S. WROE METHODS

STRATIGRAPHY AND PHYLOGENY OF THE GIANT EXTINCT RAT KANGAROOS (PROPLEOPINAE, HYPSIPRYMNODONTIDAE, MARSUPIALIA) s. WROE

Wroe, S. 1997 06 30: Stratigraphy and phylogeny of the giant extinct Rat-kangaroos (Propleopinae, Hypsiprymnodontidae, Marsupialia). Memoirs of the Queensland Museum. 41(2): 449-456. Brisbane. ISSN 0079-8835.

The Giant Rat-kangaroos were placed in the Propleopinae by Archer & F\annery (1985) and in the Hypsiprymnodontidae by Ride (1993). Cladistic analysis of Ekaltadeta material from Riversleigh, northwestern Queensland (Wroe, 1996) suggested that a middle to late Miocene dichotomy in Ekaltadeta may have produced two lineages of Plio-Pleistocene Propleopus. indicating polyphyly for Propleopus and paraphyly for Ekaltadeta. Metrical data for propleopines and stratigraphic information support Wroe's (1996) cladistic analysis of propleopines. D Propleopinae. Hypsiprymnodontidae, Riversleigh, Ekaltadeta, cladistics.

S. Wroe. School of Biological Science, University of New South Wales, N.S. W. 2052 Australia; received 4 November 1996.

Giant Rat-kangaroos (Hypsiprymnodontidae: However, a more general consideration of stratig- Propleopinae) may be the plesiomorphic sister raphy in phylogenetic analysis may be appropri- group of potoroids (Flannery, 1987). Archer & ate in association with cladistic treatment where Flannery ( 1985) considered Ekaltadeta ima,(Fig. specimens are stratigraphically disjunct or 1) the sister group to Propleopus De Vis, 1888 sparsely distributed (Gingerich, 1990). with P. oscillans De Vis, 1888 (Fig. 2) the more Sites with Ekaltadeta are late Oligocene to plesiomorphic and P. chillagoensis Archer et al. early late Miocene (Archer et al., 1989, 1994, ( 1978) (Fig. 2) the more apomorphic within Pro- 1995). A number of characters were analysed to pleopus (Fig. 3). Propleopine species described assess the development of time-dependent since 1985 are Ekaltadetajamiemulvaneyi Wroe, changes. Specimens were ranked to indicate rel- 1996, (Fig, 4) and Jackmahoneya toxoniensis ative age (Appendix 1). Stratigraphic levels are Ride, 1993. Wroe (1996) suggested another pos- from Archer et al. (1989, 1995): level l=late sible phylogeny for the Propleopinae with E. ima Oligocene early Miocene; level 2=early and P. chillagoensis forming the sister group to Miocene; level 3=late early Miocene; level another clade containing a new species, E. 4=mid Miocene; level 5=late mid Miocene; level jamiemulvaneyi, as the sister taxon to P. 6=early late Miocene; level 7=Pliocene; level wellingtonensis and P. oscillans.(Fig. 5). As an 8=Pleistocene. adjunct to the cladistic analysis (Wroe, 1996), metric and stratigraphic data for propleopines are I included all propleopines possible, although used to clarify intrasubfamilial relationships. Pliocene and Pleistocene Jackmahoneya and Dental homology for premolars follows Flower Propleopus are known from material often lim- (1867) and Luckett (1993) for molars. Higher ited to portions of upper and/or lower dentitions. level systematics of kangaroos follows Flannery Most Propleopus are from the Pleistocene, al- ( 1987) and Ride ( 1993). Specimens are housed in though material has been recorded from early the Queensland Museum (QMF). Other prefixes Pliocene local faunas (Archer & Fllinnery, 1985). Propleopus chillagoensis was described as include; UCM (University of California Mu- Pleistocene (Archer et al., 1985), but could be seum), NMV (Museum of Victoria). older, possibly late Miocene or early Pliocene (Archer pers. comm.). Jackmahoneya toxonien- METHODS sis is Pliocene (Ride, 1993). Differences in molar gradient were used by Specimens of Ekaltadeta from Riversleigh rep- Archer & Flannery (1985) and Wroe (1996) to resent 30 individuals from several stratigraphic distinguish propleopine species. Molar gradient levels. The relative paucity of specimens and reflects both the surface area and length of the chronological data precludes a strictly strato- molar tooth row. In propleopines a high molar phenetic approach (sensu Gingerich, 1976, 1979; gradient correlates with a reduction in both molar Bown & Rose, 1987) to propleopine phylogeny. surface area and the length of the tooth row. 450 MEMOIRS OF THE QUEENSLAND MUSEUM

Reducing the distance between condyle and sectorial tooth maximizes leverage applicable to the tooth (Young et al., 1989). Through shortening the molar row, leverage on the large shearing P313of pro- 1 pleopines is increased. This ef- fect is achieved at the cost of molar length. Relative P3 size and molar gradient for upper and lower dentitions has been quantified. Distinct reduction in tooth size I posteriorly occurs in upper and lower dentitions of E. ima. In the upper dentition this steep gradient begins with a reduced posterior width (pw) relative to the anterior width (aw) of M2 which then ramifies through M3-4. In E. ima M4 pw is <1/2 M2 aw. The upper dentition of P. chillagoensis is similar to that of E. ima. Lower dentition is not known for P. chillagoen- 2 sis. For P. oscillans M3-4 are missing but M2 pw is only slightly less than M2 aw sug- \ gesting a less extreme gradient. ~ This supposition is strongly supported by the lower dentition in which molar gradient ~ contrasts strongly with E. ima. 3 MI-4 tooth widths decrease steadily anteroposteriorly in E. .. ima but are reversed in p os- FIG. I. EkaltadetaIma, x 2. A, occlusal vIew of QMFI2436 (uppers).B, .II h t th .dth..buccal view ofQMF12435,left dentarycontaining II, alveolusforh,P2-3, CI ans were .00 WI m- MI-4.C,occlusalviewofQMFI2435. creases posterIorly for MI-3, with only a slight decreasein M4. distinct and 2 molars were required to demonstr- Several methods to quantify molar gradient ate a gradient. Measurements were made using a have been considered. Accurate determination of Wild MMS 235 Digital Length-Measuring Set individual molar surface areas and comparisons attached to a Wild M5A Stereomicroscope. Ab- between teeth would be useful but would require breviations are: 1=length, w=width, aw=anterior 2 or more teeth/ specimen, greatly limiting data width, pw=posterior width, dd=depth of dentary, sets, particularly for upper dentitions. Molar gra- G-value=ratioof anterior to posterior tooth width. dient might also be estimated geometrically by determining the angle at which a line drawn bucc- RESULTS ally or lingually through the faces of the crown intersects the mid-line of the dentary or skull. In this study the clear initiation of a marked M2 aw 1 M2 pw VS STRA TIGRAPmC LEVEL. molar gradient at M2 in the upper dentitions of E. (Fig. 6). For upper dentitions the ratio M2 aw: M2 ima and P. chillagoensis permitted estimation of pw (G-value) was used as an arbitrary measure of the gradient from a single molar by comparing aw molar gradient, with M2 being common to the to pw. In lower dentitions the gradient is less largest number of specimens. STRATIGRAPHY AND EKALTADETA 451

Po "' 0: .~ ~ .~ 0 § ~ ~ ~ ~ 00 ~ .5 .e. .~ ~ ~ .§ ~ ~ ~ 0£ ~ ~ ~ ~

FIG. 2. Propleopuschillagoensis, x 2. A, occlusal view ofNMV P15917, right maxillary fragment (juvenile)J, containing unerupted p3, partial MI, M2-3, partial M (cast of holotype). B, Propleopus oscillans, x 2, occlusal view of ~MF6675, left maxillary fragment, containing p3, M -2.

A trend is apparent in this scatter graph of G-value against stratigraphic level. P. chillagoensis and P. oscillans represent 2 extremes with G-values of 1.23 and 1.06 respectively, with the lower number indicating a lesser molar gradient. Ekaltadeta ima from levels 3 and 4 has a limited range ofG-values (1.09-1.15). The 2 Ekaltadeta from level 6 both fell outside the range of E. ima from older strata. E. jamiemulvaneyi (QMF24212; Cleft of Ages 4 Site) had a low G-value of 1.05, slightly less than that of P. oscillans. E. ima (QMF24211; Henk's Hollow Site) had a relatively high G-value of 1.19 approaching that of P. chillagoensis. These results indicate a divergence in the Ekaltadeta lin- FIG. 3. Cladogram for the propleopines from Archer & eage with one population leading to P. oscillans Flannery ( 1985). Character states at nodes: A=gain of and another leading to P. chillagoensis. an anterior cristid emanating from the metaconid of MI, gain of derived II morphology; B=incorporation of the protolophid into the anterior lophid of MI loss MI pw / M2 pw VS STRATIGRAPHIC LEVEL. of P2 with eruption of P3, dentary deeper anteriorly (Fig. 7). The molar gradient of the dentary was than posteriorly; C=reduction of metacone/en- estimated by dividing Mi pw by M2 pw (G- toconid, P3 hypertrophy. value). P. oscillans had the lowest G-value at 0.93. The G-values for P. wellingtonensis and J. the species with low molar gradients (1. toxonien- toxoniensis were slightly higher at 0.96. At levels sis. P. oscillans and P. wellingtonensis). 3 and 4 the G-values for Ekaltadeta were 1.01- 1.08. The G-value for E. jamiemulvaneyi. from P3 w/MIpw VS STRATIGRAPHIC LEVEL. level 6 (QMF24200, Encore Site) was 0.97. This (Fig. 8). In P. oscillans P3 width was small com- placed E.jamiemulvaneyi about halfway between pared to Mi posterior width (1.09). For J. tox- the lowest G-value from levels 3 and 4 and P. oscillans. Again the highest degree of divergence oniensis relative P3 width was greater (1.27). E. among Ekaltadeta was for the E. jamiemulvaneyi ima from levels 3 and 4 had ratios of P3 w / Mi from level 6, possibly indicating a trend toward pw of 1.35-1.52. E. jamiemulvaneyi from Encore 452 MEMOIRS OF THE QUEENSLAND MUSEUM

site (QMF24200) again posi- tioned between E. ima from lower strata and J. toxoniensis I P. oscillans, with a ratio of 1.28.

DEPTH OF DENTARY 1 AGAINST STRA TI- GRAPmC LEVEL. Depth of dentary against stratigraphic 2 level (Fig. 9). Dentary depth was measured from the alveo- lar margin ofMl to the ventral margin of the dentary perpen- dicular to the molar row. Vari- ation in the depth of dentaries was small for E. ima from levels 3 and 4 (19.3-21.lmm). E. jamiemulvaneyi (QMF24200) was much larger with a dentary depth of 28.8mm approaching P. oscillans (32.6mm). J. toxoniensis between E. ima and E. jamiemulvaneyilP. osc illans 5 with a dentary depth of 23.3mm. 4

STATISTICAL ANAL YSIS. (Table 1 ). Because all Ekaltadeta material has come \} from a relatively small area 6' (Riversleigh), a regional population of potoroids was consid- 7 ered an appropriate control. Sixteen specimens from the 8 9 Australian museum of Potor- ous tridactylus collected ar<:>undHobart were us.ed, this FIG. 4. Ekaltadetajamiemulvaneyi x 2. A, occlusal viewofQMF24200,left ~emgthelargestp.otoroldsp<:c- denlary containing P3, M1-3, hololype. B, buccal view ofQMF24200. C, l1.11ens.ample avaIlable. Vana- lingual view of QMF24200. D, occlusal view of QMF24212,left maxillary tlon In the G-values of fragment, containing p2, dP3, MI-2, referred specimen. E, buccal view of Ekaltadeta from levels 3 and 4 QMF24212. F, lingual view of QMF24212. G, occlusal view of approached that of P. QMF20842, left p3, referred specimen. H, buccal view of QMF20842. 1, tridactylus. When G-values lingual view ofQMF20849. from the 2 Ekaltadeta from level 6 were included the variation fell well out- (:Abbie, 1939), while a more extensive molar side that of the local P. tridactylus population. ,lrray may indicate a more herbivorous diet (Wells et al., 1982). DISCUSSION Species with a large molar surface area and low molar gradient (P. oscillans, P. wellingtonensis, J. toxoniensis) have relatively small premolars. Increases in premolar and molar shear within the Propleopinae appear to be mutually exclusive Species with high molar gradients and reduced and their relative importance probably reflects molar shear (E. ima, P. chillagoensis) are dietary preference. A requirement for high pre- characterised by P3 hypertrophy. The extraordi- molar shear might be associated with carnivory nary change in function for P2 shown by individ- STRATIGRAPHY AND EKALTADETA 453

g. TABLE I. Statistical summaries for M2 aw / M2 pw ~ .~ .~ (G-value) for propleopines and a local P. tridactylus ~ .:2 " E ~ ~ ~ '::i population. ~ E ~ ~ .~ ~ .9- ~ :§ § ~ ~ ] .~ ~ (oj ~ ~ Q; Q;

was largely if not wholly herbivorous. Other derived features interpreted as adaptations to herbivory for P. oscillans include a large dia- stema between P3 and 11, and large spatulate lower incisors (Wroe, 1996). Regarding dentary depth E. ima is the smallest propleopine with a general increase in depth for taxa at higher stratigraphic levels probably reflecting a general in- D crease in body size. Stratigraphic and metric analysis support the proposal of a late Miocene dichotomy in Ekaltadeta producing 2 lineages of Propleopus, and a reversal of previous assumptions on relative apomorphy within Propleopus, with P. oscillans considered the most derived and P. chillagoensis the most plesiomorphic (Wroe, 1996). However, 3 broad trends suggestedin this study are not interpreted here chronoclines in the stratophenetic sense (sensu Bown & Rose, 1987). The scarcity of material and uncertain chronology of both the ~ Oligo-Miocene Riversleigh deposits and the Plio- Pleistocene local faunas from which most propleopine specimens are known necessitates caution in the interpretation of results. A consid- FIG. 5. Minimal tree produced by Wagner analysis for erable temporal gap exists between estimated the Propleopinae (from Wroe, 1996). Character states at nodes: A = gain of an anterior cristid emanating agesof the most recent Ekaltadeta specimens and from the metaconid of Mi; basally broad conical all other propleopines. As noted by Ride ( 1993), upper molars; B = presence of lingual cingula on the the period separating the latest known incidence upper molars; C = reduced molar gradient, reduced of Ekaltadeta from Plio-Pleistocene Propleopus P3; D = incorporation of the protolophid into the and Jackmahoneya may be sufficient to have anterior lophid of MI, a dentary deeper posteriorly permitted a secondary reversal of character states than anteriorly. within Propleopus to produce P. chillagoensis. Many questions remain concerning the age, ual E. ima (Wroe & Archer, 1995) probably con- stratigraphy and method of deposition of stitutes a response to the increased loading placed Riversleigh's Oligocene-Miocene limestone de- on P3. Regarding molar gradient and relative size posits (Archer, 1994, 1995; Megirian, 1992, of the P3, E.jamiemulvaneyi is intermediate, fall- 1994). If the phylogeny for propleopines sug- ing between E. ima specimens from lower levels gested by Wroe (1996) reflects evolutionary and P. oscillans/P. wellingtonensis/J. toxonien- events, then it provides tacit support for Archer sis. Using the same criteria J. toxoniensis lies et al.'s (1989) proposed stratigraphy, with an between E. jamiemulvaneyi and P. oscillans/P. agreement of hypothesised superpositional and wellingtonensis. In terms of variation in P3 size phylogenetic patterns. and molar gradient P. chillagoensis and P. os- The capacity of stratigraphic occurrence to ex- cillans represent opposite extremes in propleop- plicitly mirror phylogenies is questionable (En- ine evolution and it is suggested that P. oscillans gelmann & Wiley, 1977). Although strong 454 MEMOIRS OF THE QUEENSLAND MUSEUM

101

R osciUans QM F3302 R weUingtonensis 81 " D UCM P45l71 ~ D J. toxoniensis AR 17579 ~ '-' D E. jamienwlvaneyi i 61 I! QMF"'200 .. E. ima QM PI2423 .E ~ 4i D ~ E. ima QMPl243S QMP24198 QMP24199QMP24197 .~ D D D D .., ~ 2 DO E. ima QM P24196 QM P2419S

0 I. .. 0.9 1.0 1.1

M2aw/M2pw MI pw/M2pw

FIG. 6. M2 aw I M2 pw vs relative stratigraphic level FIG. 7. MI pw I M2 pw vs relative stratigraphic level for propleopines. Ekaltadeta ;ma from level 3, left to for propleopines. right, QMF24207, 24204, 24205. 12436, 24203, 24208, 24209, and 24206.

10,

EosciUans 1! oscillans D QM F3302 D QMF3302 J. Ioxoniensis Bi J. toxoniensis AR 17579 D AR17579 ~ D E. jamiemulvaneyi ~ ~ " !d D E . jamiemulvaneyi 6 D QM F2A200 ~ .. QMF24200 ~ ~ ~ .= M 'E QM FI2424 QM FI2423 E. ima ~ " " QMFI2A2A QMFI2A23 E. ima ~ DD "' ~ ,~ .. D QM FI243S E. ima .~ " QMFI2A3S E. ima '03 "0! ~ ~ 2i QMF24201 E. ima D DQMF1~E. ima

, 1.1 1.2 1.3 1.4 1.5 1.6 0 I. ... 10 20 30 40

P3w/Mlpw Dentary Depth (mm)

FIG. 8. P3 W I MI pw vs relative stratigraphic level for FIG. 9. Depthof dentaryvs relative stratigraphiclevel propleopines. for propleopines.

congruence between cladistic and stratigraphic may be useful as a test of stratigraphic interpreta- arrangements has been demonstrated for many tions. vertebrate taxa by Norell & Novacek ( 1992a,b), The propleopine phylogeny of Wroe (1996) is the sameauthors advised that correlation between based on analysis of an incomplete data matrix, the 2 diminishes rapidly where cladistic or strati- with important characters unknown for several graphic data is poorly resolved. Debate over con- species. Consequently the cladistic data pre- formity of age and cladistic information commonly centres around the value of super- sented cannot be viewed as a robust basis for positional data as an adjunct to phylogenetic re- testing superpositional pattern. However, thepro- construction. Where cladistic analysis is sound it ductivity of the Oligocene-Miocene deposits of STRATIGRAPHY AND EKALTADETA 455

Riversleigh engenders reasonable expectation for Partridge, T.C. & Burkle, L.H. (eds) 'Paleocli- the reliable resolution of phylogenies. mate and evolution, with emphasis on human origins'. (Yale University Press: New Haven). BOWN, T.M. & ROSE, K.D. 1987. Patterns of evolu- ACKNOWLEDGEMENTS tion in early Eocene anaptomorphine primates (Omomyidae) from the Bighorn basin, Wyoming. M. Archer, H. Godthelp and W. D. L. Ride joumal of Paleontology 23: 1-128. provided constructive criticism and comment. DE VIS, C. W. 1888. On an extinct genus of the marsu- Vital support for this research has been given by pials allied to Hypsiprymnodon. Proceedings of the Australian Research Council (to M. Archer); the Linnean Society of N.S. W. 3: 5-8. the National Estate Grants Scheme (Queensland) ENGELMANN, G.F. & WILEY, E.O. 1977. The place (grants to M. Archer and A. Bartholomai); the of ancestor-descendantrelationships in phylogeny Department of Environment, Sports and Territo- reconstruction. Systematic Zoology 26: 1-11. ries; the Queensland National Parks and Wildlife FLANNERY, T. 1987. The relationships of the Service; the Commonwealth World Heritage macropodoids (Marsupialia) and the polarity of some morphological features within the Unit (Canberra); the University of New South Phalangeriformes. pp. 741-747. In Archer, M. Wales; ICI Australia Pty Ltd; the Australian Geo- (ed.) 'Possums and opossums: studies in graphic Society; the Queensland Museum; the evolution' .(Surrey Beatty & Sons and Royal Zoo- Australian Museum; Century Zinc Pty Ltd; Mt logical Society of New South Wales: Sydney). Isa Mines Pty Ltd; Surrey Beatty & Sons Pty Ltd; A..OWER, W.H. 1867. On the development and suc- the Riversleigh Society Inc.; the Royal Zoologi- cession of teeth in the Marsupialia. Philosophical cal Society of New South Wales; the Linnean Transcripts of the Royal Society of London 157: Society of New South Wales; and many private 631-641. supporters. Skilled preparation of most of the GINGERICH, P. D. 1976. Cranial anatomy and evolu- Riversleigh material has been carried out by tion of North American Plesiadapidae Anna Gillespie. (Mammalia, Primates). University of Michigan Papers on Paleontology 15: 1-140. 1979. The stratophenetic approach to phylogeny LITERA TURE CITED reconstruction in vertebrate paleontology. (Co- lumbia University Press: New York). ABBIE, A. 1939. A masticatory adaptation peculiar to 1990. Stratophenetics. Pp. 437-442. In Briggs, some diprotodont marsupials. Proceedings of the D.E.G. & Crowther, P.R. (eds), Palaeobiology, a Zoological Society 109: 261-279. synthesis. (Blackwell Scientific: Oxford). ARCHER, M. 1994. Introduction: background and con- LUCKErr, P. 1993. An ontogenetic assessment in troversy about the Riversleigh geological and pal- dental homologies in the therian mammals. pp. aeontological resource. Riversleigh Symposium 182-204. In Szalay, F.S., Novacek, M.j. & Mc- 1994, Abstracts: p 5-6. Kenna, m.c. (eds) 'Mammal Phylogeny: ARCHER, M., BARTHOLOMAI, A. & MARSHALL, Mesozoic Differentiation, Multituberculates, L.G. 1978. Propleopus chillagoensis, a new north Monotremes, Early Therians, and Marsupials'. Queensland speciesof extinct giant Rat-kangaroo (Springer-Verlag: New York). (Macropodidae: Potoroinae). Memoirs of the Na- MEGIRIAN, D. 1992. Interpretation of the Carl Creek tional Museum of Victoria 39: 55-60. Limestone, northwestern Queensland.The Beagle ARCHER, M. & FLANNERY,T.F. 1985. Revision of 9: 219-248. the extinct gigantic Rat-kangaroo (Potoroidae: Marsupialia), with description of a new Miocene 1994. Approaches to marsupial biochronology in genus and species, and a new Pleistocene species Australia and New Guinea. Alcheringa 18: 259- of Propleopus. Journal of Paleontology 59: 1331- 274. 1349. MYERS, T. & ARCHER, M. 1997. Ku1erintja ngama ARCHER, M., GODTHELP, H., HAND, S.J. & (Marsupialia, llariidae): a revised and extended MERGIRIAN, D. 1989. Fossil mammals of systematic analysis based on fossil material from Riversleigh, northwestern Queensland: prelimi- the late Oligocene of Riversleigh, northwestern nary overview of biostratigraphy, correlation, and Queensland, Australia. Memoirs of the Queens- environmental change. Australian Zoologist 25: land Museum 41: 379-392. 29-65. NORELL, M.A. & NOV ACEK, M.j. 1992a. Congru- ARCHER, M., GODTHELP, H. & HAND, S.J. 1994. ence between superpositional and phylogenetic 'Riversleigh. The story of animals in ancient rain patterns: comparing cladistic patterns with fossil forests of inland Australia, SecondEdition' .(Reed records. Cladistics 8: 319-337. Books Pty Ltd. Sydney). 1992b. The fossil record and evolution: comparing 1995. Tertiary environmental and biotic change in cladistic and paleontologic evidence for verte- Australia. pp. 77-90. In Vrba, E.S., Denton, G.H., brate history. Science 255: 1690-1693. 456 MEMOIRS OF THE QUEENSLAND MUSEUM

RIDE, W .D.L. 1993. Jackmahoneya gen. nov ., and the Hypsiprymnodontidae, Marsupialia). Journal of genesis of the macropodifornl molar. Memoirs of Paleontology 70: 681-690. the Association of Australasian Palaeontologists WROE, S. & ARCHER, M. 1995. Extraordinary 15: 441-459. WELLS, R.T. MORTON, D.R., & ROGERS, P. 1982. diphyodonty-related change in dental function for a tooth of the extinct marsupial Ekaltadeta ima Thylacoleo carnifex Owen (Thylacoleonidae): Marsupial carnivore ? Pp. 573-586. In Archer, M. (Propleopinae, Hypsiprymnodontidae). Archives (ed.) 'Carnivorous Marsupials'. (Surrey Beatty of Oral Biology 40: 597-603. and Sons Pty Ltd and the Royal Zoological Soci- YOUNG, W.G., JUPP, R. & KRUGER, B.J. 1989. ety of New South Wales: Sydney). Evolution of the skull, jaws, and teeth in verte- WROE, S. 1996. An investigation of phylogeny in the brates. (University of Queensland Press: Bris- Giant Rat-kangaroo Ekaltadeta (Propleopinae, bane).

APPENDIX TO FIGURES 6-9

Data for Fig. 6. M2 aw divided by M2 pw (G-value) vs. relative stratigraphic level for propleopines. Measure- ments in mm. * = skull, (R) = right tooth row, (L) =

Data for Fig. 8. P3 W divided by Mi pw (G-value) vs. stratigraphic level for propleopines. Measurements in mm.

Species Cat. no.

II:.. ima QMF24201 II:.. ima QMF12435 QMF12424 11:..ima , ima QMF12423 I Imulvaneyi1:..jamie QMF24200

I r. oscillans QMF3302 I J. toxoniensis AR17579