This article was downloaded by: [UNSW Library]On: 05 March 2014, At: 23:31Publisher: Taylor & FrancisInforma Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House,37-41 Mortimer Street, London W1T 3JH, UK

Journal of Vertebrate PaleontologyPublication details, including instructions for authors and subscription information:http://www.tandfonline.com/loi/ujvp20

Earliest modern bandicoot and bilby (Marsupialia,Peramelidae and Thylacomyidae) from the Mioceneof the Riversleigh World Heritage Area, northwesternQueensland, AustraliaK. J. Travouillon a b , S. J. Hand b , M. Archer b & K. H. Black ba School of Earth Sciences , University of Queensland , St. Lucia, Queensland , 4072 ,Australiab School of Biological, Earth and Environmental Sciences , University of New South Wales ,New South Wales , 2052 , AustraliaPublished online: 04 Mar 2014.

To cite this article: K. J. Travouillon , S. J. Hand , M. Archer & K. H. Black (2014) Earliest modern bandicoot and bilby(Marsupialia, Peramelidae and Thylacomyidae) from the Miocene of the Riversleigh World Heritage Area, northwesternQueensland, Australia, Journal of Vertebrate Paleontology, 34:2, 375-382, DOI: 10.1080/02724634.2013.799071

To link to this article: http://dx.doi.org/10.1080/02724634.2013.799071

PLEASE SCROLL DOWN FOR ARTICLE

Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”) containedin the publications on our platform. However, Taylor & Francis, our agents, and our licensors make norepresentations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of theContent. Any opinions and views expressed in this publication are the opinions and views of the authors, andare not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon andshould be independently verified with primary sources of information. Taylor and Francis shall not be liable forany losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoeveror howsoever caused arising directly or indirectly in connection with, in relation to or arising out of the use ofthe Content.

This article may be used for research, teaching, and private study purposes. Any substantial or systematicreproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in anyform to anyone is expressly forbidden. Terms & Conditions of access and use can be found at http://www.tandfonline.com/page/terms-and-conditions

Journal of Vertebrate Paleontology 34(2):375–382, March 2014© 2014 by the Society of Vertebrate Paleontology

ARTICLE

EARLIEST MODERN BANDICOOT AND BILBY (MARSUPIALIA, PERAMELIDAE,AND THYLACOMYIDAE) FROM THE MIOCENE OF THE RIVERSLEIGH WORLD

HERITAGE AREA, NORTHWESTERN QUEENSLAND, AUSTRALIA

K. J. TRAVOUILLON,*,1,2 S. J. HAND,2 M. ARCHER,2 and K. H. BLACK2

1School of Earth Sciences, University of Queensland, St. Lucia, Queensland 4072, Australia, k.travouillon@uq.edu.au;2School of Biological, Earth and Environmental Sciences, University of New South Wales, New South Wales 2052, Australia

ABSTRACT—Recent molecular phylogenies of peramelemorphians suggest that thylacomyids (bilbies) and peramelids(modern bandicoots) diversified sometime in the late Oligocene or early Miocene. Until now, however, the earliest fossilevidence of thylacomyids and peramelids was from the Australian Pliocene. Here we describe the oldest peramelid and thyla-comyid from the middle Miocene of the Riversleigh World Heritage Area, northwestern Queensland. The peramelid, Crashbandicoot, gen. et sp. nov., is represented by a single maxilla containing M1–3 that exhibits peramelid synapomorphies, in-cluding development of a metaconular hypocone, an incomplete centrocrista, and well-developed anterior cingulum. Thethylacomyid, Liyamayi dayi, gen. et sp. nov., is represented by M2 and m1, which show thylacomyid synapomorphies includ-ing a conical entoconid, a conical stylar cusp B (StB) and StD, and reduced distance between the metastyle and StD. Theresults of our phylogenetic analysis indicate that both species are part of crown-group Peramelemorphia.

SUPPLEMENTAL DATA—Supplemental materials are available for this article for free at www.tandfonline.com/UJVP

INTRODUCTION

Peramelemorphians (bandicoots and bilbies) are small- tomedium-sized omnivorous, terrestrial marsupials present indiverse modern habitats from rainforest to deserts in Australia,New Guinea, and surrounding islands. Modern bilbies arerepresented by two (one extant but vulnerable and one recentlyextinct) desert-adapted (but not -restricted) species, whereasmodern bandicoots are far more speciose and ecologicallydiverse. They are characterized by having syndactylous hindlimbs (a derived feature shared with diprotodontians) andpolyprotodonty (a plesiomorphic condition shared with dasyuro-morphians, notoryctemorphians, and extant ameridelphids). Onmorphological grounds, all bandicoots have traditionally beenregarded to be members of the Peramelidae; however, Archerand Kirsch (1977) argued that the bilbies (species of Macrotis andIschnodon) merited being treated as a family distinct from per-amelids and peroryctids. Groves and Flannery (1990) proposedthat the eight modern genera constituted just two families: Per-oryctidae (species of Peroryctes, Microperoryctes, Rhynchomeles,and Echymipera) and Peramelidae (species of Macrotis, Ischn-odon, Chaeropus, Perameles, and Isoodon). This latter taxonomicarrangement was supported by the morphological analysis ofTravouillon et al. (2010), which included Miocene taxa fromthe Riversleigh World Heritage Area (WHA) of northwesternQueensland (Fig. 1) described by Muirhead (2000) as membersof the Yaralidae, an extinct family of peramelemorphians.

In contrast, the latest molecular phylogeny of peramele-morphians (Westerman et al., 2012), based on nuclear andmitochondrial DNA sequences, recognizes three families: Thy-lacomyidae, containing species of Macrotis; Chaeropodidae,containing Chaeropus ecaudatus; and Peramelidae, containing allother living bandicoots. Within Peramelidae, Westerman et al.(2012) recognize three subfamilies: Peroryctinae (species of Per-

*Corresponding author.

oryctes), Echymiperinae (species of Echymipera, Rhynchomeles,Microperoryctes), and Peramelinae (species of Isoodon, Perame-les). Westerman et al. (2012) estimate divergence times fromrelaxed-clock Bayesian methods and suggest that crown-groupbandicoots originated in the late Oligocene or early Miocene,much earlier than currently suggested by the fossil record.The earliest known fossil peramelids are Perameles bowensisfrom the Bow Local Fauna (LF; see Fig. 1), Big Sink LF, andChinchilla LF (Muirhead et al., 1997; Mackness et al., 2000) andPerameles allinghamensis from the Bluff Downs LF (Archer andWade, 1976) are Pliocene in age and as such at least 15 millionyears younger than the estimated antiquity for crown-groupbandicoots (Westerman et al., 2012). The earliest thylacomyid,Ischnodon australis from the Palankarinna LF (Fig. 1), SouthAustralia (Stirton, 1955), is also Pliocene in age.

Here we describe two new peramelemorphians from the mid-dle Miocene of the Riversleigh World Heritage Area that appearto represent the earliest known representatives of the Perameli-dae and Thylacomyidae.

MATERIALS AND METHODS

Collecting and Processing

Specimens described here were collected from the RiversleighWorld Heritage Area, northwestern Queensland, by Archer et al.between 1983 and 2011, and are registered in the fossil collec-tion of the Queensland Museum (QM F), Brisbane, Australia.All specimens were obtained by acid-processing freshwater lime-stones recovered in the field.

Anatomical Nomenclature

Dental terminology follows Turnbull et al. (2003) and Luckett(1993) for the homology of molariform teeth. Peramelemorphiansystematics follows Muirhead (1994, 2000), Muirhead and Filan

375

Dow

nloa

ded

by [

UN

SW L

ibra

ry]

at 2

3:31

05

Mar

ch 2

014

376 JOURNAL OF VERTEBRATE PALEONTOLOGY, VOL. 34, NO. 2, 2014

FIGURE 1. Map of the location of referred Australian fossil deposits.Abbreviations: LF, Local Fauna; WHA, World Heritage Area.

(1995), Groves (2005), and Travouillon et al. (2010). Higher-levelmarsupial systematics follows Aplin and Archer (1987).

Body Mass Estimates

Tooth lengths and widths were measured for each specimen(Table 1). These were then used to estimate individual body massusing Myers’ (2001) regression equations, which correlate dentalvariables with marsupial body mass. We used the highest pos-sible ranked equation from the ‘all species excluding dasyuro-morphians’ data set from Myers (2001:table 3) as follows: equa-tion ‘3UMA’ (3rd upper molar area) for QM F56245; equation‘2UMA’ (2nd upper molar area) and equation ‘1LMA’ (1st lowermolar area) for QM F56250.

Phylogenetic Analysis

We used the data matrix by Travouillon et al. (2013) andGurovich et al. (2013), revised from Travouillon et al.’s (2010)matrix, to assess the phylogenetic relationships within Peramele-morphia of the two species described here. The matrix comprises42 taxa and 156 qualitative morphological characters, of which33 are cranial and 123 dental (see Supplementary Data 1 and 2).Several characters from Travouillon et al.’s (2010) matrix weredeleted (to be replaced by more comprehensive characters) andsome modified (see Supplementary Data 1). Because they rep-resent putative morphoclines, 71 characters were ordered in all

analyses (see Supplementary Data 1). The matrix was completelyrescored to avoid repeating mistakes that may have been presentin earlier iterations of this matrix. The following fossil bandicootswere added to the analysis: cf. Peroryctes tedfordi, cf. P. sp.,Perameles sobbei, P. bowensis, P. allinghamensis, Ischnodonaustralis, Yarala burchfieldi, Y. kida, Galadi speciosus, G. gran-dis, G. amplus, G. adversus, and Bulungu palara. These specieswere omitted from previous studies because they contained toomuch missing data and some of them weren’t yet described.Microperoryctes ornata was removed from the matrix becausespecimens used to score this taxon previously have subsequentlybeen reassigned by Australian Museum staff to Microperorycteslongicauda (Sandy Ingleby, pers. comm.). In addition to the fossildasyuromorphians Barinya wangala and Mutpuracinus archibaldiand the early Eocene stem australidelphan Djarthia murgonen-sis, five extant outgroup species from the family Dasyuri-dae were added: Dasyurus hallucatus, Dasyuroides byrnei,Phascogale tapoatafa, Antechinus stuartii, and Sminthopsismacroura.

Parsimony Analysis

A parsimony analysis was performed using PAUP∗ 4.0b10(Swofford, 2002) using a two-stage search strategy followingWorthy et al. (2006) and Beck et al. (2008). An initial search com-prising 1000 heuristic replicates, saving 10 trees per replicate, wasfollowed by a second heuristic search within the saved trees. Mul-tiple most parsimonious trees produced were summarized usingstrict consensus. Bootstrap values for each node were calculatedusing 1000 bootstrap replicates of 10 random addition sequencereplicates. Decay indices were also calculated using TreeRot.v3(Sorenson and Franzosa, 2007).

A second analysis was performed repeating the same steps butwith a ‘molecular scaffold’ as a ‘backbone’ constraint. For thebackbone, relationships among extant peramelemorphians wereconstrained to reflect the molecular phylogeny of Westermanet al. (2012). Only taxa analyzed by Westerman et al. (2012)were included in the molecular scaffold, and only those cladesthat received ≥70% partitioned maximum likelihood bootstrapsupport and ≥0.95 partitioned Bayesian posterior probability inthe analyses of Westerman et al. (2012:table 2) were enforced asmonophyletic. The five extant dasyurid outgroup taxa were alsoincluded in the molecular scaffold, with their relationships con-strained to match those recovered by Krajewski et al. (2007:fig.5) and Westerman et al. (2008:fig. 2). The molecular scaffoldtopology is included in Supplementary Data 3.

SYSTEMATIC PALEONTOLOGY

Order PERAMELEMORPHIA (Kirsch, 1968)Aplin and Archer, 1987

Family PERAMELIDAE (Gray, 1825) (sensu Groves, 2005)

TABLE 1. Measurements (in mm) of the upper dentition of type and referred material of Crash bandicoot, gen. et sp. nov., and Liyamayi dayi, gen.et sp. nov.

M1 M2 M3

Taxa Specimen Locality L W L W L W Body mass estimate (g)

Crash bandicoot QM F56245 AL90 Site 3.97 2.84 3.81 3.36 3.4 4.00 1076.24Liyamayi dayi QM F56246 Rick’s Sausage Site 3.46 3.63 966.5Liyamayi dayi QM F56247 Rick’s Sausage Site 3.22 3.1 674.18

m1

L AW PWLiyamayi dayi QM F56250 Rick’s Sausage Site 3.34 1.89 2.26 879.45

Abbreviations: AW, maximum width of trigonid; L, maximum anteroposterior length; M, upper molar; m, lower molar; PW, maximum width oftalonid; W, maximum lingual-buccal width.

Dow

nloa

ded

by [

UN

SW L

ibra

ry]

at 2

3:31

05

Mar

ch 2

014

TRAVOUILLON ET AL.—EARLIEST PERAMELID AND THYLACOMYID BANDICOOTS 377

CRASH BANDICOOT, gen. et sp. nov.(Fig. 2)

Type Species—Crash bandicoot by monotypy.Generic and Specific Diagnosis—Crash species differ from all

other bandicoots in the following combination of features: poste-rior cingulum present on M1 but not M2 or 3; centrocrista incom-plete on M1–3; small bridge connects parastylar shelf to metasty-lar shelf on M1 or 2 but not on M3; metaconular hypocone largeand distinct on M1–3; anterior cingulum complete on M1–3; StBconical on M2 and 3; StD conical on M3; StD1 present on M2; StCpresent on M3; preparacrista connects to StA on M2 and 3; StCpresent as a distinct cusp on M2 only; and stylar crest connectsmetastyle to StD on M1 and 2 only.

Etymology—Named after Crash Bandicoot, the popular videogame character created by Andy Gavin and Jason Rubin, with arelated inference that this was the start of a new radiation of moremodern bandicoots that ‘crashed’ through to dominate younger,drier ecosystems of Australia. The genus is here assigned mascu-line gender.

Holotype—QM F56245, right maxilla containing M1–3 and thealveoli for M4 (Fig. 2).

Type Locality and Age—Alan’s Ledge 1990 Site (AL90),Riversleigh World Heritage Area, northwestern Queensland,

Australia. AL90 Site is interpreted to be part of Riversleigh’sFaunal Zone C and as such middle Miocene in age (Archer et al.,1997; Arena, 2004; Travouillon et al., 2006, 2011).

Description

The M1 is longer than wide (Fig. 2). Stylar cusp A (StA) isa low cusp on the anterobuccal corner of the tooth. A crestruns from the tip of StA posteriorly and ends at the anteriorflank of StC. StC (Fig. 2B) is on the parastylar region of M1,directly posterior to StA. StB, if present, is indistinguishablefrom StC. The preparacrista is curved and connects StC to theparacone anterolingually. The postparacrista runs from theparacone posterobuccally, almost parallel to the preparacrista,and ends at the base of the posterolingual flank of StC. Itdoes not form a centrocrista with the premetacrista. Instead,a small bridge connects the parastylar region to the metastylarregion at the ectoloph and probably represents a remnant ofthe centrocrista, a condition also seen in Peroryctes broadbenti(see Aplin et al., 2010). The metacone is situated posterolingualto the paracone. The metacone is the tallest cusp on the crown,followed in decreasing order by StD, paracone, StC, metastyle,StA, protocone, and metaconular hypocone. The premetacrista,which is longer than the postparacrista, runs from the metacone

FIGURE 2. Crash bandicoot, gen. et sp. nov., right maxilla with M1–3 (QM F56245). A, occlusal view; B, lingual view. Abbreviations: ac, anteriorcingulum; ect, ectoloph; mcl, metaconular hypocone; me, metacone; mst, metastyle; M1–M3, upper molar 1 to upper molar 3; pa, paracone; pc,posterior cingulum; pmc, premetacrista; pomc, postmetacrista; popc, postparacrista; ppc, preparacrista; pr, protocone; St A, St B, St C, St D1, St D, StE, Stylar cusps A, B, C, D1, D, and E. Scale bar equals 3 mm.

Dow

nloa

ded

by [

UN

SW L

ibra

ry]

at 2

3:31

05

Mar

ch 2

014

378 JOURNAL OF VERTEBRATE PALEONTOLOGY, VOL. 34, NO. 2, 2014

anterolingually and ends at the base of the anterolingual flankof StD. The postmetacrista, the longest crest on the crown,connects the metacone to metastyle posterobuccally. A stylarcrest connects the metastyle to the tip of StD anteriorly (Fig.2B). StE, if present, is not distinguishable from this stylar crest.StD is a large round cusp directly buccal to the metacone. StD1is absent. The protocone is posterolingual to the paracone.The preprotocrista runs from the protocone anterobuccally andends at the lingual flank of StA, forming a complete anteriorcingulum. The postprotocrista is almost straight and connects theprotocone to the metaconular hypocone posteriorly. This crestthen continues posterobuccally to end below the postmetacrista,past the midpoint between the metacone and metastyle, forminga posterior cingulum. The metaconular hypocone is large and islevel with the protocone lingually.

The morphology of M2 is similar to that of M1 except as fol-lows (see Fig. 2A and B). The crown is much wider, elongatingthe preparacrista, postparacrista, premetacrista, postmetacrista,and preprotocrista. StA is taller, on the parastylar shelf, and con-nects to the paracone via the preparacrista. StC is absent. Instead,StB is a large conical cusp, located posterior to StA. StB remainstaller and bigger than StA. A short crest runs from the tip of StBanterolingually to the base of StB. A small crest is also presenton the posterior side of StB and may represent a remnant of StC.StE is present as a small cusp on the stylar crest midway betweenthe metastyle and StD. StD1 is a miniscule cusp and directly an-terior to StD. A short crest connects StD1 anterolingually to thebridge connecting the parastylar and metastylar regions. The an-terior cingulum is a larger shelf that is widest just lingual to StA.The metaconular hypocone is smaller than in M1 but more cusp-like on the lingual margin of the tooth. The postprotocrista endsat base of the posterior flank of the metacone; hence, the poste-rior cingulum is incomplete.

The morphology of M3 is similar to that of M2 except as fol-lows (Fig. 2). The crown is wider than long, with the protoconeand paracone positioned more lingually. StA is taller, almostlevel with StD. The preparacrista is straighter and longer. StBis smaller, but StC is present as a small cusp directly posterior toStB. The parastylar and metastylar regions are not connected bya bridge because the ectoloph is breached. StE is absent and nostylar crest connects the metastyle to StD. StD is smaller. StD1 isabsent. The metaconular hypocone is smaller.

Measurements of Crash bandicoot are presented in Table 1.

Family THYLACOMYIDAE (Bensley, 1903); sensu Archerand Kirsch (1977)

LIYAMAYI DAYI, gen. et sp. nov.(Fig. 3)

Type Species—Liyamayi dayi by monotypy.Generic and Specific Diagnoses—Liyamayi species differ from

all other peramelemorphians in the following combination of fea-tures: posterior cingulum absent on M2; metaconular hypoconepresent but small and anteriorly positioned; anterior cingulum in-complete on M2; centrocrista incomplete on M2; preparacristaconnects to StA and StB; StB and StD conical but retain sty-lar crests; StD1 and StC absent; distance between metastyle andStD short; remnant of anterior cingulid present on m1; paraconid,metaconid, and protoconid equidistant; small cusp present on hy-poflexid region; entoconid conical; and preentocristid present butobliquely directed.

Generic Etymology—‘Liya’ is from the Waanyi word in theRiversleigh District of northwestern Queensland meaning roundand ‘mayi’ meaning tooth (Breen, 1985), in reference to therounded shape of the teeth and cusps. The genus is here givenmasculine gender.

Specific Etymology—The species name honors geologist andpalaeontologist Dr. Robert Day, who generously supported thisresearch.

Holotype—QM F56246, RM2 (Fig. 3A, B).Paratypes—QM F56247, RM2; QM F56250, Rm1 (Fig. 3C, D).Type Locality and Age—Rick’s Sausage Site (RSS), River-

sleigh World Heritage Area, northwestern Queensland, Aus-tralia. This deposit is probably middle Miocene in age (see Dis-cussion below).

Description

Upper Dentition—The M2 is wider than it is long with thebuccal and lingual sides of the tooth rounded (Fig. 3A, B).The anterior cingulum is present as a short, low shelf lingualto StA. StA is a large cusp at the most anterobuccal corner ofthe tooth. StA connects to the paracone posterolingually via thepreparacrista. The preparacrista is almost straight, curving onlya third of the way to the paracone where another crest connectsthe preparacrista to the tip of the large conical StB posterobuc-cally. This crest continues posterobuccally from the tip of StBand ends before breaching the ectoloph. StC is absent. The para-cone is posterolingual to StA and anterolingual to StB. The post-paracrista runs from the paracone posterobuccally to the base ofthe posterolingual flank of StB. The metacone, which is directlyposterior to the paracone, is the tallest cusp on the crown, fol-lowed in decreasing order by StD, paracone, StB, StA, metastyle,protocone, and metaconular hypocone. The premetacrista runsfrom the tip of the metacone anterobuccally to the anterolin-gual flank of StD. The postmetacrista connects the metacone tothe metastyle posterobuccally. A short stylar crest connects themetastyle to the tip of the large conical StD anteriorly. StE, ifpresent, is indistinguishable from this stylar crest. Just anterior toStD, StD1 is absent, but in its place a short crest runs anterolin-gually and ends before breaching the ectoloph. The protocone isdirectly lingual to the paracone. The preprotocrista is short andconnects the protocone to the anterolingual flank of the paraconeanterobuccally. The postprotocrista runs from the protocone pos-terobuccally, and then curves posterolingually at the anteriorbase of the metaconular hypocone. This crest continues throughthe metaconular hypocone and ends at the posterolingual flankof this cusp. The metaconular hypocone is small and situated an-terolingual to the metacone. There is no posterior cingulum.

Lower Dentition—The anterior cingulid of m1 is highly re-duced, with only a small remnant present just anterior to theparaconid (Fig. 3C, D). The protoconid is the tallest cusp, fol-lowed in decreasing order by the hypoconid, metaconid, en-toconid, paraconid, and hypoconulid. The trigonid is narrowerthan the talonid. The protoconid, paraconid, and metaconid areequidistant. The paraconid is anterobuccal to the metaconid andanterolingual to the protoconid. The hypoconid is posterobuc-cal to the protoconid. The cristid obliqua is a concave crest thatruns from the tip of the hypoconid to the posterobuccal flankof the protoconid. A minute cusp is present on the shelf inthe hypoflexid region. The oblique posthypocristid connects thehypoconid to the hypoconulid. The entoconid is a large conicalcusp directly posterior to the metaconid and directly lingual tothe hypoconid. A short preentocristid runs obliquely towards theprotoconid from the entoconid.

Measurements of Liyamayi dayi are presented in Table 1.

RESULTS

The unconstrained parsimony analysis of our 156 cranio-dental character matrix recovered 21 most parsimonious treesof 835 steps. The strict consensus is illustrated in Figure 4A,with bootstrap values above branches and decay indices belowbranches. Crash bandicoot is recovered within crown-groupPeramelemorphia, unresolved within a clade containing speciesof Isoodon and Perameles, with weak support (bootstrap <50%;decay index +1; Fig. 4A). Liyamayi dayi is also recovered withinthe crown-group Peramelemorphia, sister to a clade containing

Dow

nloa

ded

by [

UN

SW L

ibra

ry]

at 2

3:31

05

Mar

ch 2

014

TRAVOUILLON ET AL.—EARLIEST PERAMELID AND THYLACOMYID BANDICOOTS 379

FIGURE 3. Liyamayi dayi, gen. et sp. nov., right M2 (holotype QM F56246), A, occlusal view; B, Liyamayi dayi, right M2, buccal view; C, Liyamayidayi, right m1 (paratype QM F56250), occlusal view; D, Liyamayi dayi, right m1, buccal view. Abbreviations: ac, anterior cingulum; acd, anteriorcingulid; end, entoconid; hyd, hypoconid; hyld, hypoconulid; mcl, metaconular hypocone; me, metacone; med, metaconid; mst, metastyle; pa, paracone;pad, paraconid; pr, protocone; prd, protoconid; St A, St B, St D, stylar cusps A, B, and D. All scale bars equal 1.5 mm.

thylacomyids (Macrotis + Ischnodon), Chaeropus and Perameles+ Crash + Isoodon (bootstrap <50%; decay index +2; Fig. 4A).

The constrained parsimony analysis, enforcing a molecularscaffold as a ‘backbone’ constraint, recovered 478 most parsi-monious trees of 864 steps. The strict consensus is illustrated inFigure 4B, with bootstrap values above branches. Again, Crashbandicoot is recovered within the crown-group Peramelemor-phia, and unresolved within the same clade as in the uncon-strained analysis. However, the rest of the crown group is fairlyunresolved and fails to identify the relationship between Liya-mayi dayi and other peramelemorphians.

DISCUSSION

Until now, the oldest fossil members of the family Peramel-idae were the Pliocene Perameles bowensis from Big Sink andBow LFs, New South Wales (Muirhead et al., 1997), and Chin-

chilla LF, Queensland (Mackness et al., 2000), and Peramelesallinghamensis from Bluff Downs LF, Queensland (Archer andWade, 1976). The most recent molecular phylogeny of peramele-morphians (Westerman et al., 2012) suggests that peramelids di-versified in the middle Miocene, and estimated that the split be-tween the three peramelid subfamilies Peramelinae (species ofIsoodon and Perameles), Peroryctinae (species of Peroryctes),and Echymiperinae (species of Echymipera, Rhynchomeles andMicroperoryctes) occurred at ∼13.8 million years ago. Crashbandicoot (Fig. 2), from AL90 Site in the Riversleigh World Her-itage Area, is here recognized as the oldest member of this fam-ily, and our phylogenetic analysis and shared synapomorphies(see below) suggests that it belonged to the subfamily Perameli-nae. AL90 Site is estimated to be middle Miocene in age (Archeret al., 1997; Travouillon et al., 2006, 2011) and radiometric datingfor this site (work in preparation) is congruent with this assess-ment (Woodhead et al., 2011). This species pushes back the first

Dow

nloa

ded

by [

UN

SW L

ibra

ry]

at 2

3:31

05

Mar

ch 2

014

380 JOURNAL OF VERTEBRATE PALEONTOLOGY, VOL. 34, NO. 2, 2014

FIGURE 4. Phylogenetic relationships of Crash bandicoot and Liyamayi dayi, gen. et sp. nov., based on our 156 character craniodental matrix. Fossiland recently extinct taxa are indicated by †. Crash bandicoot and Liyamayi dayi are highlighted in bold. Crown-group Peramelemorphia is bracketed.A, strict consensus of 21 most parsimonious trees (tree length = 835; consistency index excluding uninformative characters = 0.30; retention index =0.66) from unconstrained parsimony analysis of the matrix. Numbers above branches represent bootstrap values (1000 replicates). B, strict consensusof 478 most parsimonious trees (tree length = 864; consistency index excluding uninformative characters = 0.293; retention index = 0.639) that resultwhen the matrix is analyzed using maximum parsimony and enforcing a ‘backbone’ molecular scaffold based on Westerman et al. (2012).

occurrence of the family in the fossil record at least 5 to 10 millionyears earlier than previously thought. AL90 Site is considered torepresent a rainforest community (Travouillon et al., 2009; Blacket al., 2012) based on the high proportion of arboreal folivoreswithin the assemblage, including five species of pseudocheiridpossums, two species of phalangerid possums, one small speciesof koala, and the arboreal diprotodontid Nimbadon lavaracko-rum. The presence of C. bandicoot, the earliest peramelid, inAL90 Site suggests that peramelids evolved in rainforest beforesubsequently diversifying into other types of environments.

Crash bandicoot (Fig. 2) shares a number of synapomorphieswith peramelines. For example, the position and development ofthe metaconule on M2 and 3 is similar to that of Perameles gun-nii and P. bougainville; the centrocrista is incomplete on all mo-lars and StD and StB are conical and not connected by any sty-lar crests; the preparacrista connects to StA on M2 and 3; andStE is absent on M3. Crash bandicoot also retains a number ofplesiomorphies shared with species of the extinct genus Galadi.These include the presence of a stylar crest connecting StD tothe metastyle on M1 and 2 and the absence of a posterior cin-gulum on M2 and 3. Travouillon et al. (2010, 2013) identified

a number of synapomorphies (e.g., well-developed lingual shelfand larger major cusp on P3; large metaconule; incomplete cen-trocrista on M3; and paraconid-metaconid distance is reduced onposterior molars) shared between species of Galadi and crown-group peramelemorphians (and C. bandicoot), suggesting a closerelationship between the two clades. It is possible that C. bandi-coot shared a common ancestor with species of Galadi.

Thylacomyids are distinct from all other peramelemorphians inpossessing the following combination of molar synapomorphies:metaconule absent; posterior cingulum absent; conical StB andStD; lingual displacement of the metacone (although less extremein M. leucura than M. lagotis) such that it comes to function asa topographic ‘hypocone’; shortened distance between metastyleand StD; preparacrista connects paracone to base of StB; con-ical entoconid; markedly reduced paraconid; and presence of acusp in the hypoflexid region. Ischnodon australis shares all lowermolar synapomorphies with species of Macrotis, although theparaconid is less reduced than in Macrotis lagotis, but closer insize to Macrotis leucura (no upper dentition has yet been recov-ered for I. australis). Liyamayi dayi shares some of those synapo-morphies with species of Macrotis and Ischnodon australis:

Dow

nloa

ded

by [

UN

SW L

ibra

ry]

at 2

3:31

05

Mar

ch 2

014

TRAVOUILLON ET AL.—EARLIEST PERAMELID AND THYLACOMYID BANDICOOTS 381

shortened distance between metastyle and StD; posterior cingu-lum absent; conical StB and StD (but stylar crests are still presenton both stylar cusps); preparacrista connects paracone to baseof StB (but also connects to StA); conical entoconid; and pres-ence of cusp in hypoflexid region. However, some of these fea-tures are also present in peramelids, such as species of Isoodon(shortened distance between metastyle and StD; conical StB andStD; conical entoconid; presence of cusp in hypoflexid region)and Chaeropus ecaudatus (shortened distance between metastyleand StD; conical StB and StD; conical entoconid). These featuresprobably represent homoplasies related to similarities in diet (i.e.,omnivory). The shortened distance between the metastyle andStD is seemingly more significant because it is only present inspecies of Macrotis, Isoodon, and Chaeropus. This distance islong in species of Perameles, all New Guinea genera, and allother fossil peramelemorphians. However, Liyamayi dayi lackskey synapomorphies linking it to species of Isoodon (presence ofposterior cingulum; well-developed metaconular hypocone) or C.ecaudatus (absence of anterior cingulum; well-developed meta-conular hypocone), therefore supporting its placement withinthylacomyids. Liyamayi dayi exhibits many plesiomorphic fea-tures (e.g., presence of metaconule; stylar crest connecting StDto metastyle; connection of preparacrista to StA; incomplete an-terior cingulum) features that might be expected to be presentin a Miocene peramelemorphian. The earliest fossil notoryctid,Naraboryctes philcreaseri (Archer et al., 2011), retains a numberof dental plesiomorphies, such as the presence of a paracone, StB,StC, and StD on the upper molars. However, the paracone in N.philcreaseri is highly reduced, and absent in all other notoryc-tids (species of Notoryctes). In late Cenozoic thylacomyids, themetaconule is absent, but it might be expected that a more ple-siomorphic thylacomyid would retain a small metaconule as L.dayi does. Interestingly, both notoryctids and thylacomyids havemodern desert-adapted species (Notoryctes and Macrotis, respec-tively) that evolved in rainforest.

The age of the Riversleigh deposit that has produced L. dayiis not clear. Rick’s Sausage Site is an isolated deposit that oc-curs on the southern Gag Plateau, west of AL90 Site (FaunalZone C, middle Miocene) and Encore Site (Faunal Zone D, earlylate Miocene). The fauna is as yet poorly understood but the as-semblage includes the bandicoot Bulungu palara and the pseu-docheirid Pildra sp. 1, which are found in Faunal Zones A–C, apseudocheirid species, found in Faunal Zones A–D, and Wan-buroo hilarus, a Miocene macropodid only found in Faunal ZoneC sites (Archer et al., 2006; Travouillon et al., 2011). The presenceof the latter (also found in AL90 Site) suggests a middle Mioceneage for Rick’s Sausage Site.

The earliest thylacomyid, Ischnodon australis (Stirton, 1955)from the Palankarinna LF, South Australia, is Pliocene in age.However, Westerman et al. (2012) estimated a late Oligocenemolecular divergence date of ∼25.8 million years for the splitbetween thylacomyids and peramelids. The presence of a thyla-comyid in the Miocene of Riversleigh would not be inconsistentwith this estimate. This is around 10 million years before the oc-currence of L. dayi in the Riversleigh deposits. The absence ofthylacomyids in earlier Faunal Zones of Riversleigh suggests thateither thylacomyids first evolved elsewhere on the continent be-fore colonizing the Riversleigh rainforest, or that earlier River-sleigh thylacomyids had not evolved their dental synapomorphiesand were dentally indistinguishable from other peramelemorphi-ans until the middle Miocene.

CONCLUSIONS

The oldest peramelid and thylacomyid are described from mid-dle Miocene deposits in the Riversleigh World Heritage Area.Although both retain a number of plesiomorphies, they display anumber of apomorphies linking them to crown-group peramele-

morphians. The presence of the two new species presented inthis paper is congruent with the most recent peramelemorphianmolecular divergence times and suggest that peramelids and thy-lacomyids diversified in middle Miocene rainforests.

ACKNOWLEDGMENTS

Support for research at Riversleigh has come from theAustralian Research Council (LP0989969, LP100200486, andDP1094569 grants to M. Archer, S. J. Hand, and K. H. Blackat the University of New South Wales); XSTRATA Commu-nity Partnership Program (North Queensland); the Universityof New South Wales; P. Creaser and the CREATE Fund; theQueensland Parks and Wildlife Service; Environment Australia;the Queensland Museum; the Riversleigh Society Inc.; Outbackat Isa; Mount Isa City Council; and private supporters, includingK. and M. Pettit, E. Clark, M. Beavis, and M. Dickson. Assis-tance in the field has come from many hundreds of volunteers aswell as staff and postgraduate students of the University of NewSouth Wales. We thank R. Day for providing funding to the Uni-versity of Queensland to create a postdoctoral position for K. J.Travouillon. We thank S. Ingleby and A. Divljan from the Aus-tralian Museum, H. Janetzki from the Queensland Museum, andC. Stevenson from the Western Australian Museum for provid-ing access to modern bandicoot specimens. We thank the Univer-sity of New South Wales Palaeosciences Lab and the Universityof Queensland Palaeo Hub for their support and anonymous re-viewers for helpful comments.

LITERATURE CITED

Aplin, K. P., and M. Archer. 1987. Recent advances in marsupial system-atics with a new syncretic classification; pp. xv–lxxii in M. Archer(ed.), Possums and Opossums, Studies in Evolution. Surrey Beattyand Sons, Sydney, Australia.

Aplin, K. P., K. M. Helgen, and D. Lunde. 2010. Review of the morphol-ogy, distribution and conservation status of Peroryctes broadbenti(Ramsay, 1879), the giant bandicoot of south-eastern Papua NewGuinea. American Museum Novitates 3696:1–41.

Archer, M., and J. A. W. Kirsch. 1977. The case for the Thylacomyidaeand Myrmecobiidae, Gill, 1872, or why are marsupial families so ex-tended? Proceedings of the Linnean Society of New South Wales102:18–25.

Archer, M., and M. Wade. 1976. Results of the Ray E. Lemley expedi-tions, part 1: the Allingham Formation and a new Pliocene verte-brate fauna from northern Queensland. Memoirs of the QueenslandMuseum 17:54–58.

Archer, M., S. Hand, H. Godthelp, and P. Creaser. 1997. Correlation ofthe Cainozoic sediments of the Riversleigh World Heritage FossilProperty, Queensland, Australia. Memoires et travaux de l’Institutde Montpellier 21:131–152.

Archer, M., R. Beck, M. Gott, S. Hand, H. Godthelp, and K. Black. 2011.Australia’s first fossil marsupial mole (Notoryctemorphia) resolvescontroversies about their evolution and palaeoenvironmental ori-gins. Proceedings of the Royal Society B 278:1498–1506.

Archer, M., D. A. Arena, M. Bassarova, R. M. D. Beck, K. Black,W. E. Boles, P. Brewer, B. N. Cooke, P. Creaser, K. Crosby, A.Gillespie, H. Godthelp, S. J. Hand, B. P. Kear, J. Louys, A. Mor-rell, J. Muirhead, K. K. Roberts, J. D. Scanlon, K. J. Travouillon,and S. Wroe. 2006. Current status of species-level representation offaunas from selected fossil localities in the Riversleigh World Her-itage Area, northwestern Queensland. Alcheringa, Special Issue 1:1–17.

Arena, D. A. 2004. The geological history and development of the terrainat the Riversleigh World Heritage Area during the middle Tertiary.Ph.D. dissertation, University of New South Wales, Sydney, NewSouth Wales, Australia, 275 pp.

Beck, R. M. D., H. Godthelp, V. Weisbecker, M. Archer, and S. J. Hand.2008. Australia’s oldest marsupial fossil and their biogeographicalimplications. PLoS ONE 3:e1858.

Bensley, B. A. 1903. On the evolution of the Australian Marsupialia: withremarks on the relationships of marsupials in general. The Transac-tions of the Royal Society of London, Zoology 9:83–217.

Dow

nloa

ded

by [

UN

SW L

ibra

ry]

at 2

3:31

05

Mar

ch 2

014

382 JOURNAL OF VERTEBRATE PALEONTOLOGY, VOL. 34, NO. 2, 2014

Black, K. H., A. B. Camens, M. Archer, and S. J. Hand. 2012. Herds over-head: Nimbadon lavarackorum (Diprotodontidae), heavyweightmarsupial herbivores in the Miocene forests of Australia. PLoSONE 7:e48213.

Breen, J. G. 1985. Waanyi Dialect. School of Australian Linguistics, Dar-win Community College, Darwin, Northern Territory, Australia.

Gray, J. E. 1825. Outline of an attempt at the disposition of the Mam-malia into tribes and families with a list of the genera apparently ap-pertaining to each tribe. Annals of Philosophy (New Series) 10:337–344.

Groves, C. P. 2005. Order Peramelemorphia; pp. 38–42 in D. E. Wilsonand D. M. Reeder (eds.), Mammal Species of the World: A Tax-onomic and Geographic Reference, third edition. Johns HopkinsUniversity Press, Baltimore.

Groves, C. P., and T. F. Flannery. 1990. Revision of the families andgenera of bandicoots; pp. 1–11 in J. H. Seebeck, R. L. Wallis,P. R. Brown, and C. Kemper (eds.), Bandicoots and Bilbies. SurreyBeatty and Sons, Sydney, Australia.

Gurovich, Y., K. J. Travouillon, R. M. D. Beck, J. Muirhead, andM. Archer. 2013. Biogeographical implications of a new mouse-sized fossil bandicoot (Marsupialia: Peramelemorphia) occupying adasyurid-like ecological niche across Australia. Journal of System-atic Paleontology. DOI:10.1080/14772019.2013.776646.

Kirsch, J. A. W. 1968. Prodromus of the comparative serology of Marsu-pialia. Nature 217:418–420.

Krajewski, C., R. Torunsky, J. T. Sipiorski, and M. Westerman. 2007.Phylogenetic relationships of the dasyurid marsupial genus Murexia.Journal of Mammalogy 88:696–705.

Luckett, W. P. 1993. An ontogenetic assessment of dental homologies intherian mammals; pp. 182–204 in F. S. Szalay, M. J. Novacek, and M.C. McKenna (eds.), Mammal Phylogeny: Mesozoic Differentiation,Multituberculates, Monotremes, Early Eutherians and Marsupials.Springer-Verlag, New York.

Mackness, B. S., S. Wroe, J. Muirhead, C. Wilkinson, and D. Wilkinson.2000. First fossil bandicoot from the Pliocene Chinchilla local fauna.Australian Mammalogy 22:133–136.

Muirhead, J. 1994. Systematics, evolution and palaeobiology of recentand fossil bandicoots (Peramelemorphia, Marsupialia). Ph.D. dis-sertation, University of New South Wales, Sydney, New SouthWales, Australia, 463 pp.

Muirhead, J. 2000. Yaraloidea (Marsupialia, Peramelemorphia), a newsuperfamily of marsupial and a description and analysis of the cra-nium of the Miocene Yarala burchfieldi. Journal of Paleontology74:512–523.

Muirhead, J., and S. L. Filan. 1995. Yarala burchfieldi, a plesiomorphicbandicoot (Marsupialia, Peramelemorphia) from Oligo-Miocenedeposits of Riversleigh, northwestern Queensland. Journal of Pa-leontology 69:127–134.

Muirhead, J., L. Dawson, and M. Archer. 1997. Perameles bowensis,a new species of Perameles (Peramelemorphia, Marsupialia) fromPliocene faunas of Bow and Wellington caves, New South Wales.Proceedings of the Linnean Society of New South Wales 17:163–174.

Myers, T. J. 2001. Prediction of marsupial body mass. Australian Journalof Zoology 49:99–118.

Sorenson, M. D., and E. A. Franzosa. 2007. TreeRot version 3.0. Depart-ment of Biology, Boston University, Boston, Massachusetts.

Stirton, R. A. 1955. Late Tertiary marsupials from South Australia.Records of the South Australian Museum 11:247–268.

Swofford, D. L. 2002. PAUP∗. Phylogenetic Analysis Using Parsimony(∗And Other Methods), version 4 (updated to 10 beta). Sinauer As-sociates, Sunderland, Massachussets.

Travouillon, K. J., M. Archer, S. J. Hand, and H. Godthelp. 2006. Mul-tivariate analyses of Cenozoic mammalian faunas from Riversleigh,north-western Queensland. Alcheringa, Special Issue 1:323–349.

Travouillon, K. J., S. Legendre, M. Archer, and S. J. Hand. 2009. Palaeoe-cological analyses of Riversleigh’s Oligo-Miocene sites: implicationsfor Oligo-Miocene climate change in Australia. Palaeogeography,Palaeoclimatology, Palaeoecology 276:24–37.

Travouillon, K. J., Y. Gurovich, R. M. D. Beck, and J. Muirhead. 2010.An exceptionally well-preserved short-snouted bandicoot (Marsupi-alia; Peramelemorphia) from Riversleigh’s Oligo-Miocene deposits,northwestern Queensland, Australia. Journal of Vertebrate Paleon-tology 30:1528–1546.

Travouillon, K. J., G. Escarguel, S. Legendre, M. Archer, and S. J. Hand.2011. The use of MSR (Minimum Sample Richness) for sample as-semblage comparisons. Paleobiology 37:696–709.

Travouillon, K. J., Y. Gurovich, M. Archer, S. J. Hand, and J.Muirhead. 2013. The genus Galadi: three new bandicoots (Marsupi-alia; Peramelemorphia) from Riversleigh’s Miocene deposits, north-western Queensland, Australia. Journal of Vertebrate Paleontology33:153–168.

Turnbull, W. D., E. L. Lundelius Jr., and M. Archer. 2003. Dasyurids, per-ameloids, phalangeroids, and vombatoids from the early PlioceneHamilton fauna, Victoria, Australia. Bulletin of the American Mu-seum of Natural History 279:513–540.

Westerman, M., J. Young, and C. Krajewski. 2008. Molecular relation-ships of Pseudantechinus, Parantechinus, and Dasykaluta (Marsupi-alia: Dasyuridae). Australian Mammalogy 29:201–212.

Westerman, M., B. P. Kear, K. Aplin, R. W. Meredith, C. Emerling, andM. S. Springer. 2012. Phylogenetic relationships of living and re-cently extinct bandicoots based on nuclear and mitochondrial DNAsequences. Molecular Phylogenetics and Evolution 62:97–108.

Woodhead, J., M. Archer, S. J. Hand, H. Godthelp, I. Graham, P.Creaser, K. Black, R. Arena, and E. Price. 2011. Preliminary radio-metric ages for Cenozoic deposits in the Riversleigh World Heritagefossil area, NW Queensland; p. 87 in Record of Geological Surveyof Western Australia 2011, 13th Conference on Australasian Ver-tebrate Evolution Palaeontology and Systematics, Perth, Australia,27–30 April 2011. Geological Survey of Western Australia, Record2011/9, 104p.

Worthy, T. H., A. J. D. Tennyson, M. Archer, A. M. Musser, andS. J. Hand. 2006. Miocene mammal reveals a Mesozoic ghost lin-eage on insular New Zealand, southwest Pacific. Proceedings of theNational Academy of Sciences of the United States of America103:19419–19423.

Submitted December 20, 2012; revisions received April 14, 2013;accepted April 15, 2013.Handling editor: Thomas Martin.

Dow

nloa

ded

by [

UN

SW L

ibra

ry]

at 2

3:31

05

Mar

ch 2

014