified in fragments, but there is variation in the position of anterior alveoli (even between sides of an individual). Values in brackets are minima for measurements affected by damage. Palatine (ventral view): ptl=length of palatine from anterior tip of dentigerous process to posterior tip of tooth row spine or choanal process; pcl=base length of choanal process from intersection of anterior edge with dentigerous process to apex of posterior notch; pl 1 1=length from anterior tip to anterior edge of 11th alveolus; ptw=width across choanal and maxillary processes; pew=width in same line of choanal process; prw=width in same line of tooth row bar; pnw=width in same line of maxillary process. Pterygoid (ventral view): ttl=length from anterior spine (in plane of alveoli, not dorsal lappets) to rear of 9th alveolus; trl=tooth row |st-9th alveolus; tte=from anterior spine to furthest point of ectopterygoid process; tI5= length across most posterior 5 alveoli (5-9); taw=width between near-parallel edges anterior to ectopterygoid process; tpw=width from basipterygoid facet to intersection of ectopterygoid process and dorsolateral edge of posterior lamina; ttw=width from basipterygoid facet to furthest point of ectopterygoid process. Maxilla: mtl=length; map=length from anterior tip to posteromedial angle of palatine process; m 712=length from anterior edgeS of 7th-13th alveolus; mpw=width across palatine process; mph=depth at prefrontal process. Dentary: mff=number of mental foramina; dtl=straight-line length; dl 15=length to anterior tip of 15th alveolus; dif=length to lateral fossa; d4t= posterior edge of 4th alveolus to posterior extremity; d4 15=posterior edge of 4th to anterior edge of 15th; d4f= posterior edge of 4th to lateral fossa; dl 7=anterior tip to anterior edge of 7th alveolus; dmd=depth from dorsolateral to ventromedial edge in middle part of bone; dpp=depth of upper 23066 UIL 31393 C4 R | | | | | pel 4.5 ptw 4.0 | pew 1.9 2.0 orw 0.9 m 1 oe aa ao o9 | aera ag T 25 n tpw 2.4 2.4 z : 42 41 p (3.9) mtl - (16.3) - - maj - 7.4 - - m712 - 5.4 - - li $ 5.2 mpw 23 i (2.8) | 28 2.8 s mph 3 3.9 3.6 3.3 3 QMF | 31383 23056 | 31389 | 2 Ind. C4R | UIL mff dtl dlis NEW MIOCENE MADTSOILD SNAKE FROM RIVERSLEIGH posteriorly about half the length of the tooth row hehind the ectopterygoid process. Basipterygoid wcular surlace opposite ectopterygoid process, an oval facel facing dorsally and slightly medi- ally, beginning level with front of 8th alveolus und extending to beyond 9th, only slightly dis- tinct in outline from the rest of the medial edge. Apart from the anterior foramen mentioned ubove, 3 foramina dorsally, anterior, lateral and posiener to the facet; anterior 2 near the midline ofthe bone, posterior foramen close to the medial edge. A shallow but distingit transverse groove on the dorsal surface of the ectopterygoid process, Left plerypoid retaining posterior ¥ alveoli. which are slightly smaller and more closely Spaced than on the right: possibly a 10th alveolus or longer edentulous gap anteriorly. Maxilla. Alveoli 23, varying only sligmly in size; row curved medially antenorly, straight pus- teriorly, Anterior alveoli elongate anterolateral- posteromedially; anterior of maxilla wider thon deep, with dorsomedial edge forming a crest above Ist-3rd alveoli, with slight coneayitics dor- sal and medial to it. In lateral view, ventral margin slightly convex up ta lOth alveolus, nearly straight posteriorly; dorsal edge rising smoothly und ineréasingly steeply from the anterior tip to between 6th and 7th alveoli: highest part of the dorsal process. (7th to 9th) forming the dorsomed- ial surface for articulation with the prefrontal, On the posterior slope of the process, a low promi- nence above the 11th alveolus probably the inser- tion site for the postorbital ligament, but may also mark the anterior extent of the jugal; by 13th bone very shallow, continuing so to the posterior ex- tremity, Large lateral (trigeminal) foramen open- ing anteriorly above the Sth-6th alveoli; two smaller foramina, equally close to ventral edge, above 7th-8th and 9th-10th, and 3 small foramina higher on the prefrontal process, Medial edge forming a shelf-like ‘septomaxillary’ process from 2nd to 4th alveolus, separated trom the palatine process which widens gradually from 7th and then sharply at 10th, then gradually ap- proaches maximum width at a sharply obtuse posteromedial angle between | Lh and 12th, Me- dial shelf narrowing steeply from this point, then very gradually, but with a step-like inflexion at level of 18th alveolus (marking location of ante- rior ip of ectopterygoid), Large foramen entering maxilla at broadest part of the palatine process. ubove 11ih alveolus, and a smaller foramen exils at the same level above the 7th. Tooth row fol- luwing lateral margin closely tony stl thalys- oli, then gradually crossing Over with | hrc closer to medial edge; lateral edge forming a low dorsolateral crest (possibly homologous with More prominent crests or bulges in snakes such as Dinilysiu and pythons), Lateral as well as medial parts of posterior maxilla apparently over- lapped by the ectoperygoid, forming slight cun- cavities on either side of a slight dorsal crest. Between ectopterygoid facet and prefrontal pru- cess, the suborbital surface with a shallow longi tudinal groove which probably either was, or bounded, a facet for the juga! (an element lust in extant snakes bul probably retained in Dinilysia and madtsoiids, including Wonanbi;, Estes ct ala 1970, Scanlon, 1996), PARATYPES. Right and left mandibles (QMF3 1383, 31384), each compound and den- lary, in loose articulation, lacking the splenial, angular and coronoid of each side (Figs 3, 4), Right: Tooth row incomplete posteriorly, bro ken through 15th alveolus; no sign of ankylosed teeth. 4th to 8th largest alveoli, subequal, size reducing posteriorly and anteriorly, In lateral view, dorsal edge convex dorsad from 1st ta Sth alveolus, concave dorsad for rest of length, Ven- tral edge slightly concave anteriorly, remainder convex but samewhat worn, Three mental foram- ina, below 3rd, 41h and 6th alveoli, opening an- terodorsad. Posterior lateral fossa (compound notch) extending to between 10th and | 1th. Låt- eral face smooth hut with dorsolateral ridge de- fined by slight longitudinal concavity through feraminit. In dorsal view, tooth row concave tne- diad, slightly more so anteriorly; alveoli round or squarish except first two which are somewhat elongate transversely. 15th alveolus on a narrow process distinguished by an angular concavity from the expanded dorsomedial shelf. The medial ridge forming the upper edge of Meckel’s groove overhanging the groove distinctly up to the &th alveolus; the overhanging edge of the upper facet for the splenial beginning below the 8th bul more dorsally, forming with a slightly acute, pointed posteroventral process separated by a right-angle notch (in medial view) from the dorsal shelf. Meckelian groove narrowing anteriad, anterior end slightly expanded, communicating by a fora- men with alveolus of Ist tooth. Smooth bulb-like swelling overhanging the groove medial to the 1st and 2nd alveoli, Left: Two mental foramina, between 3rd and 4th, and 5th and 6th. Posterior lateral fossa ex- tending to between | 1th and !2th alveoli. Right compound. Elongate, shallow, 18.8mm long, 16.8mm from anterior tip to dorsal extrem- 400 ity of articular facet. Sur- angular lamina low but con- cave above, forming low coronoid process posterior to articulation with dentary, about 1/3 of length from an- terior tip. Maximum depth of compound Jess than depth of dentary at articulation (suggesting that the coronoid extended dorsal to compound, forming most of the coronoid process by it- self). Ventral edge, and lat- eral in dorsal view, nearly straight, but posterior end (below articular facet and retroarticular process) de- flected slightly ventrad and mediad from main shaft. Ar- ticular facet dorsal and me- dial in position, not extending to lateral face, reaching to middle of medial face, and as far anteriad as ventrad from dorsal extrem- ity; facet defined posteriorly by a raised transverse lip, followed by a groove ante- rior to the sigmoid dorsal edge of the retroarticular process. Slight ventrolateral and deeper ventromedial concavities defining a ven- tral ridge on the retroarticu- lar process. Shaft of compound nearly cylindri- cal just anterior to articular facet; a small dorsolateral foramen in this region. Man- dibular fossa narrow, begin- ning posteriorly at level of foramen, curved slightly mediad, and extending to half way between posterior edge of coronoid facet and top of coronoid process. Fossa partly surrounded by the facet for the coronoid anteriorly; anterior half opening below into mandibular foramen, Surangular lamina curved, overhanging the mandibular fossa for most of its length; reducing in height anterior to coronoid process in two steps, reaching a hori- zontal or somewhat dorsally concave shelf re- ceiving the posterior part of the dentary; lateral MEMOIRS OF THE QUEENSLAND MUSEUM FIG, 4. Nanowana godthelpi sp, nov., QMF31383, 31384, paratypes, compound lower jaw bones, CS Site. A-C, left compound in lateral, dorsolateral, and medial views (note missing articular), D-F, right compound in medial, dorsal, and dorsolateral views. Scale=Smm. edge expanded anterodorsally, for anterior 1/3 of length anterior to the coronoid process. Surangu- lar foramen, opening anteriad, not exposed later- ally or medially, in a shallow dorsal trough between lowest point of surangular lamina and edge of coronoid facet. Facets for coronoid and angular meeting at a very small angle below this point; their line of contact nearly horizontal, only NEW MIOCENE MADTSOIID SNAKE FROM RIVERSLEIGH FIG. 5. Nanowana godthelpi sp, nov., paratype, maxillae, CS Site, A-D, QMF31386, in ventral, medial, dorsal, and lateral views. E-H, QMF31380, in ventral, medial, dorsal, and lateral views, Scale=Smm. a short section preserved on either side, In lateral view the anterior edge of the compound rounded dorsally, separated by aright angle from a deeper ventral concavity. In medial view. a long, taper- ing notch enclosed in the facet for the angular, nearly reaching its posterior end (just posterior to middle of length of compound). A medial anterior process (defined by dorsal and ventral longitudi- nal fissures) bearing the continuation of facets for the coronoid and angular, and probably also con- 401 lacting the splenial and den- tary, broken on both sides. Left compound similar to the right, but broken posteri- orly through the articular facet. When placed in articula- tion, the right compound and dentary forming a smoothly curved structure, with total Straightline length approxi- mately 29.5mm. Other paratypes and re- ferred jaw elements (partial dentaries, maxillae, pala- tines, pterygoids) show some individual variation (Figs 5, 6) and probably on- togenetic changes of propor- tions (allometry): the smallest dentary, QMF 23056, is relatively deeper than. larger specimens (Table 1), while the largest, QMF23076, is relatively slender except for a particu- larly deep upper posterior process. Vertebrae. In shape and proportions, vertebrae sim- ilar to, and intermediate be- tween Yurlunggur and Patagoniophis and differ conspicuously from AL amitophis, Wonambi and Madtsoia. Typical anterior, middle and posterior trunk vertebrae recognised (cf. LaDuke, 1991, Scanlon, 1992, 1993): most anterior vertebra possibly 3rd cervi- cal (cf. Y. cantfieldensis Scanlon, 1992, fig. LA). Centrum in ventral view rel- atively long, similar in pro- portions to Patagoniophis sp. but with the sub- central ridges nearly straight rather than strongly concave. Cotyle slightly wider than the zygosphene, which ts wider than the neural canal (all about equal in the most anterior vertebra); condyJe and cotyle wider than deep, ventral mar- gins flattened in anterior and middle trunk, rounder posteriorly. Zygapophyseal facets inclined at about 20° from the horizontal (at mid-trunk; fatter anteri- 402 orly. slightly steeper posteri- orly), defining planes pass- ing through the internal lateral ridges of the neural canal and intersecling just above its base. Facets broader and more angular in outline (especially the pre- zygapophyses) in the largest midirunk vertebrae, with long axes inclined at about 45° from the sagittal plune (somewhat more longitudi- nal in Most anterior and pos- terior elements). Prezygapo physeal accessory processes lacking, outer face of the prezygapophysis with a but- tress-like ridge extending anterolaterally to or slightly beyond the edge of the facet. Zygosphene shallower than the neural canal, with = facets defining planes inter- / secting below the floor of the canal: dorsal edge in an- terior view flat, slightly arched or arcuate; below it are shallow concavilies de- fining a dorsal ridge and lat- eral lobes, with sharp ridge : separating the anterior face ler of the zygosphene from the MEMOIRS OF THE QUEENSLAND MUSEUM \tetne ens = Ej \ = yy ae internal roof of the neural FIG. 6. Nanowana godthelpi sp. nov., referred elements from Upper Site canal. In dorsal view the an- teriorly convex dorsal ridge and lateral lobes distinct in mid-trunk vertebrae, but in the most anterior and posterior elements median prominence less developed and zygosphene broadly concave. Paradiapophyses similar to Yurlunggur or Patagoniophis, extending laterally beyond the zygzapophyses only in the most anterior and most posterior vertebrae. Roof of zygantrum horizontal, either uniform in depth or thickening laterally, demarcated from the concave lateral parts of the neural arch by angular ‘shoulders’, with concavity directed more dorsally than laterally in the most posterior vertebrae because of the shallower neural arch and steeper postzygapophyses. One or two small paracotylar foramina on ei- ther side of the cotyle, usually 2 lateral foramina on either side posterior to the diapophyses. Sub- possibly from a single individual. A-D, left palatine, QMF23066 in lateral, dorsal, dorsomedial, and ventral views. E-H, right pterygoid, QMF23067, in ventral, lateral, dorsal and medial views, Scale=Smm. central foramina usually single on each side, small. Parazygantral and zyganiral foramina larger, usually single on each side, frequently in distinct fossae, Some vertebrae with small foram- ina on the anterior face of the prezygapophysis below the facet. Ventral face of centrum concave between the haemal keel and subcentral ridges. In the anterior trunk hypapophysis projecting well below cen- trum from its posterior half, with either an angular or sinuous anteroventral edge, and near- vertical posterior edge; in more posterior vertebrae the keel weakly sinuous to nearly straight in lateral profile. Haemal keel with median, keel-like hypapophysis reducing in depth from the cervical to mid-trunk regions; lateral ridges on the keel (initially just posterior to the subcentral foram- NEW MIOCENE MADTSOIID SNAKE FROM RIVERSLEIGH 403 gon > Ey Sau, Ra T ye Ay a a ee A) im oe eons © ig -A 4 \ 2 i LA yl ak JAIAN, i i ( ZA 3 CAD Em MMR Sie im A W P ‘i A if ; f ao Ry Cae Cia FIG. 7. Nanowana godthelpi sp. nov., QMF19741, series of vertebrae from CS Site, possibly from the same individual as the holotype (QMF31379). ina) from the approximate location of the largest vertebrae in the skeleton, ridges increasing in size in more posterior vertebrae and posterior point of the median keel fading away, leaving the ridges as paired hypapophyses, ventrolateral swellings of the keel. Haemal keel defined by smooth de- pressions in the anterior trunk, these becoming better defined more posteriorly and approaching the cotylar rim. More posterior vertebrae with distinct channels between keel and subcentral ridges (subcentral paramedian lymphatic fossae, LaDuke, 1991). Most vertebrae from all regions of the body with swellings on the neural arch roof on either side of the spine, forming short longitudinal ridges. Similar features in some Wenambi from Riversleigh are associated with smal! foramina (not the case here), Vertebrae similar to these and referred to Nanowana sp. (most of them probably N. godthelpi) from numerous sites at Riversleigh, including well-preserved examples from Wayne’s Wok, Wayne's Wok 2, Mike’s Me- nagerie, and Upper Site. Vertebrae of the cloacal region (Fig. 8) proba- bly from a single individual with short centrum, broad zygosphene, and condyle smaller than neu- ral canal (regional features allowing increased flexibility in this region), Haemal keel smooth (lacking the median ridge of Wonambi spp.), not or barely projecting below the centrum posteri- orly. Two largest vertebrae with paradiapophyses indicating articulated ribs, but on one side of one of them the articular surface is expanded and roughened suggesting an immobile cartilaginous attachment (i.e. transitional to fixed lymph- apophyses), Three others with lymphapophyses (broken distally); another with stumps of cylin- drical fixed ribs, possibly forking more distally. Nanowana schrenki sp. nov. (Figs 9-12, Table 2) MATERIAL. Holotype QMF31395, a right palatine from early Miocene Upper Site, Godthelp Hill. Other Material: Upper Site: Maxilla fragments QMF 31390, 31391, 31394. Mike's Menagerie Site: Dentary QMPF31392 and vertebra QMF23043. Camel Sputum Site: Dentary QMF23051; maxilla fragments QMF23082, 31388. all TABLE 2. Measurements of ETYMOLOGY. For Nanowana schrenki šp. nov., Friedemann Schrenk, holotype and referred jawel- He ssisches ements. Abbreviations as in Landesmuseum, Table |, with addition of dd8 Darmstadt, for his en- inane al&thalve- couragement and fi- olus. nancial assistance for palaeontological co- operdilion helween mofu utu Germany and Aus- [a7 | tralta. DIAGNOSIS. Lit- eral process of pala- line about as long as Pew Dia 4 alveoli (3-6), with dorsolateral margin strongly notched; ventral ridge of pal- atine maxillary pro- cess without distinct angular concavity, matching smooth edge of maxillary palaline process. Maxilla estimated to have about 19 tooth | positions; palatine with 11, pterygoid unknown, dentary 18 (or 17-18). Teeth ankylosed normally; 2nd to 4th of den- | tary, and 4th to 7th Jor 8th of maxilla, | much larger than others. Dentary [tooth row nearly straight in dorsal view. Three mental foramina, the third posterior to the 7th tooth. DESCRIPTION. Holotype. Alveoli 1L, teeth an- kylosed in 1, 3,4, 5,6, 8.9. 10, Llp only 8, 10 and 11 complete. Teeth with a simple curve, directed posteriorly. Tooth row deflected slightly medi- ally anteriorly, laterally posteriorly. Maxillary process slightly wider than the tooth-bearing bar, extending from between 2nd and 3rd to between 6th and 7th teeth, with an anteriorly sharp lateral notch, and sharp posterolateral angle, Ventral surface of the process with a diagonal ridge from the rear of the 4th tooth to the posterolateral angle, defining an anterolaterally concave facet to articulate with the palatal process of the max- illa. Anteriorly, the edge of the lateral process MEMOIRS OF THE QUEENSLAND MUSEUM continuous with a dorsolateral ridge extending to the antenor tip of the tooth-bearing process, A second ridge diverging medially from the antera- lateral comer of the process, bearing, a distinct knob above the tooth row and continuing onto the anterior edge of the choanal process, level with the rear of the 4th alveolus. Anteromedial comer of choanal process (to articulate with posterior process of vomer and possibly parasphenoid) missing. Medial edge intact, and smoothly con- vex, from level of 7th alveolus to rear of tooth row, but posterior process broken off. Cusp de- fining lateral edge of choanal trough diverging posteromedially from the 6th tooth, disappearing level with the 8th; 2 foramina close together in the space between and medial to 7th and 8th alveoli, one of them piercing the choanal plate to emerge dorsally in a more medial position and opening medially. Tooth-bearing bar pointed posteriorly, tapering from the 9th tooth, a broad parabolic surface for the retractor pterygoidei on the ventral face with its apex beside the 9th, becoming less distinct posterolaterally, Deep notch to articulate with the pterygoid on the dor- sal side between the tooth row and postenor process, extending to above the anterior edge of the 10th tooth. Distinct growth lines through the translucent choanal plate parallel to its curved medial edge. Referred material. Maxilla represented by sev- eral fragmentary specimens from different sized individuals (Fig. 10). Tooth row curves mediad anteriorly (QMF23082), with a strong gradient of increasing alveolar diameter from 1 to 5; 5 and 6 subequal. Dorsal edge is a sharp, concave dorsomedial crest, extending to a high dorsal process, levelling off above 6th alveolus; this crest divides anteriorly, enclosing a shallow trough above the first two alveoli (thus, maxilla partially flooring narial cavity). Lateral face mostly convex, with a shallow longitudinal trough including a large foramen (opening an- teriad and yentrad) above rear of the 4th tooth; a smaller foramen near the dorsal edge above the Sth. Medial face concave, with a trough just below the dorsomedial ridge containing a small foramen just anterior to the medial one. Middle part of maxilla (QMF3 1394) with distinct knob- like posterior part of prefrontal process and slop- ing suborbital portion, becoming more rod-like and wider than high posteriorly. Tooth size de- creasing sharply, with increased alveolar spacing, just behind prefrontal process; longest (7th or 8th?) 2.2mm long, curved at middle but straight NEW MIOCENE MADTSOLID SNAKE FROM RIVERSLEIGH distally, with medial and lat- eral cutting ridges (like longest tooth of dentary QMF31392, see below); more posterior teeth (broken before drawing) with simple curve, about half as long. Palatine process diverging /% from tooth row at last large © iooth and reaching maxi- mum width between the next 2 alveoli. Medial edge of jhe palatine process quite smooth, matching the con- cavity of the maxillary pro- cess in the holotype; large opening on dorsal face of process for palatine nerve and blood supply through several foramina on lateral surface. Teeth on posterior part of maxilla (QMF31391) still reducing in size from anterior to posterior, and X ar with slight double curve. / “a * Posterior part triangular in | section, with near vertical lateral and oblique dorsomedial faces both slightly concave, meeting at a dorsolateral ridge. Lateral CEER FE hee BUA 4ns edge straight, medial edge FIG. 8, Nanowana godrhelpi sp. nov., series of most posterior trunk and cloacal produced as ridge with con- vexity probably marking an- terior limit of ectopterygoid, Dentaries. Two right dentaries, differing con- siderably in size (Fig. 11), represent the lowerjaw in this species. QMF31392 with complete row of 18 alveoli, teeth ankylosed in 1 (possibly), 3, 6, 8. 10, 11, 13, 15, 16, and 18; 10th broken, other teeth in tact, and a replacement tooth apparently in situ behind 15th. QMF23051 has 17 alveoli, but another may have been present posteriorly; 1, 4,5, 6.7.9, 11, 12, 13, 14. and 15 ankylosed, but ull teeth broken near base (the jaw has also been broken through 3rd alveolus and subsequently healed in life), Ist alveolus approximately same size as Sth, but 2nd to 4th considerably enlarged; 3rd nearly twice diameter of Sth, size decreasing gradually more posteriorly; in the small speci- men, lengths of teeth from anterior edge of base to tip (mm) -, -, 1.26, - -, 0.61, -, 0.63, -, -, 0.55, -, 0.52, -, 0.40, -, 0,37, 0.28. Anterior alveoli (1-3) deflected venirad and mediad relative to rest of woth row, which is moderately concave dorsad yertebrae from Upper Site, possibly from the same individual as jaw elements in Fig. 6. Lateral, posterior, dorsal, and ventral views, but only very weakly concave mediad, Third tooth directed slightly laterad as. well as poste- riad; other teeth mediad, more strongly towards the rear of the tooth row. Each tooth with a weak lateral and medial cutting edge near the tip. Den- tary deepens gradually from anterior to posterior. Three menta! foramina open anteriad below alve- oli 4, 7 and 9 (QMF31392) or 3, 6-7 and & (QMF23051), decreasing in size posteriorly. A shallow dorsal trough medial to 3rd and 4th alve- oli defined by a dorsomedial crest. Lateral fossa extends as fat anteriorly as the rear of the 13th tooth, blunt in outline; posterior edge of the ver- tical Intramandibular septum smoothly concave, extending forward to between the 14th and | Sth teeth. Differences between the two include shape of Meckel’s groove (tapering more strongly in the small jaw, dorsal edge composed of two sharply defined sections separated by a short gap below Sth-9th alveoli, but no gap in the larger speci- 406 FIG, 9. Nanowana schrenki sp. nov., holotype, QMF31395 from Upper Site, palatine in ventral (A), dorsal (slightly lateral) (B), dorsomedial (C), and lateral (D) views, Scale bar=2mm. men), upper facet for splenial (with posteromed- ial angle in the smaller, buta free-ending process in the larger), and lateral fossa (constricted jn the smaller by deepened upper posterior process); both narrowest in the region of the 6th to 8th alveoli; but the larger specimen is relatively broader posteriorly. Vertebra (Fig. 12) from mid-trunk ofa juvenile, with short broad centrum, large neural canal, and condyle and cotyle much wider than deep. Weakly defined subcentral ridges narrow only slightly behind the parapophyses, posterior half of centrum nearly parallel-sided except for a shal- MEMOIRS OF THE QUEENSLAND MUSEUM low, short precondylar constriction. Blunt haemal keel extending from just behind the cotylar rim, posteriorly forming a slightly prominent single hypapophysis extending below the condyle. Keel defined laterally by broad shallow depressions. Comparisons with Yurlunggur or Patagoniophis would imply that a haemal keel of this form indicates a vertebra from close to the cardiac region (transitional between prominent single hypapophysis anteriorly and flattened or double keel posteriorly), and would thus be among the largest in the skeleton. Condyle and cotyle about twice as wide as deep, slightly oblique in lateral view; cotyle wider than the neural canal bul not as wide as the zygosphene, Zygapophyseal facets inclined at less than 20° above the horizontal, defining planes which intersect near the middle of the neural canal. Prezygapophyseal facets ob- ovate, with transverse anterior edge; postzyg- apophyseal facets more smoothly oval, and somewhat prominent posteriorly in dorsal view. Both pairs of facets are elongate anteropost- eriorly, with long axes at about 35° to the sagittal plane (as in anterior, but not middle trunk verte- bra of Patagontophis sp. cf. P. parvus; Scanlon, 1993). No prezygapophyseal processes. Paradiapophyses directed ventrolaterad, slightly wider than prezygapophyses, not extend- ing ventral to colylar rim. Interzygapophyseal ridge smoothly concave laterally, only slightly wider than the centrum, and weakly defined in lateral view. Zygosphene thin, slightly arched; anterior edge smoothly but weakly concave (again, like ante- rior rather than middle vertebrae of Paragoni- ophis). Zygosphenal facet (preserved on left only) dorsoventrally shallow, with dorsally con- vex upper and lower edges, inclined at about 45° from vertical; a plane tangent to the facet would pass close to the centre of the neural canal, Neural canal arched, about as high as wide, lacking internal lateral ridges. Neural arch low, with shallow concavities above and below the level of the zygosphene and extending to the posterior edge. Zygantral roof arched, thickness uniform across its width, In dorsal view, rear of neural arch forming a broad concavity above the zygantrum, interrupted by the neural spine. Low neural spine formed by a narrow, but sharply defined anterior lamina rising from the rear of the zygosphene and applied to a higher, columnar portion posteriorly, overhanging the zygantrum. Dorsal surface of column broken off, with a sinus wisible within the neural arch. Lateral and sub- NEW MIOCENE MADTSOIID SNAKE FROM RIVERSLEIGH 407 centra] foramina present, any other obscured by den- drites, TROPHIC SPECIALISATIONS OF NANOWANA N. godthelpi sp. nay. The homogeneity in size, mor- phology and approximate stratigraphic position of these toothless but otherwise well-preserved jaws makes it appear probable that the lack of ankylosed teeth is a natural (and apomorphic) characteristic. To quote Owen’s (1840) conclusion on the ‘dislocated’ tail of ichthyosaurs, the toothless condition *... is too uniform and common to be due en- lirely to an accidental and extrinsic cause’. Variation in the shape and size of alve- oli along the tooth rows, and the presence of ‘frothy’ bone similar to bone of attach- ment in some cases, indi- cates that different stages of replacement are repre- sented, so that absence of teeth is not explained by synchronised replacement. Some of these specimens are practically intact, preserving delicate processes, and not worn in such a way as to account for the absence of even stumps of teeth; in most other specimens from the same deposits, parts of teeth are typically retained even after heavy wear. The alveoli are shallow, rather rectangular pits, so that a thecodont type of implanta- tion is not indicated as an alternative to ankylosis. Failure of teeth to anky- el ae J SoA FIG. 10. Nanowana schrenki sp. nov., maxillary fragments, A-C, QMF31394, Upper Site, middle part of right maxilla in lateral, dorsal, and medial views. D-G, QMF23082, CS Site, anterior right maxilla in lateral, dorsal, medial, and ventral view. H-J, QMF31391, Upper Site, posterior left maxilla in medial, ventral, and lateral views. Not to same scale. lose at any stage is rare among squamates, first living Alethinophidia; Cundall et al. 1993), ap- reported by Savitzky (1981). Anomochilus parently has fibrous tooth attachment rather than weberi, a small fossorial ‘anilioid’ (An- ankylosis (Cundall & Rossman, 1993). There are omochilidae 1s possibly the sister taxon to other also several lineages of snakes, and one genus of 408 A JE TT AEE A A ; io 3 HPT OCs e i et ee === i FIG. 11. Nanowana schrenki sp. nov.. nghi dentaries. A-C, QMF23051, CS Site, in medial, đorsal, and lateral views. D_F, QMF31392, MM Site, in medial, dorsal, and lateral views, Scale bars=2 mm. lizards, where the attachment i$ not only fibrous but forms a functional hinge allowing each tooth to fold posteriorly under pressure and return up- right when released (Savitzky, 198], 1983; Patchell & Shine 1986c; cf,Edmund, 1969:141). This hinge mechanism has been interpreted in each case as an adaptation to feeding on scincid or gerrhosaurid lizards in which the seales are underlain by osteoderms; the hinged teeth are MEMOIRS OF THE QUEENSLAND MUSEUM thought to act as a ratchet mechanism, folding back rather than penetrating the dermal armour, and locking in an upright position against the edges of the scales when the prey is oriented head-first for swallowing. In extant snakes other functionally associated apomorphies also occur: the teeth are small and numerous, often with a spatulate rather than conical tip, and lack enamel on the posterior surface; and the levator anguli ors muscle (inserting on a long upper posterior process of the dentary) is enlarged (Sayitzky. 1981), In the pygopodid Lialis teeth are of similar form, and instead of increased intramandibular kinesis there is pronounced kinetic ability at the frontoparietal joint (mesokinesis: Patchell & Shine 1986b). Both types of kinesis allow the jaws more effectively to surround and compress a cylindrical prey item, immobilising or even asphyxiating it An equivalent adaptation for prey-holding (without hinged teeth) is seen in the largely scincivorous bolyeriid snakes, in which the required kinesis is provided by the uniquely derived intramaxillary joint (Cundall & Irish, 1989), Savitzky (1983) described this set of adapta- tions to feeding on skinks, which has evolved independently in several lineages, as an instance of a ‘coadapted character complex’, among other cases of ‘durophagy” (feeding on hard-bodied prey). Other durophagous snakes have distinct specialisations, and feed on other kinds of ‘hard’ prey such as snails (pareine and dipsadine colu- brids) or crabs (the homalopsine Ferdonia). *‘Durophagy’ is thus a broad concept. I introduce arthrodonty’to refer specifically to the “hinge- toothed” mode of durophagy. While soft-tissue structures such as fibrous hinges cannot be observed in fossils, absence of ankylosis implies thatattachment was fibrous and potentially flexible. N, godthelp? jaw material is similar to that of extant arthrodont species alter maceration, especially Xenopelris (Savitzky, pers. comm.). Hutchinson (1992) demonstrated that scincid lizards were abundant and diverse in the Tertiary at Riversleigh; skinks today repre- sent a major food source for small terrestrial predators, including mostextant Australian snake species (Shine, 1991). As functional arthrodonty has evolved in several lineages in association with predation on skinks, its presence in N. godthelpi is a plausible explanation for the lack of ankylosis. N. gadthelpi appears to be less specialised than each of the extant arthrodont snake lineages in some respects. The high number of nearly uni- NEW MIOCENE MADTSOUD SNAKE FROM RIVERSLEIGH FIG, 12. Nanowana schrenki sp, noy,, vertebrae, QMF23043, MM Site, probably same juvenile as dlentary QMF3] 392 (Fig, 11) in anterior(A), poste- rior( B), lateral(C), dorsal(D) and ventral (E) views. torm maxillary alveoli is typical of an arthrodont species, but a similarly long Looth row is present in Wonambinaracoortensis (Barrie 1990), and is therefore likely to be a retained ancestral condi- tion rather than a specialisation. The overlap be- tween dentary and compound is moderate, without any great elongation of the tooth-bearing posterior process; the extreme condition in extant arthrodont lineages may be precluded by the probable insertion ofm. levator anguli oris on the relatively large coronoid rather than the dentary, but the overlap is actually shorter in this species than in Wenambi, The dentary alveoli of N, godthelpi are consid- erably larger, especially in the middle part of the row, than those of the maxilla, so the lower teeth may have functioned differently, and possibly Jacked a functional hinge. In Xenopelris, rela- lively large teeth are present on the middle part of the palate (posterior palatine and anterior pter- yeoid), bul these appear to be fully hinged. While fibrous attachment of ‘sessile’ teeth has been 409 reported only in one highly unusual extant taxon, Anomochilus, it is possibly a necessary precursor or incipient stage of arthrodonty (see below), and a mixed or ‘semi-arthrodont’ condition in N. godthelpi seems possible. Nanowana schrenki sp. nov. In the absence of articulated or strongly associated material, refer- ral of jaw elements described here to a single taxon can only be provisional. In particular, the 2 near-compleie dentaries differ in several respects which make their assignment to the same specics doubtful: in QMF23051 the upper edge ol the Meckelian groove is a continuous ndge and ex- tends posteriorly as a free-ending process, while in QMF31392 itis interrupted at the 9th alveolus, and appears to end abruptly, (Additionally, the larger specimen broadens more posteriorly, while the small one is widest at the 3rd tooth, but this difference may be allometric.) The teeth of snakes play several roles in the capture, subdual, puncturing or laceration, and swallowing of prey; in general they will be adupted for a combination of functions, but often either a single function is dominant, or certain stages are either not required (e.g. because inat- tive or defenceless prey is taken) or carried out extra-Orally (e.g. constriction). Teeth specialised for different functions are often separated be- tween the front and rear of the mouth, in some cases with diastemata between teeth of different morphology (Prazzetta, 1966; Scanlon & Shine, 1988; Cundall & Irish, 1989). Numerous terms have been introduced for dit- ferent patterns of tooth size and fang location (Smith 1952). Primitive snakes (Dinilysia, un- ilipids) are isodont or mesodont, with relatively few, stout teeth; while also capable of constric- tion, they use a powerful ‘crushing’ bite in suh- duing prey (Frazzetta, 1970; Greene, 1983). Such a ‘crushing’ method seems possible lor Madtsoia cf. M. bai, in whieh the dentary is heavily built and bears relatively few teeth (Hofstetter, 1960), but not for Australian madtsoiids, Different pat- terns of tooth-size variation in upper and lower jaws are known in each of the 4 best- represented taxa: In Wonambi naracoortensis the very numerous teeth (25 in the dentary, 22 or 23 in the maxilla) are proterodont, sharp and strongly inclined pos- teriorly and medially (Barrie, 1990); the jaws are shallow, suggesting a limited role in subduing prey, and more emphasis on holding and swal- lowing functions. This implies that an extra-oral method of subduing prey (probably constriction} was well-developed, When the upper and lower 410 jaws are both proterodont, teeth often have a sigmoid curvature with the tips directed some- what anteriorly as in many pythons (Frazzetta, 1966), and seems to be associated with relatively soft-bodied prey such as mammals, birds, earth- worms (McDowell, 1969) and eels (Smith, 1926; Cogger et al., 1987). Nanowana godthelpi apparently had a nearly isodont marginal dentition. No complete tooth crowns have been reported for this species, but based on alveolar sizes it was weakly proterodont on the maxilla and mesodont on the dentary (Figs 1, 3). The condition in Yurlunggur is less clear but apparently the opposite; a dentary with well-pre- served teeth (Archer et al., 1991:71) is proterod- ont, while the maxilla was apparently mesodont (Scanlon, 1996). N. schrenki can be described as megadont (Smith, 1952), having regions of distinctly en- larged teeth. Otherwise it has the same pattern of enlargement as Yurlunggur, opposite to that of N. godthelpi, being mesomegadont on the maxilla and promegadont on the dentary. The dentary is relatively longer and less robust than in Madtsoia or Dinilysia, but not depressed as in Wonambi; the teeth are intermediate in number and in mor- phology (stouter and more erect than Wonambi, but not so much as in Dinilysia or anilioids); and the enlarged teeth are a uniquely derived condi- tion within Madtsoiidae (albeit convergent with many other lineages of snakes). Many snakes share this pattern of enlarged teeth at the front of the dentary and the part of the maxilla below the prefrontal articulation, whether or not they are set off by diastemata or local minima of tooth size. On the basis of occur- rence in scincivorous colubroids such as Lycodon, Glyphodon, Demansia, and Hemiaspis signata (but not the anurophagous H. dameli; Boulenger, 1896; Worrell, 1961; Shine, 1991; Cundall & Irish, 1989; pers. obs.), this is here tentatively considered an adaptation to hard-bod- ied prey, often skinks. Snakes with enlarged teeth offset between upper and lower jaws are able to trap hard, cylindrical prey items between a notch in one tooth-row and one or more enlarged fang- like teeth (sometimes true fangs) in the other (Cundall & Irish, 1989). As well as this ‘trapping’ function, having only a few long teeth in each jaw maximises the probability of hard-bodied prey being deeply punctured, whereas this is avoided in arthrodont forms. MEMOIRS OF THE QUEENSLAND MUSEUM EVOLUTION OF TEETH AND ATTACHMENT Snake teeth are slender compared to other ver- tebrates; they break frequently during normal use and are quickly replaced (Edmund, 1969). The have reduced occlusal area (sacrificing strength) to increase sharpness and depth of penetration. Tooth form is a compromise betwen competing selective forces defining a ‘fitness landscape’ over attainable phenotypes (Wright, 1932), and local optima will be attained only if intermediate states are evolutionarily stable. If the rate of breakage is too high, prey capture or swallowing efficiency (and consequently fitness) will be low. During the stages of feeding on a given range of prey types with given neuromuscular reper- toires, forces on the tooth come from particular directions with greater or lesser frequency and magnitude, so it will generally be favourable for the tooth to be asymmetrical rather than a simple cone. The orientation of ‘cutting ridges’ (which function as buttresses as well as blades), curves in the shaft, and the shape of the tooth base, will confer maxima of resistance in one or more direc- tions, at the expense of minima elsewhere. Horizontal components of pressure (shear stress) at the tip of an approximately conical tooth are converted to bending stresses at the base, i.e. compression at one side and tension at the other. The magnitudes of these forces will depend on base diameter, but only tension and shear will tend to either break the shaft or disrupt the attach- ment of tooth to bone, Bone of attachment can apparently withstand such stresses within a wide range of values of the ratio of tooth length to basal diameter. A fibrous connection will remain stable at low values of this ratio (short, broad teeth as in Anomochilus), and at intermediate values will have enough elasticity to return the tooth upright after displacement (functional arthrodont condi- tion). At high values (longer, slender teeth) a fibrous attachment would merely bend passively, without developing enough tension to right the tooth; the orientation of the teeth would then not be precisely controllable, and during prey capture and ingestion they would more often encounter shear stresses at unfavourable angles, leading to rupture. Such a condition (elongate, slender teeth with fibrous attachment) is unknown in any living snakes, and would presumably be evolutionarily unstable for most diets and feeding methods. This consideration of the forces applied at the tooth tip and base suggests that arthrodonty and elongate teeth are mutually exclusive conditions. NEW MIOCENE MADTSOIID SNAKE FROM RIVERSLEIGH Thus the specialisations of dentition and jaw mor- phology in Nanowana are most likely to be inde- pendently derived from the neatly isodont, ankylosed condition of other madtsonids, and ap- parently represent alternative solutions to the problem of feeding on hard- scaled lizards. Healed breaks of the jaw elements (particularly dentaries) are not uncommon in snakes (pers. bs.), and presumably result in most cases from allempts to capture or subdue relatively large and powerful prey. Sublethal trautna associated with particular morphological specialisations may be an indicator of mechanisms of selection; there are upper limits to prey size and strength for every species of predator, and both prey selection and behavioural aspects of prey-handling, as well as morphology, will be subject to selection. The break through the third dentary alveolus of QMF23051 (N. schrenki sp. nov.) would have occurred most easily (i.e. greatest stress would occur) while the 3rd alveolus was unoccupied, and while a prey item was held by the enlarged 2nd tooth, but not the smaller posterior tecth. Fractures of this kind could be expected to be less common (all else being equal) with a more uni- form dentition. but this possible disadvantage of megadonty may have been outweighed by an increased rate of capture success, or of retention once a prey item was secured behind (or impaled on) the enlarged dentary teeth, The ribbon-like posterior maxilla of N. godthelpi presents an even more fragile appear- ance, but no specimens suggest breaks during lite, While this is negative evidence, the rarity of such breaks would tend to support the presence of a jugal in the suborbital region, Presence of a jugal in Wonambi naracoortensis can similarly be in- ferred from the oblique trough crossing the max- ila (Barrie, 1990; Scanlon, 1996) which would otherwise be an obvious point of fragility. SYMPATRY OF RELATED SPECIES WITH SIMILAR DIETS The two species of Nanowana occur together in at least 3 Sites, existing sympatrically for a significant period. They are thought to have had similar diets (skinks), and similaradult size, They thus occupied quite similar niches, and were strictly equivalent ecologically. They may have differed in aspects of behaviour which would not be discernible in the fossil record, but at least a difference in habitat can be suggested. The different representation of the Iwo species when found together (minimum number ot inii- Viduals, number of identifiable elements, and quality of preservation) implies that N. godihelpi was more abundant close to the sites of deposi» tion, whereas N. schrenki may have been less abundant locally, and the more damaged remains transported from further afield (cf. LaDuke, 1991). Thus N. godrhelpi lived near water (possi- bly riparian, probably closed forest), whereas W, sohrenki may have lived further from water, pos- sibly in more open or drier areas such as clearings or mucky hills. Most sites where Nanowand vertebrae have been found have not produced jaw clements di- agnostic lo species. The genus as defined here. therefore provides a convenient level of deserip- tion which can be applied t0 a larger set of sites, bul as yel all specimens referred to Nanewana ure from Riversleigh ACKNOWLEDGEMENTS I thank Mike Archer for the opportunity to study Riversligh fossils under his supervision, and Henk Godthelp, Sue Hand, Anna Gillespie, Jeanette Muirhead, Syp Prasouthsouk, and Sic- phan Williams for preparation and other activities in field and lab which made this work possible, T also thank Mike Archer, John Barrie, Dino Frey, Mark Hutchinson, Mike Lee, Ralph Molnar, Jean-Claude Rage, Alan Savitzky, Rick Shine, Zbigniew Szyndlar, and Paul Willis, for insights, discussions, and access 10 material; and Wighardt von Koenigswald and Fricdemann Schrenk tor facilitating visits to Germany where much of the paper was written. Support for research al Riversleigh has come from the Australian Re- search Grant Scheme; the National Estate Grants Scheme (Queensland); the University of New South Wales; the Commonwealth Department of Environment, Sports and Territories, the Queens- land National Parks and Wildlife Service; the Commonwealth World Heritage Unit, ICI Aus- tralia; the Australian Geographic Society; the Queensland Museum; the Australian Museum, the Royal Zoological Society of NSW; the Linnean Society of NSW; Century Zinc; Mount Isa Mines; Surrey Beatty & Sons; the Riversleigh Society; and private supporters including Elaine Clark, Margaret Beavis, Martin Dickson, Sue & Jim Layarack and Sue & Don Scon-Orr, Vital field assistance came from many hundreds of volunteers as well as stal! and postgraduate stu- dents of the University of NSW. Skilled prepari- tion of mast of the Riversieigh maternal has been carned oul by Anna Gillesme. AJI LITERATURE CITED ARCHER, M., GODTHELP, H., HAND,S.J. & MEG- IRIAN, D. 1989. Fossil mammals of Riversleigh, northwestern: Queensland: preliminary overview of biostratigraphy, correlation and environmental change. Australian Zoologist 25; 27-65. ARCHER, M.. GODTHELP, H. & HAND, S.J. 1991. Riversleigh. 2nd Edition, (Reed Books; Sydney). BARRIE, D.J. 1990, Skull clements and associated remains of the Pleistocene boid snake Wonambi naraceortensix. Memoirs of the Queensland Mu- seum 28; 139-151, BOULENGER, G.A. 1896, Catalogue of the snakes in the British Museum (Natural History, HI, (Taylor & Francis: London), COGGER, H.G., HEATWOLE, H., ISHIKAWA, Y., MCCOY, M., TTAMIVA, N. & TERUUCHI, T- 1987. The status and natural history of the Rennel Island sea Krait, Laticauda crockeri (Serpentes: Laticaudidae), Journal of Herpetology 2): 255- 260. CUNDALL, D. & IRISH, F.J. 1989. The function of the intramaxillary joint in the Round Island boa, Casarea dussumieri. Journal of Zoology, London 217: 369-598. CUNBALL, D, & ROSSMAN, D.S. 1993. Cephalic analomy of the rare Indonesian snake Anomochils weber. Zoological Joumal of the Linnean Society 109: 235-273, CUNDALL, D., WALLACH, V. & ROSSMAN, D.S. 1993, The systematic relationships of the snake genus Anomochilus. Zoological Journal of the Linnean Society 109: 275-299, EDMUND, A.G. 1969. Dentilion. Pp. 117-197. Tn Gans, C. & Parsons, T.S. (eds), Biology of the Reptilia |. (Academic Press: London). ESTES, R.. FRAZZETTA, T.H, & WILLIAMS, E.E. 1970. Studies on the fossi) snake Dinilysia patagonica Woodward. Part 1. Cranial morphol- ogy. Bulletin af the Museum of Comparative Zo- ology, Harvard 140; 25-74. FRAZZETTA, T.H. 1966. Studies on ihe morphology and function of the skull in the Boidae (Serpentes), Part Ul. Morphology and function of the jaw appa- Titus in Python sebae and Pyrhon molaras. Journal of Morphology 118: 217-296. 1970. Studies on the fossil snake Dinilysia patagonica Woodward. IL. Jaw machinery in the earliest snakes. Forma et Functio 3: 205-22 1. GREENE, H,W, 1983, Dielary correlules of the urigin and radiation of snakes. American Zoologist 23; 431-441. HOFFSTETTER, R. 1960. Un dentaire de Madtsoiu (serpent géant du Paléocene de Patagonia), Bulle- tin du Muséum national d* Histoire naturelle, Paris (2) 31: 379-386. HUTCHINSON, M.N. 1992. Origins of the Australian scincid lizards; a preliminary report on the skinks of Riversleigh, The Beagle 9; 61-69, MEMOIRS OF THE QUEENSLAND MUSEUM LADUKE T.C. 199) The fossil snakes of Pit 91, Rancho La Brea, California. Natural History Mu- seum of Los Angeles County, Contributions in Science 424: 1-28. MCDOWELL, S.B. 1969. Texicocakunus, a New Guinea genus of snakes of the family Elapidae. Journal of Zoology, London 159: 443-511. OWEN, R. 1840. Note on the dislocation of the tail ata certain point observable in the skeleton of many ichthyosauri, Transactions of the Geological 50- ciety of London 5: 511-514, PATCHELL, F, & SHINE, R. 19864. Feeding mecha- nisms in pygopodid lizards; how can Lialis swal- low such large prey? Journal of Herpetology 20: 59-64. PATCHELL, F.C, & SHINE, R. 1986b. Hinged teeth for hard- bodied prey: a case of convergent evolu- tion between snakes and legless lizards. Journal of Zoology, London 208: 269- 275. SAVITZKY, A.H. 1981, Hinged teeth in snakes: an adaptation for swallowing hard-bodied prey, Sci- ence 312: 346-349. 1983 Coadupted characler complexes among snakes! fossoriality, piscivory, and durophagy. American Zoologist 23, 397-409. SCANLON, J.D. 1992. A new large madisaiid snake from the Miocene of the Northem Ternary. The Beagle: 9: 49-60. 1993_ Madtsoiid snakes from the Eocene Tinga- mara Fauna of eastem Queensland. Kaupia: Darmstädter Beiträge zur Naturgeschichte 3: 3-8, 1995, First records from Wellington Caves, New: South Wales, of the extinct madtsoiid snake Wonambi naracoortensis Smith, 1976. Proceed- ings of the Linnean Society of New South Wales 115; 233-238. 1996, Studies in the palacontology and systematics of Australian snakes. PhD thesis, University of New South Wales. (Unpubl.). SCANLON, J.D. & SHINE, R. 1988. Dentition and diet in snakes: adaptations to oophagy in the Austra- lian elapid genus Simoselaps. Joumal of Zoology, London 216; 519-528, SHINE, R. 1991, Australian Snakes: a natural history, (Reed: Sydney), SMITH, M.A. 1926. Monograph of the Sea-snakes (Hydrophiidae), (British Museum (Natural His- tory); London). SMITH, MJ, 1976, Small fossil vertebrates Irom Vic- toria Cave, Naracoorte, South Australia. FY. Rep- tiles. Transactions of the Royal Society of South Australia 100: 39-5), WORRELL, E. 1961. A new generic name for a nomi- nal species of Denisonia. Proceedings of the Royal Zoological Society of New South Wales 1958-1959; 54-55, WRIGHT, 5. 1932. The roles of mutation, inbreeding, crossbreeding and selection in evolution, Proceed- igs of the 6th International Congress of Genetics: 356-36. CAINOZOIC TURTLES FROM RIVERSLEIGH, NORTHWESTERN QUEENSLAND A. W. WHITE White, A.W. 1997 06 30: Cainozoic turtles from Riversleigh, northwestern Queensland. Memoirs of the Queensland Museum 41(2): 413-421. Brisbane. ISSN 0079-8835. The Chelidae and Meiolaniidae occur in the Oligo-Miocene at Riversleigh with the former dominant and short-necked Elseya/Emydura accounting for over 90% of turtle material. Chelodina and Pseudemydura are present with several undescribed forms. Chelid diversity increased from the Oligocene to the Miocene, while average size decreased, suggesting a change in the aquatic habitats. I propose that large late Oligocene river and overflow systems were replaced by smaller, slower-flowing waterways that by the mid-Miocene had developed numerous small, stationary aquatic habitats, some occupied by dwarf turtles. The Miocene appearance and radiation of the large terrestrial meiolaniid turtles and their presumed thermal and dietary requirements suggest that Riversleigh’s gallery forests were heterogeneous and punctuated by open clearings. Absence of trionychid turtles and dearth of long-necked chelids may be related to the general unsuitability of habitats at Riversleigh, particularly during the Oligocene. L] Australia, Riversleigh, turtles, palaeoecology. A.W. White. School of Biological Sciences, University of New South Wales, New South Wales 2052; received 24 March 1997, Riversleigh’s Oligocene and Miocene turtles (White, 1988, 1990, 1992; Gaffney et al., 1989; White & Archer, 1989) represent the highly aquatic side-necked Chelidae and the large, ter- restrial, horned Meiolaniidae. Chelid remains are common in Systems A, B and C of Archer et al. (1989, 1994) and are known from Pleistocene gravels (Terrace Site) adjacent to the Gregory River (White & Archer, 1994). Living specimens of the presumed extinct Elseya lavarackorum, have been collected from nearby Lawn Hill Creek and subfossil material from near the Gregory River (Thomson et al., 1997). Meiolanid remains are confined to System B and one site in System C (Gaffney et al., 1992). Riversleigh turtle material including skulls is well preserved (White & Archer, 1993) but no articulated remains have been found. Relation- ships of extant chelids have been based almost exclusively on cranial anatomy (Gaffney, 1977, 1979) but biochemical techniques are establish- ing alternative phylogenies (Georges & Adams, 1992) and derived shell features have been used to define fossil species (White & Archer, 1994). FAMILY CHELIDAE The extant chelids Pseudemydura, Emydura/ Elseya and Chelodina have been reported from the Oligocene-Miocene of Riversleigh (White, 1988; Gaffney et al., 1989; White & Archer, 1989). Celid taxonomy is based heavily of features of the skull (Gaffney, 1977, 1979). There are no derived skull characters known that can be used to distinguish Emydura Elseya despite these gen- era being electrophoretically distinct (Georges & Adams, 1992). Shell features are not normally used because they have been presumed to be variable in chelid lineages (Gaffney, 1977) but some diagnostic shell features are now known (White & Archer, 1994; Thomson & Georges, 1996). I adopt the more conservative position of using skull features to distinguish short-necked chelids. Gaffney et al. (1989) reported on Emydura/Elseya skull material from Riversleigh although shells and shell pieces were available. Scute features are used to identify modern taxa but these have little or no phylogenetic value (Gaffney, 1979). Riversleigh Oligocene-Miocene turtles are dominated, both in abundance and diversity, by short-necked chelids of which Emydura/Elseya were most common (White & Archer, 1989), representing over 90% of identifiable material, and are known from 10 fossil sites. These two genera contain nearly 60% of extant chelid spe- cies (Georges & Adams, 1992). Two mandibular pieces from Riversleigh’s Bob’s Boulders (Sys- tem C) are the only remains assignable directly to Elseya. They have alveolar ridges typical of the Elseya dentata species complex. A single skull fragment and partial plastron of Pseudemydura is known from Ringtail Site (Sys- tem C; Gaffney et al., 1989). This genus contains 414 FIG. |, Skull of Chelodina sp., QMF31303, Quentin’ s Quarry, Riversleigh. A, dorsal. B, ventral, a single extant species, the Western Swamp Tor- toise (P. umbrina), an endangered species con- fined to two small swamps north of Perth (Cogger, 1992), The Riversleigh specimen is the first fossil record of the genus and indicates Pseudemydura was once much more widespread. Pseudemydura is the most derived extant chelid (Gaffney, 1977, 1979) and is unusual among pleurodires in having little temporal or posterior emurgination of the skull. This results in a skull with a near complete roof above the cranium. The Riversleigh skull fragment consists of a mas- sively expanded supraoccipital indicating that en- closure of the hindmost portion of the skull occurred betore the Miocene. Similarity of the Riversleigh specimen to the modern species sug- gests that Pseudemydura diverged from other short-necked chelids early in the Australian radi- ation, a conclusion supported by cladistic analy- sis of modern chelid skull features (Gaffney, 1977). Electrophoretic data (Georges & Adams, 1992) did not include Pseudemydura and Gaffney’s (1977) hypothesis that Psendemydura is the sister-group to all other Australian chelids remains (but see Manning & Kofron, 1996). Long-necked chelids arose before the Oligocene (Manning & Kofron, 1996). Georges & Adam’s (1992) indicate that Chelodina is the sister group to all Australian short-necked turtles MEMOIRS OF THE QUEENSLAND MUSEUM (excluding Pseudemydura), An almost complete plastron and sore carapace bones of a small long-necked Chelodina are known (Gaffney et al., 1989). A species of Chelodina from Quentin's Quarry Site is described below, An almost entire carapace of a diminutive chelid from Melody’s Maze Site (System C) and a partial skull from CMP Site (System C) White (1992, 1993) are described below. SYSTEMATICS Order TESTUDINES Linnaeus, 1758 Infraorder PLEURODIRA Cope, 1863 Family CHELIDAE Gray, 1825 Subfamily CHELINAE Gray, 1825 Chelodina sp. (Fig. 1) MATERIAL. QMF31303, dorsal skull elements con- sisting of a fused frontal, paired parietals and prootics and supraaccipital bones from Quentin's Quarry Site, middle Miocene, System C. DESCRIPTION. Frontal fused, detached from parietals, with suture undamaged. Posterior tip of supraoccipital missing. Parietals joined, sutured to their respective prootics and the supraoccipital, Frontal flat, roughly triangular, with maximum width behind rim of the orbit near the suture for the postorbital bones, with long and tapered an- terior aspect, with the suture for the prefrontal as well as the mm of the orbit. Suture with the parietals horizontal. Parietals wider than frontal at their suture, widest in the posterior section of the orbit near their sutures with the postorbital bones, with dorsal aspect drawn into a tapering sagittal crest terminating with the intrusion of the supracecipital to form the very tip of the crest. Supraoccipital forming roof of the posterior era- nium: parietals forming roof for the majority of the vault. Cranium widest near the parietal- pro- otic suture. Prootics large, inflated, with suture attachments for the quadrate and basisphenoid intact, with latter suture uniting a broad fout-like plate of the quadrate with the basisphenoid, DISCUSSION. This specimen is referred to Chelodina on the basis of the fusion of the frontal bones, absence of temporal skull rooting and lack of contact between the parietals and squamosals (Gaffney, 1977). Goode (1966) and Legler (1985) divided Chelodina into: the C. longicollis group (A) containing C. longicollis, C. steindachneri and C. nevaeguinea and the C. CAINOZOIC TURTLES FROM RIVERSLEIGH FIG. 2. Carapace of small turtle, QMF31304, Melody’s Maze, Riversleigh. A, dorsal. B, ventral. expansa group (B) containing C. expansa, C. rugosa, C. oblonga and C. parkeri. Goode (1966) separated short-necked species (Group A) where neck length is less than shell length, the posterior skull is not disproportionately extended, the basiocciptial is large and expanded anteriorly, while the squamosal is reduced and lacks a pro- truding lateral process. Group B long-necked tur- tles were longer-necked where neck length is often longer than shell length, the posterior skull is markedly elongated, the basioccipital is small and squarish, and the squamosal has a prominent lateral process. Electophoretic data (Georges & Adams, 1992) support these groupings but not their generic status. In particular, they disagree with the placement of C. oblonga and its relation- ship to the C. longicollis group. The Quentin’s Quarry shell cannot be readily allocated to either group although the skull shows no marked posterior elongation. The skull is unique in: 1. Contribution of the frontal bone to the orbit. In living Chelodina frontals are fused, forming a flat, dorsal plate behind the orbit with a thin medial process extending to the nasal re- gion. The anterior process is intimately fused to the prefrontals which make up most of the dorsal orbit. The Quentin’s Quarry skull has a thin me- dial process but the suture points with the prefron- tals are minimal and anterior. The frontal process bears the rim of the orbit behind the prefrontal sutures and makes up the majority of the dorsal orbit. This means that the eyes were relatively closer together and directed more upwards than sideways. 2. Expanded anterior parietals. In mod- ern long-necked turtles the anterior parietals are flat dorsally but curve steeply down to form the walls of the anterior cranium. In the Riversleigh specimen, there is a parietal shelf formed by the extension of the parietals. The postorbital bones are fused to this shelf which forms part of the roof of the post-orbital canal). 3. Triangular sagittal crest. In extant long-necked turtles the parietals are flattened dorsally before being drawn into a narrow, elongate mid-cranial crest that extends to the rear of the skull. This leaves a massive canal for the neck and jaw musculature. In the Quentin’s Quarry skull, the dorsal parietals taper evenly to the back of the skull creating the most 416 robust crest of any known Chelodina.4, Enlarged prootic bones. In modern Chelodina the prootics are fused laterally to the parietals. As such, they form a horizontal beam connecting the cranium to the external ear. In the Quentin’s Quarry spec- imen the prootics are inflated producing 4 gradual sloping from the side of the skull down to the quadrate. The temporal canal is therefore triangu- lar in profile with the majority of the canal space being more lateral and closer to the quadrate. 5, Restricted posterior extension of the skull. All modern long-necked turtles have low, elongate skulls. This is achieved by the extension of the mid and hind skull regions in ihe horizontal plane. In particular, the supraoccipital, quadrates, squa- mosals and opistotics are markedly elongated. The Quentin’s Quarry skull has an almost rectan- gular supraoccipital that would scarcely have ex- tended beyond the external ear, 6. Shape of cranial vault. In modern Chelodina the cranium is widest near the frontal-parietal suture, In the Riversleigh long-necked turtle, the vaultis widest in the mid-parictal, close to the prootic contact zone. Genus indeel. A (Figs 2, 3) MATERIAL. QMF31304, an almost entire carapace und ne femur from Melody’s Maze, Gag Plateau, Riversleigh, System. C, middle Miocene (Archer etal., 1989). DESCRIPTION. Carapace elongate, oval, {00mm Jong, 75mm wide across pleurals 4. Shell without keeling along the midline, lacking ser- rated posterior peripherals, relatively low and flat in profile, with a longitudinal vertebral depres- sion running the length of the shell, without fe- nestra, Pygal without division or indentation, Pleurals 11, relatively small. Peripherals increas- ing in size towards the posterior, with peripherals 8 and 9 the largest; peripherals 3, 4 and 5 and part of 6 curved ventrally, forming rounded shell mar- gin associated with the bridge; anterior and pos- terior to the bridge, peripherals flattened, forming lateral platforms around the shell. On the inside of the carapace, rib heads on each of the pleurals. Pelvic scars well-developed on pleurals 8 and the suprapygal bones. Transverse ridge running across the Moor of pleural 1 from the raised rib head to the recess for the bndge. Recess curved anteriorly, meeting the peripherals near the suture between peripherals 2 and 3. MEMOIRS OF THE QUEENSLAND MUSEUM FIG, 3. Scute boundaries (dorsal) of QMF31304, turtle carapace from Melody's Maze. DISCUSSION, The carapace has almost all of the right side intact, lacking only the third peripheral and nuchal bones, The left side is damaged and lacks peripherals | to 7 and portions of pleurals l, 2 and 3. The Melody’s Maze carapace may be tenta- lively assigned to Emydura on the curved bridge recess and raised transverse ridge running across the floor of the first pleural bone (White & Archer, 1994). However, the pelvic scars on pleu- rals 8 and the suprapygals are more typical of Elseva. The shell is very small although itis trom an adult animal. Its adult features include the closure of the carapacial fenestral, the tight fusion of the penpherals to the pleurals and the absence of features such as keeling of the carapace and expansion and flaring of the peripherals. The shell shows signs of advanced uge: the extended growth of the pleurals during adult life has cre- ated a vertebral groove running the length of the carapace, This shel! cannot be placed in any of the known species of Emydura because of its small adult size and the unusual formation of the mid-peripheral bones. Among extant Emydura, the smallest known adults are in E. signata (Cogger, 1992) from the coastal rivers of northern NSW. Adult E. signata from the MacLean River show similar CAINOZOIC TURTLES FROM RIVERSLEIGH FIG. 4. Turtle skull, QMF31305, CMP Site, Riversleigh. features to those of the Melody’s Maze shell with carapace lengths of approximately 150mm. Curving of the mid-peripheral bones associated with the bridge occurs elsewhere only in Elseya dentata-like turtles, In modern short-necked tur- tles, the peripherals are flattened and project lat- erally to form distinct ledges around the shell margin. The posterior peripherals are often flared to produce ‘wings’ that cover the hind limbs. There is virtually no flaring of the peripherals in the Melody’s Maze shell; the peripherals are small compared to the pleurals (an adult growth trend in all chelids exaggerated in this species). Peripherals make up only 14% of the dorsal sur- face of the mid-carapace whereas Emydura (e.g., E. kreffti) the peripherals account for 20%. A, dorsal. B, ventral. C, right lateral. D, posterior. The flattened shell is unlike most Emydura species which have domed shells. Only in Pseudemydura, Chelodina, Elseya latisternum- like turtles and some Elseya dentata-like species are the shells flattened and the central crest lost. Genus indet. B (Fig. 4) MATERIAL. QMF31305, a partial skull comprising the paired frontals, parietals, post-orbitals, quadrates, right squamosals and supraoccipital, with paired pter- ygoids, quadrates, prootic bones, opistitic bones, ex- occipitals, basisphenoid and basioccipital from CMP Site, Gag Plateau, Riversleigh, middle Miocene, Sys- tem C. 418 DESCRIPTION. Medium-sized, with extensive dorsal roofing, lacking temporal and posterior emarginations that typify most chelids (Gaffney, 1979a). Parietals expanded, forming the bulk of the dorsal roof; squamosals broad, forming the temporal bridge; supraoccipitals only contribut- ing to the mid-dorsal section of the skull roof, weakly expanded in the dorsal plane. Frontals paired, with small, anterior, medial projections dividinge the prefrontals, contribut- ing minimally to the hind orbit. Postorbitals mainly in the dorsal plane, with a descending strut forming the posterior wall of the orbit. Sutures for the jugal evident. The descending strut from the postorbital and ascending process from the pteryoid in broad contact. Alary process of the pterygoid reduced, may not descend ventrally below the level of the palate, Postorbital canal correspondingly more obvious in lateral view. Dorsal skull sloping downwards anteriorly. Squamosal broad, forming a broad temporal arch between the quadrate and the parietals. Quadrate large, witha deep angular ventral base continuous with the articulation facet for the lower jaw. Quadrate with a deep posterior vertical groove. Floor of the skull widening markedly where the parietals sweep laterally to unite with the quad- rates. Articulation facet for the lower jaw well below the level of the palate, with a broad area formed by the lateral extension of the pterygoid and the quadrates. Pterygoids with a transverse suture with the pajantines. Basisphenoid triangu- lar, with a broad sutural contact with the basioc- cipital. Occipital condyle below the level of the foramen magnum. DISCUSSION. The right side of the skull is rel- atively intact whereas the left parietals and squa- mosals are broken. The skull is heavily impregnated with a dark mineral. The CMP skull cannot be assigned to any known infraorder or genus. It is particularly un- usual in the structure of the skull roof. The only other chelid that lack posterior and temporal emarginations on this scale is Pseudemrydura in which the roof is extensive, the posterior emargi- nation is replaced by a posterior dorsal extension of the supraoccipital, and the temporal emargina- tions are replaced by the extension of the squa- mosal. In the CMP skull, the supraoccipital forms relatively little of the skull roof and some poste- rior emargination is evident. The squamosal is expanded, as in Pseudemydura. As a conse- quence, the temporal roof is broad. MEMOIRS OF THE QUEENSLAND MUSEUM In Pseudemydura, expansion of the supra- occipital causes the hindmost portion of the skull to be lower than the parietals. In the CMP skull, the parietals extend back to the posterior margins of the skull and overlie most of the supraoccipital. This results in the skull having an anterior slope with its highest point behind the level of the quadrates. The quadrates have an unusually deep ventral footing that forms a thick foundation for articula- tion with the lower jaw. The articulation facet is dislocated laterally and lies beneath the most lateral edge of the quadrate. These features, com- bined with the reduction of the alary process of the pterygoid indicate an unusual distribution of muscles between the upper and lower portions of the skull. The ventral portion of the cranium and the exoccipital regions are very similar to the arrangement found in all Chelininae. PALAEOECOLOGY Living Australian freshwater turtles have a va- riety of life history strategies and survival mech- anisms (Kennett et al.,1993; Grigg et al.,1986; Georges, 1982; Georges et al.,1986; Georges & Kennett,1989; Heaphy,1990; Kennett & Georges, 1990; Georges, 1988; Thompson, 1988). Several major trends which may have ecologi- cal significance are apparent in Oligocene- Miocene turtle assemblages at Riversleigh. Only 2 families, the Chelidae and Meiolaniidae are present and of these, chelids make up about 98% of material recovered, Chelids dominate these fossil assemblages in the same way that they dominate modern Australian freshwater systems (Legler, 1985). Only in a few, far northern local- ities are non-chelid freshwater turtles present; Carettochelys insculpta, the Pig Nose Turtle (a carettochelid), is found in a few NT rivers (Georges & Kennett, 1989; Heaphy, 1990). The earliest Chelids known are from the Cretaceous of Patagonia (de Broin, 1994) and the family is thought to have evolved in southern Gondwana (South America and Australia. A sister-group, the pelomedusids, evolved at the same time in north- ern Gondwana and fossils occur in Africa, Mad- agascar and South America. Other turtles lived in Australia during the Oligocene-Miocene including the giant horned meiolaniids and soft-shelled trionychids (Gaffney, 1979, 1981). Another feature of the Riversleigh fossil turtle fauna is the dominance of the plesiomorphic short-necked chelids. Emydura/Elseya turtles ac- CAINOZOIC TURTLES FROM RIVERSLEIGH count for over 85% of turtle remains. Modern EmyduralElseya are predominantly herbivarous and Occur in coastal and inland rivers, creeks and lagoons, especially those with a well-developed aquatic flora (Cann, 1978; Legler, 1985). They are not abundant in muddy or stagnant water. Chelids in System A are typically large with shells up to 500mm long and equivalent in size to the largest extant chelids. Chelid fossils from Systems B and C are smaller (shell lengths 200- 300mm) and thinner-shelled, System C turtle is a dwarf with adult shell 100mm long. The large System A turtles occur at Site D (Archer et al., 1994) which yields many broken sections of un- usually thick turtle shell, Carapacial plates 10- 20mm thick are typical. From the larger shell pieces I estimate shell lengths of 350-450mm. The largest éxtantchelids are the northern snap- ping turtle (Elseya dentara) and gulf snapping turtle (Elseya lavarackorum) which inhabit large flowing rivers or deep still water bodies (Cogger. 1992). The only other fossil chelids from Riversieigh that approach the dimensions of the Systems A turtles are late Pleistocene Terrace Site (White & Archer, 1994), Species diversity increases from Systems A lo C with | chelid species in System A, 2 chelids and 2 meiolanids in System B and 6 chelids and 1 meiolanid in System C, The increase in diversity suggests an increase in small, shallow or slow- flowing aquatic habitats from late Oligocene ta middle Miocene. Maximum turtle size rs a useful indicator of water depth and flow rate. Riverine species that occur in deep or relatively fast flowing water are typically large and capable of sustained swim- ming, Smaller species are excluded from such sites and often confined to fringe water bodies such as side-streams, overflows or ponds (Priteh- ard, 1979), The range of small-shelled chelids in System C sites suggests that a variety of shallow waler or slow moving habitats. were available at the time. System C turtles typically had shell lengths 150-250mim. The smallest extant turtle is the bog turtle, Clemmys muhlengergi, an emydid (Pritchard, 1979) which is 76-114mm long. These turiles occur in extremely shallow, still water habitats: in some cases free water is not available, None of these sites are necessarily clear water sites (Behler & King, 1979). In the Gregory River at Riversleigh there are 5 chelids; these, in order of abundance, are Emydura sp. aff. subglobosa, Emydura sp. aff. victoriae, Elseya latisternum, Elseya lav- 419 arackorum and Chelodina rugosa. The latter two are uncommon. This level of diversity is reason- ably high formodern freshwater habitats in Aus- tralia. Species diversity increases as mean temperature and habitat variation increase (Obst, 1986). Only Systems B and C contain large, terrestrial turtles (meiolaniids) with shell lengths up ta 1 m long, Using Metolania platyceps as a madel, these creatures would have had average boy masses of 150-200 kg. There are a number of large, land turtles alive today. The hest known are the various Galapagos tortoises (Geochelone elphantepus) and the Al- dabran tortoise (Geochelone gigantea). Both reach body sizes and misses ċonsidetably greater than that calculated for Riversleigh’s Miocene mejolaniids. They are regarded as examples of island endemism leading to gigamtism (Pritchard. 1979). The majority of the extant large terrestrial turtles inhabit hor savanna regions of the world For example, both species of large African Land turtles (Geochelone sulcata and G. pardalis) occur in northern and central Africa. Neither spe- cies is found in dense forest. G. sulcata’s distri- bution is along the southern Sahara into central Africa (Pritchard, 1979). Savannah habitat verg- ing onto treeless plains is the preferred habitat for all African land tortoises. During the Miocene, Riversleigh was covered by wel forest communities (Archer et al, 1994) but the large Jand turtles seem incongruous in this hahitat, Land turtles such as the South American G. denticulata and G. carbonaria are generalist herbivores and live in a range of habitats, includ- Ing open savanna, closed woodland and rainforest (Bjorndal, 1989), Moskevits, 1985; Moskovits & Kiester, 1987). While both species will venture deep into rainforest and feed on fungi, fallen fruit and herbs, they do not permanently reside in the closed forest, Both prefer to reside near the for- est-savanna interface (Moskovits & Bjorndal, 1990). Flowers und grasses form. a major part of the dictof South American land turtles in savanna and swamp habitats. They consume sand which is thought to assist digestion of coarse fibrous matter; they are hmd-gut fermenters and need a certain amount of fibre in their diet. In the raintor- est, the turtles get fibrous foodstuffs in the form of vines and shoots, but these are usually not plentiful. In addition, the tortoises seek out clear- ings (e.g, sites of recent tree falls) to bask. Hind- gul digestion requires fairly high and constant temperatures (of 30° or more) and the turtles need to have periodic access to direct sunlight, Youn, 420) ger and smaller turtles need access to sunhghton amore regular basis than large adults (Moskovits, 1985). Tfthese limitations for giant land tortoises apply also lo meiolaniid turtles living in or near forested habitats, then there must have been clearings where the animals could get sufficient volume of high fibre food and have direct access to sunlight. The rotund body form of these land turtles means. that large animals would have a high thermal inertia. At air temperatures above 30° they may not need to seek additional external heat sources. However, should their body temperature fall below this level they would need to reach ex- posed sites quickly to restore core temperatures (Swingland & Pazier, 1979). Riversleigh’s merolaniids are low in abundance despite their high diversity; 4 species in 2 genera are known from 5 individuals, Chelids, in con- Irusl, are represented by hundreds of specimens, Nothing is known about dictary requirements of horned turtles. Large extant terrestrial turtles are opportunistic herbivores and have requirements for fibrous matter intake (Pritchard, 1976). The unusual structure of the jaws of meiolaniids indi- cates that some dietary selectivity may have been possible, but the nature of their dict is unknown, The absence of soft-shelled tnonychids from Riversleigh’s Oligocene Miocene coupled with the very low occurrence of long-necked turtles may be ecologically significant. Both types of turtles are ‘ambush predators’ (Pritchard, 1976). Both groups flourish in situations where they cannot be easily seen and this usually means turbid, muddy or dark water, Their absence and (he preponderance of Emydura/Elseya suggests relatively clear water aquatic environments. Chelodina in System C coincides with the reduc- tion in average chelid shell size, indicative of shallow, turbid lagoons during this period of the Miocene. The complete absence of trionychids suggests that the lagoons did not have silty bot- toms and were unsuitable for concealment. ACKNOWLEDGEMENTS The Riyersleigh project is supported by the Australian Research Council, the Department of the Environment, Sport and Territories, National Estate Programme Grants (Queensland), Queens- land National Parks and Wildlife Service, the Australian Geographic Society, the Waanyi peo- ple and Carpentaria Land Council, the Linnean Society of New South Wales, ICI, the Queens- land Museum, the University of New South MEMOIRS OF THE QUEENSLAND MUSEUM Wales and the Riversleigh Society. I thank Anna Gillespie who prepared the specimens and took (he photographs. LITERATURE CITED ARCHER, M, GODTHELP, H., HAND, S.J. & MEGIRIAN, D. 1989 , Fossil mammals of Riversleigh, northwestern Queensland = prelimi- nary overview of biosiratigraphy, correlation and eavimnenia, change. Australian Zoologist 25: 29-65. ARCHER, M., HAND, S.J.. & GODTHELP, H. 1994, Riversleigh. 2nd ed. (Reed: Sydney), BEHLER, J,L,, & KING, F.W. 1979. The Audibon Society heid guide to North American reptiles and amphibians. (A.A. Knopf: New York). BJORNDAL, K.A. 1989. Flexibility of digestive re- sponses in two generalists herbivores, the tortoises Geochelane carbonaria and Geochelone denticulata, Oecologia 78: 317-321. CANN, J. 1978, Tortoises of Australia. (Angus & Robe erison: Sydney). COGGER, H.G. 1992. Reptiles and amphibians of Aus tralia, Sth ed, (Reed; Sydney). de BROIN, F- 1994. Examples of parallel evolution in Pleurodira from the Cretaceous to the present. Abstracts of the Second World Congress of Her- petology: 69-70. GAFFNEY, E.S: 1977. The side-necked turtle Family Chelidae: a theory of relationships using shared derived characters. American Museum Novitates 2620; 1-28. 19794, Comparative cranial morphology of Recent and fossil turtles. Bulletin of the American Mu scum of Natural History 164: 65-375, 1979b, Fossil chelid turtles of Australia. American Museum Novitates 2681; 1-23, 1981, A review of the fossil turtles of Australia. American Museum Novitates 2720; 1-38, 199], The fossil turtles of Australia, Pp 704-720. In Vickers-Rich, P., Monaghan, J.M., Baird, R.F. & Rich, T.H. (eds). Vertebrate palaeontology of Australasia. (Pioneer Design Studio: Mel- bourne). GAFFNEY, E.S., ARCHER, M, & WHITE, A.W. 1989. Chelid turtles from the Miocene freshwater limestones of Riversleigh Station, northwestern Queensland, Australia, American Museum Novitates 2959; 1-10, 1992. Warkalania, a new meiolaniid turtle from the tertiary Riversleigh depusils of Queensland, Aus- tralia. The Beagle, Records of the Northem Ter- ritory Museum of Arts & Science 9: 35-48. GEORGES, A. 1982. Diet ofthe Australian Freshwater turtle Emydura kreffti (Chelonia:Chelidae), in an unproductive lentic environment, Copeja 1982; 331-336. 1988. Sex dererminanon ts independent of incuba- tion temperature in another chelid turtle, Chelodina longicollis, Copeia 1988; 331-350. CAINOZOIC TURTLES FROM RLIVERSLEIGH GEORGES, A. & KENNETT, R. 1989. Dry season distribution and ecology of Carettachelys in- sculpta (Chelonia;Carettochelydidae) in Kakadu Nanional Park, northern Australia, Australian Wildlife Research 16: 323-335, GEORGES, A., NORRIS, R,H, & WENSING, L. 1986. Diet of the freshwater turtle Chelodina longicollis (Testudines:Chelidae) from the coastal dune lakes of the Jervis Bay Territory. Australian Wildlife Research 13; 301-308. GEORGES, A. & ADAMS, M. 1992. A phylogeny for Australian chelid turtles based on allozyme elec- trophoreses, Australian Journal of Zoology 40: 435-476, GOODE, J, 1976. Freshwater tortoises of Australia and New Guinea. (Lansdowne: Melbourne), “GRIGG,G.C., JOHANSSEN,K., HARLOW, P., BEARD, L. & TAPLIN, L. 1976. Facultative aestivation in a tropical freshwater turtle Chelodina rugosa. Comparative Biochemistry Physiology 83A: 321- 323. HEAPHY, L. 1990. The ecology of the Pig-Nosed Turtle, Carertochelys insculpta, in northern Aus- tralia, Ph.D. Thesis, University of New South Wales, Sydney.(Unpubl.). KENNETT, R. & GEORGES, A. 1990, Habitat utilisa- tion and its relationship to growth and reproduc- tion of the eastern long- necked turtle, Chelodina longicollis (Testudinata; Chelidue) from Aus- tralia. Herpetologica 46; 22-33. KENNETT, R., CHRISTIAN, K. & PRITCHARD, D. 1993, Underwater nesting by a freshwater turtle Chelodina rugosa from tropical northern Aus- tralia, Australian Journal of Zoology 41: 47-52. LEGLER, J.M. 1985, Australian chelid turtles: repro- ductive patterns in wide-ranging taxa. Pp. 117- 123. In Grigg, G, Shine, R. & Ehmann, H. (eds), Biology of Australasian frogs and reptiles. (Royal Zoological Society of New South Wales/ Surrey Beatty & Sons: Chipping Norton, NSW). MANNING, B., & KOFRON,C.P. 1996. Evolution and zoogeography of Australian freshwater turtles. Memoirs of the Queensland Museum 39; 319- 331. MOSCOVITS, D.K. 1985. The beħaviour and ecology of the two Amazonian tortoises, Geochelone car- bonaria and Geochelone denticulata, in north- western Brazil, PhD thesis, University of Chicago, Chicago.(Unpubl.). MOSCOVITS, D.K.& KEISTER, A.R. 1987. Activity levels and ranging behaviour of the Amazonian tortoises, Geochelone carbonaria and Geochelone denticulata in northwestern Brazil, Herpetologica 46: 207-218. OBST, FJ. 1986. Turtles, tortoises and terrapins, (Druckerei Fortschnift Ertur Leipzig). PRITCHARD, P-C.H, 1979. Encyclopaedia of turtles. (T.F.H. Publications: Hong Kong), SWINGLAND, LR. & FRAZIER, J.G. 1979. The con- flict between feeding and overheating in the Al- dabran giant tortoises. Pp. 611-615. In Amlaner, C.J.W, & MacDonald, D.W.W. (eds), Handbook in biotelemetry and radio tracking. (Pergammon Press: London). THOMPSON, M.B. 1988, Population of the Murray tortoises Emydura (Chelodina): the effect of egg predation by the red fox, Vulpes vulpes. Australian Wildlife Research 10: 363-371. THOMSON, S., WHITE, A.W. & GEORGES, A. 1997. Re-evaluation of Emydura lavarackorum: identi- fication of a living fossil. Memoirs of the Queens- land Museum 42: 327-336. WHITE, A.W, 1988, Riversleigh’s turtles - all shell be revealed. Riversleigh Notes 4: 3-4. 1992. Riversleigh’s midget submarine. Riversleigh Notes 17: 2-3. 1993, Riversleigh’s researchers, part 3. Riversleigh Notes 19: 2-5, WHITE, A.W. & ARCHER, M. 1989. Riversleigh tur- tles - so far. Conference on Australian Vertebrate Evolution, Palaeontology and Systematics, Syd- ney, Abstracts: 17. WHITE, A,W. & ARCHER, M. 1994, Emydura lav- arackorum, a new pleistocene turtle (Pleurodira: Chelidae) from fluviatile deposils at Riversleigh, northwesternQueensland. Records of the South Australian Museum 27: 159-167, NEW CROCODILIANS FROM THE LATE OLIGOCENE WHITE HUNTER SITE, RIVERSLEIGH, NORTHWESTERN QUEENSLAND PAUL M.A. WILLIS Willis, Paul M.A. 1997 06 30: New crocodilians from the late Oligocene White Hunter Site, Riversleigh, northwestern Queensland. Memoirs of the Queensland Museum 41(2): 423-438. Brisbane. ISSN 0079-8835. Four new species of crocodilian are identified from the late Oligocene White Hunter Site, Riversleigh, northwestern Queensland, one of which is also found in other System A sites at Riversleigh. All four species are assigned to known genera and some revision of two generic diagnoses is required. Two different forms of posterior cranium are also identified from White Hunter Site and retained in open nomenclature. Palaeoecological significance of four crocodilians in a single site are interpreted as a sympatric assemblage because they have different head shapes, However, the diversity in these crocodilians could also suggest a thanatocenosis involving taxa from different hydrodynamic regimes with differing degrees of forest canopy cover. [_] Riversleigh, Baru, Quinkana, Mekosuchus, Oligocene. Paul Michael Arthur Willis, Quinkana Pty Ltd, 3 Wanda Cres., Berowra Hts, NSW, 2082; received 4 November 1996. The fossil assemblage from White Hunter Site at Riversleigh, NW Queensland contains skull fragments and postcranial material of crocodil- ians and other vertebrates. The fragments repre- sent at least 4 crocodilian species. Three different maxillae are assigned to new species of known genera, Of 4 mandibles identified 3 are assigned to 3 of the species identified by maxillae; the fourth belongs to a species better known from Riversleigh’s D, Sticky Beak and Pancake Sites (Willis et al., 1990). Cranial material is described but not assigned to any of the 4 new species. Mekosuchus Balouet & Buffetaut, 1987 and Baru Willis et al., 1990 were previously mono- typic and their generic diagnoses require revision in the light of new species assigned below. Quink- ana was revised by Willis & Mackness (1996) and their expanded generic diagnosis (based on Molnar, 1981) encompasses the new species de- scribed here. Mekosuchus was known from Recent cave de- posits in New Caledonia. It has a unique au- tapomorphy: the maxilla participating in the orbit. M. whitehunterensis sp. nov. is the first pre-Pleistocene record of this genus, Quinkana has 3 species: Q. fortirostrum Mol- nar, 1981 from E Queensland; Q. timara Megir- ian, 1994 is more slender-snouted from late middle Miocene of Bullock Creek, NT; and Q. babarra Willis & Mackness, 1996 is from early Pliocene at Allingham Creek, Queensland. Quinkana is distinguished by a suite of ziphodont features and is unique among mekosuchines (sensu Willis et al., 1993) in being a broad- snouted ziphodont. Quinkana meboldi sp. nov. is the third pre-Pliocene record after Q. timara and a species from the late Miocene Ongeva Local Fauna, Alcoota, NT (Murray & Megirian, 1992; Murray et al., 1993; Megirian, 1993). Baru darrowi (Willis et al., 1990) was de- scribed from middle Miocene of Bullock Creek, NT and Site D, Riversleigh. Two species of Baru (Willis et al., 1990) are recognised from White Hunter Site. One species is particularly small for the genus and the other species is based on mate- rial from a number of Riversleigh’s System A sites (sensu Archer et al., 1989), including White Hunter Site. Some of the material assigned here to the second species of Baru was previously assigned to B. darrowi. Baru species are broad, moderately deep-snouted mekosuchines with moderately compressed teeth and a distinctive ridge on the posterior of the maxilla and the jugal. Mekosuchus, Quinkana and Baru can all be shown to be mekosuchines. A more detailed phy- logenetic analysis of these three genera forms part of amore comprehensive investigation of the phylogeny of mekosuchines (Salisbury & Willis, 1996). The ecological implications of four crocodil- ians in the same deposit invites an investigation of the possible structure of crocodilian faunas. There would appear to be no ecological conflict between the sympatric existence of all four spe- cies because their different morphologies suggest the exploitation of different habitats. This is con- sistent with modern analogies such as some parts of the Amazon River Basin and with other fossil deposits such as Messel and Geisaltal in Ger- many, the Bridger Basin in the U.S.A. and the La 424 FIG. 1. Mekosuchus whitehumerensis n.sp., QMF31051, holotype, right maxilla, ventral view. Scale = 5mm., Venta fauna in Colombia, Alternatively, the dif- ferent crocodilians in White Hunter Site may be from different habitats and have been collected together in a thanatocenosis, This publication was the content of a seminar presented at the Conference on Australia Verte- brate Evolution, Palaeontology and Evolution (CAVEPS) in Alice Springs, March, 1991 and published as an abstract (Willis, 1992), Mekosuchus Balouet & Buffetaut, 1987 TYPE SPECIES. Mekosuchus inexpectatus Balouet & Buffetaut, 1987. DIAGNOSIS (translated from French). Eu- suchians with chounae relatively little displaced posteriad; wings of pterygoids strongly devel- oped posteriorly; skull deck very broad; maxilla participating in lower border of the orbit; external nares opening to the side and the front (an- terolaterally); nasals not reaching external nares; palatines very narrow in their posterior part; quadratojugal lacking a spine; snout short and FIG, 2. Mekesuchus whitehunterensis, QMF31051, holotype, right maxilla. Dorsal view with white arrow showing portion of the maxilla that participates in the orbit. Scale = 2mm, MEMOIRS OF THE QUEENSLAND MUSEUM FIG, 3. Mekosuchus whitehunterensis, QMF31052, partial frontal, dorsal view. Scale = 5mm. deep; splenial does not participate in the mandib- ular symphysis; posterior crushing teeth; 13 man- dibular teeth; lower teeth occlude medial to upper series; vertebrae procoelous with strong neural spines in the cervical region; limb bones showing strong muscle insertions; presence of dorsal scutes. My diagnosis includes the tollowing features (apomorphies indicated by ‘a’ ) are: 1, (a) maxilla participating in lower border of the orbit. 2, snout shortand deep. 3, (a) no conspicuous gap between the sixth and seventh maxillary alveoli. 4, high. narrow alveolar process. 5, symphyseal region very shallow dorsoventrally. 6, splenial anteriad to the level of the seventh dentary alveolus. 7, external mandibular fenestrae strongly reduced. 8, (a) out-turned flange on the angular and sur- angular. The type species diagnosis is; palatal fenestrae reaching anteriorly to the level of the sixth max- illary alveoli; posterior teeth of rounded, crushing form; symphysis reaching posteriorly to the level of the seventh dentary alveoli. Some features of the original diagnosis are synapomorphies of wider groups and others are of uncertain value so they are not employed herein. The character ‘nasals not reaching external nares’ is equivocal on available material so is not employed pending more complete material, Mekosuchus whitehunterensis sp. nov. (Figs 1-5) MATERIAL. Holotype. QMF31051, right maxilla (Figs 1, 2). Paratypes QMF31052, partial frontal; QMF31053, almost complete mandible; QMF31054 and QMF31055, anterior portions oF dentaries. All from late Oligocene White Hunter Site, Riversleigh. NEW CROCODILES FROM THE LATE OLIGOCENE FIG. 4. Mekosuchus whitehunterensis, QMF31053, left mandible, lateral view. Scale = lcm. DIAGNOSIS. Longitudinal sulcus below the orbit; palatal fenestrae reaching anteriorly to the level of the seventh maxillary alyeoli; posterior teeth compressed and blade-like; and symphysis extending posteriorly to the level of the sixth dentary alveoli. ETYMOLOGY. From White Hunter Site. DESCRIPTION. Maxilla broad, deep-snouted, with moderately high, narrow alveolar process (sensu Molnar, 1981). Lateral wall steeply in- clined to the palate, with longitudinal sulcus ven- tral to the orbit. Small portion of the maxilla participating in the orbit, separating lacrimal from jugal. (Full extent to which the maxilla participated in the orbit cannot be deduced be- cause the posterior portion is missing in this spec- imen.). Alveoli ovate, slightly compressed laterally, close to each other so excluding the lower series from resting between them; fifth alveolus largest; first and seventh alveoli small- FIG. 5. Mekosuchus whitehunterensis, QMF31053, left mandible, dorsal view. Scale = 1em. 426 MEMOIRS OF THE QUEENSLAND MUSEUM FIG. 6. Quinkana meboldi n. sp., QMF31056, holotype, est, almost equal in size. Only two pits for recep- tion of dentary teeth medial to the upper alveoli, between the sixth and seventh alveoli, and a dis- proportionately large pit posterior and medial to the seventh alveolus. Palatal fenestra reaching level of the seventh alveolus. Frontal. Closely resembles frontals of M. in- expectatus, very wide between the orbits; orbit margins raised, giving a concave transverse sec- tion to the dorsal surface. crania cristae frontalis shallow, close together leaving a wide, thin shelf between them and the orbit margins. Mandible and dentary fragments. Pseudoheterod- ont with an undulating tooth row; tooth row shorter with respect to the whole mandible than in other crocodiles. Posteriorly, dentary strongly compressed laterally and deep dorsoventrally. Dentary thick around the base of each alveolus; buttressing variable in proportion to size of alve- olus, not strongly developed. Slight ridge defin- ing a groove lateral to the 12th through to the 16th alveoli probably received posterior maxillary left maxilla, lateral view. Scale = lem. teeth. Splenial extending anteriorly to 7th alveo- lus; symphysis extending to 6th, Low ridge on the external surface of both dentaries running from below the 8th dentary alveolus onto the angular, probably strengthened the dentary. Teeth 16, with unserrated anterior and posterior carinae, becoming laterally compressed posteri- orly so that. posterior teeth are blade-like. QMF31053 with large, out-turned flange on the posterior and ventral margins of angular and out- turned flange on the dorsal margin of the external surface of the surangular; flanges joining at the posterior margin of the mandible, marking boundary between the sculptured surfaces and the smoother surfaces for muscle attachment. Articular broad, expanded medially; articular portion of retroarticular process shallowly con- cave. Retroarticular process short and steeply inclined. Medial side of condylar surface re- duced, strongly buttressed ventrally. External mandibular fenestra reduced, almost closed. Sculpture of indistinct scarring on the external FIG. 7. Quinkana meboldi n. sp., QMF31056, holotype, left maxilla, ventral view. Scale = Icm. NEW CROCODILES FROM THE LATE OLIGOCENE 427 FIG. 8. Left dentary of Quinkana meboldi (QMF31059). Lateral view, scale = 1cm. surfaces of the dentary and a well-developed mosaic of pits on the angular and surangular. DISCUSSION. This mandible is very derived and shows distinct similarities to M. inexpectatus. Both are distinguished by: external fenestra re- duced or closed; anterior edge of surangular forming distinct step dorsal to the dentary; angu- lar and surangular flange; posterior portion pro- portionally short and deep; symphysis very shallow; similar sculpture. Maxillae of M. in- expectatus and M. whitehunterensis exhibit the apomorphic condition of contacting the orbit. Thus, it is most parsimonious to associate the derived mandible and maxilla from White Hunter Site, both of which most closely resemble M. inexpectatus. The association of mandibles and maxillae of M. inexpectatus is not in doubt (Balouet pers. comm.). Quinkana Molnar, 1981 TYPESPECIES. Quinkana fortirostrum Molnar, 198] DIAGNOSIS. See Willis & Mackness (1996). Quinkana meboldi sp. nov. (Figs 6-9) MATERIAL. Holotype QMF31056, left maxilla (Figs 6, 7). Paratypes QMF31057, almost complete left maxilla; QMF31058, right maxillary fragment; QMF31059, dentary fragment. All from late Oligocene White Hunter Site, Riversleigh. DIAGNOSIS. Small to moderate-sized, with 14 max- illary alveoli; palatal fenestra extending anteriorly to the level of the 8th maxillary alveolus; teeth partially interlock; snout narrower than in Q. fortirostrum; mild festooning; carinae of teeth without serrations. ETYMOLOGY. For Ulrich Mebold, Max Plank Institiit für Radioastronomie. DESCRIPTION. Maxilla. Teeth 14, compressed and blade-like with anterior and posterior carinae. Alveoli compressed to varying degrees; anterior 6 teeth directed slightly posteriorly. Alveolar ridge low, mildly undulating, uninter- rupted laterally but medial side interrupted by pits for the reception of dentary teeth. Palatal fenestra extending anteriorly to the 8th alveolus. Midline palatal suture straight to the level of the 7th alveolus, then diverting laterally to accommodate a short, pointed anterior palatal process. FIG. 9. Quinkana meboldi, QMF31059, left dentary, dorsal view. Scale = 1cm. 428 |! i ts rey t Oo 5 a t G S p E t FIG. 10. Baru huberi, QMF31060, holotype, snout, dorsal view, with line interpretation. F=frontal; J=jugal: L=lacrimal; Mx=maxilla: N=nasal; Pf=prefrontal; Pmx=premanxilla. Scale in cm. Dorsal surface steep-sided, indicating a deep, moderately broad snout. Preorbital or lacrimal ridge giving the snout a trapezoidal cross section anterior to the orbits, Margins contacting nasals straight. Sculpture of distinct pits anteriorly, de- generating to pitted scars posteriorly. Alveoli with a sharply defined groove running medial to the alveoli, This appears to have been derived from.the line of foramina normally found in other crocodilians in a homologous position. DISCUSSION. The lateral compression of both the dentary and the dentition as well as the lack of festooning indicates that the dentary fragment (QMF31059) belongs to Q, meboldt. The single tooth is identical to the posterior teeth of QMF31056. This dentary form differs from M. whitehunterensis in which the posterior-most al- veoli are interconnected. It also differs from dentaries attributed to Baru which lacks strongly OT Pe TETE ‘ Siw. MEMOIRS OF THE QUEENSLAND MUSEUM compressed alveoli and does not have a laterally compressed mandibular body, Baru Willis, Murray & Megirian, 1990 TYPE SPECIES. Baru darrowit Willis et al., 1990. DIAGNOSIS. Broad, moder- ately deep snout; reduction of the second premaxillary tooth during growth sometimes re- sulting in four premaxillary teeth in adults; premaxillary l and anterior six maxillary teeth directed posteriorly; tooth crowns moderately com- pressed laterally; tooth crown and socket dimensions highly differentiated along both upper and lower tooth rows with cor- respondingly wide, deep alve- olar processes; conspicuous maxillary reception pits corre- sponding to dentary tooth crowns situated medial to the upper tooth row; anterior mar- gins of the palatal fenestrae ex- tending to the level of the seventh maxillary tooth; ante- rior palatine process absent; splenial terminates anteriorly at the level of the seventh den- tary tooth and does not enter symphysis; external nares terminal and broadly ‘apple’ -shaped; distinctive bony crest arches pos- teriorly from the maxillae and jugals, extending to the quadratojugals. REMARKS. The 2 new species are most closely related to Miocene B. darrowi from the NT (Wil- lis et al., 1990). Some of the material attributed here to B. wickeni sp. nov. was previously re- ferred to B. darrowi. ‘Internal nares with raised rim’ was included in the original diagnosis of Baru but its status is now uncertain. Baru darrowi Willis, Murray & Megirian, 1990 DIAGNOSIS, Snout broad, deep; rounded pre- maxillae; 13 maxillary teeth; nasals excluded from external nares; anterior termination of nasal NEW CROCODILES FROM THE LATE OLIGOCENE 429 FIG. 11, Baru huberi, QMF31060, holotype, snout, ventral view, with line interpretation. Mx=maxilla; Pa=palatine; Pmx=premaxilla. Scale in cm. is a short, broad wedge; serrated carinae; mandib- ular symphysis extends posteriorly to between the sixth and seventh dentary teeth. Baru huberi sp. noy. (Figs 10, 11) HOLOTY PE, QMF31060, fragmentary snout (Figs 10, 11). Paratypes QMF31061, right premaxilla and an- terior portion of right maxilla; QMF31062, relatively complete premaxilla; QMF31063, partial maxilla: QMF31064, maxillary fragment; QMF31065, maxil- lary fragment; QMF31066, maxil- lary fragment; QMF31067, den- tary; QMF31068, dentary; and QMF31069, pair of dentaries with splenial fragments, All from late Oligocene White Hunter Sile, Riversleigh. DIAGNOSIS, Snout broad, not as deep as in other species; rounded premaxillae, 14 maxillary teeth; nasals contact external nares; lat- eral border of the nasals without angulation at the maxilla-premax- illa boundary; non-serrated cari- nae; mandibular symphysis extending posteriorly to the Sth dentary teeth. ETYMOLOGY. For Professor Huber, Rektor of the Friedrich Wilhelms Universitit, Bonn DESCRIPTION. Skull anterior to orbit. Snout low, broad; pre- maxillary alveoli circular, 4th largest, 5 in juvenile, with 2nd alveolus reduced and almost lost, 4 in adult (2nd lost). Inci- sive foramen broad, tear- shaped; external nares apple-shaped, with a short an- terior process of the premaxilla and nasals on its posterior mar- gin. Deep reception pit for the first dentary tooth not reaching dorsal surface. Fourth dentary tooth reception notch promi- nent, with a secondary pit me- dial to it on the palate. Premaxillary-maxillary suture relatively straight, with a slight posterior convexity. Maxillary alveoli 14, arranged in a typi- cally crocodyline enlargement sequence with the Sth largest, laterally compressed particu- larly the posterior-most 5, Low alveolar process on the anterior 6 maxillary alveoli. Dentary tooth reception pits 5, well-developed, medial to the upper series, between 6th-1]th alveoli. Anterior teeth moderately robust, ovate in cross section. Posterior teeth with low, rounded crowns. All teeth with distinct anterior and posterior carinae. Palatal fenestra extending anteriorly to between the 7th and 8th maxillary teeth; straight suture with the palatine forming a short palatal process reaching anteriorly to 6th alveolus. Broad shelf 430 MEMOIRS OF THE QUEENSLAND MUSEUM FIG. 12. Baru wickeni, QMF 16822, holotype, dorsal view, snout, with line interpretation. F=frontal; J=jugal; L=lacrimal; Mx=maxilla; N=nasal; Pf=prefrontal; Pmx=premaxilla. Scale in cm. between palatal fenestra and posterior alveoli, rounded dorsally into the palatal fenestra and on to the internal surfaces of the maxilla. Sharp- crested ridge on the external surface of the max- illa at the line of the palate, starting above 10th alveolus, running off the posterior border of the maxilla. Nasals broad, contacting the external nares, gradually widening to the lacrimal-maxilla-nasal triple junction, then tapering more sharply. Short, pointed anterior process of the frontals dividing the posterior extremities of the nasals.Lacrimals with low, rounded canthi rostrales, about twice the size of the prefrontals. NEW CROCODILES FROM THE LATE OLIGOCENE 431 FIG. 13. Baru wickeni, QMF16822, holotype, snout, ventral view, with line interpretation. Ect=ectopterygoid; Mx=maxilla; Pa=palatine; Pmx=premaxilla. Scale in cm. Dentary fragments. Dentary pseudoheterodont, with an undulating tooth row. Symphyseal region deeper and larger than in Mekosuchus. Dentary built up around the base of each alveolus, with this buttressing variable in proportion to size of the alveolus and not strongly developed. Slightly raised area on the dorsal surface medial to the 4th and Sth alveoli. Dorsal margin of dentary be- tween and lateral to the 2nd and 3rd alveoli and between and lateral to the 7th, 8th and 9th alveoli with indentations for reception of teeth in the upper series, indicating that the upper series oc- cluded lateral to the lower series. Splenial extend- ing anteriorly to the 7th alveolus; symphysis extending to 5th alveolus. Sculpture of indistinct b tad te MEMOIRS OF THE QUEENSLAND MUSEUM FIG, 14. Baru wickeni , QMF 16822, holotype, partial left maxilla and premaxilla, lateral view. Scale = 2em. scarring on the ventral surfaces of the dentary merging to well developed pitting dorsally. DISCUSSION. This mandible is assigned to B. huberi because itis the only unassigned mandible of appropriate proportions and size range from White Hunter Site. QMF31068 is. an almost exact fit for OMF3 1060. The other 3 mandibular forms from White Hunter Site can be shown to belong to other taxa. Baru wickeni sp. noy. (Figs 12-17) MATERIAL. Holotype. QMF16822 (Figs 12-14) as- sociated posterior cervical and lumbar vertebrae and a calcaneum, Paratypes QMF31070, anterior portions of mandibles; NTM P8738-1, posterior right skull frag- ment and associated right anterior dentary fragment, NTM P8681-14, lett mandibic lacking the articular and adjacent angular and surangular posterior to the lateral foramen and a small portion of the dentary at the level of the third tooth; NTM P8738-1, right jugal, pterygoid, ectopterygoid and posterior maxilla and an associated dentary fragment; QMF16823, jugal fragment: OMF16824, premaxillary fragments; QMF16825, right dentary; QMF16826, right dentary. All from Oligocene (System A) Sile D, Riversleigh. QMF3107! and QMF31072, posterior portions of large mandibles and QMF31073, anterior dentary frag- ment. All from late Oligocene White Hunter Site, Riversleigh. SAMP27866, right premaxilla from late Oligocene Pancake Site, Riversleigh. QMF31074, right maxillary fragment and fragment of skull roof from late Oligocene Sticky Beak Site, Riversleigh. ETYMOLOGY. For ‘Tony Wicken, University of NSW, for supporting the Riversleigh Research Proj- cet. DIAGNOSIS. Snout narrower than B. huberi or B. darrow1, deep; anteriorly pointed premaxillae; 13 maxillary teeth; nasals entering external nares; anlerior termination of nasals long and thin, strongly constricted by premaxillae; non-serrated carinae; mandibular symphysis extending poste- riorly to 6th or 7th dentary teeth. DESCRIPTION. Skull material. Snout narrower than B. darrowi or B. huberi, similar in depth to B. darrowi. Premaxilla pointed anteriorly rather than rounded as in the other 2 species, Teeth similar to B. darrowi, lacking serrated carinae. Premaxillary alveoli in adults 4, in juveniles 5, with 2nd alveolus lost during growth. Maxillae with short posterior process medially invading lacrimal (not present in B. darrowi and unknown in B. huberi). Maxillary alveoli 13; palatal fenes- ta extending anteriorly to the level of 7th maxil- lary alveolus, Nasals entering premaxillae unlike B. darrowi but similarly to B. huberi, Anterior nasals distinctive in B. wickeni, being thin slivers strongly constricted between the premaxillae. Lacrimal with distinct, rounded canthus rostralis, extending for a short distance onto the maxilla. Jugal with well-defined, arched ridge on the ex- terior surface. Mandibles, Alveoli 15, subcircular except for the 4 slightly laterally compressed most posterior. Alveoli 3rd-6th on an alveolar process most strongly developed around the 4th alveolus. No NEW CROCODILES FROM THE LATE OLIGOCENE 43 ut FIG. 15. Baru wiekeni, NTM P8738-1, portions of the right side of the skull with associated dentary fragment, lateral view. Scale = 2cm. FIG. 17. Baru wickeni, QMF31070, dentary, dorsal view. Scale = 2cm. reception pits for teeth from the upper series but spacings and a lateral sulcus between 2nd and 3rd alveoli, 7th and 8th alveoli and between the 8th and 9th alveoli. Symphyseal region narrow. Sym- physis extending posteriorly to 6th alveolus; splenial reaching anteriorly to 7th. External mandibular fenestrae ovate, of moder- ate size and inclined posteriorly. Surangular nar- row dorsoventrally, inclined posteriorly with 434 MEMOIRS OF THE QUEENSLAND MUSEUM — FIG. 18. Dorsal views of reconstructed snouts of Baru huberi (top), B. wickeni (middle) and B. darrowi (bottom) showing differences in sutural relations, par- ticularly in the nasal-premaxillae sutures, and general proportions. Baru huberi based on QMF31060 (holo- type), B. wickeni based on QMF16822 (holotype) and B. darrowi based on NTM P8695-8 (holotype). Scale = 5cm. dorsal margin not parallel to the dentary. Angular slender, inclined. Smooth region for attachment of the posterior pterygoideus musculature sharply demarcated from the heavily sculptured areas of the angular and surangular by a low ridge. Artic- ular and retroarticular process short, broad and steeply inclined. FIG, 19. Ventral views of reconstructed snouts of Baru huberi (top), B. wickeni (middle) and B. darrowi (bottom) showing differences in general proportions. Baru huberi based on QMF31060 (holotype), B. wickeni based on QMF16822 (holotype) and B. darrowi based on NTMP8695-8 (holotype). Scale = Sem. DISCUSSION. This new species is based primar- ily on material from Site D. In describing B. darrowi, Willis et al. (1990) recognised that specimens from Bullock Creek differed from specimens from Riversleigh. How- ever, at that stage there was insufficient material to separate 2 species. Since then a large portion of a snout from Riversleigh (part of QMF16822) a fragment of which was in the original descrip- tion of B. darrewi has been rediscovered and NEW CROCODILES FROM THE LATE OLIGOCENE 435 prepared, This and other new material allows the material from Riversleigh to be allocated to a third species of Baru (Figs 18, 19). TWO CRANIAL FORMS White Hunter Site has produced sev- eral posteriors of crocodilian skulls and skull decks representing 2 similar forms. No specimen duplicates portions of other specimens so although the cranial forms almost certainly pertain to 2 of the taxa described above, they cannot be as- signed. WHITE HUNTER CRANIAL FORM 1 (Figs 20-22) MATERIAL. QMF31075, 31076 posterior of skulls; QMPF31077, skull fragment; QMF31078, isolated parietal. FIG. 20, Cranial form |, QMF31075, posterior portion of skull, dorsal view. Scale = lem. DIAGNOSIS. Supratemporal fenestrae tear-shaped with point directed an- lerolaterally and with posterior shelf formed by the squamosal; prominent expression of supra- occipital on skull deck; postorbital bar slender and round in section; postorbital-frontal suture twice the length of postorbital-parietal suture; foramen magnum wider than occipital condyle; width of supratemporal fenestrae greater than width of postorbital: sculpture of more or less regular pits closely spaced. DESCRIPTION. Wide across the skull deck, high with the quadrate tucked under the squamo- sals. Supraoccipital prominent on the dorsal sur- face, forming a broad triangle almost excluding parietals from posterior margin of skull deck. Supratemporal fenestrae with an anterior point, teardrop-shaped, with much of the posterior and medial portions closed by a floor formed by the squamosal and parietals inside the supratemporal fenestrae. Posterior face of the skull with pro- nounced concavities on exoccipitals and squamo- sals for attachment of mandibular depressor muscles. Paroccipital process encroaching ven- trally onto the quadrate. Foramen magnum sub- triangular, wider than the occipital condyle. Basioccipital with pronounced keel ventral to the occipital condyle. Quadrates steeply inclined. Pterygoids forming large portion of the posterior margin of the palatal fenestrae; internal nares, although not preserved, must have been well to- ward the posterior of the pterygoids. Otic meatus and foramina for the trigeminal nerve proportion- ally large. Laterosphenoids with a pronounced longitudinal crest medially on the ventral surface. Sculpture on the skull deck distinctive, deep and well-defined pits separated by equally distinct, uniform walls. Pits close spaced. WHITE HUNTER CRANIAL FORM 2 MATERIAL. QMF31079, anterior fragment of skull deck: OMF3 1080, right postorbital. DIAGNOSIS. Small supratemporal fenestrae lat- erally compressed, shallowly floored by squamo- sals; postorbital bars inset from skull deck margin, robust and triangular in section; postor- bital-frontal suture equa! in length to postorbital- parietal suture; width of postorbital greater than width of supratemporal fenestrae; sculpture of irregular shaped pits with irregular distribution. DESCRIPTION. WH 2 is described where: it differs from WH 1. Supratemporal fenestrae narrower; squamosal flooring making supratemporal fenestrae shal- lower posteriorly. Sculpture pits on WH 2 are small and irregular, separated by thick, irregular walls. Sculptured skull deck overhanging postor- bital bar on WH 2 but in WH 1 postorbital bar 436 marginal. Postorbital bar mod- erately robust, with triangular cross section. Postorbital very large compared to. the supra- temporal fenestrae. Triple junction between the postor- bilal, frontal and parietal dis- tant from margins of supratemporal fenestrae. DISCUSSION, The frontals associated With QMF3 1076 are very different from those re- ferred to Mekosuchus (QMF31052) in being nar- rower and flat between the or- bits. They are also deeper and have better defined crania cris- tae frontalis thin QMF31052. Thus cranial form | can be con- fidently excluded from Mekosuchus (but not Baru or Ouinkana). Although there are no fron- lals unambiguously associated with cranial form 2, the difference in sculpture (compared with QMF3 1052) and the thickness of the orbit margins. of the postorbital make it un- likely that this cranial form represents Mekosuchus, FIG. PALAEOECOLOGY Four crocodilians have not previously been found in a single fauna in Australia, However, compared with world faunas, this is notan unusu- ally high diversity of crocodilians, particularly when the 4 species have differing head shapes or when the site perhaps represents a thandlocenosis collected from 2 or more habitats, Among extant crocochilians, many species have ranges that overlap but true sympatry is not com- mon, In parts of South America 5 or 6 crocodilian ranges overlap but rarely do 3 or more share the same habitat (Gorzula, 1987; Magnusson & Lima, 1991), The range of Crocodylus porosus encompasses the ranges of C. johnstani, C. novaeguinede, C, mindorensis, C. siamensis, Tomistoma sċhlegelii and parts of the range of C. palustris and Gavialis gangeticus (Ross & Mag- nusson, 1989; Groombridge, 1987) but rarely do any of these species existin true sympatry, Where C. porosus and C. johastoni have been found in sympatry the larger C. porosus tends to exclude C. johnstoni to the margins of the habitat or MEMOIRS OF THE QUEENSLAND MUSEUM 21. Cranial form 1, QMF31075, posterior portion of skull, posterior view. Scale = lem. sympatry is restricted by the need for different nesting substrates (Webb et al., 1983). Modern studies of crocodilians in the Amazon Basin indicate that larger watercourses are occu- pied by larger, generalised crocodilians such as Melanosuchus niger and Caiman crocodilus as well as Paleosuchus palpebrosus, while smaller watercourses im closed canopy forests are occu- pied only by the more derived, deep-headed P, trigonatus (Magnusson, 1987, Magnusson & Lima, 1991). This could suggest that the variety of crocodilian head shapes at White Hunter Site is the result of a thanatocenosis collected fram 2 or more different habitats, Theoretically, 2 crocodilians may live sympat- rically when there are differences in head shape (implying exploitation of different prey) or where small, broad-snouted species can evade larger broad-snouted species by escaping t0 marginal habitats (Meyer, 1984). In no extant crocodilian fauna, do 2 species share the same head shape (Meyer, 1984), The most diverse fossil crocodilian fauna is from the La Venta fauna m Colombia which consisted of § species (4 broad-snouted, 1 duck- bill, 2 narrow-snouted and | ziphodont; Lang- ston, 1965), However, thal fauna is a thanato- coenosis from over 240m stratigraphically. Sym- patry was not demonstrated. NEW CROCODILES FROM THE LATE OLIGOCENE 437 The Messel fauna of Ger- many is more likely a bioceno- sis and has 6 crocodilian spe- cles including large and small broad-snouted forms, a short- snouted form and two ziphodonts. A similar assem- blage has been recovered from Geisalltal (Kuhn, 1938; Haubold. 1983; Haubold & Krumbiegel, 1984), There are two possible expla- nations for the diversity of crocodilians in White Hunter Site. Baru wickent is a large broad-snouted form while B. huberiis a much smaller broad- snouted form; Mekosuchus whitehunterensis is a small, short-snouted form; and Quinkana meboldi is a ziphod- ont, Compared to other fossil sites around the world and to modern analogues, the White Hunter assemblage differ enough to be sympatric ex- ploiting different niches. Alter- natively, they may indicate a thanatocenosis from two or more different habitats. Arrangements of differing ecomorphs of groups of mammals from sites at Riversleigh support the hypothesis that these faunas represent biocoenoses, A complex fauna of 8 species of bandicoots, belonging to clearly defined guilds, has been recovered from Upper Site (J. Muirhead, pers. comm.). Similarly, several species of ring- tail possums of differing ecomorphs have also been found in many sites at Riversleigh (M. Archer, pers. comm.). This pattern is apparently repeated in several other groups of mammals currently being investigated. That the Riversleigh mammalian faunas repeatedly show sympatry be- tween several closely related taxa supports the hypothesis that Riversleigh sites preserve bioce- noses rather than thanatocenoces. This supports the hypothesis that the White Hunter crocodilians were also sympatric. CONCLUSIONS The 4 crocodilians from White Hunter Site include the first record of Mekosuchus outside New Caledonia and demonstrates a surprising morphological diversity suggesting significant FIG, 22. Cranial form 1, QMF31076, posterior portion of skull, dorsal view Scale = lem. niche separation, This is the first record of such a diverse crocodilian fauna from Australia but it is consistent with the structure and complexity of crocodilian faunas known from elsewhere, By comparison with other Riversleigh faunas the crocodilian fauna of White Hunter Site was prob- ably typical for Oligo-Miocene Australia. ACKNOWLEDGEMENTS I thank John Scanlon, Mike Archer, Mark Norell and Ralph E. Molnar for constructive com- ments. I thank Mike Archer, Suzanne Hand and Henk Godthelp for access to specimens. This study was part of a PhD Studentship at the Universily of New South Wales. Financial sup- port was provided by the University of New South Wales, Friedrich-Wilhelms-Universitat, Bonn and the Rektor of that institution. LITERATURE CITED ARCHER, M., GODTHELP, H., HAND, SJ, & MEGIRIAN D. 1989. Fossil mammals of Riversleigh, northwestem Queensland: Prelimi- hary overview of biostratigraphy, correlation and cavironmental change. Australan Zoologist 25: 27-65, ARCHER, M., HAND, S.J. & GODTHELP, H. 1991. Riversleigh, 2nd edition. (Reed Books: Sydney) BALOUET, J.C.. & BUFFETAUT, E. 1987. Mekosuchus inexpectatus n. g. n. sp, Crocodilian nouveau de l'Holocene be Nouvelle Caledonie. Comptes Rendus Hebomadaires des Séances de l'Académie des Sciences Paris 304; 853-57. BERG, D.E. 1966. Die Krokodile, insbesondere As- tatosuchus und aff. Sebecus?, aus dem Eozän von Messel bei Darmstadt/Hessen, Abhandlungen des Heanjòchern Landesamtes für Bodenforschung 52: USS. BUFFETAOT, E. 1982. Un probléme de paléobjogéographie continental: Jes crocodilies mésosuchiens ziphodontes de |'Eocéne curopéen, Bulletin de la Société Géologique de France 24: 5-6, 1937. The ziphodont mesosuchian crocodile from Messel: reassessment, Intemational Messel Symposium, Frankfurt, Abstracts. BUSCALIONI, AD, SANZ, LL. & CASANOVAS, ML 1992. A new species of the eustichian croc- odile Diplocynedon from the Eocene of Spain. Neues Jahrbruch fiir Geologie und Paldontologie Abhunlungen 187; 1-29, FREY, E. VON, LAEMMERT, A, & RIE, J. 1987, Baryphracta depaniae n. g. n. sp. (Reptilia, Crocodylia), ein neues krokodil nus der Grube Messel bei Darmstadt (Hessen, Bundesrepublik Deutschland). Neues Jahrbruch für Geologie und Paliiontologie Monatshefte 1: 15-26. GORZULA, S. 1987. The management of crocodilians in Venezuela, Pp, 91-101, In Webb, GJ,W,, Manolis, S.C. & Whitehead, P.J. (eds), Wildlife management; crocodiles and alligators. (Surrey Beatry & Sons: Sydney). GROOMBRIDGE, B. 1987. The distribution and status of world. ctocodilians,Pp, 9-21, In Webb, GJW., Manolis, S.C, & Whitehead, P.J. (eds), Wildlife Management; crocodiles and alligators, (Surrey Beatty & Sons: Sydney), HAUBOLD, H. 1983. Wiereltiere (Vertebrata) tn Krombiegel, G., Riitfle, L. & Haubold.H.V. (eds), Das Euciine Geiseltal; ein Mitteleüropäisches Braunkohenvorkommen und seine Pflanzen- und Teirwelt, (A, Ziemsen Verlag: Lutherstadt), HAUBOLD, H. & KRUNBINGEL, G. 1984, Typenkatalog der Wierbeltiere aus dem Eozan des Geiseltals. (Martin-Luther-Universitit Halle: Wittenberg). KUHN, O, 1938, Die crocodilier aus dem mittleren Eozén des Geiseltales be: Halle. Nova Acta Leopoldina 6: 313-328, (968. Die Voreeitlichen Krokodile, (Ocben: Mu- nich). LANGSTON, W. 1965. Fossil crocodilians from Co- lombia and the Cenozoic history of the Crocodilia in South America, University of California Publi- cations in Geological Science 52: 1-157. MEMOIRS OF THE QUEENSLAND MUSEUM MAGNUSSON, W.E, 1987, Diets of Amazonian croc- odiltans. Journal of Herpetology 21: 85-95, MAGNUSSON, W.E. & LIMA, A. 1991. The ecology ofa cryptic predator, Paleosuchus trigenams, in a topical rainforest. Joumal Of Herpetology 25; 41-48. MEGIRIAN, D, 1994. A new species of Ouinkana Molnar (Eusuchia: Crocodylidac) from the Miocene Camfield Beds of Northern Australia, The Beagle |}: 145-160, MEGIRIAN, D. MURRAY, P.F. & WELLS, R.T, 1993. The Late Miocene Ongeva Local Fauna from the Waite Formation of central Australia, CAVEPS ‘93, Adelaide, Abstract, MEYER, E.R, 1984, Crocodilians os living fossils Pp, 105-131. In Etheridge, N, & Stanley, S.M. (eds), Living fossils. (Springer Verlag: New York) MOLNAR, R.E. 1981. Pleistocene ziphodont crocodil- ins of Queensland, Records of the Australian Museum 33; 803-834. MURRAY, P., & MEGIRIAN, D. 1992. Continuity and contrast in Middle and Late Miotene vertebrate communities from the Northern Territory. ‘The Beagle 9: 195-218. MURRAY, P., MEGIRIAN, D. & WELLS, R. 1993, Kalopsis yperus sp. nov, (Zygomaturinae, Marsupialia) from the Ongeva Local Fauna: New evidence for the age of the Aleoota fossil beds of Central Australia, The Beagle 10; 155-172. ROSS, C.A. & MAGNUSSON, W.E. 1989. Living crocodilians, In Ross, C.A. (ed.) Crocodiles and alligators. (Golden Press: Sydney). SALISBURY, S.W., & WILLIS, P.M.A, 1996. A new crocodylian from the early Eocene of south- easi¢m Queensland and a preliminary investiga tion of the phylogenetic relationships of erocodyloids, Alchering 20: 179-226. WEBB,G,J.W,,MANOLIS,S,C. & SACK, G.C. 1983 Crocodylus johnstani and C. porosus cocxisting in a tidal river, Australian Wildlife Research 10: 639-50, WILLIS, P.M.A. 1992. Four new crocodilans from early Miocene sites at Riversleigh Station, north- westem Queensland. The Beagle 9: 269 WILLIS, P.M.A,, MURRAY, P.F, & MEGIRIAN, D. 1990. Bark darrowi veh. et. spi nov., a large, broad-snouted crocodyline (Eusuchia: erocodylidac) from mid-Tertiary freshwater lime- stones in Northern Australia, Memoirs of the Queenshind Museum 29: 521-540, WILLIS, P.M.A., MOLNAR, R.E. & SCANLON, J.D. 1993. An early Eocene crocodilian from Murgin, southeastem Queensland. Kaupia: Darmstädter Beltrtige zur Naturgeschichte 3: 25-32. WILLIS, PMLA, & MACKNESS, B, 1996, Quinkana babarra,anew species of ziphodont mekosuchine crocodile from the early Pliocene Blulf Downs Local Fauna, northem Australia with arevision of the genus; Joumal and Proceedings of the Linnean Society of New South Wales 116: 143-151. MAYIGRIPHUS ORBUS GEN, ET SP. NOV., A MIOCENE DASYUROMORPHIAN FROM RIVERSLEIGH, NORTHWESTERN QUEENSLAND S. WROE Wroe, S. 1997 0630: Mayigriphus orbus gen, et sp. nov- a Miocene dasyuromorphian from Riversleigh, northwestern Queensland, Memoirs of the Queensland Museum 41(2): 439-448. Brisbane. ISSN 0079-8835. Mayigriphus orbus gen. èt sp. nov., an enigmatic Miocene dasyuromorphian from Riversleigh, is described from dental material. The tiny Mayigriphus orbus shows a number of derived character-states for Dasyuromorphia and two of these derived features may signify a special relationship with Planigale (Dasyuridae). However, no specialised features shown by M. orbus are unique to dasyurids within the order and M. orbus also possesses derived characters shown by basal thylacinids. Because previous investigation has indicated that Dasyuridae is not currently defined by any dental synapomorphies, caulion is demanded regarding allocation of M. orbus at the family level. Problems associated with the phyloge- netic placement of M. orbus portend a story of growing complexity for dasyuromorphian phylogeny — astory progressively being revealed in the Tertiary limestones of Riversleigh, (\Mayigriphus, dasyuromorphan, Miocene, Riversleigh. 5. Wree, School of Biological Sciences, University af New South Wales, Sydney, New South Wales 2052, Australia; received 4 November 1996, The fossil record for Dasyuromorphia is re- viewed by Wroe (1996b, 1997b). Until recently the pre-Pliocene fossil record for Dasyuromorphia was limited to five described taxa, all from deposits in central Australia (An- kolarinja tirarensis and Keeuna woodburnei (Archer, 1976a), Wakamatha tasselli (Archer & Rich, 1979), Dasylurinja kokuminola (Archer 1982a) and Thylacinus potens (Woodbume, 1967 ), With the exception of Thylacinus potens, inves- tigation has failed to unequivocally link these fossil taxa with elements of modern dasyuromorphian radiations. More recently, the fossil-rich middle to late Tertiary deposits of Riversleigh have yielded six new thylacinid spe- cies: Nimbacinus dicksoni (Muirhead & Archer, 1990), Thylacinus macknessi (Muirhead, 1992; Muirhead & Gillespie, 1995); Wabulacinus ridei (Muirhead, 1997), Ngamalacinus timmulvaneyi (Muirhead, 1997), Badjcinus turnbulli (Muirhead & Wroe, in press), and Murthacinus gadiyuli {Wroe, 1996b). However, only two un-named taxa have been assigned to Dasyuridae: a possible phascogaline taxon known from a single M ! or M 2 (Archer, 1982a) and an un-named ‘An- techinus-like’ species from Riversleigh (Van Dyck, 1989). Wroe (1996b, 1997b) investigates problems with dasyurid phylogeny, concluding that Dasyuridae is currently defined by possibly three basicranial, but no dental synapomorphies. A new dasyuromorphian described here shows an enigmatic combination of features within Dasyuromorphia and can not be unequivocally assigned at the family level, Dental nomenclature follows Flower (1867) and Luckett (1993). Taxonomic terminology for Dasyuromorphia follows Wroe (19965), wiih three subfamilies recognised within Dasyuridac (Sminthopsinae, Phascogalinae [including Murexia], Dasyurinae [including Neephascogale and Phascolosorex]) and the following taxa con- sidered Dasyuromorphia incertae sedis: Anm- kotarinja tirarensis, Keeuna woodburnei, Wakamatha tasselli and Dasylurinja kokauninola. Higher level marsupial systematics follows Mar- shall et al. (1990). Material is housed in the Queensland Museum (QMF)}, SYSTEMATICS Order DASYUROMORPHIA Gill, 1872 Family INCERTAE SEDIS Mayigriphus gen. nov, TYPE AND ONLY SPECIES. Mayigriphus orbus gen, etsp. nov. GENERIC DIAGNOSIS. Mayigriphus orbus differs from all dasyurids in the following combination of features: Premolar row compressed longitudinally. Pi very small; P3 reduced but with two roots; M] corn pressed on long axis with protoconid central on long and transverse axes with paraconid tiny, Mj.4 metaconids and metacristids reduced; M] metaconid not differentially reduced relative to M2-4 metaconids; fingual anterior termination of cristid obliyul on Mig, with M3 cristid obliqua terminating beneath metacnstid carnassial notch, Mj.4 protoconids lin- gually shifted and recurved; M1-3 entoconids small to tiny, M4 talonid reduced with entoconid absent. May- igriphus orbus can be distinguished from known thy- lacinids by the following combination of features: M1 shows agreatly reduced paruconid butunly moderately reduced! metuconid; clearly detined hypocunulid notch in anterior cingulid of lower molars; very small size; reduction of P3 relative to Py, lack of diastema between Pı and Pz. Mayigriphus orbus differs from known bandicoots in possession of the above combination of vhoracters, a well-defined posterior cingulid and more buccally shifted hypoconulid. ETYMOLOGY. Wanyi mayi tooth; Latin griphus, puzzle; refers to the enigmatc combination of dental features, Masculine, Mayigriphus orbus sp. nov. (Fig. 1) ETYMOLOGY, Latin orbus, orphan, refers 10 ils un- certaln phylogenetic posiuon. MATERIAL. Holotype, QMF23786, right dentary with partial anterior alveolus of Pi, Pi posterior root, P23, M1-4; Paratype QMF22791, right dentary fragment containing M3 and alveolus for Ma. All from early late Miocene Encore Site, Riversleigh. DESCRIPTION, Dentary broken away anteriorly from midpoint of P; anterior root alveolus and posteriorly from abont Imm along ascending ramus; dentary almost uniform in depth, shght tapering anteriorly from beneath P3 protoconid; mental foramen beneath Mı hypoconid. Pi. P; crown missing, only posterior half of anterior rool alveolus and posterior root remain; bused on root and alveolus size Pi small, less than half Po length: amerior alveolus buceally dis- placed. Pa No diastema between P) und Ps; twin tooled; Py largest premolar in height and length; protoconid moderately worn; buccal cingulid runs posteriorly from midpoint between anterior and posterior roots to small posterior central cuspule un heel; another cingulid circumscribes the lingual crown base from this cuspule to ante- nor margin of posterior root. P3. no diastema between Pz and P3; twin rooted; P4 morphology similar to Pa but differs in posses- sion of continuous cingulid circumscribing base MEMOIRS. OF THE QUEENSLAND MUSEUM of entire crown and smaller size (P3 around 30 percent smaller in height an length). Mı. no diastema between P3 and Mi; Mı un- worn; principal cusps in order of decreasing height, protoconid, metaconid, paraconid. hypoconid and entoconid; entoconid tiny, closely abutting posterior face of trigonid adjacent to metaconid; paraconid damaged, but from basal dimensions was clearly small; metaconid small and shifted posteriorly; protoconid dominant cusp, lingually recurved, occupying almost cen- tral position on tooth; Mı reduced on the long axis; talonid small, slightly wider transversely than trigonid bul shorter on long axis; paracristid parallel to, and cristid obliqua shows slight lin- gual inflection at anterior end; metacristid and hypoeristid parallel and angled at about 20° to Lransverse axis of dentary; cristid obliqua termi- nates beneath apex of protoconid; anterior cingulid runs basally from paraconid to beneath protoconid; posterior cingulid weakly developed. Mz. Me differs from Mı as follows; M2 much larger; paracomd much larger; metaconid relä- tively and absolutely Jager though still small compared to protoconid; talanid shorjer on long axis; cristid obliqua terminates against posterior face of trigonid in more lingual position, with angle formed between cristid obliquu and hypocristid more acute; metacristid and hypoeristid run closer to transverse axis of den- lary; entoconid relatively larger than in Mi though still small; posterior cingulid more strongly developed; Mp paracristid runs at about 30° to long axis of dentary, with angle between paracristid and metacristid slightly less than 90°, M3. M3 differs from Mz as follows; protoconid larger; on transverse axis trigonid wider and talonid shorter; entoconid on Ma smaller; cristid obliqua terminates in a more lingual position against posterior face of trigonid beneath carnas- sial notch of metacristid, Ma. Ma similar to M3 except: metaconid smaller than paraconid, protoconid less lingually re- curved; talonid greatly reduced, entoconid ab- sent, hypoconid and hypoconulid small; no posterior cimgulid. Meristic gradients from Myj-4, orientation of metacristid and hypoeristid to long axis of den- tary increasingly transverse from My-2, departs away from transverse from M24, then back to more transverse orientation from M3.4: orienta tion of paracristid to long axis of dentary increas- FIG. 1. Mayigriphus orbus sp. nov., QMF23780, holotype. A, buccal view. B and D, sterea-pair, occlusal view of Mı-4. C and E, stereo-pair, occlusal view of Pi posterior alveolus, P2.3, M1-4. MAYIGRIPHUS ORBUS, ENIGMATIC DASYUROMORPHIAN MEMOIRS OF THE QUEENSLAND MUSEUM TABLE I. Mayigriphus orbus, dental measurements (mm). l=anteroposterior dimensión, w l=maxi muni trans- Sera erie jen [wid | es f wid | i Tor [w|i] verse dimension of trigonid, W2=maximum transverse dimension of talonid. MAMAE 087 raas| 126] o60|067| 1a [ues] ose[1s2| nes] om 1.33 MF 2 MF? Cam T = OI T T T basioslos] ingly transverse from Mı-3, termination of cristid obliqua against posterior face of thgonid increas- ingly lingual Mj.4; reverses departing from trans- verse for Ma-4; protoconid and metaconid height increases M).3, decreases Ms.4; paracanid height increases Mj.4; talonid width increases M).2, de- creases Mag. CHARACTER ANALYSIS Wroe (1996b, 1997b) discusses taxa consid- ered appropriate in the reconstruction of a dasyuromorphian morphotype (i.c., peradectids, didelphoids, microbiotheriids, peramele- morphians), Similar methodology is used for characters treated here to assess character-stale polanties (Tables 2, 3), Within Dasyuromorphia Ankotarinja tirarensis is the least derived taxon and a possible outgroup fo remaining dasyvromorphians, Muribacinus gadtyuli and Badjcinus turnbulli are the least derived thy- lucinids. For Dasyuridae the following laxa are considered plesiomorphic for their respective subfamilies: Murexia longicaudata (Phascogalinae), Neophascogale lorentzii (Dasyurinae) and Sminthopsis leucopus (Sminthopsinue). P3, A P3 larger than P2 is plesiomorphic for outgroups to Dasyuromorphia (Wroe, 1996b, 1997b). For A. rirarensiy Ps is slightly smaller than P2. Wakamatha tasselli shows P3 larger than Pa. The basal thylacinid Badjcinus turnbulli has P3 slightly smaller than Pa Wroe (1996b) inter- prets. similar P} morphology for Muribacinus gadiyuli on the basis of alveolar dimensions, For all remaining thylacinid taxa P3 is larger than Pa, with the largest P3 in the most derived laxa (7hy- lacinus), Within Dasyuridae wide variation is apparent for this feature, Taxa treated here as basal to their respective subfamilies show a P3 larger than Po (Murexia, Sminthopsis leucopus), or slightly reduced (Neophascagale), However, within each subfamily some taxa show marked reduction or absence of P3, P3 reduction in May- igriphus orbus exceeds that shown by all taxa considered Dasyuromorphia incertae sedis, all Thylacinidae and basal taxa for dasyurid sub- families. MI paraconid. The Mj paraconid is not reduced in dasyuromorphian oulgroups and A: rrarensts, W. tasselli, moderate reduction 1s shown by thy- lacinids (excepting B. turnbulli which shows marked reduction) and plesiomorphic Sminth- opsinae (Sminthopsis leucopus), Phascogalinac (Murexia) and Dasyurinae (Neophascegale). Mj paraconid reduction in M. orbus is less marked than in all Dasyurinae excepting Neophascogale. For M. orbus Mı shows greater reduction of the paraconid than for all Phascogalinue and Sminthopsinae except Planigale. Mg talonid. The Mg talonid is unreduced in basal taxa for Dasyuromorphian outgroups except han- dicoots (e.g., Yarala burelifieldi, Muirhead & Filan, 1995). Within Dasyuromorphia A, riruren- sis, basal phascogaline (Murexia) and dasyurine (Neophascogale) taxa show slight reduction for this feature. The My talonid is greatly reduced on the plesiomorphic dasyuromorphian condition for W. tasselli, most dasyurines and a rok ht and all sminthopsines. For thy- acinids the Mytalonid is unreduced for basal taxa (Muribacinus, Badjcinus), but significantly re- duced for derived species (Thylacinus, Wabulacinus). Even for specialised Thylacinus Msg talonid reduction does not approach that of derived dasyurids Which show far greater dimi- nulion on the transverse axis, The degree of re- duction for this feature in M. orbus is closest to that shown in Phascogale, but less than for most Dasyurinae, and all Sminthopsinae. Cristid obliqua orientation, A buccal position for the anterior termination for the cristid obliqua relative to the carnassial notch of the metacristid is common lo most dasyuromorphian outgroup taxa. This feature is associated with eristid ob- liqua orientation and formation of a right angle between the cristid obliqua and hypoeristid. Most outgroup taxa to Dasyuridae and Thylacinidae show a cristid obliqua aligned closely with the long axis of the dentary and a 90° angle is formed between the cristid obliqua and hypocristid. In the MAYIGRIPHUS ORBUS, ENIGMATIC DASYUROMORPHIAN character-analysis (Table 2) only M3 is consid- ered, Buccal termination is shown by A. firarensis and, to a lesser degree, by K. woodburnei and W. tasselli. Basal thylacinids show relatively lingual anterior termination for the cristid obliqua, but a more buccal position is apparent in Thylacinus. Basal dasyurines show lingual termination for the cristid obliqua (Neophascogale, Myoictis, Dasyurus hallucatus), but more specialised taxa show buccal termination (other Dasyurus, Dasycercus, Dasyuroides, Sarcophilus). All phascogalines show lingual termination. Sminthopsines show buccal termination. May- igriphus orbus shows lingual termination. Orientation of the cristid obliqua correlates with other features of both the upper and lower molars. These include the angle between the postparacrista and premetacrista (together termed the centrocrista), the relative size of the pro- toconid and metaconid and, the occlusal surface area presented by the protocone and talonid basin. Scoring of character states for cristid obliqua withoutconsideration of these associated features may be phylogenetically misleading. For exam- ple, derived Thylacinus and some dasyurines show longitudinal alignment for the M3 cristid obliqua (unspecialised didelphids, microbio- theriids and A. tirarensis), but for these derived dasyuromorphians this feature correlates with protoconid hypertrophy, metaconid reduction or loss, and a linear centrocrista. These character- states are all associated with the dominance of longitudinally oriented vertical shearing crests. The basal position for Sminthopsinae within Dasyuridae indicated by molecular analyses (Kirsch et al, 1990; Krajewski et al., 1993; 1994) supports the contention that a buccal point of termination for the cristid obliqua is a plesiomor- phy for the clade. However, dental features of Sminthopsinae are products of a different selec- tive regime and transverse rather than longitudi- nal vertical shearing crests dominate. Archer (1976) noted that a buccal position for the cristid obliqua may be associated with reduction of the paracone or a lingual shift in the carnassial notch (of the metacristid), Both derived features are shown by sminthopsines and it is probable that cristid obliqua position represents a correlated apomorphic feature. A further derived feature shown by sminthopsines is gross reduction of the talondis on the anteroposterior axis which may also impact on cristid obliqua orientation. For sminthopsines and derived dasyurines and thy- lacinids, a buccal position for the cristid obliqua is treated as derived relative to that of microbio- 443 theriids, unspecialised didelphids, bandicoots and A. tirarensis (Tables 2, 3). A relatively more lingual termination for the cristid obliqua, as shown by most dasyurids and basal thylacinids, is also considered derived. The character com- plex associated with most dasyurids (a relatively lingual anterior termination point and acute angle formed between the cristid obliqua and hypo- cristid) is scored as ‘a’. Buccal termination and formation of 90° between the cristid obliqua and hypocristid may be associated with increased transverse vertical shear (b) or increased longitu- dinal vertical shear (c). Angle between paracristid and metacristid. For dasyuromorphian outgroups an acute angle is formed between the paracristids and metacristids (mirrored by an equivalent angle formed between the postmetacristae and preprotocristae with which they occlude in the upper dentition). Sim- ilar morphology is shown by A. tirarensis, W. tasselli and sminthopsine dasyurids. All dasy- urines, phascogalines, basal thylacinids for which a metacristid is retained, and M. orbus show a relatively obtuse angle between paracristids and metacristids. Widest paracristid-metacristid an- gles are in Sarcophilus, Glaucodon and D. maculatus among dasyurids and Ngamalacinus among thylacinids. This phenomenon is corre- lated with carnivory and the development of lon- gitudinally aligned vertical shearing crests. Hypoconulid notch, Many marsupials have a dis- tinct notch in the anterior cingulae of their lower molars to receive the hypoconulid of the preced- ing tooth. Outgroup data for Dasyuromorphia regarding this feature is equivocal. Some out- group taxa (e.g., some peradectids) show a well- developed hypoconulid notch, but among other possible outgroups this feature is absent (e.g., peramelemorphs). Within Dasyuromorphia this feature is well-developed for Ankotarinja tirarensis and Keeuna woodburnei, but poorly defined for Wakamatha tasselli (see Wroe (1996b) re arguments for possible bandicoot af- finities of this taxon). Among thylacinids, a well- developed hypoconulid notch is present for Muribacinus and Badjcinus, weakly-defined in Neamalacinus, and absent in all other taxa. A well-developed hypoconulid notch occurs in all dasyurids excepting Dasyurus maculatus (re- duced), and Glaucodon and Sarcophilus (absent). Mayigriphus orbus has a well-defined hypo- conulid notch. Wroe (in press b) infers that loss of the hypoconulid notch is a function of ad- 444 MEMOIRS OF THE QUEENSLAND MUSEUM TABLE 2, Characters and character-states used in phylogenetic analysis. C1. P; size relative to Po. 0. larger. t. reduced, 2 . intermediate. 3, tiny. 4. absent, C2, Mi paraconid size, 0. large. 1. reduced, 2. tiny. 3, absent. C3. M; talonid size. 0. large. 1. moderately reduced. 2. markedly reduced. 3. Way, C4. Ma cristid obliqua morphology +. P. plesiomorphic. a, lingual. b. trans. shear. c- long. shear- CS. Angle between paracristid-metacristid. 0, acute. L. intermediate. 2. obtuse. C6. Hypoconultd notch. 0. well developed. 1. intermediate. 2. absent. C7. Metaconid morphology. 0. no clear differential between M| and M2-4. |. clear differential. C8. Mz metaconid size. O. large. 1. reduced. 2. greatly reduced. 3. absent. C9, M3 entoconid size. 0. large. 1. reduced, 2. absent. * Three derived states recorded for this character (sce text). vanced camussialisation for derived dasyurids and thylacinids, noting that the shift to a predom- inance of longitudinal vertical shear in marsupial carnivores diminishes the requirement for a brace ugainst transverse forces (i.e, the likely role for the hypoconulid notch). Archer (1982b) regarded a hypoconulid notch in the anterior cingulum as a possible dasyurid synapomorphy. But Wroe (1997) concludes that it may have been in the cammon ancestor of Dasyuromorphia and was almost certainly in the common ancestor af Dasyuridae-Thylacinidac, For specialised dasyurids (Sarcophilus and Glaucadean) and thylacinids (Thylacinus, Wabul- acinus) loss Of the hypoconulid notch correlates with advanced carnussialisation. Metaconid . A well-developed metaconid on all lower molars occurs in all putative Dasyuro- morphian outgroups, The same is so for A, tyrarensis, K. woodburnel and W. tasselli. All thylacinids show marked reduction or loss of the inetaconid on all lower molars. Among Dasyoridae this feature ts variable. Plesio- morphic taxa for each subfamily show no reduc- lion of the metaconids. However, specialised Dasyurinae and Sminthopsinae show derived character-states. Derived dasyurines show a re- duced Mı metaconid, but Jess reduction for M2-4 metaconids (Dasycercus, Dasyuroides, Dasy- urus, Sarcophilus). In Planigale metacomid dim- inution is less advanced on M1 and more uniform through M24. This phenomenon shown for Planigale is also common to thylacinids which relain Metaconids (excepl Badjcinus turnbulli which shows the typital dasyurine condition), Mayigriphus orbus shows uniform reduction of Mı-4 metaconids, the character-state common to Planigale among dasyurids and Muribacinus, Nimbacinus and Ngamalacinus among thy- laçiwids, Localised metaconid reduction (i.e., a clear dit- ferential shown between Mı and Mz-4 metaconid reduction) as shown by some dasyurines, sminthopsines and Badjcinus, is probably related to brachycephalisation, shortening of the tooth row on the anteroposterior axis and concomitant premolarisation of Mi (Archer, 1976), Generalised metaconid reduction (Mi-4) corre- lates With increased size of the protoconid and primacy of the paracristid and postmetacristae in vertical shear, For carnivorous das yurids and thy- Jacimids these derived features arc associated with alignment of the vertical shearing crests with the long axis of the tooth row, Entoconid. Large entoconids are plesiomorphic for dasyuromorphian outgroups, A. rirarensis, K. woodburnei, basal dasyurines and sminthopsines, and phascogalines. Entoconids are reduced or absent in some Sminthopsis and Antechinomys among sminthopsines and Parantechinus, Pseudantechinus, Dasyuroides, Dasycercus, Pasykaluta, most Dasyurus (excepting D, hallucatus), Sarcophilus and Glaucodon. All thy- lacinids shaw some entoconid reduction, with the least reduction in Muribacinus and the greatest by Thylacinus. Mayigriphus orbus shows moderate reduction for this feature, Archer (1981) and San- son (1985) note that no clear form-function rela- tionship explains the distribution of entoconid reduction and loss among dasyurids, but note that this reduction is greatest for arid-adapted species (some Sminthopsis. Antechinomys, Dasyuroides, Dasycercus). However, considerable reduction or loss is shown for some species found in less extreme environments and for large dasyurid and thylacinid camivores this phenomenon is likely associated with carnivory. More form-function data is required here, MAYIGRIPHUS ORBUS, ENIGMATIC DASYUROMORPHIAN ‘TABLE 3, Character/taxon matrix, Z11a2 00000 00000 10000 00100 00200 10000 ??7al 07200 11002 110a2 70122 701e¢1 O77? G2 O0lc? 00262 10260 21360 313b0 213b0 Ollel l0lal LOlal L21al 223cl 32281 32202 3232 Mayigriphus or bis Alphadon marshi Marmosa 3p Didelphis marsupialis Dròmiciops australis Yarala burchfieldi Ankotarinja tirarensis Keeuna woodburnei Wakamatha tasselli Muribacinus gadiyuli Badjainus turnbulli Nimbacinus dicksoni Ngamalacinus timmidvaneyi Wabulacinus ridet Thylacinus mackness: Thylacinus cynocephalus Sminthopsis leucapus Planigale maculata Planigale gilesi Planigale tenurostriy Murexia longicaudata Phascogale tapotafta Neophascogale lorenizii Myoictis melas Parantechinus apicalis Dasyurus hallacatws Daysurus maculatus Sarcophilus harrisié DISCUSSION BIOSTRATIGRAPHY AND ECOLOGY. To date M. orbits is restricted to early late Miocene Archer et al. (1995) Encore Site at Riversleigh. Encore has produced a fauna that includes several unique taxa, including a large dasyuromorphian of uncertain affinity (unpubl, data), a giant Ekaltadeta (Wroe, 1996a), a derived koala (Black, pers. comm.), a palorchestid structurally intermediate between species from Riversleigh System C and the late Miocene Palorchestes inei of Alcoota (Black, 1997) and a Warenja- ike wombat (Archer et al., 1995). The rootless teeth of this wombat (unknown for other species at Riversleigh) and the relatively low abundance of the frog Lechriodus intergervis, common in other Miocene Riversleigh deposits (Godthelp, pers. comm.) indicate that climatic conditions may have been drier for the depositional episode 445 during which Encore was produced. Tentative support for a relatively late age for Encore site, is also forwarded by Wroe (1997a), IEM. orbus is a dasyurid then the derived dentition (relative to other Miocene dasyurids) might also suggest a late age for Encore site. As noted above for small dasyunds, circumstantial evidence correlates en- toconid reduction with adaptation to relatively dry environments. Mayigriphus orbus is the smallest dasyuro- morphian from the Oligocene and Miocene of Riversleigh and is comparable to Planigale maculatus in size. Only one other marsupial in- seclivore has been identified that might have competed closely with M. orbus, the diminulive bandicoot Yarala burchjieldi (Muirhead. 1995), As with modem Planigale (Denny, 1982) the die of M. orbus probably included invertebrates, frogs, small lizards and/or small mammals. PHYLOGENY. Mayigriphus orbus shows a unique mosaic of fealures among dasyuro- morphians. Two features of M. orbus may ihdi- cate a relationship with Planigale (the greatly reduced Mı paraconid concurrent with a moder- ately reduced Mj metaconid, and relatively uni- form diminution of the M14 metaconids). Although a comparable degree of Mı paraconid reduction is also common to many derived dasyurines (¢,z,, Pseudantechinus), in these taxa diminution of the Mı metaconid is far more ad- vanced and a clear differentia) is produced be- tween thal shown by My and M24. Additional apomorphies shared by M. orbus and Planigale (e.g., reduction of Mg talonid, entoconid and P4), are also found in other specialised dasyurid taxa. On the basis of cytochrome-b data, Painter et al, (1995) estimate the oldest branchings within Planigale at 11-15 mya, thus the possibility that M. orbus represents an early branch of this radi- ation can not be discounted. However, M. arbus shows at least 2 derived features not in Planizgale (wide angle formed between the paracristid, metacristid, a buccal shift in the point of termi- nation of the cristid obliqua). The oldest maternal cleariy attributable to Planigale is Pliocene (Archer, 1982a), Based on available data, a sister taxa association for M. orbus with Planigale is considered equivocal. Even at the family level, the phylogenetic pù- sition of M. orbus is considered uncertain, be- cause it has a number of features that might be interpreted as synapomorphies for either de- rived urid or thylacinid clades, but no un- equivocal synapomorphies (within Dasyuro 446 morphia) for either family. Two synapomorphies for the Dasyuridae are in the lower dentition of M. orbus: reduction of P3 (Tate, 1947; Archer, 1982b; Marshall, 1990) and the hypoconulid notch in the lower molars (Archer, 1982b). Status of both as shared-derived features for Dasyuridae is questioned by Wroe (1996b, 1997b). Reduc- tion of P3 is certainly common within Dasyuridae which culminates in the loss of this tooth among specialised taxa. However, reduction or loss of P3 may have occurred independently at least 3 times in the Dasyuridae (Archer, 1981). A further argu- ment against the phylogenetic value of this char- acter at the family level is the P3 smaller than P2 in 2 thylacinids from Riversleigh (Muirhead & Wroe, in press; Wroe, 1996b, 1997b). The status of the hypoconulid notch as a shared derived feature for dasyurids has been undermined by the discovery of plesiomorphic thylacinids with a well-detined hypoconulid notch in the lower mo- lars (Wroe, 1997b; Muirhead & Wroe, in press). Marked reduction of the Mı paraconid in M., orbus is common to specialised dasyurids but not thylacinids, excepting Badjcinus turnbulli (Muirhead & Wroe, in press). None of these features represent unequivocal synapomorphies for Dasyuridae and each have been independently derived within specialised dasyurid lineages. At least 3 features in M. orbus suggest a possible alliance with thylacinids. Firstly, the lack of a clear differential between metaconid reduction on M; and M24 in M. orbus is known only for Planigale among dasyurids but common to plesiomorphic thylacinids. In all dasyurids ex- cept Planigale, Mı metaconid reduction clearly exceeds that of Mo.4. Although reduction of the Mo2.4 metaconids is less pronounced in M. orbus than in known thylacinids, excepting Muribacinus, it is greater than for most dasyurids except D. maculatus, Sarcophilus and Planigale. Secondly, the wide angle formed between the paracristid and metacristid in M. orbus is found in basal thylacinids, but only D. maculatus, Sarcophilus and Glaucodon among dasyurids. Thirdly, among specialised dasyurids, reduction of the Mj-4 talonids and metaconids is commonly associated with a buccal shift in the point of termination of the cristid obliqua. In both M. orbus and plesiomorphic thylacinids this is not the case, with the cristid obliqua terminating in a relatively lingual position. Ultimately, this may be related to Ride’s (1964) observation of a dif- ference between Thylacinus and specialised dasyurids in the composition of the principal pos- terior shearing crest. Ride pointed out that, in MEMOIRS OF THE QUEENSLAND MUSEUM Thylacinus, the posterior shearing crest runs from the protoconid directly to the hypoconid, while in derived dasyurids (especially Sarcophilus) the posterior shearing crest connects the protoconid and metaconid. CONCLUSIONS If M. orbus is a dasyurid it represents the most derived member of the family known from pre- Pliocene times, with the possible exception of the Miocene Dasylurinja kokuminola (Archer, 1982a) from Lake Yanda in central Australia. A special relationship between these two taxa can not be discounted, with both showing specialisations associated with carnassialisation (the much larger size of D. kokuminola precludes the possibility that the 2 taxa are conspecific). D. kokuminola is known only from a single upper molar and direct comparisons with M. orbus can- not be made. Among known dasyurids M. orbus shares the greatest number of derived features with Planigale: two character-states (disparate reduction of the M; paraconid and metaconid and uniform reduction of the Mj-4 metaconids) sug- gest the possibility of a special relationship for the 2 taxa. However, uniform diminution of the Miı-4 metaconids is also shown by some basal. thylacinids, Mayigriphus orbus shows no unequivocal syn- apomorphies for either dasyurid or thylacinid clades. For Dasyuridae, unique derived-features (within Dasyuromorphia) are only found in the basicranium (Wroe, 1996b, 1997b). Unique de- rived features (within Dasyuromorphia) uniting Thylacinidae have been identified only in the upper dentition (Wroe, 1996b). Neither region is known for M. orbus. Confident phylogenetic as- signment for M. orbus has been further tempered by the identification of possible thylacinid apomorphies in this taxon, which must be consid- ered in the following context: investigation of Oligocene and Miocene material from Riversleigh is revealing a complex dasyuro- morphian phylogeny dominated by a diverse thy- lacinid clade, showing greatly expanded intrafamilial variation (Muirhead, 1992, 1997; Muirhead & Archer, 1990; Muirhead & Wroe, in press; Wroe, 1996b, 1997b) a close relationship between Dasyuridae and Thylacinidae has been established by molecular studies (Lowenstein et al, 1981; Sarich et al., 1982; Thomas, 1989; Krajewski et al., 1992), and a relatively recent genesis for Dasyuridae has been suggested (Archer, 1982a; Krajewski, 1992; Wroe, 1996b, MATIGRIPHUS ORBUS. ENIGMATIC DASYUROMORPHIAN 1997b), Given this emerging climate of complex- ity for dasyuromorphian phylogeny, the curious mix of derived features in M. orbus (otherwise considered diagnostic of either specialised dasyurid or thylacinid clades) makes reliable placement within either family impossible, pàr- ticularly as this decision must currently be based solely on elements of the lower dentition. Prob- lems associated with determining the phyloge- netic position of M. orbus highlight an unexpected phylogenetic scenario, Despite an abundance of dasyuromorphian material from Oligocene-Miocene deposits of Riversleigh no taxon has yet been described which can be un- equivocally associated with elements of the now ubiquitous dasyurid radiation. ACKNOWLEDGEMENTS M. Archer, H. Godthelp, and J. Muirhead pro- vided invaluable assistance through their con- Siructive criticism and comment. J, Muirhead also kindly took the photographs presented, Vital support for this research has been given by the Australian Research Council; the National Estate Grants Scheme (Queensland); the Department of Environment, Sports and Territories; the Queens- land National Parks and Wildlife Service! the Commonwealth World Heritage Unit (Can- berra); the University of New South Wales; IC) Australia; the Australian Geographic Society: the Queensland Museum; the Australian Museum, Century Zinc; Mt Isa Mines; Surrey Beatty & Sons; the Riversleigh Society. ; the Royal Zoolog- ical Society of New South Wales; the Linnean Society of New South Wales; and many private supporters. Skilled preparation of most of the Riversleigh material has been carried out by Anna Gillespie. LITERATURE CITED ARCHER, M. 19769, Miocene marsupicaurnivores (Marsupialia) from central South Australia, An- kotarinja lirarensis gen, et, sp, nov. Keeani woodburnei gen, et, sp, nov,, and their signifi- cance in terms of carly marsupial radiations, Transactions of the Royal Society of South Aus- tralia 100: 53-73, 1976b. The dasyurid dentition and its relationship to that of didelphids, thylacinids, borhyaenids (Marsupicarnivora) and peramelids (Peramelina: Marsupialia). Australian Joumal of Zoology, Supplementary Series 39; 1-34. 1979. Wakamarha tasselli gen, et sp. now, u fassil dasyurid (Marsupiaha) from South Australia con- 4a) vergent on modern Sminthopsis. Memoirs of the Queensland Museum 19: 309-317. 1981, Results of the Archbold expeditions. No. 14. Systematic revision of the marsupial dasyrarist genus Sminthopsis Thomas. Bulletin of the American Museum of Natural History 168: 65- Sod, 19833. A review of (he dasyurid (Marsupialia) fossil record, integration of data bearing on phyloge- nenc inlerpretation, and suprageneric classifica tion, Pp 397-443. In Archer, M. (ed.) Camivorous marsupials. (Royal Zoological Soci- ety of Australia of New South Wales; Sydney) 1982b. A review of Miocene thylacinids (Thy- lacinidae, Marsupialia), the phylogenetic posi- tion of the Thylacinidae and the problem of aprionsms in character analysis. Pp. 445-476, In Archer, M. (ed.), Carmivorous marsupials. (Royal Zoological Society of New South Wales: Syd- ney), ARCHER , M. HAND, S. & GODTHELP , H. 1995. Tertiary environmental and biotic change in Aus- tralia, Pp, 77-90, In Verba, ES., Denton, GH., Partridge. T.C. & Burkle, L.H. (eds), Paleoclimate and evolution, with emphasis on human origins, (Yale University Press; New Haven), BENSLEY, B.A. 1903, On the evolution of the Austra- lian Marsupialia, with remarks on the relation- ships of the marsupials in general, Transactions of the Linnean Society of London, Zoology 9: 83- 217. BLACK, K. 1997, A new species of palorchestid (Marsupialia, Palorchesudae) fiom the late middle la early lute Miocene Encore Local Pauna, Riversleigh, northwestem Queensland, Memoirs of the Queensland Museum 41: 151-185. DENNY. M.J.S, 1982, Review of Planigale (Dasyuridae, Marsupialia) ecology. Pp. 134-138. In Archer, M, (ed,), Carnivorous marsupials (Royal Zoological Society of New South Wales: Sydney). FLOWER, W.H, 1867. On the development and sugs cession of teeth in the Marsupialia. Philosophical Transcripts of the Royal Society of London 157; 631-641, GILL, T. 1872. Arrangement of the families of mani- mals with analytical tables. Smithsoman Miscel- lancous Collection 2: 1-98. GODTHELP, H.. ARCHER, M., CIFELLI,R., HAND, S.J. & GILKESON, C.F. 1992. Earliest known Australian Tertiary mammal fauna. Nature 356; 514-51 6. KIRSCH, J.A.W., KRAJEWSKI, C., SPRINGER, MS & ARCHER, M. 1990, DNA-DNA hybridisation studies of Carnivorous marsupials. I. Retation- ships among dasyurids (Marsupialia: Dasyuridae), Australian Joumal of Zoology 38: 673-696, KRAJEWSKI, C., DRISKELL, A.C.. BAVERSTOCK, BR. & BRAUN, M.I. 1992. Phylogenetic rela tionships of the thylacine (Mammalia: Thy- 448 lacinidae) among dasyuroid marsupials: evidence from cytochrome b DNA sequences. Proceedings of the Royal Society of London 250: 19-27. KRAJEWSKI, C., PAINTER, J., DRISKELL, A.C., BUCKLEY, L., & WESTERMAN, M. 1993. Mo- lecular systematics of New Guinean Dasyurids (Marsupialia: Dasyuridae). Science in New Guinea 19: 157-166. KRAJEWSKI , C., P AINTER , J., DRISKELL, A.C., BUCKLEY, L., & WESTERMAN, M. 1994, Phylogenetic structure of the marsupial family Dasyuridae based on cytochrome b DNA se- quences. Journal of Mammalian Evolution 2: 25- 35. LOWENSTEIN, J.M., SARICH, V.M. & RICHARD- SON, B.J. 1981. Albumin systematics of the ex- tinct mammoth and Tasmanian wolf. Nature 291: 409-411. LUCKETT, P. 1993. An ontogenetic assessment in dental homologies in the therian mammals. Pp. 182-204. In Szalay, F.S., Novacek, M.J. & Mc- Kenna, M.C. (eds), Mammal phylogeny: Mesozoic differentiation, multituberculates, monotremes, early therians, and marsupials. (Springer-Verlag: New York). MARSHALL, L.G., CASE, J.A., & WOODBURNE, M.O. 1990. Phylogenetic relationships of the fam- ilies of marsupials. Pp. 433-505. In Genoways, H. (ed.), Current mammalogy. MUIRHEAD, J. 1992. A specialised thylacinid, Thy- lacinus macknessi (Marsupialia: Thylacinidae) From Miocene deposits of Riversleigh, northwest- ern Queensland. Australian Mammalogy 15; 67- 76. 1997. Two new early Miocene thylacines from Riversleigh, northwestern Queensland. Mem- oirs of the Queensland Museum 41: 367-377. MUIRHEAD , J. & ARCHER, M. 1990, Nimbacinus dicksoni a plesiomorphic thylacine (Marsupialia: Thylacinidae) from Tertiary deposits of Queens- land and the Northern Territory. Memoirs of the Queensland Museum 28: 203-221. MUIRHEAD, J. & FILAN, S. 1995. Yarala burchfieldi, a plesiomorphic bandicoot (Marsupialia, Per- amelemorphia) from Oligo-Miocene deposits of Riversleigh, northwestern Queensland. Journal of Paleontology 59: 127-134. MUIRHEAD, J. & GILLESPIE, A. 1995, Additional parts of the type specimen of Thylacinus macknessi (Marsupialia: Thylacinidae) from Miocene deposits of Riversleigh, northwestern Queensland. Australian Mammalogy 18: 55-60. MUIRHEAD , J. & WROE, S. (in press). Badjcinus turnbulli gen. et sp. nov. (Thylacinidae, MEMOIRS OF THE QUEENSLAND MUSEUM Marsupialia) from the late Oligocene of Riversleigh, northwestern Queensland. Journal of Vertebrate Paleontology. PAINTER, J., KRAJEWSKI, C. & WESTERMAN, M. 1995. Molecular phylogeny of the marsupial genus Planigale (Dasyuridae). Journal of Mam- malogy 76: 406-413. RIDE, W.D.L. 1964. A review of Australian fossil marsupials. Journal of the Royal Society of West- em Australia 47: 97-129. SANSON, G.D. 1985. Functional dental morphology and diet selection in dasyurids (Marsupialia: Dasyuridae). Australian Mammalogist 8: 239- 247. SARICH, V.M., LOWENSTIEN, J.M. & RICHARD- SON, B.J. 1982. Phylogenetic relationships of the Thylacine (Thylacinus cynocephalus, Marsupialia) as reflected in comparative serology. Pp. 707-709. In Archer, M. (ed.), Carnivorous marsupials. (Royal Zoological Society of Aus- tralia of New South Wales: Sydney). TATE, G.H.H. 1947. Results of the Archbold expedi- tions. No. 56. On the anatomy and classification of the Dasyuridae (Marsupialia). Bulletin of the American Museum of Natural History 88: 97-156. THOMAS, R.H., SCHAFFNER, W., WILSON, A.C, & PAABO, S. 1989. DNA phylogeny of the ex- tinct marsupial wolf. Nature 340: 465-467. VAN DYCK, S. 1989, Biting remarks on the Riversleigh Antechinus. Riversleigh Notes 7: 2-3, WOODBURNE, M.O. 1967. The Alcoota Fauna, cen- tral Australia: an integrated palaeontological and geological study. Bureau of Mineral Resources Australian Bulletin 87: 1-187. WROE, S. 1996a. An investigation of phylogeny in the giant extinct Rat-kangaroo Ekaltadeta (Pro- pleopinae, Potoroidae, Marsupialia). Journal of Paleontology 70: 677-686. 1996b, Muribacinus gadiyuli (Thylacinidae: Marsupialia), a very plesiomorphic thylacinid from the Miocene of Riversleigh, northwestern Queensland, and the problem of paraphyly for the Dasyuridae (Marsupialia). Journal of Paleontol- ogy 70; 1032-1044. 1997a. Stratigraphy, phylogeny and systematics of the giant extinct Rat-kangaroos (Propleopinae, Hypsiprymnodontidae, Marsupialia). Memoirs of the Queensland Museum 41: 449-456. 1997b, A re-examination of proposed morphology- based synapomorphies for the families of Dasyuromorphia (Marsupialia): Part I, Dasyuridae. Journal of Mammalian Evolution 4: 19-52, STRATIGRAPHY AND PHYLOGENY OF THE GIANT EXTINCT RAT KANGAROOS (PROPLEOPINAE. HYPSIPRYMNODONTIDAE, MARSUPIALIA) S. WROE Wroe, S. 1997 06 30: Stratigraphy and phylogeny of the giant extinct Rat-kangaroos (Propleopinac, Hypsiprymnodontidac, Marsupialia), Memoirs ef the Queensland Museum. 41(2): 449-456, Brisbane. ISSN 0079-8835. The Giant Rat-kangaroos were placed in the Propleopinae by Archer & Flannery (1985) and in the Hypsiprymnodontidie by Ride (1993), Cladistic analysis of Ekeltadeta material from Riversleigh, northwestern Queensland (Wroe, 1996) suggested that a middle to late Miocene dichotomy in Bkaltadeta may have produced two lineages of Plin-Pleistocene Propleapus, indicating polyphyly for Prepleopus and paraphyly for Ekalradeta. Metrical data for pro- pleopines and stratigraphic information support Wroe’s (1996) cladistic analysis of pro- pleopines. C] Propleapinae, Hypsiprymnadontiidae, Riversleigh, Ekaltadeta, cladisties. S. Wroe, School of Biological Science, University of New Somi Wales, N.S.W, 2052 Australia; received 4 November 1996, Giant Ral-kangaroos (Hypsiprymmnodontidae: Propleopinac) may be the plesiomorphic sister group of potoroids (Flannery, 1987). Archer & Flannery (1985) considered Ekaltadeta ima,(Fig. 1) the sister group to Propleopus De Vis, 1888 with P. oscillans De Vis, 1888 (Fig. 2) the more plesiomorphic and P. chillagoensis Archer et al, (1978) (Fig. 2) the more apomorphic within Pro- pleopus (Fig. 3). Propleopine species described since 1985 are Ekaltadeta jamiemulvaneyi Wroe, 1996, (Fig, 4) and Jackmahoneya toxoniensis Ride, 1993, Wroe (1996) suggested another pos- sible phylogeny for the Propleapinae with £. ima and P. chillagoensis forming, the sister group Lo another clade containing a new species, E. jamiemulvaneyi, as the sister taxon to P wellingtonensis and P. oscillans.(Fig. 5). As an adjunct to the cladistic analysis (Wroe, 1996), metric and Stratigraphic data for propleopines are used to clarify intrasubfamilial relationships- Dental homology for premolars follows Flawer (1867) and Luckett (1993) for molars. Higher level systematics of kangaroos follows Flannery (1987) and Ride (1993), Specimens are housed in the Queensland Museum (QMP). Other prefixes include; UCM (University of California Mu- seum), NMV (Museum of Victoria). METHODS Specimens of Ekaltadeta from Riversleigh rep- resent 30 individuals from several stratigraphic levels. The relative paucity of specimens and chronological data precludes a strictly strato- phenetic approach (sensu Gingerich, 1976, 1979; Bown & Rose, 1987) to propleopine phylogeny. However, amore general consideration of stratiz- raphy in phylogenetic analysis may be appropri ate in association with cladistic Lreatment where specimens are stratigraphically disjunct or sparsely distributed (Gingerich, 1990). Sites with Ekaltadeta are late Oligocene to early late Miocene (Archer et al., 1989, 1994, 1995). A number of characters were analysed lo assess the development of time-dependent changes. Specimens were ranked to indicate rel- ative age (Appendix 1). Stratigraphic levels are from Archer ct al. (1989, 1995): level I=lute Oligocene early Miocene; level 2=early Miocene; level 3=ldle carly Miocene; level 4=mid Miocene: level 5=late mid Miocene; level 6=carly late Miocene; level 7=Pliocene: level $=Pleistocene. I included all propleopines. possible, although Phocene and Pleistocene Jackmationeya und Propleopus are known from material often lim- ited to portions of upper and/or lower dentitions. Most Prepleepis are from the Pleistocene, al- though material has been recorded from early Pliocene local faunas (Archer & Flannery, 1985). Propleopus chillagoensis was described as Pleistocene (Archer et al., 1985), but could he older, possibly late Miocene or carly Pliocene (Archer pers. comm.). Jackmahoneya toxenien- sis ts Pliocene (Ride. 1993). Differences in molar gradient were used by Archer & Flannery (1985) and Wroe (1996) to distinguish propleopine species. Molar gradient reflects both the surface area and length ol the molar tooth row. In propleopines a high molar gradient correlates with a reduction in both molar surfave area and the length of the tooth row. 450 Reducing the distance between condyle and sectorial tooth maximizes leverage applicable to the tooth (Young et al., 1989). Through shortening the molar row, leverage on the large shearing P?/3 of pro- pleopines is increased. This ef- fect is achieved at the cost of molar length. Relative P3 size and molar gradient for upper and lower dentitions has been quantified. Distinct reduction in tooth size posteriorly occurs in upper and lower dentitions of E. ima. In the upper dentition this steep gradient begins with a reduced posterior width (pw) relative to the anterior width (aw) of M? which then ramifies through M+ In E. ima M+ pw is <1/2 M? aw. The upper dentition of P. chillagoensis is similar to that of E. ima. Lower dentition is not known for P. chillagoen- sis, For P. oscillans M** are missing but M? pw is only slightly less than M? aw sug- gesting a less extreme gradient. This supposition is strongly supported by the lower denti- tion in which molar gradient contrasts strongly with E. ima, Mı- tooth widths decrease steadily anteroposteriorly in £. ima but are reversed in P. os- cillans where tooth width in- creases posteriorly for Mj.3, with only aslight decrease in Ma. Several methods to quantify molar gradient have been considered. Accurate determination of individual molar surface areas and comparisons between teeth would be useful but would require 2 or more leeth/ specimen, greatly limiting data sets, particularly for upper dentitions. Molar gra- dient might also be estimated geometrically by determining the angle at Which a line drawn bucc- ally or lingually through the faces of the crown intersects the mid-line of the dentary or skull. In this study the clear initiation of a marked molar gradient at M? in the upper dentitions of £. ima and P. chillagoensis permitted estimation of the gradient from a single molar by comparing aw to pw, In lower dentitions the gradient is less MEMOIRS OF THE QUEENSLAND MUSEUM FIG. 1. Ekaltadeta ima, x 2. A, occlusal view of QMF12436 (uppers). B, buccal view of QMF1 2435, left dentary containing 1l, alveolus for 12, P23, M -4. C, occlusal view of QMF12435. distinct and 2 molars were required to demonstr- ate a gradient. Measurements were made using a Wild MMS 235 Digital Length-Measuring Set attached to a Wild MSA Stereomicroscope. Ab- breviations are: =length, w=width, aw=anterior width, pw=posterior width, dd=depth of dentary, G-value=ratio of anterior to posterior tooth width, RESULTS M? aw / M? pw VS STRATIGRAPHIC LEVEL, (Fig. 6). For upper dentitions the ratio M? aw: M? pw (G-value) was used as an arbitrary measure of molar gradient, with M? being common to the largest umber of specimens. STRATIGRAPHY AND EKALTADETA A * Beem , { = n FIG. 2. Propleopus chillagoensis, x 2. A, occlusal view of NMV P15917, right maxillary fragment (juvenile) containing unerupted P°, partial M!, MŽ, partial M4 (cast of holotype). B, Propleopus oscillany. x 2, oc- clusal view of A ines left maxillary fragment, containing P?, A trend is apparent in this scatter graph of G-value against stratigraphic level. P. chillago- ensis and P. oscillans represent 2 extremes with G-values of 1.23 and 1.06 respectively, with the lower number indicating a lesser molar gradient. Ekaltadeta ima from levels 3 and 4 has a limited range of G-values (1.09-1.15). The 2 Ekaltadeta from level 6 both fell outside the range of E. ima from older strata. E. Jamiemulvaneyi (QMF24212; Cleft of Ages 4 Site) had a low G-value of 1.05, slightly less than that of P. oscillans. E. ima (QMF24211; Henk’s Hollow Site) had a relatively high G-value of 1.19 approaching that of P. chillagoensis. These re- sults indicate a divergence in the Ekaltadeta lin- eage with one population leading to P. oscillans and another leading to P. chillagoensis. Mı pw/ M2 pw VS STRATIGRAPHIC LEVEL. (Fig. 7). The molar gradient of the dentary was estimated by dividing Mi pw by M2 pw (G- value). P. oscillans had the lowest G-value at 0.93. The G-values for P. wellingtonensis and J. toxoniensis were slightly higher at 0.96. At levels 3 and 4 the G-values for Ekaltadeta were 1.01- 1.08. The G-value for E. jamiemulvaneyi, from level 6 (QMF24200, Encore Site) was 0.97. This placed E. jamiemulvaneyi about halfway between the lowest G-value from levels 3 and 4 and P. oscillans. Again the highest degree of divergence among Ekaltadeta was for the E. jamiemulvaneyi from level 6, possibly indicating a trend toward 451 g a a“ z — $ 2 2 z © a = x = = 2 = Š 2 Š = Se A = = S Š A Š 3 3 = EN a S = Ss a ki a a Q Ly FIG. 3. Cladogram for the propleopines from Archer & Flannery (1985). Character states at nodes: A=gain of an anterior cristid emanating from the metaconid of Mj, gain of derived I; morphology; B=incorporation of the protolophid into the anterior lophid of Mj loss of P2 with eruption of P3, dentary deeper anteriorly than posteriorly; C=reduction of metacone/en- toconid, P3 hypertrophy. the species with low molar gradients (J. toxonien- sis, P. oscillans and P. wellingtonensis). P3 w/Mipw VS STRATIGRAPHIC LEVEL. (Fig. 8). In P. oscillans P3 width was small com- pared to Mı posterior width (1.09). For J. tox- oniensis relative P3 width was greater (1.27). E. ima from levels 3 and 4 had ratios of P3 w / Mı pw of 1.35-1.52. E. jamiemulvaneyi from Encore site (QMF24200) again posi- tioned between E. ima from lower strata and J, toxoniensis / P. oscillans, with a ratio of 1.28 DEPTH OF DENTARY AGAINST STRATI- GRAPHIC LEVEL. Depth òf dentary against stratigraphic level (Fig. 9). Dentary depth was measured from the alveo- lar margin of M; to the ventral margin of the dentary perpen- dicular to the molar row, Vari- ation in the depth of dentaries was small for E. ima from lev- els 3 and 4 (19,3-21. 1mm). E. Jamiemulvaneyi (QMF24200) was much larger with adentary depth of 28.8mm approaching P. oscillans (32.6mm). J. tox- oniensis between E. ima and E. jamiemulvaneyi/P, oscillans with a dentary depth of 23.3mm. STATISTICAL ANALYSIS. (Table 1). Because all Ekaltadeta material has come from a relatively small area (Riversleigh), a regional popu- lation of potoroids was consid- ered an appropriate control, Sixteen specimens from the Australian museum of Poror- ous tridactylus collected around Hobart were used, this being the largest potoroid spec- imen sample available. Varia- tion in the G-values of Ekaltadeta trom levels 3 and 4 approached that of P. tridactylus. When G-values from the 2 Ekalradeta from level 6 were included the variation fell well out- side that of the local P. tridactylus population. DISCUSSION Increases in premolar and molar shear within the Propleopinae appear to be mutually exclusive and their relative importance probably reflects dietary preference, A requirement for high pre- molar shear might be associated with carnivory MEMOIRS OF THE QUEENSLAND MUSEUM FIG. 4. Ekaltadeta jamiemulvaneyi x 2. A, occlusal view of QMF24200, lett dentary containing P3, Mj-3, holotype. B, buccal view of QMF24200. C, lingual view of QMF24200. D, occlusal view of QMF24212, left maxillary fragment, containing P*, dP?, M'~, referred specimen. E, buccal view of QMF24212. F, lingual view of QMF24212. G, occlusal view of QMF20842, left P*, referred specimen. H, buccal view of QMF20842, |, lingual view of QMF20849, (Abbie, 1939), while a more extensive molar array may indicate a more herbivorous diet (Wells et al., 1982), Species with a large molar surface area and low molar gradient (P. oscillans, P. wellingtonensis, J. toxoniensis) have relatively small premolars, Species with high molar gradients and reduced molar shear (E. ima, P. chillagoensis) are characterised by P3 hypertrophy. The extraordi- nary change in function for P2 shown by individ- STRATIGRAPHY AND EKALTADETA Janlemulvanent Ji\psiprymnedan sp. P clutlagoenses Ë wellingranensts P ayetllans E ime FIG. 5. Minimal tree produced by Wagner analysis for the Propleopinae (from Wroe, 1996), Character states at nodes: A = gain of an anterior cristid emanating from the metaconid of Mı; basally broad conical upper molars: B = presence of lingual cingula on the upper molars; C = reduced molar gradient, reduced P3; D = incorporation af the protolophid into the anterior lophid of My, a dentary deeper posteriorly than anteriorly, E f! C ual E. ima (Wroe & Archer, 1995) probably con- stitutes a response to the increased loading placed on P3. Regarding molar gradient and relative size of the P3, E. jamiemulyaneyi is intermediate, fall- ing between £. ima specimens from lower levels and P. ascillans/P. wellingtonensis/J. toxanien- sis. Using the same criteria J. toxoniensis lies between E. jamiemulvaneyi and P. oscillans/P. wellingtonensis. In terms of variation in P3 size and molar gradient P, chillagoensis and P. os- cillans represent opposite extremes in propleop- ine evolution and it is suggested that P. oscillans TABLE 1, Statistical summaries for M? aw / M? pw (G-value) for propleopines and a local P, tridactylus population. 2.07 | 7.71 B-3 1.70 | 4.61 B-3 was largely if not wholly herbivorous. Other de- rived features interpreted as adaptations to herbivory for P. oscillans include a large dia- stema between P3 and h, and large spatulate lower incisors (Wroe, 1996), Regarding dentary depth E. ima is the smallest propleopine with a general increase in depth for taxa at higher strati- graphic levels probably reflecting a general in- crease in body size. Straigraphic and metric analysis support the proposal of a late Miocene dichotomy in Ekaltadeta producing 2 lineages of Propleopus, and a reversal of previous assumptions on relative apomorphy within Propleopus, with P. oscillans considered the most derived and P. chillagoensis the most plesiomorphic (Wroc, 1996). However, broad trends suggested in this study are not inter- preted here chronoclines in the stratophenetic sense (sensu Bown & Rose, 1987). The scarcity of material and uncertain chronology of both the Oliga-Miocene Riversleigh deposits and the Plio- Pleistocene local faunas from which most pro- pleopine specimens are known necessitates caution in the interpretation of results. A consid- erable temporal gap exists between estimated ages of the most recent Ekalradera specimens and all other propieopines. As noted by Ride (1993), the period separating the latest known incidence of Ekaltadeta from Plio-Pleistocene Propleopus and Jackmahoneya may be sufficient to have permitted a secondary reversal of character states within Propleapus to produce P. chillagoensis. Many questions. remain concerning the age, stratigraphy and method of deposition of Riversleigh’s Oligocene-Miocene limestone de- posits (Archer, 1994, 1995; Megirian, 1992, 1994). If the phylogeny for propleopines sug- gested by Wroe (1996) reflects evolutionary events, then it provides tacit support for Archer et al.’s (1989) proposed Stratigraphy, with an agreement of hypothesised superpositional and phylogenetic patterns. The capacity of stratigraphic occurrence lo ex- plicitly mirror phylogenies is questionable (En- gelmann & Wiley, 1977). Although strong 454 9 P oscillans P chillagoensis 8 a. br QM P6675 NMV PL5917 3 E 7 fr E. jamiemulvaneyi E. ima = 6 QM F24212 DQM F24211 E 20 = S n » 2 & OQM FNO E. ima K] e amo om E. ima {see caption fornos) —+ T T 7 1.1 1.2 1.3 M2 aw / M2 pw FIG. 6. M? aw / M? pw vs relative stratigraphic level for propleopines. Ekaltadeta ima from level 3, left to right, QMF24207, 24204, 24205, 12436, 24203, 24208, 24209, and 24206. 17 d P oscillans a 87 D QMFS3302 J roxaniensis g 3 D ARIST H 6 E, jamiemulvaneyt = QM F24200 S QM F12424 QMF12423 E, ima na 4 ao a 2 3 o QMF12435 £ ima * ad 2 n | QM F2420) E. ima a oore e 1.1 1.2 1.3 1.4 1.5 1.6 P3 w/Mi pw FIG. 8. Pa w / Mı pw vs relative stratigraphic level for propleopines, congruence between cladistic and stratigraphic arrangements has been demonstrated for many vertebrate taxa by Norell & Novacek (1992a,b), the same authors advised that correlation between the 2 diminishes rapidly where cladistic or strati- graphic data is poorly resolved. Debate over con- formity of age and cladistic information commonly centres around the value of super- positional data as an adjunct to phylogenetic re- construction. Where cladistic analysis is sound it MEMOIRS OF THE QUEENSLAND MUSEUM 107 P oscillans , ; 1 QM F3302 P wellingtonensis 84 = Á UCM P4517! 3 1 J. toxaniensis 5 J AR 17579 = . . . EŻ 6 o E. jamiemulvanevi = QM F24200 2 - = E. ima QM F 12423 nA 44 o z 4 E. ima QMF12435 QM F24198 QM F24199 QM F24197 = og o Oo 3 me 2 oo E. ima QM F24196 QM F24195 0 i T — 0.9 1.0 1.1 M1 pw / M2 pw FIG. 7. Mı pw / M2 pw Ys relative stratigraphic level for propleopines. P oscillans 87 J, toxoniensis o QM F3302 E AR 17579 Z a =, s y 4 E, jamiemulvaneyi 2 6 BD QM F24200 [es E d Z l 77) 4 DOOM FI2424 UM FI203 E. ima = o QM F12435 E. ima E 2 oQmM Fi22420 E. ima 0 ~— a or T T ——" 1 10 20 30 40 Dentary Depth (mm) FIG. 9. Depth of dentary vs relative stratigraphic level for propleopines. may be useful as a test of stratigraphic interpreta- tions. The propleopine phylogeny of Wroe (1996) is based on analysis of an incomplete data matrix, with important characters unknown for several species. Consequently the cladistic data pre- sented cannot be viewed as a robust basis for testing superpositional pattern. However, the pro- ductivity of the Oligocene-Miocene deposits of STRATIGRAPHY AND EXALTA DETA Riversleigh engenders reasonable expectation for the reliable resolution of phylogenies. ACKNOWLEDGEMENTS M. Archer, H. Gouthelp and W, D. L. Ride provided constructive criticism and comment. Vital support for this research has been given by ihe Australian Research Council (to M. Archer), the National Estate Grants Scheme (Queensland) (grants to M, Archer and A. Bartholomai); the Department of Environment, Sports and Territo- rics; the Queensland National Parks and Wildlife Service; the Commonwealth World Heritage Unit (Canberra); the University of New South Wales; ICT Australia Pty Ltd; the Australian Geo- graphic Society; the Queensland Museum; the Australian Museum; Century Zinc Pty Lid; Mt Isa Mines Pty Ltd; Surrey Beatty & Sons Pry Lid: the Riversleigh Society Inc.; the Royal Zoologi- cal Sociely of New South Wales; the Linnean Society of New South Wales; and many private supporters. Skilled preparation of most of the Riversleigh material has been cared out by Ania Gillespie, LITERATURE CITED ABBIE, A. 1939. A masticatory adaptation peculiar lo some diprotodont marsupials Proceedings of the Zoological Sociery 109: 261-279, ARCHER, M. 1994. Introduction: background and eon- troversy about the Riversleigh geological and pal- acontological resource. Riversleigh Symposium 1994, Abstracts; p 5-6, ARCHER, M., BARTHOLOMALT, A. & MARSHALL, L.G. 1978. Propleopus chillagoensis, anew north Queensland species of extinel giant Rat-kangaroo (Macropodidae! Potoroinae). Memoirs of the Nu- tional Museum of Victoria 39: 55-60, ARCHER, M. & FLANNERY, T.F. 1985 Revision of the extinct gigantic Rat-kangaroo (Potoroidae: Marsupialia), with description of a new Miocene genus and species, and a new Pleistocene species of Propleopus. Journal of Paleomology 59: 1331- 1349, ARCHER, M., GODTHELP, H., HAND, S.J. & MERGIRIAN, D, 1989, Fossil mammals of Riversleigh, northwestern Queensland: prelimi- nary Overview of biostratigraphy, correlation, and Suro gatas change. Australian Zoologist 25, 29-65. ARCHER, M., GODTHELP, H. & HAND, S.J. 1994. ‘Riversleigh: The story of animals in ancient rain forests of inland Australia, Second Edition’, (Reed Books Pry Lid. Sydney), 1995, Tertiary environmental and biote change in Australia, Pp, 77-90, In Vrba, E.S., Denton. G-H.: 455 Partridge, T.C. & Burkle, L.H, (eds) “Paleocli- mate and evolulion, with emphasis on human origins’. {Yale University Press: New Haven), BOWN, T.M- & ROSE, K D. 1987. Patterns of evolu- tion in early Eocene anaplomorphine primates (Omomyidae) from the Bighorn basin, Wyoming. Joumal af Paleontology 23: 1-128. DE VIS, C.W. 1888. On an exunet genus of the mursu- pluls allied to Hypsiprymnodon. Proceedings of the Linnean Society of N.S.W. 3: 5-8. ENGELMANN, G.F. & WILEY, E.O. 1977. The place ofancestor-descendant relationships in phylogeny reconstruction. Systematic Zoology 26: 1-11 FLANNERY, T. 1987. The relationships of the maeropodoids (Marsupialia) and the polarity of some morphological features within the Phalangeriformes.. Pp. 741-747. In Archer, M. (ed.) ‘Possums and opossums: studies in evolution’, (Surrey Beatty & Sons and Royal Zop- logical Society of New South Wales: Sydney). FLOWER, W.H. 1867. On the development and suc- cession of teeth in the Marsupialia. Philosophical Transcripts of the Royal Society of London 157. 631-641. GINGERICH, P, D. 1976, Cranial anatomy and evolu- tion of North American Plesiadapidae (Mammalia, Primates), University of Michigan Papers on Paleontology 15. 1-140. 1979, The stratophenctic approach to phylogeny reconstruction in Vertebrate paleontology. }Co- lumbra University Press; New York), 1990, Stratophenctics, Pp, 437-442. In Briggs. D.E,G, & Crowther, P.R. (eds), Patacobidlogy, a synthesis, (Blackwell Scientific: Oxford), LUCKETT, P. 1993, An ontogenetic assessment in dental homologies in the therian mammals, Pp- 182-204. In Szaliy, F\S,, Novacek, M.J. & Mce- Kenna, m.c. (eds) ‘Mammal Phylogeny: Mesozoic Differentiation, Multituberculates, Monotremes, Early Theriens, and Marsupials’, (Springer-Verlag: New York). MEGIRIAN, D, 1992, Interpretation of the Curl Creek Limestone, northwestern Queenslund, The Beagle 9. 219-248, 1994, Approaches to marsupial biochronalogy in Australia and New Guinea, Aleheringa 18° 259 274, MYERS. T. & ARCHER, M. 1997. Kuterintja ngama (Marsupialia, Mariidae): a revised and extended systematic analysis based on fossil matenal from the late Oligocene of Riversleigh, northwestem Queensland, Australia. Memoirs of the Queens- land Museum 41° 379-392, NORELL, M.A. & NOVACEK, M.J. 1992a, Congru- ence between superposilional and phylogenetic patterns: comparing cladistic pattems With fossil records. Cladistics X: 319-337, [992b, The fossil record and evolution: comparing cladistic and paleontologie evidence lur verie bnite history. Science 255: 1690-1693, 456 RIDE, W.D.L. 1993. Jackmahoneya gen. nov., and the genesis of the macropodiform molar. Memoirs of the Association of Australasian Palaeontologists 15: 441-459. WELLS, R.T. HORTON, D.R., & ROGERS, P. 1982. Thylacoleo carnifex Owen (Thylacoleonidae): Marsupial carnivore ? Pp. 573-586. In Archer, M. (ed.) ‘Carnivorous Marsupials’. (Surrey Beatty and Sons Pty Ltd and the Royal Zoological Soci- ety of New South Wales: Sydney). WROE, S. 1996. An investigation of phylogeny in the Giant Rat-kangaroo Ekaltadeta (Propleopinae, MEMOIRS OF THE QUEENSLAND MUSEUM Hypsiprymnodontidae, Marsupialia). Journal of Paleontology 70: 681-690. WROE, S. & ARCHER, M. 1995. Extraordinary diphyodonty-related change in dental function for a tooth of the extinct marsupial Ekaltadeta ima (Propleopinae, Hypsiprymnodontidae). Archives of Oral Biology 40: 597-603. YOUNG, W.G., JUPP, R. & KRUGER, B.J. 1989. Evolution of the skull, jaws, and teeth in verte- brates. (University of Queensland Press: Bris- bane). APPENDIX TO FIGURES 6-9 Data for Fig. 6. M? aw divided by M? pw (G-value) vs. relative stratigraphic level for propleopines. Measure- ments in mm. * = skull, (R) = right tooth row, (L) = Data for Fig. 7. Mı pw divided by M2 pw (G-value) vs. stratigraphic level. Measurements in mm. Data for Fig. 8. P3 w divided by Mı pw (G-value) vs. stratigraphic level for propleopines. Measurements in Fa 2 Species Cat. no. yi HA ae è veil Species Cat. no. ee oe ey Level E. ima QMF24203 | 6.80 | 6.10 | 1.12 | 3 E. ima QMF24195 | 6.70 | 6.30 | 1.06 2 E. ima QMF24204 | 6.80 | 6.20] 1.10} 3 E. ima QMF24196 | 6.30 | 5.90 | 1.07 2 E. ima * (R)QMF12436 | 6.40 | 5.80 | 1.10 | 3 E. ima QMF24197 | 6.50 | 6.00 | 1.08 3 E. ima * (LJ)QMFI12436| + | 5.90 3 E. ima QMF24198 | 5.70 | 5.50 | 1.04 3 E. ima QMF24205 | 6.50 | 5.90 | 1.10 | 3 E. ima QMF12435 | 6.50 | 6.20 | 1.05 3 E. ima QMF24206 | 6.80 | 5.90} 1.15 | 3 E. ima QMF24199 | 6.50 | 6.10 | 1.07 3 E. ima QMF24207 | 7.20 | 6.60 | 1.09 | 3 E. ima QMF12423 | 7.00 | 6.90 | 1.01 4 E. ima QMF24208 | 6.70 | 5.90 | 1.14} 3 a acl QMF24200 | 8.20| 8.50|0.97| 6 E. ima QMF24209 | 6.20 | 5.40} 1.15 | 3 N E. ima QMF24210 |740| 6.60} 1.12; 4| (Spyelinson | ucmp4s171 | 9.20 | 9.60 | 0.96 | 8 E. ima QMF24211 | 6.90 | 5.80 | 1.19 | 6 P. oscillans QMF3302 | 9.70 | 10.4 | 0.93 Famien ivaneyi QMF24212 | 6.90 | 6.60 | 1.05 | 6 J. toxoniensis AR17579 7.00 | 7.40 | 0.96 z. P. oscillans QMF6675 9.20 | 8.70 | 1.06 | 8 P. chillagoensis | NMVP15917 | 10.7 | 8.70 | 1.23 | 8 Data for Fig. 9. Depth of dentary vs. stratigraphic level for propleopines. Measurements in mm. mm. Species Cang. P3 w Mı vi [Level Species Cat. no. Dentary depth | Level pwo vane E. ima QMF24201 19.3 l E. ima QMF24201 | 10.3 | 6.80 | 1.56] 1 E ima QMF12435 19.4 3 E. ima QMF12435 | 8.70 | 6.50 | 1.34| 3 E. ima QMF12424 21.1 4 E. ima QMF12424 | 8.80 | 6.50 | 1.35] 4 E ima QMF12423 203 4 E. ima QMF12423 | 9.60 | 7.00 | 1.37 | 4 fF tna eh QMF24200 | 28.9 6 AEA i QMF24200 | 10.5 | 8.20 | 1.28 | 6 4 i y P. oscillans QMF3302 32.6 8 P. oscillans QMF3302 10.6 | 9.70 | 1.09 | 8 Paoxbnizasis ARI7579 23.3 7 J. toxoniensis AR17579 8.9 | 7.00 | 1.27 ti CONTENTS (continued) CREASER, P. Oligocene-Miocene sediments of Riversleigh: the potential significance of topography .,... .. 303 DAVIS, A.C. & ARCHER, M. Palorchestes azael (Mammalia, Palorchestidae) from the late Pleistocene Terrace Site Local Fauna, Riversleigh, northwestern Queensland ........... 0660. e eee eee es 315 GILLESPIE, A. Priscileo roskellyae sp. nov. (Thylacoleonidae, Marsupialia) from Riversleigh, northwestern Queensland .. 2... .... cece cece e tte beet betray eth ebeeesee 321 GODTHELP, H. Zyzomys rackhami sp, noy. (Rodentia, Muridae) a rockrat fromRackham’s Roost Site, Riversleigh, northwestern Queensland 1.2.0.0... 0. ccc e eee eee eee e ete eees 329 HAND, S. New Miocene leaf-nosed bats (Microchiroptera, Hipposideridae) from Riversleigh, northwestern Queensland ....... 0... ccc cece eee eee teen eee eee e eee beens 335 HAND, S. Miophyllorhina riversleighensis gen. et sp. nov., a Miocene leaf-nosed bat (Microchiroptera, Hipposideridae) from Riversleigh, Queensland .........-.2..-.4. 351 HUTCHINSON, M. The first fossil pygopod (Squamata, Gekkota), and a review of mandibular variation in living species cassas taisen eee eke eee ae ay ae ad Medea ee eh cede beled 355 MUIRHEAD, J. Two new early Miocene thylacines from Riversleigh, northwestern Queensland .......-.... 367 MYERS, T.J. & ARCHER, M. Kuterintja ngama (Marsupialia, Ilariidae): a revised systematic analysis based on material from the late Oligocene of Riversleigh, northwestern Queensland .............-+--- 379 SCANLON, J.D. Nanowana gen, nov., small matsoiid snakes from the Miocene of Riversleigh: sympatric species with divergently specialised dentition .............. 2022s eens 393 WHITE, A.W. Cainozoic turtles from Riversleigh, northwestern Queensland ..........-..2.+-+-+-+e4e5 413 WILLIS, P.M.A. New crocodilians from the late Oligocene White Hunter Site, Riversleigh, northwestern Queensland 2... 06. e eee ete ee ee tees ebb eee eee eee ew es 423 WROE, S Mayigriphus orbus gen. et sp. nov., a Miocene dasyuromorphian from Riversleigh, northwestern Queensland ©... . 2... 2c pe pee eee etre eee ee bee eee narani nenu 439 WROE, S. Stratigraphy and phylogeny of the giant extinct rat kangaroos (Propleopinae, Hypsiprymnodontidae, Marsupialia} 0... eee eee eee eens 449 CONTENTS ARENA, D.A. The palaeontology and geology of Dunsinane Site, Riversleigh ...............000.0200005 171 BLACK, K. A new species of Palorchestidae (Marsupialia) from the late middle to early late Miocene Encore Local Fauna, Riversleigh, northwestern Queensland ....... 181 BLACK, K. Diversity and biostratigraphy of the Diprotodontoidea of Riversleigh, Howea Queen ooo) oe sel ee chien ny ene eee pe th ene eae 187 BLACK, K.& ARCHER, M. Silvabestius gen. nov., a primitive zygomaturine (Marsupialia, Diprotodontidae) from Riversleigh, northwestern Queensland .............¢c ccs n ee eeeeeeneevees 193 BLACK, K. & ARCHER, M. 4 Nimiokoala gen. nov. (Marsupialia, Phascolarctidae) from Riversleigh, northwestern Queensland, with a revision of Litokoala ......... ccc ccc cece cee ees 209 BOLES, W.E. A kingfisher (Halcyonidae) from the Miocene of Riversleigh, northwestern Queensland, with comments on the evolution of kingfishers in Australo-Papua ..............606 229 BOLES, W.E. Hindlimb proportions and locomotion of Emuarius gidju (Patterson & Rich, 1987) A EIEE T pe) Cee ne ee Us on ON rata A patsy A Ke icel g E EOE ee 235 BOLES, W.E. Riversleigh birds as palacoenvironmental indicators ..........0.00eeceeeeeeeeneeevenens 241 BRAMMALL, J. & ARCHER, M. A new Oligocene-Miocene species of Burramys (Marsupialia, Burramyidae) from Riversleigh, northwestern Queensland ............0 00 cece crete crete eeeeee 247 COOKE, B.N. Two new balbarine kangaroos and lower molar evolution within the subfamily ............. 269 COOKE, B.N. ` New Miocene bulungamayine kangaroos (Marsupialia, Potoroidae) from Riversleigh, Deiihewesterh Queensland ee cy ols space EEE g LE A DADE BRE EE E es EEA AA E S 281 COOKE, B.N. Biostratigraphic implications of fossil kangaroos at Riversleigh, northwestern Queensland . .. . 295 (continued inside cover)