lWO NEW EARL y MIOCENE THYLACINES FROM ...Muirhead (1997...lWO NEW EARL y MIOCENE THYLACINES FROM RIVERSLEIGH, NORTHWESTERN QUEENSLAND JEANETTE MUIRHEAD Muirhead, J. 19970630: Two

lWO NEW EARL y MIOCENE THYLACINES FROM RIVERSLEIGH,NORTHWESTERN QUEENSLAND

JEANETTE MUIRHEAD

Muirhead, J. 19970630: Two new early Miocene thy1acines from Rivers1eigh. northwesternQueensland. Memoirs of the Queens/and Musewn 41(2): 367-377. Brisbane. ISSN 0079-8835.

Thylacines, Wabulacinus ridei gen. et sp. nov.and Ngamalacinus timmulvaneyi gen. et sp.nov ., are described from the early Miocene of Riversleigh. northwestern Queensland. Bothshow carnivorous adaptation intermediate between that of the plesiomorphic Nimbacinusdicksoni and derived Thylacinus. The family concept is revised to include these new taxa.All known thylacinid genera occur in late Oligocene to middle Miocene Riversleigh faunasand some may have overlapped in time followed by a decline in family diversity since theMiocene. D Thylacine, marsupial. carnivore, Miocene, Riversleigh, Queensland.

i. Muirhead, School of Biological Sciences, University of New South Wales NSW 2052Australia; received 25 June 1995.

The Thylacinidae consists of three species ofThylacinus (T. cynocephalus Harris, 1808, T.potens Woodburne, 1967 and T. macknessiMuirhead, 1992) and the monotypicNimbacinusdicksoni Muirhead & Archer, 1990 from the lateOligocene to middle Miocene of Queensland andthe Northern Territory (Muirhead & Archer,1990). It is the oldest and most primitive thY-lacinid, more closely resembling dasyurids inmany plesiomorphic features. Thylacinus potensfrom the late Miocene Alcoota Local Fauna(Woodbume, 1967) is considered (Archer, 1982)the sister species of modern T. cynocephalus andis almost as specialised. Thylacinus macknessi.from earl y to middle Miocene Ri versleigh faunas,is also a highly specialised thylacine. Because itretains some plesiomorphic features, it is consid-ered to be the sister species to the T. potens -T.cynocephalus clade (Muirhead, 1992). Two newearly Miocene thylacinids from Riversleigh aredescribed here. In many features they provide acontinuum in morphological change from theplesiomorphic dentition of N. dicksoni to that ofspecialised Thylacinus. Dental nomenclature fol-lows Aower(1869) and Ll1ckett(I993) where theadult dentition includes PI-3 and MI-4. Taxo-nomic nomenclature follows Muirhead & Archer(1990). Material is housed in the QueenslandMuseum (QMF) or Northern Territory Museum.

Wabulacinus gen. nov.

TYPE SPECIES. Wabulacinus ridei gen. et sp. nov.

ETYMOLOGY. Wanyii Wabula, long ago; GreekIcynos. dog. Masculine.

DIAGNOSIS. Infraorbital foramen surrounded wholllby the maxillary and positioned low and anterior to M ;cenb"OCrista and preparacrista parallel, forming contin-uous Sb"aight line on MI; entoconid absent (on M3);hypoconulid enlarged (on M3).

COMPARISON. Wabulacinus differ from N.dicksoni by larger size, lack of stylar cu~s BandD on MI, lack of stylar cusp Bon M2 and theminute size of St D on this toothl the straight oralmost straight centrocrista on M and M2, antc-rior cingulum of MI has no notch for placementof preceding premolar, the anterior root of M I lies

directly under the cingulum, the anterior width ofthe upper molar crowns are less than that of thebuccal lengths, wider angle of crests at theparacone and metacone, extreme reduction of thetalonid basin and protocone, particularly on MIwith concurrent loss of metaconules on this tooth,extreme reduction in size of the metaconid, ab-sence of entoconid, reduced talonid basin by themore lingual position of the hypoconid and lackof diastemata between PI and ?2.

Species of Wabulacinus differ from all speciesof Thylacinus in the extreme reduction of thetalon and protocone on MI, the more parallelalignment of the preparacrista with the cen-trocrista on MI, a small metaconid (at least on theM3), less elongate snout by lack of diastematabetween the premolars as well as between PI andthe canine. Wabulacinus ridei is similar in molar

SYSTEMATICS

Order DASYUROMORPHIA (Gill, 1872)Superfamily DASYUROIDEA (Goldfuss, 1820)Family rnYLACINIDAE (Bonaparte, 1838)

368 MEMOIRS OF THE QUEENSLAND MUSEUM

size to T. macknessi, but lacksan anterior cingulum on Mi.

Wabulacinus ridei sp. nov.

(Fig. 1)

ETYMOLOGY. For David Ridefor his long-term commitment toAustralian vertebrate palaeonto-logy.

MATERIAL. Holotype.QMF16851. right maxillary frag-mentcontainingMI-2(Fig.lA-C).Paratype. QMFl6852 left dentaryfragment with broken M3 (Fig.1 D- F) from early Miocene (Sys-tem B) Camel Sputum Site,Godthelp Hill, Riversleigh.

DIAGNOSIS. As for genus.

DESCRIPTION. Maxillapartly preserved. Infraorbitalforamen enclosed within thebodyofmaxilla,abovethepos- Itenor alveolus of p3.

Buccal crown of Mi lengthexceeds anterior width.Metacone largest cusp fol-lowed by paracone, protoconeand St E. No other cusps.Postmetacrista longest crest,curving buccally at the poste-

Inor end. Preparacrista orien-

tated almost parallel to thetooth row, terminating at theanterior tip of the crown. Pre-metacrista and postparacristaconnecting as a straight cen-trocrista which parallels the

preparacrista. Lacking pre-protocrista, postprotocrista, I

protoconule, metaconule, stY-lar shelf or stylar cusps anteriorto St E. Buccal flank of crownforming continuous slope from. ..I 2 ..paracone and metacone to low- FIG. 1. Wabulacmus ndel. A = QMF16851 (M and M ) lIngual view. B =est buccal edge of the crown QMF16851 (Mi and M2) buccal view wilh infraorbital foramen arrowed.P Il .C and CI = QMF16851 stereo occlusal views. D = QMF16852 (P2 and M3)

rMot~?n~

I Sma M.I S buccal view. E = QMF16852 (M3) lingual view showing small metaconid.Slffil ar to except: t (arrowed). F and F' = QMFI6852 stereo occlusal view.

E mInute. Stylar shelf regionhigh, of many tiny indistinct cusps and crests, and premetacrista. Postparacrista and pre-especially on the more posterior half of the metacrista forming a wide angled centrocrista.crown. Postmetacrista longest crest on the crown, Postmetacrista leaving metacone almost parallelfollowed in declining length by preparacrista, to the premetacrista, curving buccally. Pre-postprotocrista, preprotocrista, postparacrista paracrista straight, connecting to the postpara-

TWO NEW EARL Y MIOCENE THYLACINES FROM RIVERSLEIGH 369

narrower angle of centrocrista; less reduced stylarshelf with retention of prominent St B, St D andstylar shelf crests on Mi and M2; less reducedtalon basin, particularly on Mi; less anteroposter-ior elongation of the molars and associated crestlengths; larger talonids; and larger metaconid(larger than the paraconid) and with a distinctmetacristid.

Ngama/acinus further differs from w. ridei inthe more posterior position of the infraorbitalforamen, presence of an entoconid on the lowermolars and smaller hypoconulid.

Ngamalacinus timmulvaneyi sp. nov.(Figs 2, 3)

ETYMOLOGY. For Tim Mulvaney, a long-time sup-poner of research at Riversleigh.

MATERIAL. Holotype QMFI6853 right dentary withMI-4 (Fig. 2) from early Miocene (System B) Inabey-ance Site, Godthelp Hill, Riversleifh. ParatypesQMF30300 left maxillary with p2-M (Fig. 3A-C),from early Miocene (System B) Camel Sputum Site,Godthelp Hill. Referred specimen QMF16855, rightM2 (Fig. 3D), from the type locality.

DIAGNOSIS. As for genus.

crista at approximately 90° and oblique to thetooth row. No St B present. Trigon basin widerthan on MI. Lingual flank of trigon basin 'V'-shaped with a distinct ridge running verticallydown its centre. The preprotocrista andpostprotocrista prominent with a minute pro-toconule and metaconule. Ectoflexus on the buc-cal surface of this tooth slightly developed.Anterior cingulum terminating anterior to base ofthe paracone, lacking a notch.

Mental foramen under the anterior root of P2.Alveoli for PI-3, MI-2 and the anterior root of M4;M3 only molar present. Symphysis beginning ad-jacent to the anterior root of P3. No diastematabetween alveoli. All alveoli pairs orientated par-allel to the tooth row except PI oblique, indicatingsome crowding of PI against the canine. Alveolisize indicating relative lengths of P3> P2> PI,M3=M2> MI.

M3 with cusps in decreasing height paraconid,metaconid, hypoconulid, hypoconid. All cuspsprominent except minute metaconid; entoconidand related crests absent. Protocristid longestcrest, followed (in decreasing length) by theposthypocristid and cristid obliqua. Remains ofthe metacristid connect to the small metaconid.Small talonid basin open on the lingual side.Hypoconid almost medial to the trigonid basin.Posthypocristid and cristid obliqua orientatedoblique to the dentary, meeting at the hypoconidat right angles. Anterior cingulum continuingbuccally past the anterobuccal comer of tooth,with wide notch. Posterior cingulum poorly de-veloped, a small bulge in the enamel.

Ngamalacinus gen. nov.

TYPE SPECIES. Ngamalacinus timmulvaneyi et sp.nov.

ETYMOLOGY. Wanyii Ngamala. died out; Greekkynos. dog. Masculine.

DIAGNOSIS. Moderately specialised amongthylacinids in the reduced conules, reduced stylarshelf, anteroposteriorly elongated molars. Re-taining small St B and D, metaconid, entoconidand hypoconulid.

DESCRIPTION. All articulating surfaces ofdentary broken. Coronoid process rising from theramus at approximately 120°. All four molars andthe posterior alveolus of pJ present. No diaste-mata between these teeth. Degree of eruption ofM4 indicating a juvenile.

Protoconid of Mi tallest cusp, followed (in de-creasing height) by metaconid, paraconid,hypoconid, hypoconulid and entoconid. All cuspsdistinct, with crests. Paracristid longest crest oncrown followed (in decreasing length) byposthypocristid, metacristid, cristid obliqua, pre-,postentocristid. Anterior cingulum with a verysmall notch. Paracristid almost straight with avery wide angle connecting the paraconid andprotoconid. Talonid basin entirely enclosed bycrests, large and deeply concave to central point.Hypoconid more buccally positioned than pro-toconid. Cristid obliqua continuing up the poste-rior wall of the protoconid. Posterior cingulumdistinct, uniform in thickness to the base of thecrown, with a slight notch formed between it andthe hypoconulid.

M2 same as Mi except: Metaconid relativelylarge, distinct, taller than the paraconid. Allcristids higher and distinct. Anterior cingulumbroader and the notch more distinct. Angle atcrests on the protoconid approximately 100-11 0°.

COMPARISON. Ngamalacinus differs from N.dicksoni in its larger size, reduced metaconulesand protoconules, reduction of St D particularlyonM2.

Ngamalacinus differs from w. ridei and Thy-lacinus in its: smaller size; narrower angle ofcrests at the paracone, metacone and protocone;

MEMOIRS OF THE QUEENSLAND MUSEUM370

Paracristid and metacristid longer. Metacristidstraight; paracristid changing orientation at thevalley between the paraconid and protoconid.

M3 same as M2 except: the anterior half of thecrown thicker than the posterior because of themore lingual position of the paraconid andmetaconid. Paracristid and metacristid elongated.Hypoconulid and entoconid slightly more to pos-terior, with entoconid slightly smaller than onM2. Posthypocristid bending posteriorly to con-nect to the posteriorly positioned hypoconulid.Paracristid proportionally longer than on M2.

M4 same as M3 except: Talonid basin reduced,well defined and enclosed by crests. Entoconidminute; hypoconid small; hypoconulid highestcusp on talonid. Small posterior cingulum pres-ent.

No obvious sutural boundaries on the maxillaexcept a posterior suture that may have connectedto either the jugal or the lachrymal. Maxilla indi-cating that the canine was large, its root extendingdeeply into the maxilla. Infraorbital foramenabove M2. The region immediately posterior tothe infraorbital foramen damaged but a depres-sion in the maxilla here and sutural boundaries ofthe jugal indicate that the jugal is likely to havecontacted the external opening of the infraorbitalcanal. Maxilla with large extension projectingback towards and contributing to the zygomaticarch. No maxillary palatal vacuities in the regionof the premolars.

Small diastemata between the upper premolars.p2 triangular in lateral view with both an anteriorand posterior cusp, with a crest from the majorcentral cusp to the posterior cusp and a less welldefined crest anteriorly to the smaller anteriorcusp, with posterior region wider than the ante-rior, with ridges extending along both sides (lin-gual and buccal) of the posterior cusp. p3 largerthan p2 and similar except for: anterior and pos-terior cusps relatively larger, anterior cusp withridges off the lingual and buccal sides, posteriorcrest from the major cusp more prominent but notstraight, posterior half of tooth relatively widerwith enlarged crests bordering the posterolingualand posterobuccal edges of the crown, with anadditional posterobuccal cusp.

Mi damaged, with anterior cingulum, a large StD larger than the distinct St B, a stylar crestrunning posteriorly from St D to the metastylarcomer, talon broad with a possible protoconule,postmetacrista long and straight, crests at theparacone at approximately 90°, preparacristaconnecting to St B, almost perpendicular to thetooth row.

FIG. 2. Ngamalacinus timmulvaneyi lower dentition.A = QMFI6853 (dentary with MI-5) lingual view. B= QMFI6853 buccal view. C and CI = QMFI6853

M2 same as MI with: in occlusal viewposterolingual dimension longest followed bybuccal length and anterior width. Anterior cingu-lum not notched in QMF16855 but is inQMF30300, cingulum terminating at the anteriorface of the base of the paracone without connect-ing to the talon basin. No posterior cingulum.Metacone highest cusp on the crown, followed (in

TWO NEW EARL Y MIOCENE THYLACINES FROM RIVERSLEIGH 371

decreasing height) by: paracone, St B,metastylar cusp(s) and protocone.Postmetacrista longest crest on thecrown, followed by the preparacrista,premetacrista, postprotocrista, pre-protocrista and postparacrista. Allcrests relatively straight. Enamel sur-face slightly raised about the pro-toconule. Metaconule not present as adistinct cusp. Slightly raisedpostprotocrista connecting the pro-tocone to the metacone base where asharp crest runs up the lingual surfaceof the metacone. A less distinct ridgerunning down the lingual side of theparacone and protocone. Slightectoflexus at the buccal side of thecrown due to bulging of enamelaround St B. St Ea raised part of thestylar crest. Between St E and Bareminute cusps on QMF16855 but St Dis more distinct on QMF30300. Onecrest connecting the metastylarcusp(s) to the posterolingual comer ofthe crown. Talon basin large with abroad, flat base. Preprotocrista andpostprotocrista relatively low. Cen-trocrista at approximately 100°.

M3 same as M2 except: Ectoflexusstronger and all stylarcusps reduced tocuspules. Stylar crest not continuousalong the lingual edge of the crown. StB largest stylar cusp. Anterior cingu-lum with less distinct notch than in M2of QMF30300. Preparacrista andpostmetacrista longer; paracone rela-tively larger but smaller thanmetacone. Paracone more lingually lo-cated. Centrocrista at approximately900; postparacrista strongly curved.Talon narrower. Protoconule andmetaconule with ridges connecting tothe lingual face of the paracone andmetacone respectively.

FIG. 3. Ngamalacinus timmulvane);i uyper dentition. A =QMF30300 maxillary fragment with P2-M and showing infraorbi-tal foramen (arrowed). B and BI = QMF30300 stereo occlusal view

...of pJ and molars. C = QMF30300 lingual view. D and Dl =Ngamalacmus tlmmulvaneyl and W. QMF16855 (M2) stereo occlusal views.

ridei do not share any apomorphiesthat are not also found in Thylacinus.

These, therefore, cannot be considered to repre-

sent members of the same genus.Wabulacinus ridei and Ng. timmulvaneyi have

combinations of features that place them between

THYLACINID PHYLOGENY

plesiomorphic N. dicksoni and apomorphic Thy-lacinus but do not form an independent dichot-omy (Fig. 4). Neither species can be placed in aknown genus because: neither shares any

MEMOIRS OF THE QUEENSLAND MUSEUM372

TABLE 2. Character state distribution among thyla-cines. (a = either O or autapomorphic combination ofentoconid and hYJX>conulid. ? = unknown state).

()OOO()()000()(XX)

11111 11001010

1111112011010

0232803112127

72222 24023 221

1222224024222

1222224124222.-.

Dasyurids

N imbacinus dicksani

Ngamalacinus timmulvaneyi

Wabulacinus r~i

Thylacinus macknessi

Thylacinus patens

Thylacin~fY~halus

more specialised carnivores are located on theright

Wabulacinus ridei and Ng. timmulvaneyi aremore plesiomorphic than Thylacinus in the largersize of the metaconid (small on W. ridei and muchlarger on Ng. timmulvaneyl) and the lack of ex-pansion of the premaxillary region. Both species(and particularly W. ridel) are more specialisedthan N. dicksoni in the reduction of the stylar shelfand the metaconule and protoconule, talon basinand degree of ectoflexus on M3 .

WABULACINUSR/DEI. Features that are moreapomorphic than in N. dicksoni and are syn-apomorphic with Thylacinus are: the sb"aight cen-b"ocrista; the widened angle of the preparacristarelative to the postparacrista, particularly on theMI where this crest is parallel with the an-teroposterior dimension of the tooth; an increasein the size of the angle formed by crests at theparacone and metacone, thereby increasing over -all tooth length; further reduction in size of thestylar cusps than that seen in either Ngoti~ulvaneyi and N. dicksoni ; reduction in sizeof the entoconid; reduction of the metaconid to aminute cusp; reduction in size of the talonid basinby the more lingual position of the hypoconid;and reduction in size of the talon basin.

Wabulacinus ridei exhibi ts some au-tapomorphies not seen in any other thylacinid,some of which are considered specialisations be-yond that of To cynocephalus. The preparacristaon MI of W. ridei is parallel with the tooth rowand the centrocrista. The preparacfistae on N .dicksoni and Ng. timmulvaneyi show theplesiomorphic state similar to most dasyurids inwhich it lies almost perpendicular to the tooth rowand forms almost a 90° angle with respect to thepostparacrista. The morphocline otherwiseshown in thylacines from N. dicksoni through toT. cynocephalus is a widening of the angle atwhich these crests meet (Fig. 5). This elongatesthe tooth in an anteroposterior direction and pro.

TABLE I. Characters and states among thylacines.

I. Infraorbital foramen: 0. not bound by jugal. I. boundby jugal.2. fentrocrista. I. angled. 2. straight (as indicated byM ). 13. Preparacrista on M .I. angled almost perpendicularto the tooth row axis. 2. wider angle than state I. 3.straight.4. Angle of crests at paracone and metacone. I. widerthan on plesiomorphic dasyurids. 2. further widened.s. Entoconid. I. small. 2. minute. a. either absent orposteriorly positioned and combined with thehypoconulid.6. Hypoconulid size. 0. large. I. reduced. 2. minute.7. Stylar shelf size. I. crests and cusps present butreduced compared to plesioMorphic dasyurids. 2. re-duction in size of some cusps and crests13. further lossof cusps and crests (mostly absent on M ).4. completeloss on crests, only a single small cusp present on theposterior of the crown. 18. Ant('riorcingulum. 0. completeonM .1. incompleteonM .9. Metaconule and protoconule. 0. present and large. I.present and reduced. 2. further reduced or absent.10. Metaconid size. I. reduced compared toplesiomorphic dasyurids. 2. small. 3. absent but reten-sion of crest arrangement in posterior molars. 4. com-plete absence of cusp and associated crests.II. Talonid basin size. 0. unreduced. I. reduced bylingual placement of hypoconid. 2. further reduction.12. Talon size. I. reduction of talon width compared toplesiomorphic dasyurids with associated lengthening.2. loss of metaconid and further widening of the crests13. Diastemata and size of M4. 0. no diastemata inpremolar region, M4 shorter thanM3. I. diastemata andM4 equal in length to M3. 2. diastemata and M4 islonger than M3.

apomorphy with No dick.5oni that is not alsoshared with Thylacinus; to include either in Thy-lacinus would expand it beyond any otherdasyuromorphian genus. Wabulacinus rideishows character conflict in the plesiomorphicnature of the infraorbital foramen which is moreplesiomorphic than in No dicksoni and Ngoti~ulvaneyi. This character may have under-gone reversal in Wo ridei.

The single most parsimonious tree of thy lacinidrelationships was found using an ExhaustiveSearch PAUP 3.1 (Swofford, 1993) with 13 or-dered characters (Tables 1& 2) usingpleisomorphic dasyurids as the outgroup. Eachtaxon represents the sister species of all thyla-cines immediately to its right. In general, the

TWO NEW EARL Y MIOCENE THYLACINES FROM RIVERSLEIGH 373

macknessi but is similar to the condition in 7:cynocephalus.

The anterior cingulum on Mi of W. ridei isreduced compared to that of N. dicksoni (it isunknown in Ng. timmulvaney/). In W ridei it isincomplete while in N. dicksoni it continues lin-gually to join the talon basin. This feature is moreplesiomorphic than in T. cynocephalus where thecingulum is lost, but more specialised than in T.macknessi where a complete cingulum is re-tained. In addition, the anterior portion of Mi ofW. ridei is unique in that the anterior root liesmuch further forward under the crown than inother thylacines.

Another trend in thylacines is for the entoconidto become reduced. Only in W. ridei is this cuspcompletely lost.

Wabulacinus ridei is autapomorphic within thefamily in having an enlarged hypoconulid. Inother thylacines the hypoconulid shows reduction(e.g., in N. dicksoni, T. cynocephalus) and mayalso move posteriorly (e.g., in T. macknessl). Thisenlarged cusp in W. ridei may be compensate forloss of the entoconid, or alternatively, it mayrepresent a combination of the hypoconulid anda more posteriorly placed entoconid.

A feature previously used to distinguish thyla-cines from dasyurids is the posterior position ofthe infraorbital foramen posteriorly delimited bythe jugal (Muirhead & Archer, 1990). It is knownin T. cynocephalus, T. potens, Ng. timmulvaneyi(Fig. 3A) and N. dicksoni. Wabulacinus ridei hasthe infraorbital foramen anterior to Mi and welldistant from the jugal (Fig. lB). This position issimilar to dasyurids in which it most frequentlyoccurs above MI/M2 (e.g., in Dasyurus and An-techinus). The anterior position of this foramenin these dasyurids indicates that posterior place-ment near the jugal in most thylacines isapomorphic (Archer, 1976). The anterior place-ment of the jugal in W. ridei is thereforeplesiomorphic relative to other thylacinids.

Wabulacinus ridei is plesiomorphic in manyrespects to Thylacinus excluding it from Thy-lacinus. W. ridei has a number of features uniqueamong thylacinids placing it outside the range ofvariation within Thylacinus.

FIG. 4. Cladogram of thylacines showing characterstate changes. Cladogram is the single most parsimo-nious tree of 32 steps (CI = 0.906, HI = 0.094, RI =0.917, RC = 0.831 ). Striped box = unknown state ofeitherplesiomorphic or highly derived. For charactersand their distribution see Tables I & 2.

duces an enlarged longitudinal blade formedfrom the postmetacrista, centrocrista and pre-paracrista. Only on MI of w. ridei does the pre-paracrista lie parallel to the tooth row, a conditionmore derived than that in any other thylacine. Thetalon basin on the MI of w. ridei is also morederived in its degree of reduction than that of 7:

NGAMAIACINUS TIMMULVANEYl. This spe-cies shares with W. ridei and Thylacinus theapomorphic reduction in the stylar shelf com-pared to N. dicksoni (Fig. 4). This includes reduc-tion in size of St D of M2. On QMF16855. St Dis further reduced and replaced by a number ofminute cusps that border the stvlar shelf. On other

374 MEMOIRS OF THE QUEENSLAND MUSEUM

molars, size of the stylar shelf is comparable tothat in N. dicksoni.

The protoconules and metaconules of w. rideiare apomorphically reduced compared to those ofN. dicksoni. The talon basin is also slightly morereduced than that of N. dicksoni. This speciesfurther differs from N. dicksoni in the less ex-treme ectoflexus, an apomorphic feature. Thesespecialisations of Ng. timmulvaneyi compared toN. dicksoni are less marked than the degree ofspecialisation of these same features in w. rideiand Thylacinus. Ngamalacinus timmulvaneyi andN. dicksoni share several plesiomorphies and, interms of overall similarity, Ng. timmulvaneyi ismuch closer to N. dicksoni than to any otherthylacinid (Fig. 4). These two species do not shareany apomorphy not also found in other thyla-cines.

P ALAEOECOLOGY OF RIVERSLEIGH

THYLACINIDS

number of raptorial birds (Boles, pers. comm.;Archer et al. , 1994). Hence it is probable that therelatively high diversity of Riversleigh thylacinesreflects an overall higher biotic diversity in therainforests of the Riversleigh region.

Camel Sputum is the only Riversleigh site tohave produced more than one thylacine: Ng.timmu/vaneyi and W. ridei. These taxa are verysimilar in size. The maxilla of Ng. timmu/vaneyirecovered from Camel Sputum Site differs fromthe maxilla of w. ridei in the position of theinfraorbital foramen (in Ng. timmu/vaneyi it typ-ically lies above M2 and was probably boundedby the jugal while in W. ridei it lies anterior toMi) and the more plesiomorphic structure of themolars in Ng. timmu/vaneyi. These differencescannot be accounted for by intraspecific variationand the specimens clearly represent two differentspecies.11 is not clear how many of Riversleigh ' s thyla-

cines co-existed. While 2 are present in the CamelSputum assemblage, the more generalised N.dicksani may have been present throughout theOligo-Miocene (Systems A to C; Muirhead &Archer, 1990). Thy/acinus macknessi in SystemsBand C (Muirhead, 1992) suggests that by theearly to middle Miocene, all 4 genera co-existedat Riversleigh. By late Miocene Alcoola time,one lineage is known: T. patens.

In late Cainozoic deposits from other areas ofAustralia (i.e., cave assemblages in eastern,southern and western Australia; Ride, 1964;Archer, 1974, 1982; Dawson, 1982), thylacinidremains are common. Sites where thylacinid re-mains are abundant (e.g., Thylacine Hole on theWestern Australian Nullarbor; Lowry, 1972)may be interpreted to represent lairs or trapswhere carnivores were preferentially attracted,perhaps by the presence of other animals. In theRiversleigh deposits, most of which appear tohave accumulated in shallow pools withinrainforest environments (Archer et al., 1989;Archer et al., 1994), thylacinid remains are rela-tively rare and therefore may more fairly repre-sent natural frequencies.

Thylacinids described from the Riversleigh as-semblages are N. dicksoni, Ng. timmulvaneyi, W.ridei and T. macknessi. This diversity raises ques-tions about niche diversification. Although onlyone thylacine appears to have been present at anyone time in late Miocene (Alcoota, T. potens),Pliocene (Awe & Chinchilla, T. cynocephalus)and Quaternary (many assemblages, T. cyn-ocephalus) local faunas of Australia and NewGuinea (Archer, 1982; Dawson, 1982), prior tothe late Miocene, available resources enabled the'thylacine niche' to be more finely divided. Partof the explanation may be found in the apparentabsence from the Riversleigh local faunas of anylarge dasyurids as specialised for carni vory as thelate Cainozoic species of Glaucodon, Sarco-philus and Dasyurus. Presence of large carnivo-rous dasyurids appears inversely correlated withthylacinid diversity. The subsequent rise of thesedasyurines may, therefore, have accompaniedlate Miocene decline in thylacinid diversity.

Although there is a greater diversity of thyla-cines in the Oligo-Miocene Riversleigh depositsthan later, a wider range of large carnivores wasalso present in these Riversleigh local faunas. Forexample in single local faunas, there were often3 crocodilians (P. Willis, pers. comm.), at least 2large snakes (madtsoiids and pythonids; J.Scanlon, pers. comm.), 2 lineages of thy-lacoleonids (Wakaleo and a genus similar toPriscileo; Archer et al., 1989), a possibly carniv-orous kangaroo (Archer & Flannery, 1985; Wroe& Archer, 1995; Wroe, 1996) and an unknown

THYLACINID DIAGNOSIS ANDMORPHOLOGICALTRENDS

Thylacinids differ from dasyurids and otherpolyprotodont marsupials by having, in combina-tion, the following features. The premetacristaand postparacrista join as a centrocrista. Theangle formed by these crests is straight or almoststraight in occlusal view in at least M2 of the

TWO NEW EARL Y MIOCENE THYLACINES FROM RIVERSLEIGH 375

~ Dl 0 -

El ~(~gG

FIG. 5. Upper and lower dentitions of all known thylacine genera showing crest orientation compared to adasyufid. A=Dasyurus. 8=Nimbacinus dicksoni (P8695-92) C=Ngamalacinus timmulvaneyi. D=Wabulacinusfidei. El =Thylacinus macknessi E=Thylacinus cynocephalus. Upper dentitions include p3 where known. Scales= O.5mm.

upper dentition. The cristid obliqua continues upthe posterior flank of the protoconid from thetalonid region rather than terminating at the baseof the protoconid. This functions in elongatingthis crest and becomes more prominent as themetaconid is reduced (e.g., in Thy/acinus). Thestylar cusps are reduced. This occurs most prom-inently on M3 but also occurs to varying degreeson more anterior molars. The size of themetaconid is reduced on all lower molars. Thisreduction is correlated with the more posteriorplacement of the metaconid relative to the pro-toconid, functioning in widening the angle ofcrests at the protoconid and enlarging the trigonidbasin. Reduction of the metaconid is also foundto progress in degree from the more reducedcondition on anterior molars to posterior molars(Muirhead & Gillespie, 1995). The size of thetalonid basin is reduced because of the morelingual position of the hypoconid. This cusp oc-

cupies much of the surface of the talonid basinsuch that no flat surfaces occur on the basin floor.

Structural morphoclines of the family (appar-ent in more specialised forms) include the follow-ing. There is an increase in the angles formed bycrests of the paracone and metacone, increasingthe length of the postmetacrista. More antero-posterior orientation of the preparacrista. The lossof extreme ectoflexus particularly in M3 (relatedto the overall elongation of the teeth). A reductionin size of the protoconule and metaconule as wellas the entire talon basin and reduction in size ofthe stylar shelf. All of these features of the upperdentition increase the anteroposterior length ofthe molars with the entire tooth row acting as asystem of anteroposteriorly orientated blades(Fig. 5). These are typical specialisations in mam-malian carnivores.

In the lower molars, the trends are for completeloss of the metaconid and opening of the trigonid

376 MEMOIRS OF THE QUEENSLAND MUSEUM

LITERA TURE CITEDbasin. Here, like the upper molars, the lowermolar crests become orientated anteroposteriorly(Fig. 5). The paracristid becomes the anteriorcrest with the elongated cristid obliqua function-ing as the posterior crest (the postprotocrista)(Muirhead and Gillespie, 1995). The lingual sideof the talonid is also reduced through reductionof the entoconid.

Only in Thylacinus is the snout elongated byboth diastemata between the canine and premo-lars and elongation of M4 (such that it is longerthan preceding molars). Extreme posterior place-ment of the infraorbital foramen and partial en-closure by the jugal is also a possiblesynapomorphy of Thylacinus related to snoutelongation.

All thylacinids plesiomorphically retain theparaconid on MI, remnants of posterior and ante-rior cingula on the lower molars and posteriorincrease in size from PI to P3.

Variation among thylacinids that does not ap-pear to follow these 'carnivorous trends' includesposition of the infraorbital foramen which,plesiomorphically and unlike all other knownthylacinids, is more anteriorly positioned in w.ridei above p3.

ACKNOWLEDGEMENTS

The study was undertaken with support fromthe Queen Elizabeth II Silver Jubilee Trust ForYoung Australians and the Australian common-wealth Department of Employment, Educationand Training.

Material on which this study is based was dueto support from: the Australian Research GrantScheme; the University of New South Wales; theNational Estate Grants Scheme, Queensland; theDepartment of Arts, Sport, the Environment,Tourism and Territories; Wang Australia; ICIAustralia; the Queensland Museum; the Austra-lian Museum; the University of New SouthWales; the Australian Geographic Society; MIM;Ansett Wridgways; and Surrey Beatty and Sons.

I am grateful for the advice and suggestions ofreferees Ken Aplin and Mike Archer and for theassistance given by the Australian Museum (inparticular Linda Gibson), the Northern TerritoryMuseum (in particular Dr Peter Murray), RobynMurphy, Anna Gillespie, Henk Godthelp (Uni-versity of New South Wales).

ARCHER,M. 1974. New Information about the Qua-ternary distribution of the thylacine (Marsupialia,Thylacinidae) in Australia. Royal Society ofWestern Australia 57: 43-50.

1976. The dasyurid dentition and its relationship tothat of didelphids, thylacinids, borhyaenids(Marsupicarnivora) and peramelids (Peramelina:Marsupialia). Australian Journal of Zoology,Supplementary Series 39: 1-34.

1982. A review of Miocene thylacinids (Thy-lacinidae, Marsupialia), the phylogenetic posi-tion of the Thylacinidae and the problems ofapriorisms in character analysis. Pp 445-476. InArcher, M. (ed.), Carnivorous marsupials. (RoyalZoological Society of New South Wales: Syd-ney).

ARCHER, M. & FLANNERY,T.F. 1985. Revision ofthe extinct gigantic rat kangaroos (Potoroidae:Marsupialia). With description of a new Miocenegenus and species, and a new Pleistocene speciesof Propleopus. Journal of Paleontology 59: 1331-1349.

ARCHER, M., HAND, S.J. & GODTHELP, H. 1994.Riversleigh. Second edition. (Reed Books: Syd-ney).

ARCHER, M., GODTHELP, H., HAND, S.J. &MEGIRIAN, D. 1989. Fossil mammals ofRiversleigh northwestern Queensland: prelimi-nary overview ofbiostratigraphy, correlation andenvironmental change. The Australian Zoologist25: 29-65.

ARCHER, M., HAND, S.J. & GODTHELP, H. 1995.Tertiary environmental and biotic change in Aus-tralia. Pp 77-90. In Vrba, E.S., Denton, G.H.,Partiridge, T. C. & Burckle, L.H. (eds), Paleocli-mate and evolution, with emphasis on humanorigins. (Yale University Press: New Haven).

BENSLEY, B.A. 1903. On the evolution of the Austra-lian Marsupialia with remarks on the relationshipsof marsupials in general. Transactions of theLinnean Society, London (Zoology) 9: 83-217.

DA WSON, L. 1982. Taxonomic status of fossil thyla-cines (Thylacinus, Thylacinidae, Marsupialia)from late Quaternary deposits in eastern Australia.Pp 527-536. In Archer, M. (ed.), Carnivorousmarsupials. (Royal Zoological Society of NewSouth Wales: Sydney).

FLOWER, W .H. 1869. Remarks on the homologies andnotation of the teeth in the Marsupialia. Journal ofAnatomy and Physiology 3: 262-278.

HENNIG, W. 1965. Phylogenetic systematics. AnnualReview of Entomology 10: 97-116.

LOWRY, J.W.J. 1977. The taxonomic status of smallfossil thylacines (Marsupialia, Thylacinidae) fromWestern Australia. Journal and Proceedings of theRoyal Society of Western Australia 55: 19-29.

LUCKETT, W.P. 1993. An ontogenetic assessment ofdental homologies in therian mammals. Pp 182-204. In Szalay, F.S., Novacek, M.J. & McKenna,

TWO NEW EARL y MIOCENE THYLACINES FROM RIVERSLEIGH 377

APPEND IXAll measurements are actual distance between

cusps except those with '(horiz)' for which mea-surements were made from a horizontal planeabove the cusps (occlusal view).

Wabulacinus ridei dentition (~~

Mi M2

M.1-

I~~~

QMF16851

OMF16852

Para-meta

Meta-proto

Proto-para

Anterior/width

Buccal/length

Posterolingual/uppers

para-meta Qloriz)

meta-proto Qloriz)

proto-para Qloriz)

hypo-hypocooulid

~ento

~-1,Q!J

1..1Ml ~

/!,!rJ

---

N amalacinus timmul

QMF30300

-..9MF16855 I M2 I I M33.641 3.6613.341 ; --

1.1

2.8613.3613.6515.(

M.C. (008), Mammal phylogeny. (Springer-Ver-lag, New York).

MUIRHEAD, J. 1992. A specialised thylacinid, Thy-lacinus macknessi, (Marsupialia: Thylacinidae)from Miocene deposits of Riversleigh, northwest-ern Queensland. Australian Mammalogy 15: 67-76.

MUIRHEAD, J. & ARCHER, M. 1990. Nimbacinusdicksoni, a plesiomorphic thylacinid(Marsupialia: Thylacinidae) from Tertiary depos-its of Queensland and the Northern Territory.Memoirs of the Queensland Museum 28: 203-21.

MUIRHEAD, J. & GILLESPIE, A. K. 1995. Additionalparts of the type specimen of Thylacinusmacknessi (Marsupialia: Thylacinidae) fromMiocene deposits of Riversleigh, NorthwesternQueensland. Australian Mammalogy 18: 55-60.

RIDE, W .D.L. 1964. A review of Australian fossilmarsupials. Proceedings of the Royal Society ofWestern Australia 47: 97-131.

SWOFFORD, D.L. 1993. P AUP (phylogenetic analysisusing parsimony). Documentation for Version3.1. (Illinois Natural History Survey: Campaign).

WOODBURNE, M.O. 1967. The Alcoota Fauna, cen-tral Australia: an integrated palaeontological andgeological study. Bulletin, Bureau of Mineral Re-sources, Geology and Geophysics, Australia 87:1-187.

WROE, S. 1996. An investigation of phylogeny in thegiant extinct rat kangaroo Ekaltadeta (Pro-pleopinae, Potoroidae, Marsupialia).Joumal ofPaleontology 70: 681-690.

WROE, S. & ARCHER, M. 1995. Extraordinarydiphyodonty-related change in dental function fora tooth of the extinct marsupial Ekaltadeta ima(Propleopinae, Hypsiprymnodontidae ). Archivesof Oral Biology 40: 597 -603.

Para-rneta

Meta- roto

Proto- ra

~i lower dentition (~-

!!J:L-3.56

M2 M3

N amalaciJUlS timmul

MF16853

Para-meta

Meta- roto

Proto- ra

~mto

3.34

~3.62

~~

M4

)3-

: X>

3.77 5.14 5.02 5.52

2.25 4.26 4.90 7.68 7.20 8.25

5.56 7.92 8.11 9.10 8.76 8.17

8.15 10.29 ' --27._-::

~.3U 10.

3.30 3.42

2.28 3.05

3.34 3.25