UNIVERSITY OF ILLINOIS LIBRARY AT URBANA-CHAMPAIGN GEOLOGY NOTICE: Return or renew an Library Material*! The Minimum Fee for each Lost Book is $50.00. The person charging this material is responsible for its return to the library from which it was withdrawn on or before the Latest Date stamped below. Theft, mutilation, and underlining of books are reasons for discipli- nary action and may result in dismissal from the University. To renew call Telephone Center, 333-8400 UNIVERSITY OF ILLINOIS LIBRARY AT URBANA-CHAMPAIGN APR 2 8 |9<3 w # UIAN 1^8 £ ?no7 LI61— O-1096 GEOLOG ;ARY FIELDIANA Geology NEW SERIES, NO. 17 The Mammalian Fauna of Madura Cave, Western Australia Part VII: Macropodidae: Sthenurinae, Macropodinae, with a Review of the Marsupial Portion of the Fauna Ernest L. Lundelius, Jr. William D. Turnbull March 31, 1989 Publication 1399 PUBLISHED BY FIELD MUSEUM OF NATURAL HISTORY Information for Contributors to Fieldiana General: Fieldiana is primarily a journal for Field Museum staff members and research associates, although manuscripts from nonaffiliated authors may be considered as space permits. The Journal carries a page charge of $65 per printed page or fraction thereof. Contributions from staff, research associates, and invited authors will be considered for publication regardless of ability to pay page charges, but the full charge is mandatory for nonaffiliated authors of unsolicited manuscripts. Payment of at least 50% of page charges qualifies a paper for expedited processing, which reduces the publication time. Manuscripts should be submitted to Scientific Editor, Fieldiana, Field Museum of Natural History, Chicago, Illinois 60605-2496, USA. Three complete copies of the text (including title page and abstract) and of the illustrations should be submitted (one original copy plus two review copies which may be machine copies). No manuscripts will be considered for publication or submitted to reviewers before all materials are complete and in the hands of the Scientific Editor. Text: Manuscripts must be typewritten double-spaced on standard-weight, 8V2- by 1 1-inch paper with wide margins on all four sides. For papers longer than 100 manuscript pages, authors are requested to submit a "Table of Contents," a "List of Illustrations," and a "List of Tables." In most cases, the text should be preceded by an "Abstract" and should conclude with "Acknowledgments" (if any) and "Literature Cited." All measurements should be in the metric system. The format and style of headings should follow those of recent issues of Fieldiana. For more detailed style information, see The Chicago Manual of Style (13th ed.), published by The University of Chicago Press, and also recent issues of Fieldiana. In "Literature Cited," authors are encouraged to give journal and book titles in full. Where abbreviations are desirable (e.g., in citation of synonymies), authors consistently should follow Botanico-Periodicum-Huntianum and TL-2 Taxonomic Literature by F. A. Stafleu & R. S. Cowan (1976 et seq.) (botanical papers) or Serial Sources for the Biosis Data Base (1983) published by the BioSciences Information Service. References should be typed in the following form: Croat, T. B. 1978. Flora of Barro Colorado Island. Stanford University Press, Stanford, Calif., 943 pp. Grubb, P. J., J. R. Lloyd, and T. D. Pennington. 1963. A comparison of montane and lowland rain forest in Ecuador. I. The forest structure, physiognomy, and floristics. Journal of Ecology, 51: 567-601. Langdon, E. J. M. 1979. Yage among the Siona: Cultural patterns in visions, pp. 63-80. In Browman, D. L., and R. A. Schwarz, eds., Spirits, Shamans, and Stars. Mouton Publishers, The Hague, Netherlands. Murra, J. 1946. The historic tribes of Ecuador, pp. 785-821. In Steward, J. H., ed., Handbook of South American Indians. Vol. 2, The Andean Civilizations. Bulletin 143, Bureau of American Ethnology, Smithsonian Institution, Washington, D.C. Stolze, R. G. 1981. Ferns and fern allies of Guatemala. Part II. Polypodiaceae. Fieldiana: Botany, n.s., 6: 1- 522. Illustrations: Illustrations are referred to in the text as "figures" (not as "plates"). Figures must be accompanied by some indication of scale, normally a reference bar. Statements in figure captions alone, such as "x 0.8," are not acceptable. Captions should be typed double-spaced and consecutively. See recent issues of Fieldiana for details of style. Figures as submitted should, whenever practicable, be 8V2 by 1 1 inches (22 x 28 cm) and may not exceed 1 IV2 by I6V2 inches (30 x 42 cm). Illustrations should be mounted on boards in the arrangement you wish to obtain in the printed work. This original set should be suitable for transmission to the printer as follows: Pen and ink drawings may be originals (preferred) or photostats; shaded drawings should be originals, but within the size limitation; and photostats should be high-quality, glossy, black and white prints. All illustrations should be marked on the reverse with author's name, figure number(s), and "top." Original illustrations will be returned to the author upon publication unless otherwise specified. Authors who wish to publish figures that require costly special paper or color reproduction must make prior arrangements with the Scientific Editor. Page Proofs: Fieldiana employs a two-step correction system. Each author will normally receive a copy of the edited manuscript on which deletions, additions, and changes can be made and queries answered. Only one set of page proofs will be sent. All desired corrections of type must be made on the single set of page proofs. Changes in page proofs (as opposed to corrections) are very expensive. Author-generated changes in page proofs can only be made if the author agrees in advance to pay for them. THIS PUBLICATION IS PRINTED ON ACID-FREE PAPER. UNIVERSITY OF ILLINOIS LIBRARY AT URBANA-CHAMPAIGN GEOLOGY FIELDIANA Geology NEW SERIES, NO. 17 The Mammalian Fauna of Madura Cave, Western Australia Part VII: Macropodidae: Sthenurinae, Macropodinae, with a Review of the Marsupial Portion of the Fauna Ernest L. Lundelius, Jr. Professor of Geological Sciences University of Texas at Austin Austin, Texas 78713-7909 Research Associate Department of Geology Field Museum of Natural History William D. Turnbull Curator, Fossil Mammals Department of Geology Field Museum of Natural History Chicago, Illinois 60605-2496 Committee on Evolutionary Biology University of Chicago Chicago, Illinois 60637 Research Associate Texas Memorial Museum Austin, Texas 78705 Accepted March 11, 1987 Published March 31, 1989 Publication 1399 PUBLISHED BY FIELD MUSEUM OF NATURAL HISTORY © 1989 Field Museum of Natural History Library of Congress Catalog Card Number: 88-82134 ISSN 0096-2651 PRINTED IN THE UNITED STATES OF AMERICA •^ «* ^ ^J Table of Contents 'i^ '■£ Abstract 1 Introduction 1 Macropodidae Sthenurinae Sthenurus (Simosthenurus) sp 1 Macropodinae Lagorchestes hirsutus 6 Lagostrophns fasciatus 19 Onychogalea lunata 23 Incertae Sedis within the Small Walla- bies 33 Protemnodon sp 39 Petrogale sp 45 Macropus 46 Macropus fuliginosus 51 Macropus sp. (Probably M. fuligino- sus) 51 Macropus titan 57 Macropus sp. (Probably M. titan) .... 57 Macropus robustus 62 Incertae Sedis among the Large Macro- podids 63 Assessment of the Marsupial Segment of the Fauna 65 Acknowledgments 69 Literature Cited 69 List of Illustrations 1. Sthenurus (Simosthenurus) sp., near S. oreas DeVis and 5". gilli Merrilees, from Madura Cave compared with S. atlas from Wellington Caves, New South Wales, S. andersoni from Weetalibah, New South Wales, and S. brownei from Mammoth Cave, Western Australia . . . 2-3 2. Bivariate graph with length of P4 plotted against posterior width of P4 for samples of the various species of Sthenurus 5 3. Drawing of the skull and right mandible of the modern Lagorchestes hirsutus from Bernier Island, Western Australia 8-9 4. Lagorchestes hirsutus from the surface of Madura Cave 10 5. Lagorchestes hirsutus from the surface of Madura Cave 11 6. Lagorchestes hirsutus from the surface of Madura Cave 12 7. Lagorchestes hirsutus and Lagostrophus fasciatus from Madura Cave compared with a Recent specimen from the sur- face of Webb's Cave, Mundrabilla Sta- tion, Western Australia 14-15 8. Lagostrophus fasciatus from Bernier Is- land, Western Australia 20-2 1 9. Onychogalea frenata from New South Wales 26-27 10. Onychogalea unguifera from Derby, Western Australia, and Onychogalea lunata from Weeke's Cave, South Aus- tralia 28-29 1 1 . Onychogalea lunata from Snake Pit Cave, Western Australia, and Weeke's Cave, South Australia 30-3 1 12. Onychogalea lunata from Madura Cave 32 13. Onychogalea lunata from Madura Cave 34 14. Protemnodon sp. from Madura Cave compared with Protemnodon anak from Wellington Caves, New South Wales ... 40 15. Protemnodon brehus and P. roechus from Wellington Caves, New South Wales 42-43 16. Bivariate graphs showing the Madura Cave specimens of Protemnodon in comparison with certain Wellington Caves specimens and with samples of P. brehus and P. roechus from the litera- ture 48^9 17. Petrogale sp. from Madura Cave com- pared with various modern species of the genus from Wedge's Cave, Mime- gara, Western Australia; Oenpelli, East Alligator River, Northern Territory; Kimberly District, Western Australia; and Rockhampton-Atherton area, Queensland 50 18. Graphs showing teeth of Macropus from Madura Cave compared with those of M. titan, M. giganteus, and M. fuligino- sus from various localities 52-53 19. Macropus fuliginosus from Madura Cave 54-55 20. Macropus titan from Madura Cave .... 59 2 1 . Macropus titan and M. robustus from Madura Cave compared with M. robus- tus antilopinus from Kimberly District, Western Australia, and modern species, M. fuliginosus, M. (Megaleia) rufa, and M. robustus antilopinus 64 ui List of Tables 1 . Numerical data on upper dentitions of Lagorchestes hirsutus from Madura Cave 16 2. Numerical data on upper dentitions of Holocene samples of Lagorchestes hir- sutus from Western Australia 17 3. Numerical data on lower dentitions of Lagorchestes hirsutus from Madura Cave 18 4. Numerical data on lower dentitions of Lagorchestes hirsutus from Western Australia 19 5. Measurements of upper dentitions of Lagostrophus fasciatus 22 6. Measurements of lower dentitions of Lagostrophus fasciatus 23 7. Numerical data on upper dentitions of Onychogalea lunata from Madura Cave 35 8. Numerical data on lower dentitions of Onychogalea lunata from Madura Cave 36 9. Dimensions of upper dentitions of Ony- chogalea from various sources 37 0. Dimensions of lower dentitions of Ony- chogalea from various sources 38 1 . Numerical data on mandibles and lower dentitions of Protemnodon from Lake Victoria, Wellington Caves, and Madura Cave 44 12. Numerical data on upper teeth of Pro- temnodon from Wellington Caves, Lake Victoria, and Bingara, New South Wales, and Queensland in comparison with the broken tooth of Protemnodon sp. from Madura Cave 46-47 13. Numerical data on upper dentitions of a Recent sample of Macropus giganteus from New South Wales 51 14. Numerical data on lower dentitions of a Recent sample of Macropus giganteus from New South Wales 51 15. Measurements of upper and lower den- titions of Macropus fuliginosus from Madura Cave 56 16. Numerical data on upper dentitions of a Recent sample of Macropus fuliginosus from New South Wales 57 17. Numerical data on lower dentitions of a Recent sample of Macropus fuliginosus from New South Wales 57 18. Measurements of metatarsals and termi- nal phalanges of digits IV and V of the pes of Recent and fossil Macropus 58 19. Measurements of upper dentitions of Macropus titan from Madura Cave . 60-6 1 20. Measurements of lower dentitions of Macropus titan from Madura Cave .... 62 2 1 . The marsupial component of the Ma- dura Cave fauna listed by taxon and stratigraphic occurrence, compared with the fauna reported by Milham and Thompson (1976) and the Recent fauna of the Nullarbor as reported by Brooker (1977) and others 66-67 The Mammalian Fauna of Madura Cave, Western Australia Part VII: Macropodidae: Sthenurinae, Macropodinae, with a Review of the Marsupial Portion of the Fauna Abstract The Sthenurinae and Macropodinae from Ma- dura Cave consist of Sthenurus (Simosthenurus) near 5. oreas and S. gilli, Lagorchestes hirsutus, Lagostrophus fasciatus, Onychogalea lunata, Pro- temnodon near P. brehus and P. roechus, Petrogale sp., Macropus fuliginosus, Macropus titan, and Macropus robustus. With the exception of Mac- ropus robustus and M. fuliginosus, which occur only in Units 2-7, all the extant species are found in all units of the deposit. The marsupial fauna from the Pleistocene Units 2-7 is more diverse than that of the Holocene Unit 1, and contains species that are found today in more mesic areas to the east and west of the Nul- larbor Plain. These Pleistocene units also contain numerous disharmonious pairs of species that in- dicate a more equable climate than that of the present. The assemblage from Unit 1 more closely approximates the present fauna of the region, but retains a few taxa now found to the east and west. Introduction This section of the study of the Madura Cave mammalian fauna covers Sthenurinae and Ma- cropodinae and concludes the systematic treat- ment of the marsupials. It also gives a brief anal- ysis of the marsupial fauna as discussed here and in the previous sections (Lundelius & Turnbull, 1973, 1975, 1978, 1981, 1982, 1984) and its re- lationships to other major Pleistocene marsupial faunas of Australia. Scales for the drawings are indicated adjacent each object; all are in centi- meters except for Figures 1 2 and 1 3, where some are in centimeters, others, in millimeters. Scales shown along the edges of Figures 1 6 and 1 8 are in millimeters. Values given in Tables 1-20 are in millimeters. Measurements, abbreviations, and statistical and dental terminology are either those in standard use, or they have been given in the previous sec- tions of this report, or they are defined where used. The study was completed before the dental ter- minology of Archer (1978) became widely ac- cepted; hence, the older standard of Thomas (1 888) was followed. MACROPODIDAE 1839 Sthenurinae Glauert, 1 926 Sthenurus Owen, 1873 (nomen nudum), 1874 (Simosthenurus) Tedford, 1 966 Sthenurus (Simosthenurus) sp. near S. oreas DeVis, 1895, and S. gilli Merrilees, 1965 Material Trench 2, Unit 2, 2 '/a ft PM 4356, right P4 (Sthenurus sp., Lundelius; 1963, S. Igilli, Merrilees, 1965; S. cf. oreas, Tedford, 1966) (fig. 1A) Trench 4, Unit 2, Level 1 PM 38998, anterior third, left P4 or P3 (fig. IB) Trench 4, Unit 2, Level 2 PM 38996, partial crown, right upper molar (fig. ID) PM 38997, molar fragment Trench 4, Units 4-5 TMM 41 106-3500, crown, left upper molar (fig. IQ Comparative Material Sthenurus andersoni Weetalibah (Binnia Creek), New South Wales PM 4516, symphysis and rami with left I-M2, right I-M, (fig. IF) Sthenurus atlas Wellington Caves, New South Wales PM 1571, right maxillary fragment with M1-3 (fig. IE) LUNDELIUS & TURNBULL: MADURA CAVE. PART VII. Fig. 1 . Sthenurus (Simosthenurus) sp., near S. oreas DeVis and S. gilli Merrilees, from Madura Cave compared with Sthenurus sp. from other localities. Sthenurus sp. from Madura Cave: A, PM 4356, right P4 shown in labial (right), lingual, and crown views; B, PM 38998, partial left P4 or P3 shown in labial (left), lingual, and crown views; C, TMM 41 106-3500, left upper molar crown shown in crown view; D, PM 38996, partial right upper molar shown FIELDIANA: GEOLOGY in crown view. Sthenurus atlas from Wellington Caves, New South Wales: E, PM 1571, right maxillary fragment with M1"3 shown in crown view. Sthenurus andersoni from Weetalibah (Binnia Creek), New South Wales: F, PM 45 16, symphysis and part of both rami with left I-M2 and right I-M, shown in crown view. Sthenurus brownei from Mammoth Cave, Western Australia: G, PM 4414, left maxilla with P*-M4 shown in crown view; H, PM 7981, right maxillary fragment with P5 and P4 (removed from crypt) shown in crown and labial views. LUNDELIUS & TURNBULL: MADURA CAVE. PART VII. PM 39065, left P4 PM 6776, P3 (or P3) PM 6777, right P4 Sthenurus brownei Mammoth Cave, Western Australia PM 4414, left maxillary with P4-M4 (fig. 1G) PM 7891, right maxillary fragment with P3 and P4 removed from crypt (fig. 1 H) Sthenurus tindalei Lake Menindee, New South Wales PM 4529 (cast of SAM PI 3820), palatal portion of skull with adult dentition Descriptions The P4 is a two-crested tooth (fig. 1 A). Its main crest begins near the anterior end of the crown at the anterior cusp. The crest soon incorporates a second laterally compressed, in-line cusp before dividing just anterior to the midpoint of the crown. From there the main crest is notched and contin- ues as the lingual crest. It first runs diagonally posterolingually to within the posterior quarter of the crown and then turns back toward the midline and abruptly tapers down to the crown base at the posterior edge of the tooth. This lingual crest is comprised of three narrow cusps, the anterior one (which lies just behind the notch) being the most distinct. From the dividing point at the notch a lower but distinct labial crest rapidly descends, at first running transversely, then turning posteriorly and reaching its lowest point before beginning a steady rise as it continues posteriorly. Nearly at the rear of the tooth, where it is again nearly as high as the main crest, it arcs lingually across the labial half of the tooth, descends, and swings slightly forward as it enters the median valley. This forms a backward-opening posterior central basin be- tween the two crests. Anteriorly within the basin, a low, sharp ridge connects the labial crest to the anterior end of the lingual portion of the main crest. This ridge is nearly parallel to the anterior, curved portion of the labial crest. At least two other crenulations swing off from the labial crest into the shallow posterolabial side of the basin. Measurements in millimeters of the tooth are: length 14.65, anterior width 6.63, posterior width 8.47, posterior basin length 8.15, and width 3.17. The fragment of a P4 (or P3) blade (fig. 1 B) cor- responds roughly to the anterior cusp of the main crest in the complete tooth, but differs in having more pronounced relief in its grooves and ridges. The two upper molars (fig. 1 C-D) are similar in size and major morphological features, but differ in detail. Both are brachyodont with lophs that are slightly convex anteriorly. The anterior cin- gulum of TMM 41 106-3500 is small, but extends across the entire anterior face of the tooth. The posterior cingulum of both molars is very low and thin. The two teeth differ markedly in the density and coarseness of the crenulations on the faces of the lophs. In PM 38996 the crenulations are less dense and coarser than in TMM 41 106-3500. In addition, the posterior crest of the paracone of PM 38996 is more prominent than that of TMM 41106-3500, and it closes the median valley la- bially. Low extensions of the midlink can be seen on the faces of the lophs in PM 38996, but not in TMM 41106-3500. In TMM 41106-3500 both faces of both lophs and the median valley are cov- ered with fine crenulations that tend to be oriented at right angles to the crests and lophs. The longer crenulations are slightly curved, and many bifur- cate away from the main crests and lophs. TMM 41 106-3500 is less worn than the other tooth and further wear would undoubtedly simplify the or- namentation, but probably not to the extent seen in PM 38996. Discussion The dimensions of the Madura Cave specimens have been compared with measurements for var- ious species of Sthenurus given by Bartholomai (1963), Tedford (1966), Merrilees (1965, 1968a), and Marcus (1962, 1976), and with an additional specimen of S. andersoni from Weetalibah, N.S.W. (PM 4516: P4 length 14.65 mm, anterior width 6.40, posterior width 7.50) (fig. 2). This compar- ison shows that the P4 from Madura Cave is too short to be assigned to S. occidentalis, S. orientalis, S. antiquus, S. pales, S. tindalei, S. atlas, or S. notabilis. It is too wide to be assigned to S. an- dersoni and slightly too narrow to be readily as- signed to 5. brownei. Its proportions are closest to those of S. gilli and S. oreas, which agrees with the Merrilees (1965) and Tedford (1966) assign- ments. It is also close to both of these species in the size of the posterior basin and in the absence from the basin of a ridge from the anterior cusp such as is shown by Tedford (1966) in S. pales. Comparisons of the two molars from Madura Cave with those from other localities is difficult because of the uncertainty of their positions in the tooth row. Their small size (measurements, ob- FIELDIANA: GEOLOGY CO -r- C\J o T" CO T ■8 X Q) \ | \ C \ \« \ s g (0 2 OB . co t>+co in -r - 9 3-2 z LU CO . CM a> co P/d H1QIM HOIhBJLSOd 3 . M O d £ 6 LUNDELIUS & TURNBULL: MADURA CAVE. PART VII. tainable only from TMM 41 106-3500, are: length 1 1.5 mm, anterior width 10.3 mm, posterior width 10.5 mm) excludes them from Sthenurus pales, S. tindalei, S. atlas, S. andersoni (except for the M1), probably from S. notabilis and S. antiquus, and possibly from S. oreas. They are within the size range of one or more of the upper molars of S. gilli, S. brownei, and S. occidentalis. The weak development of the forelink and the presence of a buccal ridge closing the median valley in PM 38996 are features cited by Merrilees (1968a) as char- acteristic of S. brownei. Extensive fine crenulations are also cited by Merrilees ( 1 968a) as characteristic of S. brownei, but his Figures 4 and 6 are not clear enough to allow detailed comparisons with TMM 4 1 1 06-3500. Comparison of the upper molars from Madura Cave with those of S. oreas is uncertain because of the lack of upper teeth which have been unequivocally assigned to that species. Material from Queensland referred to S. oreas by Bartho- lomai (1963) is reported by him to have coarse ornamentation. Bartholomai's Figure 5 indicates coarser ornamentation than is seen in TMM 4 1 1 06- 3500, but is not unlike that of PM 38996. It is not possible to make a positive assignment to species on the basis of the available material. The Madura Cave specimens are not necessarily all from the same taxon, but not enough is known about intraspecific variation of minor morpholog- ical features to rule out their assignment to the same taxon. Flannery (pers. comm., 1983) states that the difference in ornamentation between the two molars is greater than the range of variation in known samples of Sthenurus. Milham and Thompson (1976) reported Sthenurus teeth from the south passage of Madura Cave, referring them to two species, S. gilli and an unnamed larger form, but no figures or descriptions of the specimens are given. More material is needed from Madura Cave to determine the number and identity of the species of Sthenurus from this locality. Macropodinae Thomas, 1888 Lagorchestes hirsutus Gould, 1 844 Material Surface TMM 4 1 1 06-679, skull and left ramus (figs. 4A- D, 5A) Trench 1, top 1 ft PM 4784, left ramus fragment with M,.3, alveoli for P4 and M4 PM 25540, right ramus fragment with M3, al- veoli of M2 and M4 Trench 2, 2lh ft below surface PM 25221, left ramus with dP4, M„ part of M2, part of crypt for P4 (fig. 7B). Trench 3, Unit 2, Level ? PM 39038, right P4 PM 39039, left P4 Trench 4, Unit 1 , Level 1 PM 39047, right I1 Trench 4, Unit 1 , top 1 ft TMM 41106-5130, left P4 PM 39003, skull and mandible (figs. 5B, 6A-C) Trench 4, Unit 2, Level 1 TMM 41 106-5087, right I1 PM 38916, P3 Trench ? (probably 4), Unit 2, Level ? PM 38914, left maxillary fragment with P3 (fig. 7C) Trench 4, Unit 2, Level 2 PM 38947, right M4 Trench 4, Unit 2, Level 4 TMM 41106-150, left P4 cf. Lagorchestes hirsutus Trench 3, Unit 2, Level 1 TMM 41 106-5044, left M3 (or M2 or M4) TMM 41 106-5045, right M4 or M3 TMM 41 106-5049, left M, PM 38922, left M3 or M2 PM 39035, left M4 or M3 PM 39041, right M, Trench 3, Unit 2, Level 2 TMM 41106-140, left ramus fragment with M3^ Trench 4, Unit 1, top 1 ft TMM 41 106-5059, left M4 or M3 TMM 41 106-5082, right M2 or M3 PM 38892, left M3 or M4 PM 38893, left M4 PM 38898, left P4 PM 38901, right M3 or M4 Trench 4, Unit 2, Level 1 PM 38942-38944, three left M4s Trench 4, Units 4-5 PM 36981, right I3 Comparative Material Bernier Island, Western Australia AMNH 155106 (fig. 3) Webb's Cave (surface), Mundrabilla Station, Western Australia TMM 41209-528 FIELDIANA: GEOLOGY TMM 41209-891 TMM 41209-892 (fig. 7A) Descriptions Skull— The skull has the normal macropodid shape with the braincase relatively more inflated and the rostrum relatively narrower than in the large species of Macropus. The muzzle is wider than that of most Lagostrophus or Onychogalea of comparable size (O. unguifera being one excep- tion). In dorsal view the rostrum narrows abruptly immediately anterior to the orbits. The sides of the zygomatic arches and the interorbital area are straight and parallel. The nasals taper anteriorly and extend only slightly anterior to the dorsal ends of the premax- illae. They extend posteriorly as far as the lacri- mals and are square or gently rounded posteriorly. The frontals extend approximately one-half the distance from the posterior end of the nasals to the nuchal crest. The posterior end of each is gently rounded. The frontals expand anterolaterally to contact the maxillary and lacrimal bones. Between the orbits, where the lateral edges of the frontals are parallel to each other, they are sharply angled at the junction of the median wall of the orbit and the dorsal surface of the skull. In contrast, the modern comparative specimens of Lagostrophus and Onychogalea exhibit several different frontal shapes. The frontals of Lagostrophus fasciatus are narrower and taper posteriorly nearly to a point, those of Onychogalea frenata bow outward from a narrowest point just behind the flange for the lacrimal and then taper roundly posteriorly, and those of O. unguifera are wide and more closely resemble those of Lagorchestes than those of O. frenata. The parietals make up approximately two-thirds of the dorsal part of the braincase. They extend anteriorly, lateral to the frontals, as far as the pos- terior end of the interorbital constriction. A small triangular interparietal is present in front of the supraoccipital, at the junction of the sagittal and occipital crests. The modern comparative speci- men also has the interparietal, as do specimens of Onychogalea frenata and O. unguifera; specimens of Lagostrophus fasciatus lack it. The dorsal bor- der of the temporal fossa is marked by a weak ridge which extends from the posterior end of the interorbital constriction diagonally across the dor- sal surface of the braincase to join its mate, form- ing a short sagittal crest just anterior to the inter- parietal. In a modern specimen from Bernier Island (AMNH 155106, fig. 3), the ridges converge but do not join to form a sagittal crest. Each lacrimal has a small flange that projects into the antero- dorsal part of the orbit. In lateral view the premaxilla with its incisors projects below the plane of the molars. The dorsal profile of the skull is straight from just ahead of the frontoparietal suture to the anterior end of the nasal. This profile resembles that of the modern Lagostrophus fasciatus, but differs from Onycho- galea frenata and O. unguifera, which have a more rounded and undulating profile. The muzzle is deep, with sides that are almost flat. This is also true of the modern specimen and of specimens of Lagostrophus fasciatus. There is no prominent depression on the lower part of the lateral surface of the maxilla ahead of P4, such as is seen in mod- ern Onychogalea. The anterior opening of the in- fraorbital foramen is small (~ 1 mm), as in the modern specimen and in Lagostrophus fasciatus and Onychogalea unguifera; in O. frenata it varies between ~ 1 mm and ~ 3 mm. In the Madura Cave specimens the foramen is located immediately ahead of the orbit, about halfway between the tooth row and the dorsal surface of the skull. In Lagos- trophus and Onychogalea the anterior opening of the infraorbital foramen is located nearer the tooth row. The descending process of the zygomatic arch extends to the level of the occlusal surface of the molars, as in Lagostrophus. In Onychogalea the descending process is smaller and shorter. In ventral view the tooth rows are bowed out- ward slightly, with the maximum width at the po- sition of the descending process of the zygomatic arch. The palate shows little narrowing anterior to P4; this differs from both Onychogalea and Lagostro- phus. The modern specimen from Bernier Island (AMNH 1 55 106) shows more taper than the Ma- dura Cave specimens (figs. 3A,C, 4A,C, 6A,C). The incisive foramina are elongate ovals which extend from the level of the midpoint of I3 to the suture between the premaxilla and maxilla. They are larger than those of modern Lagostrophus, ap- proaching the size of those of modern Onycho- galea. The palatal fenestrae are irregularly oval and lie across the maxillary-palatine suture, opposite M3^* or M2-3. Additional fenestration of the pal- atines is not extensive. The fenestrae of the Ma- dura Cave specimens are smaller than those of the modern specimen, and much smaller than those LUNDELIUS & TURNBULL: MADURA CAVE. PART VII. * ■ ^» B FIELDIANA: GEOLOGY Fig. 3. Drawing of the skull and right mandible of the modern Lagorchestes hirsutus, AMNH 155106, from Bernier Island, Western Australia. Views of skull: A, dorsal; B, right lateral; C, ventral. Views of jaw: D, lateral; E, dorsal; F, mesial. LUNDELIUS & TURNBULL: MADURA CAVE. PART VII. ■ Fig. 4. Four views of a skull of Lagorchestes hirsutus, TMM 41 106-679, from the surface of Madura Cave: A, dorsal; B, left lateral; C, ventral; D, posterior. 10 FIELDIANA: GEOLOGY Fig. 5. Medial, dorsal, and lateral views of the left mandibles of two specimens of Lagorchestes hirsutus from the surface of Madura Cave: A, TMM 41 106-679 Oaw belongs to skull shown in fig. 4); B, PM 39003 Oaw belongs to skull shown in fig. 6). LUNDELIUS & TURNBULL: MADURA CAVE. PART VII. 11 s: T B < ■x' Fig. 6. Three views of a skull of Lagorchestes hirsutus, PM 39003, from the surface of Madura Cave: A, dorsal; B, left lateral; C, ventral. (Jaw of this specimen is shown in fig. 5B.) 12 FIELDIANA: GEOLOGY of modern Lagostrophus fasciatus and Onycho- galea frenata, where they extend farther forward into the maxilla. The very young modern O. un- guifera specimen has no true palatal fenestrae; in- stead, a small perforation pierces each maxilla at or near the suture with the palatine, and both max- illa and palatine are riddled with a dozen similar perforations, plus so many minute perforations that the bone resembles lacework. The auditory bullae are inflated to a greater de- gree than in Lagostrophus or Petrogale and to about the same degree as in Onychogalea, but not to the extent seen in Bettongia. The bullae of Lagor- chestes are more elongate than those of Onycho- galea. The mastoid process protrudes farther ven- trally and laterally, but is not bent forward to lap around the side of the tympanic as it is in Ony- chogalea. The tympanic in each of these taxa is expanded laterally to form a considerable meatal tube. Upper Dentition— The upper incisors are ar- ranged in a smooth, broadly U-shaped arch in the fossil and modern Lagorchestes, like modern On- ychogalea (ROM 91.11.1.190; USNM 122614, 237643), and in contrast to the V-shape of this arch in Lagostrophus (AMNH 155104). The I1 is the largest incisor. It is almost oval in cross section, with a flat area on the medial side and a shallow indentation posteriorly for the I2. In Lagostrophus the I' is smaller than I2 and I3 and is triangular in cross section. In Onychogalea the I1 is larger than I2 or I3 and has a more flattened oval cross section than that of Lagorchestes. The I2 and I3 are approximately equal in size, but are different in shape. The I2 is oval in cross section except for a flattening on the outside and a broad, shallow depression posteriorly for the an- terior edge of I3. A small posterior lobe is present but is not seen in worn teeth. The I2 of Onycho- galea is slightly larger and has a better developed posterior lobe. The I2 of Lagostrophus is much larger and more elongate than that of either La- gorchestes or Onychogalea, and is divided into two lobes by a diagonal anterior-interior groove along the occlusal surface. The I3 is an elongate tooth that is divided into two lobes, a rounded antero-external (labial) one and a sigmoidal lingual one that extends to the posterior end of the tooth. The two lobes are joined at the anterior end of the tooth. The I3 of Ony- chogalea is somewhat wider anteriorly and has a less continuous lingual sigmoidal ridge. The I3 of Lagostrophus has a straight lingual lobe that is not connected to the labial lobe anteriorly. A small (1 mm diameter) canine is located at the anterior end of the maxilla, separated from the I3 and the P4 by diastemas. Two bladelike teeth (PM 389 1 4, 389 1 6) are ten- tatively identified as P3 of this species. Their struc- ture is similar to that of the P4, but they are slightly smaller and their main ridges have only three cusps; those of the P4 have four. The lingual shelf of each is weaker than that of the P4 and the posterolingual cusp is continuous with the posterior cusp of the main ridge. Marshall (1973a) found that the basic cusp number in Lagorchestes leporides was the same for P3 as for P4, but in L. hirsutus, if these identifications are correct, these two teeth differ. Their dimensions are similar to those of two mod- ern specimens (WAM 685 from Dorre Island, Ml 471 from Canning Stock Route, tables 1-2). A positively identified dP4 is not present in the Ma- dura Cave sample. The P4 is an elongate blade with four cusps aligned along its ridge. There is a low lingual ledge that varies somewhat in its development from specimen to specimen. It may be represented by one or more separate cuspules, or there may be a low ridge incorporating them. Wear on these struc- tures is also variable; in PM 39003, in which the left M, has been shed, the P4 has worn a distinct abrasion facet on the central cuspule of the lingual ledge of P4. The main ridge bends sharply lingually at its posterior end to join a large posterolingual cusp which lies in line with the lingual ledge. This last cusp is usually separated from the lingual ledge by an open valley, but they may be connected. The anterior cusp of the main ridge is the same size as the posterior one. It forms a continuation of the main longitudinal crest of the blade. The anterior cusp lacks a distinct labial ridge. A weak labial cingulum is present. The upper molars are slightly longer than wide. In the specimens from Madura Cave and in the modern specimens, mesial drift has pushed the M' against the posterior end of P4 and interdental wear has removed most of the procingulum. Remnants of the procingulum of the M1 of PM 39003 indicate that it was about as extensive originally as those on M2^*. The molars increase in size from M1 to M4. In M' the protoloph is slightly shorter than the metaloph, in M2 the two lophs are nearly equal, and in M3 and M4 the metalophs are progressively shorter than the protolophs. Unworn lophs are thin and bowed forward in the middle. Worn lophs are straight. Low crests on the paracone and meta- cone become exaggerated by wear to give the ap- pearance of a labial link. The midlinks are formed LUNDELIUS & TURNBULL: MADURA CAVE. PART VII. 13 9 Fig. 7. Lagorchestes hirsutus and Lagostrophusfasciatus from Madura Cave compared with Lagorchestes hirsutus from Webb's Cave, Mundrabilla Station, Western Australia. Lagorchestes hirsutus: A, TMM 41209-892, Recent specimen from the surface of Webb's Cave, left ramus with P3 and dP4-M2 in lateral (left), dorsal, and medial views; B, PM 25221, from Madura Cave, left ramus with dP4, M„ part of M2, and a part of the crypt for P4 in lateral (left), dorsal, and medial views; C, PM 38914, from Madura Cave, left maxillary fragment with P3 shown in lateral (left), dorsal, and medial views. Lagostrophus fasciatus from Madura Cave: D, PM 38979, left M1 shown in labial (top), 14 FIELDIANA: GEOLOGY crown, and lingual views; E, PM 38909, left upper molar shown in labial (top), crown, and lingual views; F, PM 39054, right P4 shown in labial (right), crown, and lingual views; G, TMM 41106-580, right M, shown in lingual (left), crown, and labial views; H, TMM 41 106-5067, partial left M, or M2 shown in labial (left), crown, and lingual views; I, TMM 41 106-5084, right M2 or M3 shown in lingual (left), crown, and labial views. LUNDELIUS & TURNBULL: MADURA CAVE. PART VII. 15 Table 1 . Numerical data on upper dentitions of Lag- orchestes hirsutus from Madura Cave. N OR Mean ,>. M1 M2 M' M' M'~2 M2-3 M'-3 M1^ L AW PW L AW PW L AW PW L AW PW L AW PW L AW PW L L L L 3.94-^.08 1.72-1.82 2.32-2.34 4.81-5.30 1.68-1.80 2.10-2.25 3.80-4.07 3.40-3.55 3.82-3.83 5.12-5.22 4.09^.17 3.93-3.95 5.65-5.96 4.09^.36 3.55-3.67 5.54-6.05 3.97-4.04 3.45-3.65 8.34-8.76 10.25-10.49 13.72-14.37 19.29-19.45 4.01 1.77 2.33 5.06 1.74 2.18 3.94 3.48 3.83 5.17 4.13 3.94 5.81 4.23 3.61 5.80 4.01 3.55 8.55 10.37 14.05 19.37 by spurs of unequal size that join in the central valley. The protolophs make up the larger part of the midlinks, in the form of a posterior crest of the protocone that turns sharply to meet the spur from the metaloph. Mandible— The mandible has the normal ma- cro podid form. The horizontal ramus has about the same depth from M, through M4. The pro- portions of the horizontal ramus are similar for Lagorchestes hirsutus from Madura Cave, Webb's Cave (TMM 41209-891), the modern specimen from Bernier Island, a modern specimen of La- gostrophus fasciatus, and material of Onychogalea lunata from Webb's Cave on the Nullarbor Plain. However, modern specimens of Onychogalea frenata from Queensland and New South Wales and of O. unguifera from Western Australia have a more slender and elongate ramus, possibly be- cause of their immaturity (each has the P3 and dP4 in place). The profile of the tooth row in the mod- ern specimens is slightly arched in the region oc- cupied by P4 and M,, apparently as the result of some forward molar progression. The P4 in both Madura Cave specimens and in the modern spec- imens from Bernier Island is rotated forward and downward. The diastema is relatively shorter than in Onychogalea and about the same length as in Lagostrophus. A mental foramen is located just anterior to the P4 and about halfway between the upper and lower edges of the mandible. The ascending ramus and masseteric fossa show few distinctive features. The ventral border of the masseteric fossa is located higher above the ventral margin of the jaw and the postero ventral margin of the masseteric fossa is more rounded than in Lagostrophus. The foramen into the dental canal is located lower than in Onychogalea. The condyle of the jaw is oval in Lagorchestes and the articular surface is gently convex antero- posteriorly. In Lagostrophus the condyle is more elongate transversely and in Onychogalea it is al- most round and smaller than in Lagorchestes. In the modern Lagostrophus and all three species of Onychogalea a small bony spur projects forward from the medial side of the condyle; this has not been seen in Lagorchestes. Lower Dentition— The lower incisors are slen- der, procumbent, lancelike teeth. Enamel is pres- ent all along the ventrolateral surface. In the Ma- dura Cave specimens the anterior ends are truncated by a prominent wear surface oriented at an angle of about 45° to the long axis of the tooth. Onychogalea frenata has a similarly oriented wear surface, but in Lagostrophus fasciatus the wear surface is almost parallel to the long axis of the tooth, giving the tooth a sharply pointed end. This difference in the wear of the lower incisors is re- lated to the difference in the upper incisors in these taxa. The I1 and I2 of Lagorchestes and Onycho- galea are large and occlude primarily with the ends of the lower incisors, producing wear surfaces only at the ends of the lower incisors. In Lagostrophus the I1 is small relative to the I2 and P, which are elongate teeth that occlude with a considerable length of the upper edge of the lower incisors, pro- ducing wear surfaces along the upper edges of the lower incisors. The ventromedial edge of the lower incisors shows a wear facet produced by wear be- tween the two lower incisors. This facet is better developed in Lagostrophus than in Lagorchestes. The P3s have not been identified in the Madura Cave material. The dP4 of Lagorchestes is known from only one specimen, PM 2522 1 . It is assigned to Lagorchestes on the basis of its association in a mandible with an M, and M2 which lack the complex procingula characteristic of Lagostro- phus, and of the presence under the dP4 of a crypt for an unerupted P4 larger than that of Onycho- galea. The dP4 is an elongate molariform tooth that is wider across the hypolophid than across the 16 FIELDIANA: GEOLOGY Table 2. tralia. Numerical data on upper dentitions of Holocene samples of Lagorchestes hirsutus from Western Aus- Canning Stock Route* Dorre Islandf Be AN rnier Island N OR Mean N OR Mean IMI 155106 P3 L 1 3.7 3.7 3.9 3.9 AW 1 1.5 1.5 1.8 1.8 PW 1 2.1 2.1 2.0 2.0 dP4 L 1 3.7 3.7 4.1 4.1 AW 1 3.1 3.1 3.0 3.0 PW 1 3.4 3.4 3.2 3.2 P L 2 5.0-5.3 5.15 4 4.8-5.1 4.95 1 4.7 AW 2 1.8-2.2 2.00 4 1.9-2.0 1.93 1 1.8 PW 2 2.1-2.3 2.20 4 1.9-2.3 2.18 1 2.1 M1 L 3 3.5-4.0 3.73 5 3.5^.7 4.22 1 3.2 AW 3 3.6-3.8 3.73 5 3.7-3.9 3.86 1 3.4 PW 3 3.5^.1 3.80 5 3.7^.1 4.00 1 3.9 M2 L 3 5.0-5.4 5.17 5 5.0-5.6 5.42 1 5.0 AW 3 4.2-4.5 4.40 5 4.0-4.8 4.38 1 4.3 PW 3 4.0-4.6 4.37 5 3.9^».8 4.30 1 4.2 M' L 2 5.9-6.0 5.95 4 6.0-6.2 6.08 1 5.2 AW 2 4.9-5.0 4.95 4 4.2^.6 4.33 1 4.4 PW 2 4.0-4.5 4.25 4 3.7^.1 3.88 1 4.1 M4 L 1 6.3 6.3 1 5.8 AW 1 4.6 4.6 1 4.3 4.3 1 3.8 PW 1 3.8 3.8 1 3.5 M'-2 L 1 7.5 M2~3 L 1 10.2 M'-3 L 1 13.2 M'-» L 1 20.5 20.5 1 18.1 * Sample from Canning Stock Route in The Western Australian Museum (Ml 464, Ml 465, Ml 471). t Sample from Dorre Island in The Western Australian Museum (WAM 685, 10565, 10567, 10624, 10625). protolophid. The procingulum projects forward and is narrower than in any of the molars. The metaconid and entoconid have anterior and pos- terior crests. The anterior crest of the metaconid does not join the procingulum as it does in the molars, but instead turns into the cingular basin. The forelink extends in a sigmoid curve to the anterior point of the tooth. Labial to the forelink, the cingulum is very weak. The P4 is a simple blade made up of a long, compressed cusp on either end and two lower and smaller, indistinct cusps be- tween. The anterior cusp is smooth labially and has a variably developed ridge on its lingual side. The posterior cusp is bent lingually. There is no external cingulum and only a hint of an internal cingulum. The Madura Cave specimens agree with the descriptions of P4 of Lagorchestes leporides from Lake Victoria (Marshall, 1973a). The lower molars increase in size posteriorly (table 3). The hypolophid is wider than the pro- tolophid in M,, about equal in M2, and narrower in M3 and M4. There are no accessory ridges in the midlink area. The forelink tends to have a lingual bow, as is pointed out by Marshall (1973a) for Lagorchestes leporides from Lake Victoria. Mesial Drift Mesial (anterior) drift of the cheek teeth of La- gorchestes is apparent in the crowding of the M1 against the P4 in the upper dentition and in the rotation of P4 forward and downward in the man- dible. Other indications are the tendency of the lower tooth row to be arched and for the anterior molars to show heavy wear while the posterior ones show little or none. Sanson (1983) has point- ed out that mesial drift is at a maximum in those taxa in which the premolars are either reduced in size or are quickly shed; Peradorcas is the most extreme example. LUNDELIUS & TURNBULL: MADURA CAVE. PART VII. 17 Table 3. Numerical data on lower dentitions ofLag- orchestes hirsutus from Madura Cave. N OR Mean dP4 P4 M, M2 M, M4 M,_2 M2.3 M,_3 L AW PW L AW PW L AW PW L AW PW L AW PW L AW PW L L L L 3.72 2.04 2.14 4.4-4.8 1.5-1.8 1.5-1.9 3.8-4.1 2.7-3.0 2.8-3.4 4.5-4.7 3.2-3.6 3.3-3.7 5.0-6.0 3.1^.3 3.3^.3 5.5-5.6 3.6^4.7 3.1-3.8 8.5-8.7 9.7-10.8 13.6-14.3 19.3-19.6 3.72 2.04 2.14 4.60 1.65 1.78 3.90 2.82 3.10 4.62 3.33 3.56 5.47 3.82 3.75 5.54 4.05 3.37 8.60 10.13 13.94 19.46 In Lagorchestes, P4 appears to drift more readily than P4. After drifting over the hump in the jaw, the P4 becomes the most steeply inclined of all the teeth. The P4 appears to have a firmer anchorage to the bone than P4, and its lesser drift results in a slightly different mechanism of accommodation in the upper molar dentition. Here the drift is accomplished by extreme interdental wear so that M1 soon loses the lingual two thirds of its procin- gulum and a very appreciable portion of the pro- toconal and hypoconal areas of the crown itself. The result is that the tooth occupies a space only about one half that of its original length, and comes to overlap the labial side of P4. The P4 erupts progressively, so that the level of its occlusal sur- face remains below that of M1, causing a break in the curved surface of the functional occlusal plane. This helps it to maintain its occlusal relationship with P4 and M,. Another striking indication of mesial drift is the remarkable way that the most heavily stressed of the upper molar roots are eroded and resorbed, and apparently remodeled in the area between the crown base and the alveolar surface. This results in molar roots that curve posteriorly toward their tips. This phenomenon is best seen in the hypo- conal region of the anterior molars (TMM 41 106- 679, PM 39003, AMNH 155106). As far as we can tell it is most pronounced in Lagorchestes, intermediate in Onychogalea, and absent in La- gostrophus, which shows less medial drift.. The rea- sons for these differences are not known, but one could speculate that in Onychogalea, in which me- sial drift is fully comparable to that in Lagor- chestes, P4 is a very reduced tooth which offers very little resistance to drift. Perhaps without much resistance the alveoli have never had time to be- come as extensively ossified and alveolar remod- eling is all that is necessary to permit drift to occur. Conversely, in Lagorchestes, where a larger blade resists the drift more effectively, a more complete ossification may result; in this case the remodeling, which involves both the alveolar bone and the tooth base and roots, may depend to a greater extent on tooth root changes. Discussion Mean values of dental measures of the Madura Cave sample of Lagorchestes hirsutus average slightly larger than those of Recent samples given by Tedford (1967) and Marshall (1973a), but in most cases there is extensive overlap. There also is extensive overlap in dental dimensions between the Madura Cave sample and Recent samples from the Canning Stock Route and Dorre and Bernier Islands, Western Australia (tables 1-4). The lengths of P5 are greater in the Madura Cave sample than in Tedford's and Marshall's samples, and there is no overlap in the lengths of the P4. The lengths of Mj are lower in the Madura Cave sample, with no overlap, probably because the Madura Cave sam- ple is made up of older individuals in which both occlusal and interdental wear would act to reduce the length. As pointed out by Marshall (1973a), measures of L. hirsutus and L. leporides overlap. Tedford (1967, fig. 2) separated L. leporides and L. hirsutus by plotting length of P4 against maxi- mum width of M1. This separation is actually the result of the non-overlapping values of the length of P4, which is shown in Tedford's Table 30. The Madura Cave sample extends the range of lengths of P4 upward to 5.30 mm (table 1), which is below the lowest value (5.4 mm) for the combined Re- cent samples of L. leporides of Tedford (1967) and Marshall (1973a). On the basis of their smaller size all Madura Cave P4s are assigned to L. hir- sutus. Lagorchestes hirsutus is known from a wide area in the interior of Western Australia and from Ber- nier and Dorre islands, in Shark Bay, northwestern Australia (Ride & Tyndale-Biscoe, 1959; Finlay- 18 FIELDIANA: GEOLOGY Table 4. Numerical data on lower dentitions of Lagorchestes hirsutus from Western Australia. Canning Stock Route* Dorre Island Rprnipr IvI-.inH N OR Mean N OR Mean AMNH 155106 p, L 3.5 3.5 3.0 3.0 AW 1.4 1.4 1.6 1.6 PW 1.7 1.7 1.9 1.9 dP4 L 3.5 3.5 3.8 3.8 AW 2.3 2.3 1.8 1.8 PW 2.7 2.7 2.6 2.6 P4 L 4.6 4.6 4 4.4-4.6 4.48 I 3.8 AW 1.8 1.8 4 1.6-1.9 1.75 1 1.6 PW 1.8 1.8 4 1.7-1.9 1.83 1 1.6 M, L 3 3.5^1.0 3.70 4 4.0-4.5 4.28 1 3.5 AW 2 2.9 2.9 4 2.7-2.9 2.80 I 2.5 PW 3 2.9-3.2 3.10 4 2.9-3.2 3.03 2.8 M2 L 3 4.0-5.4 4.70 5 4.3-5.0 4.86 I 4.2 AW 3 3.4-3.5 3.43 5 3.3-3.6 3.46 I 3.0 PW 2 3.0-3.4 3.20 5 3.4-3.7 3.50 I 3.2 M, L 2 5.7-5.9 5.80 4 5.7-6.0 5.83 I 4.8 AW 2 3.7-3.9 3.80 4 3.7^.0 3.83 I 3.7 PW 2 3.6-3.9 3.75 4 3.4-3.8 3.53 I 3.6 M, L 2 5.9-6.0 5.95 2 5.7-6.3 6.00 5.5 AW 2 3.7-3.9 3.80 2 3.9-4.0 3.95 1 3.2 PW 1 3.3 3.3 2.9 M„ L 1 19.5 19.5 7.2 M2_, L I 9.6 M,_, L ... ... 1 12.9 Mw L 16.5 * Sample from Canning Stock Route in The Western Australian Museum (Ml 464, Ml 465, Ml 471). t Sample from Dorre Island in The Western Australian Museum (WAM 685, 10565, 10567, 10624, 10625). son, 1936; Glauert, 1933). Since it has previously been reported from Horseshoe Cave on the Nul- larbor Plain (although without stratigraphic con- text; Archer, 1972, 1974), its presence in the Ma- dura Cave fauna is not unexpected. Like many other species, it had a wider distribution in the past. It has been reported from deposits dated be- tween 30,000 and 35,000 b.p. in Devil's Lair, in the southwestern corner of Western Australia (Dortch & Merrilees, 1972; Baynes et al., 1975; Balme et al., 1978; Merrilees, 1979). Tedford (1967) reported a specimen, probably of Holocene age, from the lower Cooper's Creek 40 mi (6 1 km) east of Lake Eyre in South Australia. Lagorchestes hirsutus is said to have been an inhabitant of the open plains. Lagostrophus Thomas, 1887 Lagostrophus fasciatus (Peron & Lesueur), 1807 Material Trench 4, Unit 1 , Level 1 TMM 41 106-5067, left M, or M2 (fig. 7H) TMM 41 106-5068, right M2 or M3 PM 38909, left upper molar (fig. 7E) PM 38911, left M2 Trench 4, Unit 1, top 1 ft TMM 41 106-580, right M, (fig. 7G) TMM 41 106-5084, right M2 or M3 (fig. 71) PM 39054, right P4 (fig. 7F) Trench 4, Unit 2, Level 4 PM 38979, left M1 (fig. 7D) Comparative Material Lagostrophus f. fasciatus Bernier Island, Western Australia AMNH 155104 (fig. 8) Dorre Island, Western Australia USNM 218467 Descriptions Upper Dentition— The P4 is represented by the anterior half of an unrooted tooth, which preserves LUNDELIUS & TURNBULL: MADURA CAVE. PART VII. 19 • Fig. 8. Skull and left mandible of Lagostrophus f. fasciatus, AMNH 155104, from Bernier Island, Western Australia. Views of skull: A, dorsal; B, right lateral; C, ventral. Views of mandible: D, lateral; E, dorsal; F, medial. 20 FIELDIANA: GEOLOGY LUNDELIUS & TURNBULL: MADURA CAVE. PART VII. 21 Table 5. Measurements of upper dentitions of Lag- ostrophus fasciatus. Madura Cave Recent PM PM AMNH USNM 38979 38909 155104 218467 P3 L AW PW 4.5 2.6 2.6 dP4 L AW PW 4.1 3.6 3.3e P4 L AW PW 5.4 2.6 2.8 M1 L 4.2 4.2 4.2 AW 3.4 3.8 3.8 PW 3.6 4.1 3.7 M2 L 4.7 4.6 AW 4.5 4.0 PW 4.3 3.8 M3 L 5.1 4.9 AW 4.5 4.5e PW 4.5 3.9e M4 L AW PW 5.7e Molariform L tooth AW PW 3.3 e = Estimate. the anterior three of the usual 5 + cusps (fig. 7F). The tooth is bladelike, with the axial ridge con- tinuing onto the front of the anteriormost cusp. That cusp also has labial and lingual ridges that ascend posteriorly toward the tooth base. The lin- gual ridge joins a weak lingual cingulum; the labial ridge joins the outward-bulged side of the tooth about halfway toward the base of the crown. The M1 is an unworn, unrooted tooth (fig. 7D). The crown is rectangular, nearly as wide as long, and has only a weak constriction between its two parts. There is a distinct but unexpanded procin- gulum across the front of the tooth from the an- terior crest of the paracone to the anteromedial corner of the base of the protocone. A weak fore- link connects the base of the protoloph with the midpoint of the cingulum. Both lophs are bowed forward slightly, and the paracone and metacone have longitudinal anterior and posterior crests along their labial borders, which lie nearly in line. The midlink is a thin, simple, arcuate ridge that curves off" from the protocone to join the anterior side of the metaloph at about midheight. The pos- terior basin is open, and its medial and lateral margins are symmetrical. The distinctive second "postlink" is thin but pronounced. This structure is not the true postlink of Tedford (1966), which is also present. We call the tooth an M1 because it compares well in size with the M ' of the Amer- ican Museum specimen (fig. 8; table 5). In that specimen the other molars have a very similar morphology but are each successively slightly larg- er, suggesting that size is the best means of distin- guishing one tooth from another. The other upper molar is a partial tooth lacking the anterior side of the protoloph and the postero- labial corner (fig. 7E). It also appears to have been unerupted and unrooted. It has a well developed "postlink." Its only unusual feature is an anterior double wrinkle in the metaloph which is the meta- loph contribution to the midlink. Otherwise it agrees with the description of the M1. Lower Dentition— The lower molars conform to a standard pattern (fig. 7G-I). In the modern specimen used for comparison (fig. 8), all are elon- gate and rectangular with simple posteriorly bowed cross lophs, and there is a uniform gradual increase in size from M, to M4 (table 6). All lower molars of L. fasciatus have a distinctive forelink-procin- gulum pattern in which the link stands off from the cingulum, rather than merging with it as it does in most other wallabies. The hypolophid makes the greatest contribution to the midlink. The con- tribution of the protolophid to the link is variable, and sometimes absent. Other variable features are low, rounded bulges that sometimes occur on the anterior faces of the lophids medial to their mid- points. The Madura Cave teeth all conform to this pattern, and one or more exhibit each of the minor variables seen in the modern specimen. Estimates of their positions in the series are based on size. In spite of the small number of specimens of this taxon, the close agreement of morphological detail between the Madura Cave specimens and the modern comparative materials permits con- fident assignment of the Madura Cave material to Lagostroph us fasciatus. Discussion Lagostrophus fasciatus was found over much of southwestern Australia during the early period of European settlement (Shortridge, 1909; Calaby, 1 97 1 ). It may be extinct now on the mainland, but it survives on Bernier and Dorre islands in Shark Bay (Ride & Tyndale-Biscoe, 1959). It has been 22 FIELDIANA: GEOLOGY Table 6. Measurements of lower dentitions of Lagostrophus fasciatus. Madura Cave Recent TMM TMM TMM TMM AMNH USNM 41106-5067 41106-5068 41106-580 41106-5084 155104 218467 p, L ... 4.0 AW • • •• 1.8 PW •• •• 2.1 dP4 L AW PW •• •• 4.0 3.2 3.4 P4 L AW PW •• •• 4.7 2.0 2.3 M, L AW 4.< \ •• 4.3 3.1 4.2 3.2 4.5 3.4 PW ■• •• 3.3 3.4 3.4 M2 L • • • • 5.2 4.5 4.8 AW •• 3.; > 3.7 3.6 3.6 PW •• •• 3.3 3.7 3.5 M, L • • .. 4.9 AW • • • • 3.8 PW •• •• 3.8 M4 L AW •• •• 5.2e 4.9e PW e = Estimate. reported previously with no stratigraphic context from Horseshoe Cave on the Nullarbor Plain (Archer, 1972, 1974). The Nullarbor Plain rec- ords, some of which are of Holocene age (including most of the Madura Cave materials), together with its presence in archaeological sites of Holocene age along the lower Murray River in South Australia (Wakefield, 1 964), indicate a former distribution throughout South Australia. It has not been re- ported from Pleistocene faunas at Lake Menindee (Tedford, 1 967) or Lake Victoria (Marshall, 1973a). Onychogalea Gray, 1 84 1 Onychogalea lunata (Gould), 1 840 Material Trench 1 , Unit 1 , top 1 ft PM 4783, left maxilla with P4-M4 (fig. 12B) PM 4785, left ramus with I,-M4 (fig. 12A) PM 25538, right maxillary fragment with P3, dP4, P4 removed from crypt (fig. 1 2Q PM 25539, right ramus fragment with P4 re- moved from crypt, M„ alveoli for P3 and dP4, M2_3 in crypt (fig. 12D) PM 25541, left ramus with M2^,, alveoli for P4- M, Trench 2, 2'/2 ft below surface PM 25222, right ramus with part of M4 Trench 3, Unit 2, Level ? (presumably 1) TMM 41 106-183, right ramus with M,_2, P4 re- moved from crypt, M3 in crypt, alveoli for P3anddP4(fig. 13 A) TMM 41 106-184, right maxilla with P3, dP4, P4 removed from crypt, M1 2 (fig. 13B) TMM 4 1 106-506 1 , left maxillary fragment with P3, dP4, P4 removed from crypt, alveoli for M1 (fig. 13Q TMM 41 106-5088, right ramus fragment with dP4, P4 removed from crypt, alveoli for M,, crypt for M2_3 (fig. 13D) PM 34469, left maxillary fragment with P4 PM 38918, left ramus fragment with M, or dP4 PM 38926, left M, or dP4 PM 38927, trigonid, left M2 or M3 PM 39005, right maxillary fragment with P4- M1 (fig. 13E) PM 39007, right premaxillary fragment with I1 (fig. 13F) PM 39049, right maxilla with M'-\ alveoli for dP4, crypt for P4 PM 39050, left maxilla with M'"2 PM 39052, right maxillary fragment with M3-4 LUNDELIUS & TURNBULL: MADURA CAVE. PART VII. 23 Trench 3, Unit 2, Level 2 TMM 41106-141, left maxilla with unworn dP4, alveoli for P3 and M1, crypt for P4 (opened but tooth lost) (fig. 13G) TMM 41 106-142, left maxillary fragment with dP4, alveoli for P3, P4 removed from crypt (in two pieces) Trench 3, Unit 2, Level 4 PM 38919, left P4 Trench 4, Unit 1 , top 1 ft TMM 41 106-494, left maxillary fragment with M1"2, crypt for P4 TMM 4 1 1 06-495, right maxillary fragment with M1"2 TMM 41 106-5063, edentulous right premaxilla (fig. 13H) PM 38884, left premaxilla with I2 (fig. 131) PM 38894, right P4 PM 38899, left M1 PM 38900, right M3 or M2 PM 38902, left P4 PM 38915, left dP4 or M, PM 39046, left maxillary fragment with P3 Trench 4, Unit 2, Level 1 TMM 41106-315, left ramus with M3, alveoli for P4-M2 and M4 TMM 4 1 1 06-3 1 6, right maxillary fragment with dP4, P4 removed from crypt, alveoli for P3, M1 PM 38775, right dP4 PM 38940, left P4 or P3 PM 38941, left dP4 PM 38995, left M4 cf. Onychogalea lunata Trench 3, Unit 2, Level 1 TMM 41 106-5039, right M4 or M3 TMM 41 106-5040, right M3 or M4 TMM 41 106-5047, right M3 or M2 PM 39033, right M4 or M3 Trench 3, Unit 2, Level 4 PM 39016, left M1 PM 39017, left dP4 Trench 4, Unit 1, top 1 ft TMM 41 106-492, left ramus fragment with M2 or M3 TMM 41 106-493, right M3 or M2 TMM 4 1 106-542, left ramus fragment with M^ TMM 41 106-543, left ramus fragment with MM TMM 41 106-579, left dP4 or M, TMM 41 106-581, left M3 TMM 41 106-594, broken right M1 TMM 41 106-625, right dP4 TMM 41 106-627, right M, or M2 TMM 41 106-2833, right M, TMM 41 106-2834, right dP4 TMM 41106-5069, left upper molar, probably M1 or M2 TMM 41 106-5072, right upper molar, probably M2 or M3, possibly M4 TMM 41 106-5075, worn left dP4 TMM 41 106-5076, right M, or dP4 TMM 41 106-5078, right M4 TMM 41 106-5079, left M3 or M4 TMM 41 106-5080, left M3 TMM 41 106-5151, left dP4 PM 38903, left M4 Trench 4, Unit 2, Level 1 PM 38925, right ramus fragment with M2_3 PM 38980, right M2 or M3 or M4 PM 38982, right M, PM 38984, right M3 or M4 PM 38986, right M1 PM 38987, left M3 or M2 PM 38988, left M3 or M2 PM 38989, left M2 or M3 PM 38991, right M4 PM 38993, left M, Trench 4, Unit 2, Level 2 PM 38931, right dP4 PM 38932, left M1 Comparative Material Onychogalea frenata Warwick, Queensland ROM 91.11.1.190 (fig. 9C) New South Wales via National Zoological Park USNM 122614, male (fig. 9A-B) National Zoological Park USNM 219299, male (diseased and abnormal) Onychogalea unguifera Derby, Western Australia USNM 237643 (fig. 10A-B) Onychogalea lunata Weeke's Cave (surface), South Australia PM 38776, subadult (fig. 10C) PM 38777, subadult (fig. 11Q Weebubbie Cave (surface), Western Australia TMM 41 107-334, juvenile TMM 41 107-335, juvenile Jenning's Cave (surface), Western Australia TMM 42121-1, juvenile Snake Pit Cave (surface), Western Australia TMM M-937 (fig. 1 1A-B) Descriptions Upper Dentition— An I1 and an I2, both in pre- maxillary fragments, are the only upper incisors 24 FIELDIANA: GEOLOGY from Madura Cave assigned to Onychogalea. The I1 (fig. 13F) is curved along its long axis and has a flattened oval cross section. The occlusal surface makes an acute angle with the mesial surface of the tooth. The morphology is like that of modern O.frenata, but the size is smaller. The I2 (fig. 131) is almost square in cross section. It too is slightly smaller than the I2 of modern O. frenata, and is more deeply worn than the I2 of the modern Queensland specimen, as a result lacking the pos- terior lobe seen in that specimen. The occlusal surface is oriented at an angle of about 45° to the long axis of the tooth. The morphology of the in- cisors clearly marks them as Onychogalea. Their assignment to O. lunata is based on their small size (Marshall, 1973a) and the presence of other specimens clearly assignable to that species. The P3 is a variable tooth (figs. 12C, 13B-C,G). Most specimens are triangular, with two large crested cusps on the labial edge separated by a valley. A third smaller cusp is located lingual to the main posterior one and is connected to it by a weak ridge. One specimen (PM 25538; fig. 12C) has an elongate molariform P3 with four cusps and incipient transverse lophs. This tooth is narrower across the anterior loph than across the posterior loph. None of the P3s has a lingual cingulum. The dP4 is a molariform tooth that differs from M ' in its slightly smaller size, more elongate rect- angular form, and greater development of the par- astylar crest (figs. 1 2C, 1 3B-C,G). The protoloph is variably developed. Most specimens, such as TMM 41 106-5061 (fig. 13C), have a well-defined protoloph, but one (TMM 4 1 1 06-3 1 6) lacks a pro- toloph, the protocone and paracone being sepa- rated by a V-shaped valley. The procingulum is more asymmetrical than it is in the molars because of the strong development of the parastylar area. The midlink is low and extends straight from the protocone to the center of the metaloph. There is only a hint of a forelink. The P4 is smaller than the P3. Its morphology is variable (figs. 1 2B-C, 1 3B-C,E), but less so than that of the P3. All available specimens are trian- gular with three cusps, two large ones on the labial side of the tooth separated by a deep valley, and a smaller cusp lingual to the large posterior cusp and connected to it by a ridge. One specimen (PM 38919) has a small cuspule in the valley between the two labial cusps, and an incipient lingual cin- gulum. The upper molars increase in size posteriorly (table 7; figs. 12B, 13B,E). The protoloph is nar- rower than the metaloph in M1, about the same width in M2, and wider in M3 and M4. The para- cone and metacone are crested when unworn. The protoloph and metaloph are strongly bowed an- teriorly, particularly in unworn teeth. The mid- links consist of an anterior part that extends from the protocone to the center of the interloph valley, and a posterior part that extends from the center of the metaloph. Small accessory ridges are present on the midlink at the point where the two halves meet; these ridges are particularly noticeable on unworn teeth. There usually are other accessory ridges in the interloph valleys, those on the lingual side developed from the base of the tooth between the protocone and the hypocone. The procingulum is prominent, and reaches almost all the way across the front of the tooth. The anterior edge of the procingulum is joined to the paracone by a ridge. There is no forelink, but a swelling is present on the cingulum in the position usually occupied by the forelink. The postlink isolates a small pit at the back of each tooth. Mandible— The mandible of Onychogalea lun- ata from Madura Cave is similar to that of La- gorchestes hirsutus, but is slightly smaller and more delicate (figs. 12A,D, 13A,D). It also is slightly smaller than the modern comparative mandible of O.frenata. The horizontal ramus has upper and lower borders which are parallel in adults, but in one juvenile (PM 25539; fig. 12D) the horizontal ramus is deeper under the alveolus for the dP4. In profile, the tooth row is arched, with the apex of the arch located at dP4 in PM 25539, at P4-M, in PM 4785 (fig. 12A), and at M2 in PM 25541. This change in the position of the teeth with respect to the apex of the arch, together with the more an- terior position of the M4 relative to the ascending ramus in older individuals, indicates the existence of mesial drift in Onychogalea lunata, as in La- gorchestes hirsutus. A prominent rugosity is lo- cated on the side of the jaw below the ventral margin of the temporal fossa. This is not present in Lagorchestes or Lagostrophus, but is present in the modern specimens of Onychogalea frenata. A mental foramen is located just anterior to P4. Lower Dentition— Only one of the specimens (PM 4785; fig. 12A) that can be reliably assigned to Onychogalea has a lower incisor. The end is broken, but the small part of the wear surface that remains is oriented at an angle of about 45° to the long axis of the tooth. This is similar to the ori- entation of the wear surface in Lagorchestes (see the section on that taxon) and different from the wear surface on the lower incisor of Lagostrophus, in which it is oriented at a very low angle to the long axis of the tooth. No P3 has been recognized from the Madura Cave material. Only TMM 41 106-5088 (fig. 13D) has a dP4 in LUNDELIUS & TURNBULL: MADURA CAVE. PART VII. 25 26 FIELDIANA: GEOLOGY Fig. 9. Onychogalea frenata, USNM 122614, from New South Wales (via National Zoological Park): A, skull shown in dorsal, left lateral, and ventral views; B, left mandible shown in lateral (left), dorsal, and medial views. Onychogalea frenata, ROM 9 1 . 1 1 . 1 . 1 90, from Warwick, Queensland: C, right maxilla and mandibular dentitions shown in occlusal views. LUNDELIUS & TURNBULL: MADURA CAVE. PART VII. 27 Fig. 10. Onychogalea unguifera, USNM 237643, from Derby, Western Australia: A, skull shown in top dorsal, left lateral, and ventral views; B, right mandible shown in lateral (left), dorsal, and medial views. Onychogalea lunata, PM 38776, from Weeke's Cave, South Australia: C, partial skull shown in right lateral and ventral views. 28 FIELDIANA: GEOLOGY LUNDELIUS & TURNBULL: MADURA CAVE. PART VII. 29 Fig. 1 1 . Onychogalea lunata, TMM M-937, from Snake Pit Cave, Western Australia: A, skull shown in dorsal, left lateral, and ventral views; B, lower jaws shown in dorsal view, with medial view of left ramus and lateral view of right ramus. Onychogalea lunata, PM 38777, from Weeke's Cave, South Australia: C, partial skull shown in right lateral and ventral views. 30 FIELDIANA: GEOLOGY LUNDELIUS & TURNBULL: MADURA CAVE. PART VII. 31 * t# 32 FIELDIANA: GEOLOGY place. This tooth is elongate and molariform, dif- fering from the molars in being slightly smaller and relatively narrower across the protolophid. The protolophid and hypolophid are slightly bowed posteriorly. The midlink extends in a straight line from the hypoconid to a point labial to the center of the protolophid. There are no accessory ridges in the interloph valley. The procingulum is an ar- cuate ridge that connects the forelink with a crest on the anterior side of the metaconid to enclose an almost circular cingular basin. Two of the iso- lated teeth have the morphology of dP4s. However, they are slightly wider across the protolophids than is the dP4 of the one specimen which has that tooth in place (TMM 41106-5088), so it is uncertain whether they are dP4s or M,s with narrower than normal protolophids. The P4 apparently is highly variable. Two spec- imens show its well-developed form— a two-cusped tooth which is widest posteriorly. The cusps are in line with the long axis of the tooth. The anterior cusp is the smaller of the two, and may be either a simple cone (as in PM 4785; fig. 1 2A) or flattened from side to side and thus more bladelike (as in TMM 41 106-183; fig. 13A); in PM 4785, this cusp has another much smaller cusplet adhering to its labial side. The posterior cusp has a posterolingual bladelike crest, and may have other weaker crests or bulges. In two other specimens (TMM 41 106- 5088, fig. 13D; PM 25539), this tooth is only a single cusp in a crypt, at a very early develop- mental stage. Each is a minute cone, about 1 mm high and less than 1 mm in diameter, and one shows a small posterior crest. A modern young adult specimen from the surface of another Nul- larbor cave, Snake Pit Cave (TMM M-937; fig. 1 1 B), has the tooth fully erupted and shows that it may sometimes be much smaller and have a far simpler morphology, consisting of a single high cusp and a low talonid bulge. The crown of this tooth is supported by two in-line fused roots. The lower molars differ from each other only in minor details (figs. 1 2A,D, 1 3A). The size gradient is M, < M2 < M3 = M4 (table 8). The protolophids and hypolophids are slightly bowed posteriorly. In M, to M3 the widths of both lophs are very nearly the same, but in M4 the protolophid is noticeably wider than the hypolophid. The anterior crest of the hypoconid makes a smooth, slightly sigmoid sweep, first medially, then anteriorly, to join the protolophid just labiad to its midpoint. At this junction the link reaches to midheight on the near- ly vertical posterior side of the protolophid. The protolophid contribution to the midlink is very weak. There tends to be a bulge or cusplet on the lingual side of the procingulum. This feature is distinct from the bulge of the edge of the cingulum itself and varies in its expression. It is commonest on M3, often found on M4, and rarely present on the anterior molars. Discussion Onychogalea lunata has been recorded as a liv- ing species from southwestern Western Australia, inland across southern Australia to the region of the junction of the Murray and Darling rivers in western New South Wales or Victoria, and within the southern part of the Northern Territory. The Elder Expedition reported it in the Everard Range of South Australia, and the Horn Expedition col- lected it at Alice Springs (Shortridge, 1 909; Glauert, 1933; Jones, 1923-1925). Tedford (1967) gives measures for a juvenile from Rawlinna, on the Nullarbor Plain in Western Australia. This record supports the belief of Shortridge (1909) that this animal was distributed across the Nullarbor Plain in the recent past. The easternmost record from the junction of the Murray and Darling rivers is based on specimens collected by the Blandowski Expedition (Wakefield, 1 966). In eastern Australia it is known from Holocene deposits in an archae- ological site at Fromm's Landing on the lower Murray River (Wakefield, 1 964), and from Pleis- tocene deposits at Lake Menindee and Lake Vic- toria in western New South Wales (Tedford, 1 967; Marshall, 1 973a). In Western Australia it is known from undated deposits in Horseshoe Cave (Archer, 1972, 1974). It is present in both Holocene and Pleistocene deposits in Madura Cave. Comparison of dental measurements of speci- mens from different stratigraphic levels in Madura Cave (tables 7-8), from Lake Menindee (Tedford, 1967, tables 33-34), and from Lake Victoria (ta- bles 9-10) shows extensive overlap in all cases. To test for differences between the samples from Unit 2 and Unit 1, Mann- Whitney tests were run on lengths, anterior widths, and posterior widths of dP4, P4, M\ dP4, and M,. No significant differences (P > .05) were found. All samples currently avail- able are too small to demonstrate any significant geographic or chronological differences. Incertae Sedis within the Small Wallabies Material Probably Onychogalea lunata Trench 3, Unit 2, Level ? LUNDELIUS & TURNBULL: MADURA CAVE. PART VII. 33 B b 3 g. a a i m^ ^K^ \ 34 FIELDIANA: GEOLOGY Table 7. Numerical data on upper dentitions of Onychogalea lunata from Madura Cave. Unitl Unit 2 N OR Mean N OR Mean P3 L 2 2.6-2.9 2.75 2 2.7-3.0 2.85 AW 2 1.3-1.4 1.35 2 1.3-1.4 1.35 PW 2 2.0 2.0 2 2.0 2.0 dP4 L 1 3.5 3.5 7 3.7^.3 3.93 AW 1 2.7 2.7 7 2.6-3.1 2.89 PW 1 2.8 2.8 7 2.9-3.5 3.15 P4 L 3 2.7-3.1 2.86 5 2.9-3.3 3.02 AW 3 1.2-1.4 1.31 5 1.2-1.7 1.45 PW 4 1.6-2.3 1.84 5 1.9-2.4 2.05 M' L 4 3.7-4.2 3.87 4 3.4^t.O 3.76 AW 4 3.2-3.4 3.25 4 3.0-3.2 3.12 PW 4 3.3-3.5 3.38 4 3.1-3.3 3.20 M2 L 3 4.5^.6 4.54 2 4.0-^.5 4.25 AW 3 3.6-3.7 3.62 2 3.4-3.7 3.55 PW 3 3.5-3.6 3.55 2 3.3-3.6 3.45 M3 L 5.1 5.1 1 4.7 4.7 AW 3.8 3.8 1 3.6 3.6 PW 3.5 3.5 1 3.3 3.3 M4 L 5.3 5.3 1 4.8 4.8 AW 3.9 3.9 1 3.6 3.6 PW 3.0e 3.0e 1 2.8 2.8 M'-2 L 8.3 8.3 2 7.8-8.5 8.15 M2"3 L 9.5 9.5 M'-3 L 13.1 13.1 ... ... M'^ L 18.1 18.1 e = Estimate. TMM 41 106-55, left M, TMM 41 106-5041, right M' PM 39034, left M3 or M2 Trench 4, Unit 1, top 1 ft TMM 41 106-544, left M2 or M3 TMM 41 106-2835, right M1 TMM 41 106-2836, left M, TMM 41 106-5073, right M1 or M2 TMM 41 106-5074, right M'orM2 TMM 41 106-5083, right M4 or M3 Trench 4, Unit 2, Level 1 PM 38901, right M3 or M4 PM 38977, left M1 or dP4 PM 38978, right M3 or M2 Trench 5, Unit 6 PM 38885, right M2 Probably Lagorchestes hirsutus Trench 4, Unit 1, top 1 ft TMM 41 106-5081, left M2 or M3 PM 38891, right M2 Probably Lagorchestes hirsutus or Onychogalea lunata Opposite Page: Fig. 13. Onychogalea lunata from Madura Cave. A, TMM 41 106-183, right ramus fragment with P4 removed from crypt, M,_2, M3 in crypt, and alveoli of P3 and dP« shown in dorsal and lateral views. B, TMM 41 106-184, right maxilla with P3-M2 shown in lateral (left), occlusal, and medial views. C, TMM 41 106-5061, left maxillary fragment with P3, dP, P* (removed from its crypt), and alveoli of M1 shown in medial (left), crown, and lateral views. D, TMM 41 106-5088, right ramus fragment with dP4, P„ (only partly formed and removed from its crypt), alveoli of M,, and crypts for M2_3 shown in lateral (left), dorsal, and medial views. E, PM 39005, right maxillary fragment with P*-M' shown in lateral (left), crown, and medial views. F, PM 39007, right premaxillary fragment with I1 shown in lateral (left) and medial views. G, TMM 41 106-141, left maxilla with dP4 and alveoli of P3 and M1 shown in medial (left), crown, and lateral views. H, TMM 41106-5063, edentulous right premaxilla shown in lateral (left), ventral, and medial views. I, PM 38884, left premaxilla with I2 shown in lateral (left), ventral, and medial views. LUNDELIUS & TURNBULL: MADURA CAVE. PART VII. 35 Table 8. Numerical data on lower dentitions of Onychogalea lunata from Madura Cave. Unitl Unit 2 N OR Mean N OR Mean dP4 L 1 3.5 3.5 AW 2.2 2.2 1 2.0 2.0 PW 2.4 2.4 1 2.4 2.4 P. L 2.6 2.6 1 2.7 2.7 AW 1.1 1.1 1 1.0 1.0 PW 1.3 1.3 1 1.6 1.6 M, L 2 3.7-3.9 3.80 2 3.6-3.7 3.65 AW 2.6 2.6 2 2.2-2.4 2.30 PW 2.8 2.8 3 2.4-2.8 2.57 M2 L 4.2 4.2 1 4.5 4.5 AW ... 1 3.1 3.1 PW 3.1 3.1 2 2.9-3.1 3.00 M3 L 4.9 4.9 1 4.9 4.9 AW ... ... 1 3.2 3.2 PW 3.4 3.4 1 3.2 3.2 M4 L AW 2 4.8-5.0 3.3 4.90 3.3 PW 2 2.7 2.7 1 2.8 2.8 M,_2 L 7.8 7.8 M2_3 L 9.1 9.1 M,_3 L 12.7 12.7 MM L 17.2 17.2 Trench 3, Unit 2, Level ? TMM 41 106-5043, left dP4 TMM 41 106-5046, right M3 TMM 41106-5086, right ramus fragment with M,_3 or M2^, Trench 4, Unit 1, top 1 ft TMM 41 106-5058, right M2 or M3 TMM 41 106-5077, left dP4 or M, Undetermined Trench 3, Unit 2, Level ?, probably level 1 TMM 41 106-2838, right P4 TMM 41 106-5029, left I, TMM 41 106-5030, right I, TMM 41 106-5031, right I, TMM 41 106-5032, left I, TMM 41 106-5033, incisor or canine TMM 41 106-5034, badly eroded left lower mo- lar TMM 41 106-5035, left M, or M2 TMM 41 106-5036, broken right M4 TMM 41 106-5037, broken left M3 or M4 TMM 41 106-5042, broken left M4 TMM 41 106-5048, broken left M3 or M4 TMM 41 106-5050, right M, or M2 TMM 41 106-5051, worn out upper molar TMM 41 106-5052, right M2 or M3 TMM 41 106-5053, broken right M3 or M4 TMM 41 106-5054, worn out upper molar TMM 41 106-5055, broken right M3 or M4 TMM 41 106-5056, broken left M2 or M3 PM 38923, right I, PM 39008, right ramus with alveolus for dP4- M„ crypts for P4 and M2_3 PM 39009, right ramus fragment with alveolus for one molar, crypt for one molar PM 39010, maxillary fragment with two worn and broken molars PM 39036, broken right M3 or M4 PM 39037, worn out molar fragment PM 39042-39044, three right I,s PM 39045, left I, PM 39051, left M, PM 39053, right maxillary fragment with al- veolus for two molars Trench 3, Unit 2, Level 2 TMM 41 106-5121, broken left M2 or M3 Trench 3, Unit 2, Level 4 PM 38920, broken and eroded right M, or dP4 PM 39011, worn left I, PM 39012, right I, 36 FIELDIANA: GEOLOGY Table 9. Dimensions of upper dentitions of Onychogalea from various sources. O. lunata o. mm- guifera Derby, Lake Victoria, N.S.W.* Weeke's Cave, S.A ., and NMV P NMV P Snake Pit Cave, w.a. PM PM , surface TMM W.A. USNM O. frenata USNM USNM ROM 28573 28830 38777 38776 M-937 237643 219299 122614 91.11.1.190 P3 L 3.5 3.7 3.5 AW 2.0 2.0 1.8 PW 2.7 2.5 2.8 dP4 L AW PW 3.7 2.8 3.0 4.8 3.6 4.0 3.9 3.2 3.5 4.0w 3.7 3.8 P4 L AW PW 3.3 1.6 2.1 2.5 1.4 1.8 3.3 1.2 2.1 3.8 2.3 2.8 M1 L 4.0 3.7 3.6 4.3 5.6 4.8 4.5 4.7 AW 3.2 3.1 3.1 3.5 4.3 4.1 3.8 4.1 PW 3.3 3.1 3.2 3.5 4.4 4.2 3.8 4.2 M2 L 4.2 4.3 4.9 5.7 5.3 5.6 AW 3.5 3.6 3.8 4.5 4.2 4.4 PW 3.5 3.4 3.7 3.6 4.5 4.2 4.4 M3 L 5.0 4.9 5.0 5.6 6.7 5.8e 6.4 AW 3.8 3.7 4.0 4.9 4.6 4.8 PW 3.6 3.6 3.4 3.7 4.6 4.3e M4 L AW PW 4.9 3.9 3.4 ::: M'-2 L 7.9 7.7 8.9 10.4 9.9 M2"3 L 9.3 9.1 10.4 12.5 11.9 M'-3 L 12.7 12.4 14.2 16.8 16.3 * Data from Marshall (1973a, table 45). w = Worn; e = estimate. PM 39015, broken right upper molar PM 39019, broken left upper molar Trench 3, Unit 3, Level ? TMM 41106-37, left ramus fragment with al- veolus for M2^» Trench 4, Unit 1, Level ?, probably level 1 TMM 41 106-2837, broken premolar fragment PM 39137, right upper I1 PM 39142, terminal phalange Trench 4, Unit 1, Level 1 TMM 41 106-552, broken right lower molar TMM 41 106-5064, maxillary fragment with al- veolus for two molars TMM 41 106-5065, left upper molar TMM 41106-5066, eroded and broken M, or dP4 TMM 41 106-5148, broken left I, PM 38896, broken molar fragment PM 38908, incompletely formed left I1 PM 39109, anterior half, upper molar Trench 4, Unit 1 , top 1 ft TMM 41 106-624, broken lower molar TMM 41 106-626, broken upper molar TMM 41 106-5070-5071, two broken right up- per molars PM 38897, ramus fragment PM 38904, worn out M, PM 38905, worn right I, PM 38910, broken left M4 PM 38913, right ramus fragment PM 39132, right upper I PM 39133, right upper molar Trench 4, Unit 2, Level 1 PM 7986-7987, right lower I and left M„ re- spectively PM 38936, left premaxilla with I2 LUNDELIUS & TURNBULL: MADURA CAVE. PART VII. 37 Table 10. Dimensions of lower dentitions of Onychogalea from various sources. O. frenata O. lunata O. un- New South Wales (via Weebubbie Cave, Jen- viirt-i'*" ning's Snake Pit guifera National National Cave, Cave, Derby, Zoologi- Zoological Warwick, TMM TMM surface surface W.A. cal Park Park) Queensland 41107- 41107- TMM TMM USNM USNM USNM ROM 335 334 42141-1 M-937 237643 219299 122614 91.11.1.190 P3 L 2.5 2.4 3.5 3.7 3.5 AW 1.3 1. 3 1 7 2.0 1.8 PW 1.4 1. 4 2 2 2.5 2.8 dP4 L 3.5 3. 5 3.2 4 7 3.9 4.0w AW 2.1 2. 3 2.0 2 7 3.2 3.7 PW 2.5 2. 4 2.3 3 1 3.5 3.8 P4 L AW PW 1.8 1.1 3.8 2.3 2.8 M, L 3.8 3. 7 3.6 4.1 5 7 4.8 4.5 4.7 AW 2.6 2. 7 2.4 2.6 3 4 4.1 3.8 4.1 PW 2.6 2. 7 2.6 2.9 3 8 4.2 3.8 4.2 M2 L 5.0 5.9 5.3 5.6 AW 3.1 4.5 4.2 4.4 PW 3.3 4.5 4.2 4.4 M3 L AW PW 5.5 3.3 3.3 6.7 4.9 4.7 5.8e 4.6 6.4 4.8 4.3e M4 L AW PW 5.2e 3.3 2.6e M,_2 L 8.9 10.4 9.7 9.9 M2_3 L 10.2 12.5 11.0 11.9 M..3 L 14.0 16.8 15.4 16.3 MM L 19.2e w = Worn; e = estimate. PM 38975, worn out lower molar PM 38976, right M2 or M3 PM 38981, right M, or M2 PM 38983, broken upper molar PM 38985, broken lower molar PM 38990, broken left M2 or M3 PM 38992, broken left M2 or M3 PM 38994, broken right lower molar PM 391 19-39121, three terminal phalanges PM 39122-39123, two subterminal phalanges PM 39146, 39148, two anterior halves, upper molars PM 39150, posterior half, left upper molar PM 39151, posterior half, left lower molar Trench 4, Unit 2, Level 2 PM 38933, left M2 PM 38934, left M3 PM 38948, right I, PM 38954, broken left upper molar PM 39072, anterior half, left upper molar PM 39077, partial upper incisor Trench 4, Unit 2, Level 3 TMM 41 106-4A-B, right I,s Trench 4, Units 4-5 PM 38876, left I, PM 38877-38878, two maxillary fragments PM 39080, partial left upper molar PM 39090, half, molar PM 39096, anterior half, left lower molar Trench 4, Unit 7, Level ? PM 38928, right ramus fragment with alveolus for M2_3, crypt for M4 PM 38929, right ramus fragment with alveolus for M,_3, crypt for M4 38 FIELDIANA: GEOLOGY Trench 4, Unit 7, Level 2 PM 38882, left maxillary fragment with broken P4, alveolus for M1 PM 38930, broken left M' Trench 5, Unit 5 PM 38887, left I2 or I3 Trench 5, Unit 5 or 6(?) PM 39126, molar fragment Trench 5, Unit 6 TMM 41 106-641, broken left lower molar PM 38886A-B, two upper incisors PM 39125, left I1 Protemnodon Owen, 1873 (nomen nudum), 1874 Protemnodon sp. near P. brehus (Owen) and P. roechus Owen Material Trench 2, 2Vi ft PM 53920, ventral side, left I, Trench 4, Unit 2, Level 1 PM 39063, left ramus with broken P4, M,_3, roots of M4 (fig. 14 A) Trench 4, Unit 2, Level 2 PM 39089, posterior one-third, left P4 (fig. 14B) TMM 41 106-2832, anterior half, left upper M3 or M4 (fig- 14C) Comparative Material Protemnodon anak Wellington Caves, New South Wales PM 1553, left ramus with broken I, P4-M4 (fig. 14D) cf. Protemnodon brehus Wellington Caves, New South Wales PM 1 534, left ramus with I, dP4, P4 in crypt (fig. 15 A) PM 1541, right ramus with P3, dP4, and P4 and M3 in crypts (fig. 1 5B) PM 1543, right ramus with dP4-M,, broken P4 in crypt (fig. 1 5C) PM 1544, left ramus with M,_3 (fig. 15D) PM 1551, right ramus with P3, dP4, P4 in crypt PM 1557, left ramus with P3, dP4-M2 PM 1560, right mandible with P4 exposed in crypt, M,_3, M4 in crypt (fig. 15E) cf. Protemnodon roechus Wellington Caves, New South Wales PM 1570, right maxilla with P3, dP»-M2, P4 in crypt (fig. 15F) PM 1583-1584, left maxilla in two fragments with P3, dP»-M2, P4 and M3 in crypt (fig. 15G) PM 39064, right P3 (fig. 15H) PM 39066, right P4 Descriptions The horizontal ramus (fig. 14A) is moderately shallow and thick (depth at M,-M2 = 34.2 mm, thickness = 16.35 mm; depth at M2-M3 = 33.2 mm, thickness ■ 17.9 mm). From P4 to M3, it shows little change in depth, but thickens appre- ciably. The base of the symphysis rises at a low angle from the plane of the ventral edge of the horizontal ramus. Only the posterior part of the symphysis is preserved, so its shape and length cannot be determined. The shape of the preserved portion suggests that it was shallow, but it is deeper than that of a specimen of Protemnodon anak from Wellington (PM 1553; fig. 14D). The symphysis is rugose, but not ankylosed. The geniohyal pit is shallow and located at the posterior end of the symphysis. The mental foramen is located about 1 1 mm anterior to P4 and about 4 mm below the dorsal edge of the diastema. The lateral groove is shallow and is located 7 mm below the edge of the alveolus; it extends from the premolar at least to the posterior root of M2. The posterior part of the P4 is broken away. The outline of this tooth is an elongate oval. The labial surface bears an irregular wear facet for most of its length, which covers from one-third (in rear) to one-half (in front) of the crown below the crest of the occlusal surface. The anterior part of this wear surface has a broad groove across it, setting off a triangular ridge. The anterior part of the un- worn lingual surface is gently convex. The lower molars are rectangular in occlusal view with a slight constriction at the interloph valley, which is mostly confined to the labial side of the teeth. The relative sizes of the molars are: M, < M2 < M3. The protolophid is slightly narrower than the hypolophid in M, and about equal to it in M2 and M3 (table 1 1). The lophids are weakly convex posteriorly when unworn but are straight when worn. Forelinks are moderately well devel- oped, extending from the protoconid anterolin- gually and then anteriorly to join the procingulum just labiad the center line of the tooth. The pro- cingulum is prominent but narrow. It descends labially from its junction with the forelink to the LUNDELIUS & TURNBULL: MADURA CAVE. PART VII. 39 Fig. 14. Protemnodon sp. from Madura Cave: A, PM 39063, left ramus with P4-M3 shown in lateral (left), occlusal, and medial views; B, PM 39089, partial left P4 shown in labial (left), crown, and posterior views; C, TMM 41 106- 2832, anterior half of left M3 or M4 shown in crown view. Protemnodon anak from Wellington Caves, New South Wales: D, PM 1553, left ramus with I (broken)-M4 shown in lateral (left), dorsal, and medial views. 40 FIELDIANA: GEOLOGY base of the protoconid, forming an anterolabial pit. The anterolingual area of the procingulum is nearly flat. The midlink descends anterolingually from the hypoconid and then turns anteriorly to join the protolophid near its base at the midline of the tooth. The labial side of the interloph valley slopes away from the midlink more steeply than does the lingual side. Low, indistinct lingual ridges descend from the metaconid and entoconid to- ward the interloph valley. The postcingulum is very small or absent. There is no bulge at the base of the crown. The P4 is represented by the posterior third of the crown and most of the posterior root (fig. 14B). The outer blade is missing, but the posterior basin formed by the posterior cingulum, the posterolin- gual cusp, and the rear of the labial crest is well preserved, as is the posterior part of the lingual trough. The ridgelike posterolingual cusp is joined to the main crest and to the lingual crest. There is a vertical groove on the lingual surface where it joins the lingual crest. The Madura Cave specimen is similar in size and morphology to a specimen from Lake Menindee figured by Tedford (1967, fig. 25C), which he referred to Protemnodon bre- hus. Because of its large size (width across protoloph = 15.1 mm), the anterior half of the upper molar (TMM 41 106-2832) probably is an M3 or M4. Its anterior width falls within the limits of the ranges of M2, M3, and M4 of Bartholomai's (1973) Darling Downs, Queensland, sample of Protemnodon roe- chus, and just within the range of M3 of his Queensland sample of P. brehus, and of M4 of his Bingara, New South Wales sample. It also falls within the range of M3 of Marshall's (1973a) Lake Victoria, New South Wales, sample. There is no forelink, and the stout procingulum has a low angle of inclination. The protoloph is bowed anteriorly in its middle. There is a delicate but distinct pro- toconal spur. A similarly developed labial crest and spur extend from the posterolabial edge of the paracone. Both of these spurs lead into the inter- loph valley. The tooth is slightly larger than that recorded by Stirton (1963, pp. 152-153) for the type and most of the other specimens of P. brehus, as well as any of the specimens of other species measured by him. Discussion Six species of Protemnodon were named by Owen (1874, 1877) on the basis of material from Pleis- tocene deposits in Australia: P. anak, P. og, P. minutus, P. brehus, P. antaeus, and P. roechus. Stirton's review of the genus (1963) made a start at determining the relationships of these species by unraveling several taxonomic problems, but presented no definitive discussion of the validity of each species. Bartholomai's review of the genus (1973) synonymized P. og into P. anak, P. mimas into P. brehus, and P. antaeus into P. roechus, and split off P. chinchillaensis and P. devisi, both Plio- cene in age, from P. anak. The characters which have been used to differ- entiate these species show considerable intrasam- ple variability, the extent of which has been poorly understood until recently. This has made specific identification of Protemnodon specimens difficult, especially in the case of isolated specimens. Bar- tholomai (1973) analyzed large samples of Pro- temnodon from Queensland, which provided some information on intraspecific variation for material from that area. Marcus (1976) has done the same for material from Bingara. We have doubts about the usefulness of the qualitative characters suggested by Bartholomai ( 1 973) as distinguishing Protemnodon brehus from P. roechus. The tuberculation on the lingual side of the interloph valley of the upper molars is vari- able in the two comparative specimens from Wel- lington Cave. Furthermore, one of Bartholomai's figures (1973, pi. 1 3) shows a specimen of P. brehus (which is supposed to lack this feature) to have a weakly developed tuberculation on M1 and M3. The degree of labial concavity of the labial crest of P4 in P. roechus, as shown in Bartholomai's figures (1973, fig. 7, nos. 5-8), appears to be vari- able. The extent of the expansion of the bases of the lower molars also is variable. In view of these doubts and of the extensive overlap in the size ranges of virtually all metric characters (see Bar- tholomai, 1973, tables 6, 1 0, fig. 9), we have doubts, as did Flower (1884), Lydekker (1887), and Mar- shall (1973a), that the two species are distinct. Nonetheless, since we lack the comparative ma- terial to investigate this question, we will consider these to be two separate species in the following discussion. The width of the protoloph of the upper molar fragment from Madura Cave exceeds the upper limit of the observed range for all upper molars of Protemnodon anak given by Bartholomai (1973), but is within the observed range for M3 and M4 of P. brehus and M2^* of P. roechus (table 12). No qualitative character, such as the tuberculation seen by Bartholomai (1973) on the labial side of the LUNDELIUS & TURNBULL: MADURA CAVE. PART VII. 41 Fig. 15. Protemnodon sp. from Wellington Caves, New South Wales, for comparison. Protemnodon brehus: A, PM 1534, left ramus with I, dP4, and P4 in crypt shown in lateral (left), dorsal, and medial views; B, PM 1541, right ramus with P3 and dP4 shown in lingual (left), crown, and labial views; C, PM 1543, right ramus with posterior roots of P3, broken P4, and dP4-M, shown in medial (left), dorsal, lateral, and anterior views; D, PM 1544, left ramus with 42 FIELDIANA: GEOLOGY H M|_, shown in lateral and crown views; E, PM 1560, right mandible with P4-M4 shown in dorsal and lateral views. Protemnodon roechus: F, PM 1570, right maxilla with P3-M2 shown in lateral and ventral views, and P4 exposed in crypt shown in lateral, ventral, and lingual views; G, PM 1583-1584, adjoining maxillary fragments with P3-M3 shown in crown view; H, PM 39064, right P3 shown in labial (left), crown, and lingual views. LUNDELIUS & TURNBULL: MADURA CAVE. PART VII. 43 Table 1 1 . Numerical data on mandibles and lower dentitions of Protemnodon from Lake Victoria, Wellington Caves, and Madura Cave. Protemnodon cf. brehus Protemnodon brehus Lake Victoria* N OR Mean Ma- dura Cave PM 39063 Wellington Caves N OR Mean Protem- nodon anak Welling- ton Caves PM 1553 p3 L AW PW dP4 L AW PW P4 L AW PW M, L AW PW M2 L AW PW M, L AW PW M4 L AW PW Mandible below M|_2 Mandible below M2_3 Depth Depth Mandible at M,_2 Mandible at M2.3 Thickness Thickness 12.2 10.0 9.7 15.0-15.2 11.6-12.3 17.4-18.0 12.8-13.9 12.7-13.6 18.8-19.5 13.1 11.6-12.7 12.2 10.0 9.7 15.07 11.95 17.68 13.43 13.10 19.10 13.1 12.15 16.4 6.3 11.9 8.9 9.3 13.0 10.4 10.5 16.2 11.9 12.0 34.2 33.2 16.3 17.9 11.7-12.3 5.5-6.5 6.3-6.9 10.3-12.5 5.9-7.5 7.1-8.7 17.5 12.1-13.9 8.7-10.2 8.9-9.9 15.8-16.7 11.1-11.5 11.2 18.1-19.9 12.6-12.7 12.4-12.9 12.05 5.90 6.60 11.55 6.85 8.04 17.5 13.20 9.45 9.55 16.23 11.30 11.2 19.00 12.65 12.65 32.0-32.2 32.10 30.9-31.2 31.05 13.7-14.4 14.05 15.4-16.2 15.80 17.4 5.1 7.3 10.9 6.5 7.6 12.7 10.0 9.3 15.4 17.0 11.1 9.7 32.3 32.8 10.9 11.3 * Data from Marshall (1973a, table 61). interloph valley in P. roechus is preserved. It is not clear whether this molar and the ramus rep- resent the same species. However, since they are similar in size and lack features which might sug- gest a difference, the most parsimonious assump- tion is that they do represent the same taxon. The ramus from Madura Cave differs from oth- ers assigned to Protemnodon anak in a number of characters which have been cited as diagnostic by Bartholomai (1973) and Marcus (1976). The angle between the base of the symphysis and the ventral margin of the horizontal ramus rises at a steeper angle (10°-20°), and the symphysis is deeper. The geniohyal pit is prominent just behind the sym- physis. The horizontal ramus is deeper and thick- er, as is the ramus of our comparative specimen from Wellington Cave (PM 1553; table 11, fig. 1 6 A), but matches the dimensions given by Bar- tholomai for P. brehus and P. roechus and by Mar- cus for P. brehus. The dental dimensions of the Madura Cave specimen are larger than those given for Protem- nodon anak, but within the ranges given for P. brehus and P. roechus by Bartholomai (1973), Marshall (1973a), and Marcus (1976) (tables 1 1- 1 2, fig. 1 6). A comparison of the Madura Cave specimen with the qualitative dental characters given by Bartholomai is less easily made. The an- 44 FIELDIANA: GEOLOGY tenor cingular areas of the lower molars of the Madura Cave specimen are not as wide relative to the widths of the lophids as is shown in Bar- tholomai's figures of P. brehus, but are wider than is shown for P. anak. The Madura Cave specimens also differ from P. anak and are similar to P. brehus and P. roechus in most other characters, but be- cause of the doubts mentioned above it is difficult to be certain to which of these two species they belong. All measurements which could be taken on the Madura Cave specimens fall within the broad area of overlap of the size ranges of these two species (table 1 1 , fig. 1 6). The qualitative char- acter that suggests the most unequivocal assign- ment is the degree of expansion of the base of the crown in the lower molars. In P. roechus the base of the crown is expanded (Bartholomai, 1973); in P. brehus and the Madura Cave specimens the base of the crown is not expanded. Protemnodon has been reported from other lo- calities on the Nullarbor Plain. Glauert (1912) re- ported P. anak from Balladonia, but Merrilees (1968a) subsequently referred this material to Sthenurus. Protemnodon cf. brehus has been re- ported by Milham and Thompson ( 1 976) from the south passage of Madura Cave, but they did not figure the material or give the basis for their as- signment; P. brehus has also been reported from the Mammoth Cave deposits in southwestern Australia (Tedford, 1967). Petrogale Gray, 1837 Petrogale Species Indeterminate Material Trench 3, Unit 2 PM 39006, left I1 in a fragment of the premaxilla (fig. 17 A) Trench ? (probably 4), Unit 1, top 1 ft PM 39130, left I1 (fig. 17B) Trench 4, Unit 2, Level 2 PM 39068, left M4 (fig. 17C) Comparative Material Petrogale brachyotis Kimberly District, Western Australia FM 119823 (fig. 17F) FM 120577 Petrogale cf. lateralis Northwest Cape, Western Australia (Late Pleis- tocene or Holocene) PM 26694 PM 26701 PM 36718 Wedge's Cave, Mimegara (north of Perth), West- em Australia (Late Pleistocene or Holocene) PM 5749 (fig. 17D) PM5771 PM 5772 PM 5773 Petrogale inornata Rockhampton-Atherton area, Queensland FM 64360 FM 64430 (fig. 17G) Petrogale venustula Oenpelli, East Alligator River, Northern Territory USNM 284068 Petrogale pearsoni Oenpelli, East Alligator River, Northern Territory RCS London A.348.51 (fig. 17E) Descriptions The left M4 (PM 39068; fig. 1 7C) compares well with that of a Holocene specimen of Petrogale from Wedge's Cave, Western Australia (PM 5749; fig. 17D) in most morphological characters, al- though it is slightly smaller. The protoloph and metaloph are convex anteriorly, while the proto- loph is noticeably longer than the metaloph. Both lophs contribute to the midlink. A cleft divides the midlink in the median valley. Ridges from the paracone and metacone almost meet on the labial side of the median valley to form a median basin. The procingulum occupies the entire anterior bor- der of the tooth. It is connected to the paracone by a prominent ridge, but is not connected to the metacone, leaving the procingular basin open on the lingual side. The postcingulum is connected to the hypocone by a ridge, forming a posterior cin- gular basin. According to Merrilees (1979), this is characteristic of M2^*; this agrees with our obser- vations on recent specimens of Petrogale from Queensland, Western Australia, and the Northern Territory. This character distinguishes the M2~* of Petrogale from those of macropodids such as Mac- ropus irma and M. eugenii that have dentitions similar to those of Petrogale. The Madura Cave specimen differs from the Wedge's Cave specimen principally in its small size and in the form of the procingulum, which does not slope lingually to- ward the base of the tooth. The dimensions of PM 39068 are: length 7.50 mm, anterior width 5.00 mm, and posterior width LUNDELIUS & TURNBULL: MADURA CAVE. PART VII. 45 Table 1 2. Numerical data on upper teeth of Protemnodon from Wellington Caves, Lake Victoria, and Bingara, New South Wales, and Queensland in comparison with the broken tooth of Protemnodon sp. from Madura Cave. Protem- nodon sp. P. roechus Madura Cave, W.A. P. brehus Wellington Caves, N.S.W.* PM PM PM 1570 1583-84 39066 Mean Darling Downs, Qld.f N OR Mean TMM 41 m*- Bingara, N.S.W4 2832 N OR Mean P4 L 22.4 20.4 21.40 7 18.3-20.7 19.4 AW 10.3 10.4 10.35 7 9.2-10.4 9.7 PW 11.3 10.5 10.90 M1 L 13.5 13.7 13.60 4 12.9-14.0 13.4 AW 11.9 12.1 12.00 9 12.2-13.9 12.8 PW 12.0 12.2 12.10 M2 L 15.2 15.5 15.35 12 15.7-17.7 16.7 4 15.4-16.6 16.00 AW 13.5 12.9 13.20 10 13.7-15.4 14.6 15.1 4 13.3-13.9 13.60 PW 13.0 12.7 12.85 M3 L 18 17.2-19.9 18.6 1 16.7 16.70 AW (13.6) 12 14.9-16.2 15.6 15.1 PW M4 L 13 17.7-20.0 19.0 2 18.7 18.70 AW 11 14.7-16.5 15.7 15.1 2 14.4-15.2 14.80 PW 2 * FMNH specimens not previously reported, f Data from Bartholomai (1973, table 10). % Data from Bartholomai (1973, table 6). § Data from Marshall (1973a, table 61). 4.35 mm. A comparison of these dimensions with those given by Merrilees (1979) for a series of Pleistocene, Holocene, and modern samples of Pe- trogale from southwestern Western Australia shows that the length of the Madura Cave specimen is within the observed range but the widths, espe- cially the posterior width, are below the observed ranges. The I's (PM 39006, 39130; fig. 17A-B) are strongly curved, with a wear surface developed on the lingual face. There is a shallow groove near the posterior edge of the labial surface of PM 39006, which produces a notch on the cutting edge of the tooth; no trace of this groove can be found on the other specimen. A vertical buttress is present on the posterior part of the lingual surface of the tooth. None of the Madura Cave specimens shows a dis- crete cuspule arising from this lobe, as reported by Merrilees (1979) for some specimens from southwestern Western Australia. Discussion The taxonomy of this genus is confused. Tate ( 1 948) recognizes three species with eight subspe- cies. Marlow (1962) shows distributions of seven species divided into 17 subspecies, while Ride (1970) recognizes six species. Calaby (1971) has suggested that this is the result of the highly frag- mented distribution of the genus today, which has caused a large amount of local variation. The Ma- dura Cave material is inadequate for a specific assignment. Rock wallabies of the genus Petrogale inhabit cliffs, rock piles, and rocky outcrops in most parts of Australia (Calaby, 1971). The genus has not been reported from the Nullarbor Plain. The only parts of this area that appear to provide suitable habitat are the scarp that separates the Roe Plain from the Hampton Tableland, and possibly some of the karst features such as the larger dolines. Macropus Shaw, 1790 The taxonomy of the genus Macropus has long been a problem. There is no general agreement on the boundaries of the genus, although most stu- dents now place the red kangaroo in a separate genus, Megaleia (Sharman, 1961; Calaby, 1966; Frith & Calaby, 1969; Bartholomai, 1975). More 46 FIELDIANA: GEOLOGY Table 12. Extended. P. brehus Queensland Sample^ Lake Victoria§ N OR Mean N OR Mean 9 10 18.1-19.8 9.3-10.6 19.2 10.0 6 4 12.7-14.6 12.0-12.7 13.6 12.3 1 1 1 14.4 12.2 12.6 14.4 12.2 12.6 14 6 14.8-17.0 13.4-14.6 16.3 14.0 1 1 1 16.4 14.8 14.8 16.4 14.8 14.8 18 15 16.2-18.1 13.7-15.1 17.5 14.5 2 2 1 18.2-18.4 15.0-15.2 15.2 18.30 15.10 15.2 12 11 17.3-19.1 13.7-15.0 18.2 14.3 2 2 17.7-17.8 14.6-14.7 17.75 14.65 recently, Peacock et al. (1981) have suggested that there is little justification for this separation be- yond the chromosome number difference. Even with Megaleia removed, the genus is particularly troublesome for the paleontologist because so many species have been named on the basis of minor dental characters whose significance is unknown. Although it has now become possible to separate many of the large-sized species of Macropus and to separate Macropus from Megaleia on the basis of dental characters which are usable on paleon- tological materials (Tedford, 1967; Frith & Cala- by, 1969; Bartholomai, 1975; Marcus, 1976), problems remain. It has been shown by Kirsch and Poole (1967, 1972) on the basis of serological studies and by Peacock et al. (1981) on the basis of DNA se- quence studies that the living gray kangaroos are in fact two species, Macropus giganteus and M. fuliginosus. Macropus giganteus is distributed over the eastern part of Queensland, much of New South Wales and Victoria, and northern Tasmania, while M. fuliginosus is found in western Victoria, south- western New South Wales, and the southern parts of South Australia and Western Australia (Shep- herd, 1982). The two species overlap without in- terbreeding in western Victoria and southwestern New South Wales (Shepherd, 1982). In spite of rigorous attempts by Poole et al. (1980) to do so, no dental or skeletal criteria are known that will consistently separate these two species. A third fossil species, M. titan, which is morphologically similar to the living species, has been recognized on the basis of significantly larger size (Owen, 1 874; Marshall, 1973a; Marshall & Corruccini, 1978). The deposits in Madura Cave have produced remains that are referable to Macropus titan and M. fuliginosus. All material that can be confidently assigned to M. titan on the basis of both mor- phology and size comes from Units 2-7, which have radiocarbon dates ranging from 15,600 ± 250 to 37,800 ± 3520 b.p. (Lundelius & Turnbull, 1973). All material that can be assigned to M. fuliginosus is from the present surface of the cave or from Unit 1. This unit, which is 2 ft (~ 60 cm) thick, has an eroded top surface. The top 1 ft (30 cm) of the unit has been radiocarbon-dated at 7470 ± 120 b.p. (Lundelius & Turnbull, 1973). Thus, all of the confidently referred M. titan material is of Pleistocene age and the M. fuliginosus material is of modern or Holocene age. The tentatively as- signed specimens of these taxa seem to follow this same pattern, but those which lack definitive fea- tures or are of intermediate size have been as- signed to Macropus sp. This stratigraphic and chronological distribu- tion of these two species in the Madura Cave de- posits is consistent with their chronological dis- tribution in other parts of Australia as reviewed by Marshall (1973a). The close morphological resemblance of M. gi- ganteus-M. fuliginosus to M. titan has been noted by many investigators (Owen, 1874; Lydekker, 1887; Tedford, 1967; Marshall, 1973a; Marshall & Corruccini, 1 978) and has been cited by the last two authors as an example of dwarfing in a single lineage at the end of the Pleistocene. The discovery that M. fuliginosus and M. giganteus are separate species raises questions about the details of the relationship of these three taxa. Furthermore, Marshall (1973a) points out that M. titan as it is currently recognized also may have been hetero- geneous. Regardless of the exact phylogenetic relation- ship between M. titan and M. fuliginosus, the re- cord at Madura Cave shows that a larger Macropus was replaced by a smaller one with similar mor- phology after 16,000 b.p., about the same time as in other parts of Australia and at the same time as the disappearance of the extinct mammals and the disharmonious assemblages. LUNDELIUS & TURNBULL: MADURA CAVE. PART VII. 47 aa 1 1 MANDIBLE i T 1 T J— © • i co 34 ■ CM 2 © O z PM 1560 X • < 33 - * -i 111 10 P. anak PM 1553 Key to symbols x PM # = FM specimens from Wellington Caves ~if= Madura Cave Specimens Bingara Specimens © = P. brehus -Marcus • = P. brehus -Bartholomai □ = P. roechus -Bartholomai 1 — f— i Darling Downs Specimens . © PM 1543 PM1557 M/1 10 11 12 13 LENGTH 14 15 48 FIELDIANA: GEOLOGY Id 1 1 -i 1 1 1 M/2 12 0 0 I , © P. brehus 11 PM 1560 P. roechus Ox 0 PM 1544 0 * 0 • 10 PM 1553 9 P. anak _i _i 1_ i . 12 13 14 15 16 LENGTH 17 18 13 - 12 10 19 13 ■ cr O cr 10 M/3 14 e * P. anak P. brehus PM1560 T 0 . 0 P. roechus PM 1553 15 16 19 20 -13 PM 1544 -12 ■10 21 17 18 LENGTH Fig. 1 6. Bivariate graphs showing the Madura Cave specimens of Protemnodon (stars) in comparison with certain Wellington Caves specimens and with samples of P. brehus and P. roechus from the literature. Upper left, mandibular proportions at M,_2 and M2_3; lower left, length x anterior width of M,; upper right, length x anterior width of M2; lower right, length x anterior width of M3. LUNDELIUS & TURNBULL: MADURA CAVE. PART VII. 49 B D Fig. 1 7. Petrogale sp. from Madura Cave compared with various modern species of the genus. Petrogale sp. from Madura Cave: A, PM 39006, left I1 in premaxillary fragment shown in lingual (left) and labial views; B, PM 39130, left I1 shown in lingual (left) and labial views; C, PM 39068, left M4 shown in lingual (left), crown, and labial views. Petrogale sp. from Wedge's Cave, Mimegara, Western Australia: D, PM 5749, right maxilla shown in ventral view. Petrogale pearsoni from Oenpelli, East Alligator River, Northern Territory: E, RCS A.348.51, right side of palate shown in ventral view. Petrogale brachyotis from Kimberly District, Western Australia: F, FM 1 19823, right side of palate shown in ventral view. Petrogale inornata from Rockhampton-Atherton area, Queensland: G, FM 64430, right side of palate shown in ventral view. 50 FIELDIANA: GEOLOGY Table 13. Numerical data on upper dentitions of a Recent sample of Macropus giganteus from New South Wales. Table 14. Numerical data on lower dentitions of a Recent sample of Macropus giganteus from New South Wales. N OR Mean N OR Mean P3 L 2 6.4-6.8 6.60 P, L 2 5.6-6.0 5.8 AW 2 3.2-3.3 3.25 AW 2 2.6 2.6 PW 2 4.5-4.6 4.55 PW 2 3.4-3.6 3.5 dP L 2 7.9-8.0 7.95 dP4 L 2 7.8-8.3 8.05 AW 2 6.1-6.5 6.30 AW 2 4.4-5.2 4.80 PW 2 6.5-6.9 6.70 PW 2 5.3-5.8 5.55 M1 L 5 6.9-10.2 9.02 M, L 4 7.8-9.7 9.04 AW 5 7.2-8.6 7.88 AW 4 5.6-6.1 5.98 PW 5 7.5-8.7 8.14 PW 4 6.1-7.4 6.80 M2 L 4 9.9-11.2 10.63 M2 L 5 10.0-11.7 10.86 AW 4 8.5-9.5 8.93 AW 5 6.4-8.0 7.20 PW 4 8.2-9.5 8.90 PW 5 6.5-8.1 7.24 M3 L 3 12.0-13.0 12.47 M, L 3 11.5-12.7 12.20 AW 3 9.4-10.2 9.77 AW 3 8.2-8.6 8.43 PW 3 8.8-10.0 9.33 PW 3 7.6-8.4 8.03 M4 L 3 12.8-14.1 13.30 M4 L 3 12.8-13.4 13.07 AW 3 9.9-10.8 10.23 AW 3 8.3-9.2 8.67 PW 3 8.9-10.3 9.47 PW 3 7.7-8.6 8.03 Diastema L 5 53.0-63.5 58.38 Diastema L 5 43.6-51.0 46.66 M1^ L 3 42.2^»6.2 43.53 MM L 3 36.2-43.0 40.13 Macropus fuliginosus Shaw and Nodder, 1790, part; (Desmarest, 1817) part Material Surface TMM 41106-23-24, pair of rami (same indi- vidual) with left and right I,, P3-dP4, M, in crypt (fig. 19D) Trench 2, Unit 1, top 1 ft PM 6246, right ramus with I, P4-M3, M4 in crypt (fig. 19C) Trench 4, Surface and Unit 1, top 6 inches TMM 41 106-510, right maxilla with M1^ (fig. 19 A) TMM 41 106-547, posterior one-third, left dP Trench 4, Unit 1 , top 1 ft (presumably level 2) PM 39128, left upper molar (fig. 19B) PM 39134, right upper molar TMM 41106-501, distal half, right metatarsal V(fig. 19F) Macropus sp. (Probably Macropus fuliginosus) Trench 1, Unit 1, top 1 ft TMM 41 106-499, right I2 or I1 Trench 4, Surface and Unit 1 , top 6 inches TMM 41 106-551, tooth fragment PM 7983, terminal phalanx from one of the syndactylous toes PM 7984, terminal phalanx from manus PM 391 10, anterior half, left dP4 or M, PM 391 13, anterior half, lower molar or dP4 Trench 4, Unit 1 , top 1 ft (presumably level 2) TMM 41 106-500, left metacarpal II (fig. 19E) TMM 41 106-502, first phalanx, manus TMM 41 106-503, terminal phalanx, pes TMM 41 106-504, terminal phalanx, pes PM 39129, left P PM 39138, right I3 PM 39143, terminal phalanx, manus Comparative Material Macropus sp. (Probably Macropus fuliginosus) Murraelellevan Cave (surface red clay), Western Australia PM 24334, left ramus with M,^,, alveolus of P4 Hasting's Cave (surface), Western Australia PM 50847, skull and mandible with P in crypt, erupted P\ dP4, M1 in crypt; I„ P3, dP4, erupting M, Macropus fuliginosus Jurien Bay, Western Australia TMM M-925 West coast north of Perth, Western Australia TMMM-927(fig. 211) LUNDELIUS & TURNBULL: MADURA CAVE. PART VII. 51 10 12 14 -+- ** 8 ^— H 15 12 i Sh 2 P/3 Length P/4 Length 16 T 18 Key to symbols M = Madura Cave Specimens • = M. fuliginosus © = M. titan T Q = Macropus titan from Queensland T LV = M. titan from Lake Victoria T C = M. titan from Camperdown GV = M giganteus from Victoria G NSW = M. giganteus from New South Wales F Kl = M. fuliginosus fuliginosus from Kangaroo Isl. FM = M fuliginosus melanops F NSW = M. fuliginosus from New South Wales Numbers indicate sample size, bar graphs show range and mean. Scale is shown top and bottom. 6 8 10 12 14 16 18 Fig. 18. Graphs showing Madura Cave specimens (M) of Macropus sp. in comparison with comparable teeth of M. titan (T), M. giganteus (G), and M. fuliginosus (F) from various localities. Left, comparison of lower teeth; right, comparison of upper teeth. Descriptions Upper Molars— The lophs of the upper molars (fig. 1 9 A-B) are convex anteriorly when unworn, straight when worn. The anterior cingulum ex- tends across the full width of the tooth, and is tied by a low ridge to the base of the paracone. Lin- gually, it extends upward to join the protocone close to its base. A low, straight forelink connects the anterior cingulum to the protoloph and divides the cingular basin. In the unworn state, the mid- link is lower than the lophs. Both lophs contribute 52 FIELDIANA: GEOLOGY 10 M TQ TLV TC GV GNSW FKI F.M. FNSW M TQ TLV TC GV GNSW FKI F.M. FNSW M TQ TLV TC GV GNSW FKI F.M. FNSW M TQ TLV TC GV GNSW FKI F.M. FNSW M TQ TLV TC GV GNSW FKI F.M. FNSW M TQ TLV TC GV GNSW FKI F.M. FNSW M TQ TLV TC GV GNSW FKI F.M. FNSW 12 14 16 18 Key to symbols M = Madura Cave Specimens • = M. tuliginosus © = M. titan T Q = Macropus titan from Queensland T LV = M. titan from Lake Victoria T C = M. titan from Camperdown GV = M. giganteus from Victoria G NSW = M. giganteus from New South Wales F Kl = M. tuliginosus tuliginosus from Kangaroo Isl. FM = M. tuliginosus melanops F NSW = M. tuliginosus from New South Wales Numbers indicate sample size, bar graphs show range and mean. Scale is shown top and bottom. M1/ Length sr ^i ■MO M2/ Length -M7 H4 —•28 ^13 ■•5 -M1 10 12 14 16 18 to the midlink, with their junction point being marked by a cleft. A labially directed spur extends from the metaloph portion at this point. The ac- cessory cuspule on the anterolabial side of the mid- link reported by Stirton (1963, p. 121) and Mar- shall (1973a) in Macropus titan is incipient in this specimen. The posterior cingulum is formed by a prominent ridge on the posterior face of the hy- pocone and a much smaller ridge on the base of the posterior face of the metacone. Comparison with the upper molars of modern Megaleia rufa shows that the Madura Cave spec- imen differs in having a well-developed forelink and a procingulum which is tied to the paracone. LUNDELIUS & TURNBULL: MADURA CAVE. PART VII. 53 54 FIELDIANA: GEOLOGY J Fig. 19. Macropus fuliginosus from Madura Cave. A, TMM 41 106-510, right maxilla with M1-4 shown in lateral (left), ventral, and medial views. B, PM 39128, left upper molar shown in crown view. C, PM 6246, right jaw ramus with I, P4-M3, and M4 in its crypt shown in lateral (left), dorsal, and medial views. D, TMM 41 106-24, right jaw ramus with I, P,-dP4, and M, in crypt shown in lateral (left), dorsal, and medial views. E, TMM 41 106-500, left metacarpal II, a tentatively referred specimen, shown in left dorsal and ventral views. F, TMM 41 106-501, distal half of right metatarsal V shown in left ventromedial, lateral, and dorsal views. LUNDELIUS & TURNBULL: MADURA CAVE. PART VII. 55 Table 15. Measurements of upper and lower den- titions of Macropus fuliginosus from Madura Cave. TMM TMM 41106- 41106- PM PM 510 547 39128 39134 dP4 L AW PW 5.3 M1 L AW PW 10.6 8.5b 8.8b M2 L AW PW 11.9 9.6 9.5 M3 L AW PW 13.6 10.5 10.1 M4 L AW 12.7 9.7 PW 8.2b Molar L 10.1 >9.8 AW 7.6 7.8 PW 7.8 PM TMM TMM 6246 41106-23 41106-24 P3 L 6.7 6.7 6.6 AW 3.1 3.1 3.1 PW 4.4 3.8 3.7 dP4 L 9.3 8.0 7.9 AW 5.5 4.9 4.5 PW 6.7 5.6 5.3 M, L 10.8 AW 6.8 PW 7.4 M2 L AW PW 11.5 7.5 6.9? M3 L AW 14.2 PW b = broken. In both of these characters it is similar to modern Macropus fuliginosus. The posterior cingular pit is somewhat larger than is seen in specimens of M. fuliginosus available for comparison. Lower Dentition— The P3 (fig. 19D) is elon- gate with three main cusps and an anterior cuspule. The anterior cusp is laterally compressed, with a low vertical ridge on each side and a sharper ridge connecting its apex with an anterior cuspule and the posterior labial cusp. The two posterior cusps are joined to form a transverse loph. The dP4 (fig. 19D) is molariform. The proto- lophid is distinctly narrower than the hypolophid. The midlink is like that of the molars, but the forelink is incomplete even though the procingul- um is large. A shallow vertical groove is present on the posterior face of the hypolophid, as in the molars. The P4 of PM 6246 (fig. 19C) is similar to the P3 in morphology. It is triangular with the large anterior cusp joined to an anterior cuspule by a ridge. It differs from the P3 in that the posterior ridge of the anterior cusp bifurcates, joining both the posterior cusps to form a posterior basin. The two posterior cusps join to form a posterior lophid. A small ridge extends anteriorly from the middle of the posterior transverse lophid into the poste- rior basin. The lower molars are bilophodont and brachy- hypsodont. The protolophid and hypolophid are slightly concave anteriorly when unworn and straight when worn. The procingulum projects for- ward and upward, and where it is joined by the forelink it is almost as high as the protolophid in an unworn tooth. The forelink arises from the protoconid and turns sharply linguad and then anteriad to join the pro- cingulum. The large midlink is made up of con- tributions from the protolophid and hypolophid. Their junction is marked by a cleft and some over- lap in an unworn tooth. The lophids are parallel, in contrast to Megaleia rufa in which the ento- conid is located posterior to the hypoconid and the protolophid and hypolophid are not parallel. A vertical groove is present on the posterior face of the hypolophid. Postcranial Skeleton— The distal one-third of one right fifth metatarsal (TMM 4 1 1 06-50 1 ; fig. 19F) is present. The distal part of the shaft is strongly curved laterally. The distal articular sur- face is asymmetrical with the lateral border pro- jecting outward and backward. A median ridge is present ventrally on the posterior part of the ar- ticular surface. The transverse diameter of the dis- tal end is 13.0 mm, which is within the size range of two modern specimens of M. fuliginosus (TMM M-927, 13.9 mm; TMM M-925, 12.9 mm). The corresponding measurements of two modern spec- imens of Megaleia rufa from Western Australia are 8.7 mm (TMM M-939) and 7.6 mm (TMM M-928). In addition, the distal articular surface in M. rufa lacks the median ridge. A terminal phalanx of digit IV of the pes (TMM 41 106-504) is tentatively referred to M. fuligino- sus. It has the high triangular shape of the artic- 56 FIELDIANA: GEOLOGY Table 16. Numerical data on upper dentitions of a Recent sample of Macropus fuliginosus from New South Wales. Table 1 7. Numerical data on lower dentitions of a Recent sample of Macropus fuliginosus from New South Wales. N OR Mean N OR Mean P3 L 15 5.6-7.2 6.42 P, L 15 5.5-6.8 6.05 AW 15 3.5-4.2 3.87 AW 15 2.6-3.3 2.95 PW 15 4.7-5.7 5.08 PW 15 3.3-4.3 3.87 dP4 L 18 7.1-9.0 8.16 dP< L 18 7.1-9.0 8.22 AW 18 6.2-7.2 6.59 AW 18 4.6-5.5 4.91 PW 18 6.6-7.7 6.94 PW 18 5.4-6.3 5.70 P4 L 10 6.5-8.3 7.23 P< L 10 5.2-6.7 6.06 AW 10 2.8-4.9 3.26 AW 10 2.1-3.6 2.67 PW 10 3.4-5.0 4.25 PW 10 2.6-3.9 3.26 M' L 31 8.5-11.0 9.68 M, L 29 8.4-10.8 9.73 AW 31 7.2-9.0 7.86 AW 30 5.4-7.6 6.20 PW 31 7.3-9.7 8.18 PW 28 5.6-7.3 6.46 M2 L 30 7.8-12.3 10.87 M2 L 31 9.5-13.1 11.20 AW 31 7.4-10.4 8.75 AW 32 6.8-8.4 7.38 PW 27 8.0-10.6 8.95 PW 28 6.2-8.2 7.21 M3 L 18 11.5-13.5 12.32 M3 L 17 10.2-13.1 12.08 AW 19 7.4-11.1 9.42 AW 17 7.5-9.1 8.13 PW 18 8.4-10.8 9.48 PW 17 7.0-8.7 7.70 M4 L 12 12.5-13.6 12.98 M4 L 11 8.2-14.5 12.22 AW 13 7.7-11.4 9.61 AW 11 7.4-9.7 8.21 PW 10 8.6-10.9 9.39 PW 9 7.0-8.5 7.56 Diastema L 31 51.0-68.3 57.20 Diastema L 36 31.6-53.9 44.81 M'^ L 13 34.3-51.5 42.15 MM L 6 32.^46.5 40.27 ulation facet and the short, broad protruding ven- tral base typical of macropodids. It is slightly smaller than modern specimens of M. fuliginosus and Megaleia rufa (table 1 8). Discussion All of the Macropus material that can be con- fidently assigned to M. fuliginosus comes from Unit 1 or from the present surface of the deposits, and thus is either modern or Holocene in age. M. fu- liginosus is a member of the modern fauna of this region and apparently has been present throughout most of the Holocene. Macropus titan Owen, 1838 Material Trench 2, Unit 2, Level 2% ft PM 6247, right ramus with posterior half, M3, M4 (fig. 20A) Trench 3, Unit 2, Level ? TMM 41 106-5057, labial side, left P3 Trench 3, Unit 2, Level 4 PM 39021, right dP4 Trench 4, Unit 2, Level 1 PM 38974, dP4 Trench 4, Unit 2, Level 2 PM 7993, right maxillary fragment with M1-3 (fig. 20B) PM 39070, left dP4 PM 39071, left P3 Trench 4, Units 4-5 PM 7994, right P4 PM 7995, left M4 PM 7998, left ramus with dP4-M3, P4 in crypt (fig. 20E) PM 39000, left ramus with M4 (fig. 20D) Trench 4, Unit 7, Level 4 PM 7992, right maxillary fragment with P3, an- terior half of dP4 (fig. 20C) Macropus sp. (Probably Macropus titan) Trench 2, Unit 2, Level 2% ft PM 26 1 64, tip, lower incisor PM 39102, phalanx Trench 3, Unit 2, Level ? (probably 1) and Level 1 LUNDELIUS & TURNBULL: MADURA CAVE. PART VII. 57 Table 1 8. Measurements of metatarsals and terminal phalanges of digits IV and V of the pes < af Recent and fossil Macropus. Meta- Terminal Terminal Metatarsal 4 tarsal phalanx, phalanx, 5 D4, articular D5, articular Ant.-post. diam. distal Proximal Distal facet Distal facet end Length width width width Width Height Width Height Macropus fuliginosus TMM M-925 23.7 171.7 28.7 23.7 TMM M-927 20.2 166.8 27.2 23.9 16.3 16.7 TMM 41 106-501 12.9 TMM 41 106-503 9.2 6.2 TMM 41 106-504 12.0 10.8 Macropus titan PM 39002 A >17.8 173.0 >27.2 >19.7 PM 39002 B 17.9 25.9 23.0 M. (Megaleia) rufa FM 98914 19.0 165.0 29.0 23.6 13.8 13.3 FM 44274 19.7 158.0 27.5 25.0 8.7 14.7 14.2 TMM M-928 16.5 160.0 24.6 21.2 7.6 14.0 12.7 5.9 6.2 TMM M-939 8.6 12.7 13.3 Macropus robustus FM 104674 15.5 125.0 22.7 20.7 FM 104813 22.8 FM 119818 15.7 112.0 21.3 20.0 10.7 FM 120574 16.5 134.3 24.5 22.1 11.1 TMM 41 106-105 25.6 TMM 41106-103, proximal end, left humerus (fig. 2 ID) PM 39060, left upper molar, probably M4 PM 39061, left upper molar, probably M4 PM 39062, right I2 or I1 PM 39067, left I1 PM 39083, left upper molar PM 39084, anterior one-third, lower molar PM 39085, molar fragment PM 39086, metaloph, right molar PM 39149, anterior half, left dP4 or M, Trench 3, Unit 2, Level 4 TMM 41 106-144, left upper incisor TMM 41 106-146, thoracic vertebrae, about T- 6-9 TMM 41 106-147, second phalanx, digit IV, pes Trench 3, Unit 3, Level ? (probably 1) TMM 41 106-45, tip, left lower incisor Trench 4, Unit 2, Level 1 PM 7982, fragment, upper incisor PM 7985, anterior half, right lower molar PM 7988, terminal phalanx, digit III, manus PM 7989, terminal phalanx PM 39124, second phalanx, digit V, pes PM 39145, molar fragment Trench 4, Unit 2, Level 2 PM 39055, left P PM 39056, right I1 PM 39073, broken left upper molar PM 39074, broken left lower molar PM 39075, broken molar PM 39076, molar fragment PM 39088, terminal phalanx, manus (or pos- sibly of pes of a smaller form) Trench 4, Unit 2, Level 3 PM 39093, anterior one-fourth, right lower mo- lar Trench 4, Units 4-5 PM 39001, portion of midshaft, right tibia PM 39002A-B, two right fourth metatarsals (fig. 21A-C) PM 39092, anterior half, left lower molar PM 39097, posterior half, right lower molar Descriptions Upper Dentition— The P3 is a triangular tooth with the labial blade interrupted by a shallow notch (fig. 20C). A posterior lingual cusp is joined to the 58 FIELDIANA: GEOLOGY ) w H> Fig. 20. Macropus titan from Madura Cave. A, PM 6247, right ramus fragment with part of M3 and M4 shown in lateral (left), dorsal, and medial views; B, PM 7993, right maxillary fragment with M'-3 shown in lateral (left), ventral, and medial views; C, PM 7992, right maxillary fragment with P3 and part of dP1 shown in ventral view; D, PM 39000, left ramus with M4 shown in dorsal and lateral views; E, PM 7998, left ramus with dP„-M3 and P4 in crypt shown in dorsal and medial views. LUNDELIUS & TURNBULL: MADURA CAVE. PART VII. 59 Table 19. Measurements of upper dentitions of Macropus titan from Madura Cave. TMM 41106- PM PM 5057 7994 7992 PM 38974 PM 39021 PM 39060 PM 39061 PM 7993 PM 39083 P1 L AW PW ' 1.1 7.3 3.5 4.9 dP4 L AW PW . . i >.5 8.0 4.1 5.4 9.6 >6.1 7.1 ::: P L AW PW 9 3 5 .7 .9 .4 ... ... M1 L AW PW 11.2 M2 L AW PW 12.9 11.6 11.8 M3 L AW PW 12.7 9.5 9.6 14.7 11.8 11.6 M4 L AW PW >11.1 >9.0 >7.9 Molar L AW PW 14.7 9.6 8.9 posterolabial cusp just anterior to its apex by a transverse ridge, and at its posterior end by a low ridge to form a small posterior basin. A small an- terior lingual cingular cusp is present, but is not joined to the posterior one. The dP4 is a molariform tooth with a prominent procingulum that is connected to the paracone by a sharp ridge. Marshall (1973a) states that this feature, which is not present in the molars, is typ- ical of the dP4 of Macropus titan of the Lake Vic- toria sample. The small interloph cuspule on the lingual side of the tooth observed by Marshall ( 1 973a) in a Lake Victoria specimen is not present on the Madura Cave specimens. The upper molars (fig. 20B) are virtually iden- tical to those of Macropus fuliginosus. The lophs are high and convex anteriorly when unworn, with sides that are straight but converge toward the crown. The procingulum extends across the com- plete breadth of the tooth. It is joined to the pro- toloph by the forelink, which is located slightly lingual to the midline of the tooth, but it is sep- arated from the paracone and protocone by clefts. The midlink is large, with its major part from the protocone and a smaller contribution from the middle part of the metaloph. The midlink does not bow labially, and there are no accessory cus- pules associated with the midlink or the postero- labial face of the protoloph, both of which were reported by Marshall (1973a) for some specimens of M. titan from Lake Victoria, the Camperdown district of Victoria, and the Darling Downs of Queensland. The bases of the protocone and hy- pocone of PM 7993 have small cuspules in the lingual side of the median valley. The postcingu- lum is large and is formed primarily by a ridge from the hypocone. It is joined to the base of the metacone to form a posterior basin. There is no vertical groove on its posterior face. Lower Dentition— The P4 is a compressed bladelike tooth with two major cusps and a lower posterolingual cusp joined to the main posterior cusp by a ridge (fig. 20E). This agrees with the description of M. titan given by Marshall (1973a). The lower molars are bilophodont, brachyhyp- sodont teeth (fig. 20A,E). The lophids are slightly concave anteriorly when unworn and straight when worn. The procingulum projects forward and up- ward from the base of the tooth. The forelink arises from the protocone, turns sharply linguad, and 60 FIELDIANA: GEOLOGY Table 19. Extended. PM 39084 PM 39086 PM 39092 PM 39097 PM 39071 then turns anteriad to join the procingulum. In an unworn tooth, the central part of the procingulum where it joins the forelink is as high as the pro- tolophid. The protolophid and hypolophid contribute to the large midlink, and their junction is marked by a cleft. The protolophid and hypolophid are par- allel, in contrast to Megaleia rufa, in which the entoconid is located posterior to the hypoconid and the hypolophid and protolophid are not par- allel. A vertical groove is located on the posterior surface of the hypolophid. A cuspule is present in the lingual side of the valley in PM 7998 (fig. 20E), as is reported for Macropus titan by Marshall (1973a). In PM 6247 (fig. 20A) and PM 39000 (fig. 20D), the median valley and the lingual part of the procingular valley contain cement. Postcranial Skeleton— The proximal part of a left humerus (TMM 41 106-103; fig. 2 IB) from Madura Cave is the same size as the comparable portion of the humeri of modern Macropus fuli- ginosus (TMM M-925, M-927) and Megaleia rufa (98914, TMM M-928). The morphology differs only in the shape of the head; the articular surface of the head of the Madura Cave specimen is oval with an extension toward the greater tuberosity, while the articular surfaces of the modern speci- mens are almost circular. The fourth metatarsal is represented by a com- plete specimen (PM 39002 A; fig. 21 A) and the proximal half and distal articular surface of another (PM 39002B; fig. 21B-C); both are from the left foot. PM 39002A lacks the posterior part of the proximal articular surface. The anterior part of the articular surface is smoothly concave. The prox- imal articular surface of PM 39002B is also smooth, but is more deeply concave medially. Both spec- imens have a prominent ridge on the medial side of the anterior surface of the shaft that extends from 2 cm below the proximal end to the middle of the shaft. This feature is present in two modern specimens of Macropus fuliginosus (TMM M-925, M-927; fig. 211) but not on our specimens of Me- galeia rufa (98914, TMM M-928, M-939; fig. 2 1 J), and is small or absent on specimens of modern Macropus robustus (FM 1 04674, 1 048 1 3, 1 1 98 1 8 [fig. 2 IK], 120574). Both of the Madura Cave metatarsals lack the prominent rugose bulge on the proximal part of the posterior face that is seen in Megaleia rufa. Macropus fuliginosus also lacks this bulge. The condition in Macropus robustus is intermediate. The Madura Cave specimens are essentially the same size as modern specimens of Macropus fu- liginosus (including those from Western Australia) and Megaleia rufa, and are 25% larger than mod- ern specimens of Macropus robustus (table 18). Discussion All previous studies of Pleistocene Macropus have concluded that Macropus titan is morpho- logically indistinguishable from M. giganteus and M. fuliginosus and differs from them only in its 25°/o-30% larger size (Owen, 1874;Lydekker, 1887; Tedford, 1967; Marshall, 1973a; Marcus, 1976; Bartholomai, 1975; Marshall ACorruccini, 1978). All of the Macropus material from Unit 1 , which is Holocene in age, can be assigned to M. fuligi- nosus on the basis of both size and morphology. The Macropus material from Units 2-7, which is of Pleistocene age, presents a somewhat confused picture. With the exception of a dP4 (PM 7998), all the lower teeth of Macropus from the lower units whose position in the jaw can be determined fall within the size range of samples of M. titan from Lake Victoria, the Eastern Darling Downs, LUNDELIUS & TURNBULL: MADURA CAVE. PART VII. 61 Table 20. Measurements of lower dentitions of Macropus titan from Madura Cave. PM 7998 PM PM 39070 6247 PM 7995 PM 39092 PM 39084 PM 39074 PM 7985 PM 39075 PM PM 39093 39097 dP4 L AW PW 9.5 1 >5.8 7.4 5.3 5.7 P4 L AW PW 6.9 2.3 3.7 M, L AW PW >10.0 M2 L AW PW 14.0 9.0 8.6 M3 L AW PW 16.2 M4 L AW PW 16.9 9.7 9.5 18.4 10.5 9.2 Molar L AW PW >1 >9 9.4 and the Camperdown area of Victoria reported by Marshall (1973a). The dP4 is slightly below the size range reported by Marshall (1973a), but is larger than the dP4s of several samples of M. gi- ganteus and M. fuliginosus from various parts of Australia (fig. 1 8 A). The other dental dimensions are closer to those of M. titan than to those of M. giganteus and M. fuliginosus. With the exception of four P3s (TMM 41 106-5057; PM 38974, 39071, 7992) and an M1 (PM 50847), those upper teeth from Units 2-7 whose positions in the jaw can be determined fall within, but usually at the lower ends of, the observed size ranges of samples of M. titan from Lake Victoria, the Eastern Darling Downs, and the Camperdown area of Victoria re- ported by Marshall (1973a) and Bartholomai (1975) (fig. 18B). The P3s are shorter than those of M. titan and fall in the upper part of the range of several samples of M. giganteus and M. fuli- ginosus (fig. 1 8B). The postcranial material from Units 2-7 is the same size as that of modern spec- imens of M. fuliginosus. Marshall (1973a) pointed out that the size of Macropus titan increases from Queensland to Vic- toria parallel to the size change in M. giganteus and M. fuliginosus. This suggested the possibility of a Bergmann cline, but Marshall believed the data to be inadequate to demonstrate this. The somewhat small size of some of the Madura Cave specimens might cause one to speculate about a possible east-west cline for these taxa across the Nullarbor Plain, which would not be a Bergmann response. Such speculation is premature in any case, for there is inadequate data from the Madura Cave M. titan and M. fuliginosus samples to show any size trends. In addition, the M. titan sample from Units 2-7 in Madura Cave spans a significant period of time and cannot be treated as a single coherent sample; it is also of inadequate size to show change through time. Macropus robustus Gould, 1840 Material Trench 3, Unit 2, Level ? (probably 1) PM 39058, left P (fig. 21H, right) PM 39057, left I2 (fig. 21H, left) TMM 41106-105, proximal left metatarsal IV (fig. 2 IE) Trench 4, Units 4-5 PM 799 1 , proximal one-fourth, right ulna, lack- ing epiphysis of olecranon process and rim of articular facets (fig. 2 1 F) 62 FIELDIANA: GEOLOGY Comparative Material Discussion Macropus robustus cervinus Cape Range, Western Australia FM 104670 FM 104671 FM 104674 FM 104676 FM 104687 FM 104690 FM 104692 FM 104694 FM 104701 Macropus robustus National Zoological Park FM 104813 Macropus robustus antilopinus Kimberly District, Western Australia FM 119818 (fig. 21G,K) FM 120574 Descriptions Upper DENTmoN-The left I3 (PM 39058; fig. 2 1H) is a long tooth with one groove separating a stout, narrow, anterior lobe from a posteriorly flar- ing but thinner posterior lobe. This morphology is similar to that of Macropus robustus and Me- galeia rufa, but in the latter the tooth is much smaller. The P of Macropus fuliginosus differs from that of the Madura Cave specimen in having two grooves rather than only one. The I2 (fig. 21H) resembles that of modern M. robustus both in size and in having no groove on its outer face. Postcranial Skeleton— The fourth metatarsal fragment (fig. 2 1 D) appears to be referable to Mac- ropus robustus on the basis of size (table 1 8) and morphology. Seen from the front, the articular sur- face is smoothly concave and the concavity is shal- low as in modern M. robustus. The ulna (fig. 2 1 F) is slightly smaller and more delicate than any of our modern comparative spec- imens (fig. 2 1 G). The frayed edges of the articu- lation facets make detailed comparison difficult. The olecranon process is small but robust, and the facet for articulation with the radius is very small. The specimen is in the same general size range as the modern specimens of M. robustus, and the olecranon process is about 25% shorter than that of the ulna of Macropus fuliginosus (fig. 21G) and Megaleia rufa. Macropus robustus is widely distributed in Aus- tralia where suitable habitats in the form of rocky outcrops occur (Frith & Calaby, 1969). The flat topography of the Nullarbor Plain does not appear to provide suitable habitats for M. robustus. How- ever, the scarp that separates the Hampton Ta- bleland from the Roe Plain may have provided small areas of rocky outcrops suitable for M. ro- bustus, as well as for Petrogale. Incertae Sedis among the Large Macropodids Specimens which we cannot identify with cer- tainty, but which probably are referable to one or another of the large macropods are included. Trench 3, Unit 2, Level 1 TMM 41106-109-111, one subterminal phal- ange and two terminal phalanges, probably from the manus of a large species of Mac- ropus TMM 41 106-186, root and labial side, I1 or P, from a large species of Macropus PM 39 1 56-39 157, two terminal phalanges, pos- sibly from the manus of a Macropus species (had been associated with TMM 41106- 103-111) Trench 3, Unit 3, Level ?, probably 1 PM 39082, terminal phalange Trench 4, Unit 1 , Level 1 PM 39108, molar tooth fragment from between lophs PM 39131, macropodid right I1 or I2 PM 39140, macropodid partial right I2 or I3 Trench 4, Unit 2, Level 1 TMM 41106-300-303, four macropodine toe bones PM 7990, partial vertebra, either a posterior thoracic or anterior lumbar, consisting of centrum and neural arch Trench 4, Unit 2, Level 2 PM 39069, right M4, similar to that of Thylogale and Macropus irma PM 39087, terminal phalange Trench 4, Unit 2, Level 3 PM 7996, distal seven-eighths, large second phalange, digit IV, pes, probably from a large species of Macropus Trench 4, Unit 2, Level 4 PM 38999, proximal epiphysis, tibia, may be from a large Protemnodon or Macropus, or possibly a small diprotodont Trench 4, Units 4-5 PM 39098, partial right upper molar LUNDELIUS & TURNBULL: MADURA CAVE. PART VII. 63 64 FIELDIANA: GEOLOGY Assessment of the Marsupial Segment of the Fauna Fossil remains of marsupials from Madura Cave provide a good picture of Australian marsupial faunas during the late Pleistocene and the early Holocene, and the faunal changes that took place through that span of time. The late Pleistocene fauna shows a higher taxo- nomic diversity than either that of the early Ho- locene or the modern historic fauna of the Nul- larbor Plain. This diversity is characteristic of Pleistocene faunas in most parts of the world. A total of 37 Pleistocene and 24 early Holocene and modern historic taxa is recorded from the Nul- larbor Plain (table 21). Of the 37 Pleistocene taxa, all but one (Megaleia rufa) were recovered from Madura Cave. It is more difficult to determine how many species were living on the Nullarbor Plain in historic times because of inadequate surveys of the mammalian fauna prior to the habitat changes brought about by domestic and other introduced animals. Brooker (1 977) observed only four species of marsupials during his work in this area, but listed 10 species known to have occurred there prior to 1 940. Examination of distribution maps by Shortridge (1909) and Marlow (1962) suggests that there may have been as many as 1 5 species of marsupials on the Nullarbor Plain prior to the beginning of European influence. The change in diversity between the Pleistocene and historic times is the result not only of extinction of some species (4 completely extinct plus 2 extinct on the main- land), but also of the extirpation of about 1 6 extant species. In addition to the increased diversity, the Pleis- tocene marsupial assemblage from Madura Cave contains a number of species that today are allo- patric and seemingly ecologically incompatible. These types of associations, first recognized by Hibbard (1960), and termed "disharmonious" by Semken (1974), are characteristic of late Pleisto- cene faunas wherever they have been adequately studied. The presence of these disharmonious as- sociations in Australian Pleistocene faunas has been summarized by Lundelius (1983). Seventy-three pairs of species that are now allopatric were found in Units 2-7 of Madura Cave. Examples of such pairs of formerly sympatric, but now allopatric, species are: Phascolarctos cinereus and Dasycercus cristicauda, P. c. and Dasyuroides byrnei, P. c. and Caloprymnus campestris; Antechinus flavipes and Dasycercus cristicauda, A. f. and Dasyuroides byr- nei. Forty pairs of presently allopatric species were found in Unit 1. It should be pointed out that a substantial number of the disharmonious pairs in- volve Sarcophilus harrisi and Thylacinus cyno- cephalus, whose absence from the historical fauna of the mainland may have more to do with the introduction of the dingo than with the general post-Pleistocene climatic change (Archer, 1974). If the disharmonious pairs involving these two taxa are subtracted the numbers fall to 33 for the older units and 22 for Unit 1 . There is a major change in the marsupial fauna from Unit 2 (dated at 15,600 b.p. at its top) to Unit 1 (dated at 7470 b.p. at its top, but separated from Unit 2 by an erosion surface). Milham and Thompson (1976) give dates of present to ~7000 b.p. for the same unit in the South Tunnel. The faunal changes include the disappearance of most of the extinct taxa, including Sthenurus, Protem- nodon, Macropus titan, possibly Thylacoleo, and the two extant taxa Antechinus flavipes and Phas- colarctos cinereus. Milham and Thompson (1976) have reported the presence of Protemnodon sp., Sthenurus sp., and Phascolarctos sp. from the upper unit in the South Tunnel. The results of nitrogen and fluorine analyses on dentine of Protemnodon teeth from this unit reported by them suggest that the material of this taxon and probably that of Sthenurus and Phascolarctos is derived from an older deposit. If this is so, the radiocarbon dates of 3450 to 7880 Opposite Page: Fig. 21. Various specimens of Macropus, some from Madura Cave, some from the modern species from other localities: cf. Macropus titan from Madura Cave: A, PM 39002A, right metatarsal IV shown in proximal and dorsal (anterior) views; B, PM 39002B, proximal half of right metatarsal IV shown in proximal view; C, PM 39002B, distal end of right metatarsal IV shown in ventral (posterior) view (fragments in B and C are associated parts of the same bone); D, TMM 41 106-103, proximal end of left humerus shown in proximal view. Macropus robustus from Madura Cave: E, TMM 41 106-105, portion of metatarsal IV shown in proximal and dorsal (anterior) views; F, PM 7991, proximal end of right ulna lacking olecranon epiphysis shown in medial (left), anterior (dorsal), and lateral views; H, PM 39057, left I2 shown in labial view and PM 39058, left P shown in labial view. Macropus robustus antilopinus from Kimberly District, Western Australia: G, FM 1 198 18, right ulna shown in medial and lateral views; K, FM 119818, left metatarsal IV shown in anterior view. Modern Macropus fuliginosus: I, TMM M-927, right metatarsal IV shown in anterior view. Modern M. (Megaleia) rufa: J, TMM M-939, right metatarsal IV shown in anterior view. LUNDELIUS & TURNBULL: MADURA CAVE. PART VII. 65 SI 2 3 15 ^g 1 A 03 all 5 ** «i as *■ r o, a -C *h 6 M «> e «- _. ™ ■p B *a s o. -. E « 2© li *a e L- - _ u — * o 2 PQ X X X X X X x xXX X xX X XX X XXXXX Xx|XX XXx XxxX x : : x xXX XXxXxj::xXX:Xx XXxX XxxxX XXXX x XXx XxxXx xx;xxXXxX ::::: xXXX + + + + » + • + + +•&•*»+ o- o- + + + + + + -Si •t° a S ^j § ?-S £.3 £ ,« F ^ 2 <5 a t- S « §■■£ a - 3 5 a S Q 5 -c' w> -«:' & ^ -s; >2 3 Q •5 a •2 % ■si s £ v- & a 3 § a 5 3 a ^ 3 ^ 5o-3 ft, 0 oq eq 2 a c a R FIELDIANA: GEOLOGY 1 c a ^ S s ^> a E o ■a H fl § *• 6 ._ o •c 2 ii: o • LUNDELIUS & TURNBULL: MADURA CAVE. PART VII. 67 b.p. from these deposits do not apply to these taxa. Thylacoleo is present in Unit 1 . If this specimen was in primary context, then it is the youngest known specimen of this taxon. However, as dis- cussed in Part III (Lundelius & Turnbull, 1978, p. 91), the matrix adhering to the specimen suggests that it may have been eroded from older deposits near the front of the cave and redeposited farther in. Two other groups of taxa that no longer occur on the Nullarbor Plain are found in Unit 1 . One group includes Parantechinus apicalis, Sminthop- sis murina, Phascogale calura, Phascogale tapoa- tafa, and Potorous platyops, which are found today in association with Antechinus flavipes and Phas- colarctos cinereus in areas of eastern and/or south- western Australia with climates that are more hu- mid than the present-day climate in the region of Madura Cave. The other group, which consists of Dasyuroides byrnei, Myrmecobius fasciatus, and Caloprymnus campestris, is found today in areas approximately as arid as that of Madura Cave. The presence of the first group of taxa indicates a climate more humid than the present one. This agrees with information from other parts of Aus- tralia. Although the general pattern of climatic change through this period is a shift to drier con- ditions, the details are not clear. The association in the Pleistocene and early Holocene faunas of members of these two groups of species with seem- ingly disparate environmental requirements forms the disharmonious associations mentioned above. Disharmonious faunas in North America have been interpreted as indicating more equable climates during the Pleistocene (Hibbard, 1 960). The same probably is true for Australia (Lundelius, 1983). The lower number of disharmonious associations in Unit 1 (~40 vs. ~ 73 in the older units) indicates a change to less equable conditions after approx- imately 15,000 b.p. Planigale sp. indet. also disappears at the end of the Pleistocene sequence in Madura Cave. How- ever, we cannot categorize this species as readily as the groups discussed above, for too little is known about its habitat requirements to permit a gener- alization about its environmental implications. Archer (1976) lists two other occurrences of Plan- igale sp. indet. The closest to Madura Cave is a modern specimen from the eastern edge of the Nullarbor Plain, in much the same sort of arid environment. The other occurrence is from the Hammersly Range in the Pilbara, but habitat data are not given. Planigale maculata appears to be limited to wetter climates (Taylor et al., 1982). Andrews and Settle (1982) report that P. gilesi is restricted to riverine floodplains and overflows. Denny (1982) states that most planigales are found close to water, but that P. tenuirostris can be found in drier habitats. Read (1982) speculated that the drifting home ranges of P. tenuirostris are an ad- aptation to an arid environment. There are few changes in morphology or size in those taxa that persist into the Holocene or Recent. One change which did occur is the small increase in size of Dasycercus cristicauda from Unit 2 to Unit 1 . We suggested (Lundelius & Turnbull, 1 978, p. 64) that this represents an instance of character release related to the disappearance of the mor- phologically similar and closely related Dasyu- roides byrnei after about 16,000 b.p., which may have allowed Dasycercus cristicauda to broaden its niche. The absence ofMegaleia rufa from the Madura Cave deposits is puzzling, although it is not com- mon anywhere as a fossil. The species is present today on the Nullarbor Plain. It is found in the late Pleistocene fauna from Lake Menindee (Ted- ford, 1967), where it is associated with many of the same species that occur in the Madura Cave fauna. It is not recorded as having been found in the late Pleistocene fauna of the Lake Victoria region, which has produced many of the same species as Lake Menindee and Madura Cave (Mar- shall, 1973b). It is present in the late Pleistocene fauna from Unit III of Seton Rock Shelter, Kan- garoo Island, South Australia (Hope et al., 1977). Its absence from the Madura Cave fauna may be the result of a sampling accident, since the larger animals are poorly represented. This poor repre- sentation of the larger taxa suggests that most of the fossils were accumulated by owls, which could not handle the larger forms. A comparably puz- zling situation is the absence of the monotreme Tachyglossus aculeata. Macropus eugenii, M. irma, and Potorous tri- dactylus are also absent from the Pleistocene de- posits. These taxa, along with Vombatus, Phas- colarctos, and Potorous platyops, if their Pleistocene records in southwestern and southeastern Austra- lia are considered (Merrilees, 1968b), have dis- junct distributions on either side of the Nullarbor Plain, and at some time in the past these popu- lations should have been connected across this area. This expectation has been realized for Phas- colarctos cinereus (Lundelius & Turnbull, 1982), whose modern distribution is eastern and south- eastern Australia, but which is known from Pleis- tocene deposits in southwestern Australia (Mer- 68 FIELDIANA: GEOLOGY rilees, 1 968b; Balme et al., 1 978), and for Potorous platyops, known as a living animal in western Aus- tralia and as a fossil from eastern Australia (Wake- field, 1964). There are several possible explana- tions for the absence of these three taxa: (1) They may have been connected across an area north of the Roe Plain, and may not have been present in the vicinity of Madura Cave; (2) the connection or dispersal may have taken place at some time earlier than that represented by the Madura Cave deposits; or (3) their absence may be a sampling accident. Three species of marsupials, Parantechinus ap- icalis, Tarsipes spencerae, and Setonyx brachyu- rus, are restricted to southwestern Australia. Only one of these, Parantechinus apicalis, is known from Pleistocene or Holocene faunas of the Nullarbor Plain. If the absence of these taxa from the late Pleistocene fauna of that area is not a sampling accident, then the marsupial fauna from the Pleis- tocene deposits indicates an environment that was mesic and more equable than the present envi- ronment, but which lacked the dense swampy areas preferred by Setonyx brachyurus, the thickets pre- ferred by Macropus eugenii (Ride, 1970), and the forests preferred by Vombatus. A savannah or woodland is indicated. Acknowledgments We gratefully acknowledge the encouragement, support, and collecting assistance given by Judith Lundelius and Priscilla Turnbull; without their help the work would not have been accomplished. We also thank Marlene H. Werner for the fine draw- ings of both the Madura Cave fossils and the suite of modern comparative materials; and Zbigniew T. Jastrzebski, Senior Scientific Illustrator at Field Museum, who montaged the drawings, prepared the graphs, and gave advice. The original fieldwork was supported in part by a grant from the National Science Foundation (GB975), and others from the Geological Foundation, University of Texas and from Field Museum. The Foundation further sup- ported Lundelius on a study trip to Field Museum. Literature Cited Andrews, D. L., and G. L. Settle. 1 982. Observations on the behavior of species of Planigale (Dasyuridae, Marsupialia) with particular reference to the narrow- nosed planigale (Planigale tenuirostris), pp. 31 1-324. In Archer, M., ed., Carnivorous Marsupials, vol. 1, chap. 3 1 . Royal Zoological Society of New South Wales, Sydney, N.S.W., Australia. Archer, M. 1972. Nullarbor 1970. The Western Cav- er, 12(1): 21-24. 1 974. New information about the Quaternary distribution of the thylacine (Marsupialia, Thylacini- dae) in Australia. J. Roy. Soc. West. Aust., 57(2): 43- 50. . 1976. Revision of the marsupial genus Plan- igale Troughton (Dasyuridae). Mem. Queensland Mus., 17(3): 341-365. . 1978. The nature of the molar-premolar boundary in marsupials and a reinterpretation of the homology of marsupial cheekteeth. Mem. Queensland Mus., 18(2): 157-164. Balme, J., D. Merrilees, and J. K. Porter. 1 978. Late Quaternary mammal remains, spanning about 30,000 years from excavations in Devil's Lair, Western Aus- tralia. J. Roy. Soc. West. Aust., 61(2): 33-65. Bartholomai, A. 1963. Revision of the extinct ma- cropodid genus Sthenurus Owen in Queensland. Mem. Queensland Mus., 14(3): 51-76. . 1973. The genus Protemnodon Owen (Mar- supialia, Macropodidae) in the upper Cainozoic de- posits of Queensland. Mem. Queensland Mus., 16(3): 309-363. . 1975. The genus Macropus Shaw (Marsupialia: Macropodidae) in the upper Cainozoic deposits of Queensland. Mem. Queensland Mus., 17(2): 195-235. Baynes, A., D. Merrilees, and J. K. Porter. 1975. Mammal remains from the upper levels of a late Pleis- tocene deposit in Devil's Lair, Western Australia. J. Roy. Soc. West. Aust., 58: 97-126. Brooker, M. G. 1977. Some notes on the mammalian fauna of the western Nullarbor Plain, Western Aus- tralia. West. Aust. Nat., 14(1): 2-15. Calaby, J. H. 1966. Mammals of the upper Richmond and Clarence rivers, New South Wales. C.S.I. R.O. Div. Wildl. Res. Pap., 10: 3-55. . 1971. The current status of Australian Macro- podidae. Aust. Zoologist, 16(1): 17-29. Denny, M. J. S. 1982. Review of Planigale (Dasyu- ridae, Marsupialia) ecology, pp. 131-138. In Archer, M., ed., Carnivorous Marsupials, vol. 1, chap. 13. Royal Zoological Society of New South Wales, Sydney, N.S.W., Australia. Desmarest, A. G. 1817. Kanguroo, p. 35, plate 22. In Nouveau dictionaire d'histoire naturelle, n. ed.. 17. Chez Deterville, Paris. DeVis, C. W. 1895. A review of the fossil jaws of the Macropodidae in the Queensland Museum. Proc. Linn. Soc. N.S.W., 2nd series, 10: 75-133. Dortch, C. E., and D. Merrilees. 1972. A salvage excavation in Devil's Lair, Western Australia. J. Roy. Soc. West. Aust., 54: 103-1 13. Finlayson, H. H. 1936. On mammals from the Lake Eyre Basin. Trans. Roy. Soc. S. Aust., 60: 157-161. Flower, W. H. 1884. Catalogue of the Specimens Il- lustrating the Osteology and Dentition of Vertebrated Animals, Living and Extinct, Contained in the Mu- seum of the Royal College of Surgeons of England. Part II, Class Mammalia Other Than Man. Taylor and Francis, London, 779 pp. LUNDELIUS & TURNBULL: MADURA CAVE. PART VII. 69 Frith, H. J., and J. H. Calaby. 1969. Kangaroos. F. W. Cheshire, Melbourne, Victoria, XI + 209 pp. Glauert, L. 1912. Fossil marsupial remains from Bal- ladonia in the Eucla Division. Rec. West. Aust. Mus. Art Gallery, 1: 47-65. . 1926. A list of Western Australian fossils. Geol. Surv. West. Aust., Bull. No. 88: 36-71. . 1933. The distribution of the marsupials in Western Australia. Part IV. Fieldiana: Geol., n.s., 6: 1-72. 1982. The mammalian fauna of Madura Cave, Western Australia. J. Roy. Soc. West. Aust., 19: 17- 32. Gould, J. 1840. On five new species of kangaroos. Proc. Zool. Soc. London, 1840: 92-94. . 1 844. Exhibition and character of a number of animals transmitted from Australia by Mr. Gilbert. Proc. Zool. Soc. London, 1844: 103-107. Gray, J. E. 1837. Descriptions of some new or little known Mammalia, principally in the British Museum collection. Mag. Nat. Hist. (Charlesworth), n.s., 1: 577- 587. . 1841. Contributions toward the geographical distribution of the Mammalia of Australia with notes on some recently discovered species, pp. 397-414. In Grey, G., ed., Journals of Two Expeditions of Dis- covery in Northwest and Western Australia, vol. 2, appendix C. T. and W. Boone, London. Hibbard, C. W. 1960. An interpretation of Pliocene and Pleistocene climates in North America. Annu. Rep. Mich. Acad. Sci., Arts & Lett., 62: 5-30. Hope, J. H., R. J. Lampert, E. Edmondson, M. J. Smith, and G. F. van Tets. 1 977. The late Pleistocene fau- nal remains from Seton Rock Shelter, Kangaroo Is- land, South Australia. J. Biogeog., 4: 363-385. Jones, F. W. 1923-1925. The bandicoots and the her- bivorous marsupials (the syndactylous Didelphia), pp. 133-270. In The Mammals of South Australia, vol. 2. Handbooks of the Flora and Fauna of South Aus- tralia. R. E. E. Rogers, Adelaide, S.A., Australia. Kirsch, J. A. W., and W. E. Poole. 1967. Serological evidence for speciation in the grey kangaroo Macropus giganteus Shaw (Marsupialia: Macropodidae). Nature (Lond.), 215: 1097-1098. . 1972. Taxonomy and distribution of the grey kangaroos, Macropus giganteus Shaw and Macropus fuliginosus (Desmarest), and their subspecies (Mar- supialia: Macropodidae). Aust. J. Zool., 20: 315-339. Lundelius, E. L., Jr. 1963. Vertebrate remains from the Nullarbor Caves, Western Australia. J. Roy. Soc. West. Aust., 46: 75-80. — . 1983. Climatic implications of late Pleistocene and Holocene faunal associations in Australia. Al- cheringa, 7: 125-149. Lundelius, E. L., Jr., and W. D. Turnbull. 1973. The mammalian fauna of Madura Cave, Western Aus- tralia. Part I. Fieldiana: Geol., 31(1): 1-35. . 1975. The mammalian fauna of Madura Cave, Western Australia. Part II. Fieldiana: Geol., 31(2): 37- 117. . 1978. The mammalian fauna of Madura Cave, Western Australia. Part V. Fieldiana: Geol., n.s., 11: 1-32. 1 984. The mammalian fauna of Madura Cave, Western Australia. Part VI. Fieldiana: Geol., n.s., 14: 1-63. Lydekker, R. 1 887. Catalogue of the Fossil Mammalia in the British Museum, Part V. London, xxv + 345 pp. Marcus, L. F. 1 962. A new species of Sthenurus (Mar- supialia, Macropodidae) from the Pleistocene of New South Wales. Rec. Aust. Mus., 25: 299-304. . 1976. The Bingara fauna: A Pleistocene ver- tebrate fauna from Murchison County, New South Wales, Australia. Univ. Calif. Publ. Geol. Sci., 114: 1-145. Marlow, B. 1962. Marsupials of Australia. Jacaranda Press, Brisbane, Queensland, Australia, 141 pp. Marshall, L. G. 1973a. The Lake Victoria local fauna: A late Pleistocene-Holocene fauna from Lake Victo- ria, southwestern New South Wales, Australia. M.A. thesis, Monash University, Clayton, Victoria, Austra- lia. . 1973b. Fossil vertebrate faunas from the Lake Victoria region, S.W. New South Wales, Australia. Mem. Natl. Mus. Victoria, 34: 151-172. Marshall, L. G., and R. S. Corruccini. 1978. Vari- ability, evolutionary rates, and allometry in dwarfing lineages. Paleobiology, 4(2): 101-119. Merrilees, D. 1965. Two species of the extinct genus Sthenurus Owen (Marsupialia, Macropodidae) from south-eastern Australia, including Sthenurus gilli sp. nov. J. Roy. Soc. West. Aust., 48(1): 22-32. . 1968a. Southwestern Australian occurrences of Sthenurus (Marsupialia, Macropodidae), including Sthenurus brownei, sp. nov. J. Roy. Soc. West. Aust., 50(3): 65-79. -. 1968b. Man the destroyer: Late Quaternary changes in the Australian marsupial fauna. J. Roy. Soc. West. Aust., 51(1): 1-24. 1979. Prehistoric rock wallabies (Marsupialia, Western Australia. Part III. Fieldiana: Geol., 38: 1- 120. . 1981. The mammalian fauna of Madura Cave, Macropodidae, Petrogale) in the far southwest of Western Australia. J. Roy. Soc. West. Aust., 61(3): 73- 96. Milham, P., and P. Thompson. 1976. Relative antiq- uity of human occupation and extinct fauna at Madura Cave, southeastern Western Australia. Mankind, 10(3): 175-180. Owen, R. 1838. Letter, pp. 365-369. In Mitchell, T. L., ed., Three Expeditions into the Interior of Eastern Australia, vol. 1 , 2nd ed. T. and W. Boone, London. . 1 873. On the fossil mammals of Australia. Part VIII. Macropodidae: Genera: Macropus, Osphranter, Phascolagus, Sthenurus, Protemnodon [abstract]. Proc. Roy. Soc. London, 21: 128. . 1874. On the fossil mammals of Australia. Part VIII. Macropodidae: Genera: Macropus, Osphranter, Phascolagus, Sthenurus, Protemnodon. Phil. Trans. Roy. Soc. London, 164: 245-287, 8 pis. 70 FIELDIANA: GEOLOGY . 1 877. Researches on the Fossil Remains of the Extinct Mammals of Australia with a Notice of the Extinct Marsupials of England, vols. 1 and 2. J. Erx- leben, London, xv + 522 pp. Peacock, W. J., E. S. Dennis, A. Elizur, and J. H. Calaby. 1981. Repeated DNA sequences and kan- garoo phylogeny. Aust. J. Biol. Sci., 34: 325-340. Peron, F., and C. A. Lesueur. 1807. Voyage de De- couvertes aux terres Australes, atlas plate 27. (Fide Tate, 1948.) Poole, W. E., S. M. Carpenter, and H. G. Simms. 1 980. Multivariate analysis of skull morphometries from the two species of grey kangaroos, Macropus giganteus Shaw and M.fuliginosus (Desmarest). Austral. J. Zool., 28:591-605. Read, D. 1982. Observations on the movements of two arid zone planigales (Dasyuridae, Marsupialia), pp. 227-231. In Archer, M., ed., Carnivorous Mar- supials, vol. 1, chap. 21. Royal Zoological Society of New South Wales, Sydney, N.S.W., Australia. Ride, W. D. L. 1970. A Guide to the Native Mammals of Australia. Oxford University Press, London, 249 pp. Ride, W.D.L., and C.H.Tyndale-Biscoe. 1959. The mammals. In Ride et al., The results of an expedition to Bernier and Dorre islands, Shark Bay, Western Aus- tralia in July 1959. Fauna Bull. Fisheries Dept. West. Aust., no. 2: 54-97. Sanson, G. D. 1983. Evolution of feeding adaptations in fossil and Recent macropodids, pp. 489-506. In Rich, P. V., and E. M. Thompson, eds., The Fossil Vertebrate Record of Australasia. Monash University Offset Printing Unit, Clayton, Victoria, Australia. Semken, H. A., Jr. 1974. Micromammal distribution and migration during the Holocene, p. 25. Abstracts, Amer. Quat. Assoc. 3rd Bienn. Meeting, University of Wisconsin, Madison. Sharman, G. B. 1961. The mitotic chromosomes of marsupials and their bearing on taxonomy and phy- logeny. Aust. J. Zool., 9: 38-60. Shaw, G., and F. P. Nodder. 1 790. The Naturalists' Miscellany, vol. 1, text for plate 33. Nodder & Co., London. Shepherd, N. C. 1982. Extension of the known range of western grey kangaroos, Macropus fuliginosus, and eastern grey kangaroo, M. giganteus, in New South Wales. Aust. Wildl. Res., 9(3): 389-391. Shortridge, G. C. 1 909. An account of the geograph- ical distribution of the marsupials and monotremes of south-west Australia, having special reference to the specimens collected during the Balston Expedition of 1904-1907. Proc. Zool. Soc. London, 1909: 803-848. Stirton, R. A. 1963. A review of the macropodid genus Protemnodon. Univ. Calif. Publ. Geol. Sci., 44(2): 97-162. Tate, G. H. H. 1948. Studies on the anatomy and phylogeny of the Macropodidae (Marsupialia). Bull. Amer. Mus. Nat. Hist., 91(2): 233-352. Taylor, M. J., J. H. Calaby, and T. D. Redhead. 1 982. Breeding in wild populations of the marsupial mouse Planigale maculata sinualis (Dasyuridae, Marsupi- alia), pp. 83-87. In Archer, M., ed., Carnivorous Mar- supials, vol. 1, chap. 9. Royal Zoological Society of New South Wales, Sydney, N.S.W., Australia. Tedford, R. H. 1966. A review of the macropodid genus Sthenurus. Univ. Calif. Publ. Geol. Sci., 57: 1- 72. . 1967. The fossil Macropodidae from Lake Menindee, New South Wales. Univ. Calif. Publ. Geol. Sci., 64: 1-164. Thomas, O. 1887. On the wallaby commonly known as Lagorchestes fasciatus. Proc. Zool. Soc. London, 1886: 544-547. . 1888. Catalogue of the Marsupialia and Mon- otremata in the collection of the British Museum (Nat. Hist.), London. Taylor and Francis, London, i-xiii + 401 pp. Wakefield, N. A. 1964. Mammal remains, appendix 1. In Mulvaney, D. J., G. H. Lawton, and C. R. Twi- dale, Archaeological excavations of Rock Shelter no. 6, Fromm's Landing, South Australia. Proc. Roy. Soc. Victoria, 77: 494-498. . 1966. Mammals of the Blandowski Expedition to northwestern Victoria, 1856-57. Proc. Roy. Soc. Victoria, 79(2): 371-391. LUNDELIUS & TURNBULL: MADURA CAVE. PART VII. 71 Other Fieldiana: Geology Titles Available in This Series The Mammalian Fauna of Madura Cave, Western Australia. Part I. By E. L. Lundelius, Jr., and W. D. Turnbull. Fieldiana: Geology, vol. 31, no. 1, 1973. 35 pages, 13 illus. Publication 1160, $3.25 The Mammalian Fauna of Madura Cave, Western Australia. Part II. By E. L. Lundelius, Jr., and W. D. Turnbull. Fieldiana: Geology, vol. 31, no. 2, 1975. 81 pages, 21 illus., appendix. Publication 1209, $7.25 The Mammalian Fauna of Madura Cave, Western Australia. Part III. By E. L. Lundelius, Jr., and W. D. Turnbull. Fieldiana: Geology, vol. 38, no. 1, 1978. 120 pages, 27 illus., 29 tables. Publication 1290, $9.50 The Mammalian Fauna of Madura Cave, Western Australia. Part IV. By E. L. Lundelius, Jr., and W. D. Turnbull. Fieldiana: Geology, n.s., 6, 1981. 72 pages, 21 illus., 8 tables. Publication 1315, $7.25 The Mammalian Fauna of Madura Cave, Western Australia. Part V: Diprotodonta (Part). By E. L. Lundelius, Jr., and W. D. Turnbull. Fieldiana: Geology, n.s., 1 1, 1982. 32 pages, 10 illus., 4 plates, 4 tables. Publication 1332, $4.50 The Mammalian Fauna of Madura Cave, Western Australia. Part VI. By E. L. Lundelius, Jr., and W. D. Turnbull. Fieldiana: Geology, n.s., 14, 1984. 63 pages, 15 illus., 21 tables. Publication 1354, $8.25 Order by publication number and/or ask for a free copy of our price list. All orders must be prepaid. Illinois residents add current tax. All foreign orders are payable in U.S. dollar-checks drawn on any U.S. bank or the subsidiary of any foreign bank. Prices subject to change without notice. Address all requests to: FIELD MUSEUM OF NATURAL HISTORY Library— Publications Division Roosevelt Road at Lake Shore Drive Chicago, Illinois 60605-2498, U.S.A. Field Museum of Natural History Roosevelt Road at Lake Shore Drive Chicago, Illinois 60605-2496 Telephone: (312) 922-9410