FIRST RECORD OF THE GPOT.ROY (MICROCHIROPTERA : ) FROM

by

SuzanneHAND *

CONTENTS Page Abstract,Rdsum6 48 Introduction 48 Systematics 49 Comparisons with other megadermatids . . . 53 Fossil forms 53 Living taxa. . 55 Phylogeneticrelationships 57 Discussion 58 Assignment to M egaderma 58 Intrageneric relationships 60 Palaeobiogeographyand palaeoecology . 61 Acknowledgements 62 References 63 Appendix. .. . 65 Legendsofplates 66

* School of Biological Science,University of New South Wales,PO Box l, Kensington,Sydney, NSW 2033, Australia

Key-words: Chiroptera, Megaderma, Megadermatidae, , Rackham's Roost Site, Riversleigh, Australia. Mots-cl6s: Chiroptera, Megaderma, Megadermatidae,PliocEne, Localit6 Rackham's Roost, Riversleigh, Australie.

Paheovertebrara, Montpellier,24 (l-2): 4146,2 pl. (Regule 29Juillet 1993, acceptd le 23 Septembre1993, publi€ le 14Juin 1995) ABSTRACT

A new Tertiary megadermatid is described from Rackham's Roost Site, a Pliocene limestone cave deposit on Riverileigh Station, northwestern Queensland, Australia. It appears to represent the first Australian record of Megaderma GeoFrnoy, 1810, a genus otherwise known from Tertiary African and (Lyroderma) European taxa and the living Asian species M. spasma (Lu'rNeeus,1758) and M- lyra perens, 1872. Megaderma richardsi n. sp. is one of the smallest megadermatids known. It exhibits a mixture of plesiomorphic and autapomorphic features, the latter appearing to exclude it from being from the ancestralto any living megadermatid.The new speciesis one of eight megadermatidsidentified Australian fossil record, most of which are referable to Mtu-en, 1906'

RESUME

gisement Un nouveau mdgadermatid6 tertiaire est ddcrit de la localit6 Rackham's Roost, un premidre cavernicolepliocbne de RiversleighStation, Nord-Ouest Queensland, Australie' Il repr6sentela jusqu'ici decouverteaustralienne du Megaderma GeorrRov, 1810, un genre connu par des espdces et tertiairesafricaines et europdenneset par les espbcesrdcentes asiatiques M. spasma(LnNneus, 1758) M. (Lyroderma.) /yra Prrrns , 1872. Megaderma richardsi n. sp. est un des plus petits m6gadermatid6s de toute connus.Il montre un mdlangede pl6siomorphieset d'autapomorphies,ces dernibresI'excluant ascendanceh l'6gard des m6gadermatid6srdcents. L'espEce nouvelle est I'un des huit m6gadermatid6s 1906. d6nombr6sdans le r6gistre fossil australien,dont la plupart sont r6f6rablesde MacrodermaMtrt-ex,

INTRODUCTION

The bar family Megadermatidaeis an Old World tropicalgroYp Yltll five extant species.These are iommonly (e.g.Honacki et aI. 1982,Hill & Smith 1984)referred to fiur genera.Three genera,Lavia GRAY,1938, CardiodermaPETERS, 1873 and Macridermc MTLLER,1906, are consideredto be monotypic.Megaderma GEOFFROY, 1810then containstwo species,Megaderma spasrna Md M. (Lyroderma)Iyra. I-avia alrdM' spasmaNe frons andCardioderma cor Ne African species,Megaderma lyra Asian andMacroderma glgas Australian. Megadermatidsare insectivorous(e.g. L. fron! or facultativelycarniv.orous and are gene;lly characterisedby a dentitionadapted for meat-eating.The fanlly is well reprEsentedin the fossil record,being known from at least 19 fossil taxa from early fertiary to Quaternarysediments in Europe, andAustralia (e.g. Sig6 1976;Hand 1985,iable 8; Sevilla 1990;Ziegler1993). The majority of Tertiarymegadennatids are referred to the genus Megadirmd, a notable exception being the oldest known megadermatid,I{ecromantii adichasterWetrsoreR, 1887 from late EoceneFrench sediments(Remy et aI.1987). In lgi6, Sig6 examined the evolutionary history of megadermatidsand recommendedrecolnition of only threemegadermatid genera, , Invia and Megaderma,the latter containingCardioderma and Macrodermaas subgenera'He

48 arguedthat, without rarely preservedcranial material, fossil megadermatidscould not be confidently assignedto the more restrictedspecies groups commonly used for living taxa. Subsequently,however, after examination of new Australian fossil megadermatid material, Hand (1985) suggestedthat the generic narte Macroderma be retained for an Australian clade which now includes at least three Tertiary species (M. godthelpi HAND,1985, M. koppa HeNp et al., 1988 and M. malugarc HAND,submitted) as well as Australia'sonly living megadermatidM. gigas (DonsoN, 1880).She also suggested that the name Lyroderma might be resurrectedfor M. Iyra and the middle M. gaillardi (Tnourssanr, I 898). Although most Australian Tertiary megadermatids appear referable to Macrodermc, Miocene depositson RiversleighStation, northwestern Queensland, have also produced enigmatic taxa such as the Dwornamor Variant (Hand 1985) and the Henk's Hollow megadermatid indet. (Hand submitted) that lack apparent synapomorphiesfor this clade. Thesetaxa are representedby isolatedteeth and cannot be confidently assignedto any other speciesgroup. A tiny, new megadermatidfrom Riversleigh'sPliocene Rackham's Roost Site (Hand 1987,Archer et al. 1989,1991) is possiblyallied to theseMiocene forms but is betterrepresented, by a maxillary fragment and severalisolated teeth,and is tentativelydescribed here as the first representativeof the genusMegaderma from Australia. Dental terminology follows Hand (1985). The prefix QM F- indicatesthat the sourceof specimensis the QueenslandMuseum fossil collection.

SYSTEMATICS

OTdeTCHIROPTERA BLUMENB ^CH, I7 79 SuborderMICROCHIROPTERA DossoN, 1875 Superfamily Gnev, 1825 Family MEGADERMATTDAEAt-t-EN, 1 864 GenusMEGADERMA GEOTTNOY, 1810 Megaderma richardsi n. sp. (Pr.1-2)

Holotype: QueenslandMuseum F236L9,a right maxillary fragmentwith Cr and Pa. Etymology: The speciesis named after Greg Richards, one of Australia's foremost researchers,whose help in interpreting aspects of Riversleigh's complex fossil bat faunasis greatly appreciated. Paratypes (from type locality): QM F23620,a right C'; QM F23621,a left Ct. Referred material (from type locality): QM F23598, a right M'; QM F23599,a right tvlt; QM F23600, a left M '. Type locality, lithology and age: The type locality, Rackham'sRoost Site (Godthelp 1988,Archer et al. 1989,1991), occurs at 19.02.09N, 138.41.60E (GlobalPositioning

49 Satellite Device) on Riversleigh Station, northwestern Queensland, and has been described by Hand (submitted). It appearsto represent the indurated floor of a long, narrow cave developed in Thorntonia Limestone. At one end, the cave evidently opened onto a vertical cliff flanking what is now the Gregory River. More than 50 metres away in the oppositedirection, another larger, lower opening appearsto have been developed. The sedimentis a breccia of tiny, mostly fragmented bones and teeth set in a fine-grained, pink-coloured (presumably iron-stained) limestone. The deposit covers an area of approximately 2N squaremetres, with a maximum depth of 0.5 to 1.0metre. On the basis of its fauna, the Rackham'sRoost deposit is interpreted to be Pliocenein age.It containsabundant plesiomorphic murids (at least 13 species)and a macropodid similar to Protemnodon snewini BeRntoLovel, 1978 from the early PlioceneBluff Downs Local Faunaof northeasternQueensland (Archer & Wade 1976).

Associatedfauna and taphonomy: The Rackham'sRoost depositcontains the remains of crustaceans, fish, frogs, lizards, small crocodileS, Snakes, birds, dasyurids, peramelids,a pseudocheirid,an extinct macropodidand potoroid, at least 13 murids, iwo emballonurids (Taphozousspp), four hipposiderids,four vespertilionids and the large megadermatidMacroderma gigas (Archer et al. 1991, Hand submitted). The very finely broken remains of the small vertebrates and depressed fractures and impressions in the bones characteristicof Recent Macroderma gigas canines suggest thit the deposit representsthe remains of prey accumulatedby that species. Boids, the emballonuiids, hipposiderids and vespertilionids probably cohabited the megadermatid roost;the macropodidmay havecamped near the cave'sentrance.

Diagnosis: Megaderma richardsi differs from living species of Megaderma (r.e. M. spasmaand M. lyra) in its lossof P2. It differs from most speciesof Megadermain its very smallsize. However, it is similarin sizeto the TertiaryM. iaegeri andM. Iopezae. Megaderma richardsi differs ftom M. jaegeri Srcri, 1976 of the middle Miocene geni Mefal deposit in Morocco in, among other features,its slightly smaller size, Cr with taller anterior accessorycusp, taller posterior accessorycusp less separatedfrom the main cusp, Mt with shorter postmetacrista,smaller parastyle, less lingually displaced mesostyle, posteriorly opening protofossa, M 1 lower crowned with less sinuous buccal cingulum, trigonid and talonid subequal in length, more anteriorly extendedpre-entocristid, and smaller and less posteriorly displacedhypoconulid so that there is little inflexion in the hypocristid. Megaderma richardsi differs from M. lopezae SpvtLLR, 1990 of the late Carrascosadel Campo locality, Spain, in its Mt with more lingually displaced mesostyle,posteriorly openingprotofossa, and more posteriorly directed heel, and M t having a more laterally compressedtrigonid, less sinuous buccal cingulum, anteriorly extendedpre-entocristid, the metaconidcontribution to the cristid obliqua less reduced,ana tne nypoconulidless well-developed and lessposteriorly-situated so that it is less distinct from the entoconid and there is little inflexion in the hypocristid.

50 Description The right maxillary fragmentretains Ct and Pa.There is no sign of an alveolusfor P2 either in the tooth row or lingually (as in M. spasmaand M. lyra).The anterobuccal shelf of Pa restson a small posterobuccalcingular shelfin C'. C' is known from threespecimens, QM F23619,QM F23620and QM F2362I.It is massively-rootedand recurvedposteriorly with a tall, slender,dominant main cusp.In outline the tooth is semicircular,flattened on the lingual side. There is a narrow cingulum around the whole tooth. It has a large posterior accessorycusp, occuring within the cingular rim, that is poorly separatedfrom the paraconeand rises to more than half the paracone'sheight. It has a well-developedanterolingual cingular cusp. The antero- and posterolingualcorners of the paraconeare defined by vertical crests.The anterior vertical crestintersects the crown basejust buccalto the anterolingualcingular cusp. The posterolingualvertical crest extends to the anterior base of the posterior crown cusp. On the flattened lingual face of the paraconeis a vertical crest which is attenuatedtowards the base and apex of the cusp. There is a slight indentationin the buccal cingulum immediatelybuccal to the junction of the paraconeand posteriorcusp, a small cingular cusp at the most posteriorpoint of the crown, and immediatelybuccal to this a small cingularshelf on which the anterobuccalshelf of P4 rests. Pa is known from the holotype QM F23619. In occlusal view the crown is subtriangular.There is one main cusp, the paracone,which occurs midway along its length and slightly buccal to the crown's centre. There is a short, nalrow, posterolinguallydirected heel on its lingual margin. The heel is surroundedby a thin iontinuous cingulum that is thickestand deepestaround the heel. The buccal cingulum is relatively thick but is intemrptedby the flank of the paraconeabout midway along the tooth'sbuccal length.Anteriorly thereis a relativelytall cingularcusp. The preparacrista is convex in profile and runs anteriorly from the paracone'sapex towards the anterior cingular cusp from which it is separatedby a narrow valley that extendsbuccally and lingually. The postparacrista,which is slightly concaveocclusally and in profile, runs towards the posterior margin of the tooth before dropping vertically immediately before reachingthe cingulum. QM F23600,LM,, QM F23598,RMr, and QM F23599,RMz, arereferred to M. richardsi. They are of an appropriate size to belong to this taxon and exhibit features typically found in megadermatids. M, is nyctalodontwith a laterally-compressedtrigonid, reducedparaconid with the paracristidlonger than the metacristid,twinned entoconidand hypoconulid,reduced pre-entocristid such that the talonid opens lingually, and continuous but relatively poorly developedanterior, buccal and posterior cingula. The trigonid and talonid are approximatelyequal in length but the talonid is much wider than the trigonid which is very laterally compressed.The trigonid is not much taller than the talonid. Thus the protoconid is taller than the metaconid which is taller than the entoconid (and broken hypoconid) which is taller than the reducedparaconid. The paraconidcontribution to the paracristid is very short. The angle made by the para- and metacristids is very wide, being approximately 160 degrees. The cristid obliqua is much longer than the paracristid and sweepsup the posteriorflank of the metaconidalmost to its apex. The hypocristid is also long but shorterthan the cristid obliqua.The hypoconulidis located

51 on the lingual margin of the crown close to the small (worn) entoconid. The angle formed between the the cristid obliqua and hypocristid is relatively small, being approximately 60 degrees.From the entoconid, a pre-entocristid extends anteriorly only a Jfrort distance. There is a slight inflexion in the buccal cingulum where the trigonid and talonidjoin. M' is longer than wide. The tooth is worn. The metaconeis the tallest and most massive cusp and is taller than the paraconewhich is taller than the protocone.The postmetacristais much longer than the subequal premetacristaand postparacristae which are longer than the preparacrista. The mesostyle is tall, being the same topographic height as the small parastyleand the metastyle,and is slightly lingually displaced.There is a very small parastyle.There is no posterior cingulum, but a very

QMFno 23619 23620 23621 23598 23599 23600

c 1-P4 L 3.24

ct L 2.00 1.77 1.92 I 1.08 0.97 1.13 Hi 2.64 2.80 He 2.79 2.87 Hi' 1.63 1.33 1.62 He' 1.75 1.M 1.74

P4 I 1.44 I 1.31 Hi 1.89 He 1.63

M1 L 1.88 LI 1.81 Li 1.43 I 1.93

M2 L 2.08 LI 1.85 Li 1.32 I 2.13

Ml L 1.98 LT 1.01 Lt 0.97 IT 1.35 It 1.30

Table l.- Measurements (mm) of the holotype (QM F23619), Paratypes (QM F23620-23621) and referred specimens(QM F2359S-236N) of Megaderma richardsi n. sp. from the Pliocene Rackham's Roost Site, Riversleigh Station, northwestem Queensland,Australia. Measurementsand abbreviationsfollow Sig6 (1968): L, length; Li, internallength, Ll, lingual length;l, width; Hi, intemal height; He, extemal height;Hi', internal height of posterioraccessory cusp; He', extemal height of posterior accessorycusp; LT, trigonid length; Lt, talonid length; lT, trigonid width; lt, talonid width.

52 faint anterior cingulum on the buccal half of the tooth and a separatecingular swelling anterolingual to the baseof the paracone.The lingual cingulum extendsfrom the base of the metacone(posterior to the baseof the protocone)and enclosesthe very posteriorly directedheel. The heel cingulum is more robust posteriorlythan lingually and thickest at the most posterolingual corner of the heel. The protocone is a conical, puncturing cusp that is topographicallyquite low, particularly with respectto the metacone.The preprotocrista is initially directed anterobuccallytowards the base of the the paracone,a short curved crest (paraloph) extending the preprotocrista a short distance up the posterolingual flank of the paracone.The postprotocristaruns posterobuccallythen almost immediatelyvertically and posterolinguallyonto the heel but does not meet the lingual cingulum to which it briefly runs parallel. M2 is very worn but is basically similar to Mrexcept as follows. There is less difference in size between the paraconeand metacone,the preparacristais only just shorter than the postmetacrista,and the mesostyleis perhapslower. As in Mr, there is no posterior cingulum, a very feint anterior cingulum, and a separateanterolingual cingular swelling betweenthe para- and protocones.The protocone structure is very similar to that in M' but the tooth is more worn and this accentuatesthe curve in the postprotocrista.The heel is rounder, there being no elaborationor thickening of the cingulum at the posterolingualcorner of the heel. Measurementsof the teeth of Megaderma richardsl n. sp. are given in table 1.

COMPARISONS WITH OTHER MEGADERMATIDS

FOSSL FORMS

Megaderma richardsi n. sp. has been compared with all adequately described fossil speciesof Megaderma, with the following exceptions. Comparable material has not been reported for: Megadenna sp. from the late Oligocene of St Victor la Coste, France (Sig6 1967); M. janossyi of the Pliocene of Hungary (Topal D70: and M. watwat of the of Palestine (Bate 1937). A mandible and maxilla representing an undescribed Pliocene species of Cardioderma (Bfiler & Greenwood 1965 Megaderma sensu Sig€ 1976) from Olduvai Gorge, East Africa were not examined. Among fossil forms examined,Megaderma richardsi appearsmost similar to four taxa: the late Oligocene Megaderma lopezae from Carrascosa del Campo, Spain (Sevilla 1990); M. jaegeri from the middle Miocene Beni Mellal fauna of Morocco (Sig6 1976);M. lugdunensisfrom the early Miocene Vieux-Collongesdeposit, France (Gu6rin & Mein l97l), middle Miocene of Neudorf, Czechoslavakta(Zapfe 1950) and late Miocene of Lissieu, France (Mein 1964); and M. vireti of the middle Miocene La Grive St-Alban, France(Gu6rin & Mein I97I),late Miocene of Lissieu, France(Mein 1964) and Kohfidisch of Austria (Bachmayer& Wilson 1970). Megadermalopezae and M. richardsi share:almost identical Pa moqphology;Mt

53 with similar sized heel, tall mesostyle not greatly displaced lingually, no posterior cingulum, possibly a paraloph;a relatively low crowned M, with trigonid and talonid approximately equal in length and hypoconulid not greatly buccally displaced. Megaderma lopezae differs from M. richardsi (in addition to the features noted above in the diagnosis)in its M' with buccally projectingmesosfyle, unreduced postparacrista and premetacrista,general protocone morphology and small, closedprotofossa. C' and M2 are not known in M. lopezae. Megaderma jaegeri and M. richardsi share: similar general proportions of Mr, height of mesostyle(which is slightly more lingually displacedrn M. jaegeri), reduced postparacrista and premetacrista,heel direction and size, anterolingual cingulum (thoughvery faint); relativelylow crownedM r, with lingual hypoconulid,similar lateral and anteroposteriortrigonid compressionand lateral talonid compression(with more anteroposteriorcompression in jaegeri). In addition to the differencesnoted above in the diagnosis, Megaderma jaegeri differs from M. richardsi in its: proportionately longer Cr with thicker buccal cingulum and more massive root; Mr protocone morphology and weak paraloph development;M ' with trigonid longer than talonid, hypoconulid and entconidmore distinct and buccally displaced,and reducedmetaconid contributionto the cristid obliqua. Pa and M2 are not known for M. iaegeri. Megaderma lugdunensisand M. richardsi share: Mt with posteriorly directed heel and similar heel cingulum, very similar parastyledevelopment, relative size of paraconeand metacone,mesostyle height; M2 with postparacristaand premetacrista only moderately reducedand heel posteriorly directed; M ' with well developedpre- entocristid, and lingually and posteriorly situated hypoconulid. It differs ftom M. richardsi in: its C'with buccal vertical crest; P4 with slimmer heel, shorter anterior cingular cusp, broader and buccally deeper anterior cingular shelf, anteriorly poorly defined buccal cingulum; Mt with longer and broaderheel, no paraloph,more lingually displacedmesostyle; M2 with more reducedpostparacrista and premetacrista,broader styiar shelf and much lower, more lingual mesostyle;M t with trigonid less laterally compressedand reducedmetaconid contribution to the cristid obliqua. Megaderma vireti and M. richardsi share: similar Cr morphology including separationof a blade-likeposterior cusp, presence of posteriorcingular cusp, cingulum development, relative cusp heights; similar Pa morphology; M' with posteriorly directe-dheel and large difference between size of paracone and metacone; M2 with similar relative proportions, heel, cingulum, anterolingual cingulum and paraloph. Differences include: lack of tall anterior cingular cusp in Pn; M' with large, narrow heel, protocone morphology; M2 protocone morphology, mesostylemuch lower and lingually situated, postparacristaand premetacristamore reduced;long, narrow M ' with stronger lateral compression of talonid and trigonid, hypoconulid more anteriorly and buccally placed, reducedpre-entocristid and metaconidcontribution to cristid obliqua. Megaderma richardsi appears to be less similar to: Megaderma brailloni ftom the French early Miocene Bouzigues (Sig6 1968)and Serre de Verges (Meurisseer a/. 1969) localities; M. franconica from the early Miocene of Wintershof-West of Eichstritt,Germany (Ziegler 1993); M. gaillardi from the middle Miocene sites of La Grive St-Alban, France(Gu6rin & Mein I97l), Beni Mellal, Morocco (Sigd 1976)and Escobosa,Spain (Ses6 1986); and M. mediterraneumfrom the late Pliocene of Sdte,

54 southern France (Sig6 1974\. Similarities, however, include: with brailloni, the degree of separationof the posterioraccessory cusp from the paraconein Cr, in Mt general morphology of the buccal part of the crown (stylar shelf, paraconeand metacone),and in M I the degreeof lateral compressionof the trigonid and anteroposteriorcompression of the talonid; with gaillardi, Cr morphology of the posterior accessorycusp, anterior accessorycusp development and vertical lingual crest, M 2 heel size;withfranconica, al least M2 heel size and probably (as in gaillardi) C' moqphologyof the posterior accessorycusp, anterior accessorycusp development and verticallingual crest;and with mediterraneum,in Cr separationof the blade-like (but short) posterior cusp, well defined buccal cingulum, pronouncedlingual vertical crest and marked indentationin the buccal crown.

Megaderma brailloni differs from M. richardsi in its: C' with posterior cusp relatively short and conical (rather than blade-like),no posteriorcingular cusp, vertical buccal crest; M' with broader stylar shelf and shorter and more lingually situated mesostyle,paracone and metaconesubequal in height, lingually directedheel, different protocone morphology, no paraloph, parastyle relatively well developed; M ' with sinuousbuccal cingulum on trigonid, more laterallycompressed talonid, more anteriorly and buccally located hypoconulid, more robust cingulum, greater difference in trigonid and talonid heights,and reducedmetaconid contribution to the cristid obliqua. Megaderma gaillardi differs from M. richardsi in its: taller crowned teeth; C' with poorly defined buccal cingulm, buccal vertical crest; M2 with reduced postparacrista and premetacrista, broader stylar shelf, much lower, more lingual mesostyle, no paraloph, more lingually directed heel, lingual elaboration of heel cingulum, very different protocone (although the protofossa does open posteriorly slightly), lack of anterolingualcingulum; M r with very buccally situatedhypoconulid, laterally compressedtrigonid, little difference between talonid and trigonid widths, very sinuous buccal cingulum. (Note that in M, the cristid obliqua is not greatly reduced, there is a greater difference in talonid and trigonid heights, the pre-entocristid is not as anteriorly developed, the hypoconulid more buccally and posteriorly situated and the hypoconid cingulum is elaborated.)Megaderma franconica differs from M. richardsi in at least its: P4 heel morphology,Mr-2 protoconemorphology, and M r cristid obliqua and buccal cingulum moqphology. Megaderma mediterraneum differs from M. richardsi in its: C' with buccal vertical crest, no posterior cingular cusp although the cingulum may rise there very slightly; high crowned lower molars (as representedby M) and, by extrapolation from M r, posteriorly situated hypoconulid so that there might be an inflexion in the hypocristid.

LIVING TAXA

Among living taxa, Megaderma richardsi is most similar to Megaderrna spasma and Cardioderma cor. Megaderma richardsi and M. spasma share: similar overall Pa morphology; M'

55 shape and proportions, heel direction, a posteriorly-openingprotofossa (slightly in M. spasma); M 2 shape and proportions, heel direction and cingulum, relatively low protocone. Megaderma spasmadiffers in having: a weakly developedor absent anterior accessorycusp in CI; broader and shorter Pa heel; in Mr-2, more lingual and lower mesostyle, reduced postparacristaand premetacrista,a posterior cingulum, more continuous anterior cingulum; in M' slightly broaderheel, better developedparastyle, no posteriorelaboration of heel cingulum,no paraloph;M 2 with longerheel, protofossa not as open;M, with more laterallycompressed trigonid and talonid. Cardioderma cor shares with M. richardsi: in Cr, a tall, blade-like posterior accessorycusp little separatedfrom the paracone,posterior cingular cusp and buccal cingulum; losi of P2; in Pa, heel size,shape and direction,paracone position; in Mr-2, heel size, little lingual displacement of mesostyle, only moderately reduced postparacristaand premetacrista;in M ,, posteriorpostion of hypoconulidand unreduced meraconid contribution to the cristid obliqua. It differs from M. richardsi: in Cr, lacking an anterior accessorycusp, less separationof the posterioraccessory cusp from the paracone;Pa without anteriorcingular cusp (replacedby posteriorcingular cusp on C'), shallow anterobuccalcingulum; M' with broaderand lingually directed heel, no posterior elaborationof cingulum, protofossaclosed, no anterior or posterior cingulae; M2 with more lingually directed heel, protofossaclosed, anterior cingulum but no anterolingualcingulum. Lavia fronr shareswith M. richardsi: in Mr-2 the relatively buccalposition of the mesostyle,an anterolingualcingulum (also anteriorcingulum); in M,, degreeof lateral and anteroposterior compression of talonid, little difference between trigonid and talonid lengths and heights,position of hypoconulid(posterior and lingual), unreduced metaconid contribution to the cristid obliqua. It differs from M. richardsi: in C' with short posterior accessorycusp well separatedfrom the paracone; Pa morphology very distinct, with very curved postparacristafrom central paracone,small, indistinct anterior cingular cusp, but with broad anterior (=anterolingual) shelf and deep anterobuccal cingular sheif, broad and long heel that is very different from other taxa; Mr and M2 protocone morphology different again with buccal cingulum sweeping up protocone, continuous posterior and heel cingula; in M 1, pre-entocristidunreduced, extending anteriorly to metaconid. Megaderma (Lyroderma) lyra and M. richardsi share: in Ct a bladelike posterior accessorycusp; in Mr-2, lack of posteriorcingulum; and in M1, degree of iateral and anteroposterior compression of the trigonid. Megaderma lyra differs in: being high crowned; in C' the posterior accessorycusp relatively short and better separated;P4 with no anterior cingular cusp,much slimmer heel (which appearsto be posteriorly displaced),poorly developedbuccal cingulum; in Mr-2, more lingual and iow mesostyle,closed protofossa, better developed parastyle, reduced postparacrista and premetacrista,heel longer and slimmer (lingually directed in Mr), better developed posterior cingulum, more continuous anterior cingulum, no paraloph; in M 1, IIIore iaterally and anteroposteriorly compressedtalonid, greater difference in trigonid and talonid heights and lengths,little differencein widths, buccal position of hypoconulid, sinuousbuccal cingulum and reducedmetaconid contribution to the cristid obliqua. Megaderma richardsi shares with Macroderma gigas: loss of P2; Pa with

56 anterior cingular cusp and posterolingually directed heel; M'-2 with no posterior cingulum; and in M,, similar degreesof lateral and anteroposteriorcompression of the rigonid. Macroderma gigas differs from Megaderma richardsi in: being high crowned; in Ct with well separated,short and conical posterior cusp; Pa with buccal paracone,no buccal cingulum, tall anterior cingular cusp, waisted, elongated heel, pronounced vertical drop from posterior end of relatively curved postparacrista to cingulum; M'-2 with broad buccal shelf and low, lingual mesostyle,more reduced postparacristaand premetacrista,closed protofossa, long, posterolinguallydirected heel, no posterior elaborationof heel cingulum, no paraloph; M ' with strong lateral and anteroposteriorcompression of talonid. Overall, Megaderma richardsi appears to be most similar (phenetically) to M. lopezae,M. jaegeri, M. lugdunensis and M. vireti among fossil forms and M. spasma and Cardioderma cor among living species.It is perhapssignificant that these taxa are mostly small to medium-sizedmegadermatids, with very large taxa suchas Megaderma brailloni, M. gaillardi andMacroderma gigas being most dissimilar (seeDiscussion).

PHYLOGENETICRELATIONSHIPS

A subsetof the shared featuresnoted in the comparisonsabove appear to provide preliminary phylogenetic information. Polarities of character stateswere determined by outgroupcomparison (using nycterids,hipposiderids and rhinolophids;see Hand 1985) and/or commonality within the family Megadermatidae.A list of characters surveyed for phylogenetic information is given in Appendix. Potentially phylogenetically useful characters include the degree of lingual displacementof the mesostyleof Mr-2, lateral and anteroposteriorcompression of the trigonid and talonid of M 1 and buccal displacementof its hypoconulid (see also Sig6 19i6, Hand 1985, Sevilla 1990). Other features, such as Mr-2 heel direction, presence/absence of paralophs, development of cingula and probably the presence/absenceof P2, are extremely difficult to interpret phylogenetically. Many appearto representhomoplasies that provide little phylogenetic information. Becauseso few unambiguouslyuseful charactersare evident in the material of M. richardsi it is possible only to make preliminary remarks about its phylogenetic relationships. With respect to Lavia frons, Megaderma richardsi appears to be more apomorphic in its more lingually displaced and lower Mr-2 mesostyle, reduced postparacristaand premetacrista,elongated postmetacrista, transversely compressed M ' trigonid and talonid, and reducedpre-entocristid. In its tingually displacedand lower Mr-2 mesostyle,shorter postparacristaand premetacrista, longer postmetacristaand more laterally compressedM r trigonid, M. richardsi also appearsto be more apomorphic thatrM. lopezae. Megadermajaegeri exhibits a further lingual displacementof the MI mesostyle, and more derived stylar shelf as well as a longer trigonid than talonid (resulting from anteroposteriorcompression of the talonid), and is inteqpretedhere to be generally more

57 apomorphic than M. richardsi. Megaderma brailloni, M. vireti, M. lugdunensis, M. spasma and Cardioderma cor arealso generally more apomorphic,exhibiting a more derived Mr-2 morphology (i.e. stylar shelf), greater difference between M, trigonid and talonid heights and lengths, greater anteroposteriorand lateral compressionof the talonid and buccal displacement of the hypoconulid. Megaderma vireti and M. lugdunensis share some features with M. richardsi that could be potential apomorphies,such as the posterior (rather than posterolingual)heel direction in Mr-2 and a much larger metaconethan paracone. Megaderma gaillardi, M. franconica, M. lyra and fossil and Recent species of Macroderma appearmore derived than other megadermatids(including M. richardsi) in their greater M'-2 heel development,large size, high crowns and further lateral and anteroposterior talonid compression in M r. Megaderma lyra and species of Macroderma also sharea taller Pa with betterdeveloped heel and more buccally placed paracone.Because comparisonsbetween M. richardsi and M. mediterraneum were limited to Cr characters,the phylogenetic relationship of these taxa could not be satsfactorilyassessed. In essence,Megaderma richardsi appears to be more apomorphic than Lavia frons and Megaderma lopezae but more plesiomorphic than all other species of M eg ade rma, Cardiode rma andM acrode rma.

DISCUSSION

ASSIGNMENT TO MEGADERMA

The taxon described here as Megaderma richardsi exhibits a number of features that together characterisespecies of Megaderma. These features include a pronounced reductionof the M, paraconidand lateralcompression of the trigonid, a simple P4 with small heel and central paracone,and Mr-2 with small heel. It also lacks a number of featuresusually found in Macroderma speciessuch as high crowns, strong lateral and anteroposteriorcompression of talonids,Ct with a well separated,relatively short and conical posterior accessorycusp, Pa and Mr-2 with enlargedheels, tall P4 with buccally situated paracone, MI-2 with broad buccal shelf and markedly lingually displaced mesostyle. Assignment of this Australian Pliocene speciesto the genus Megaderma remains tentative. Unlike living Megadermc species,in which the tiny P2 is extruded lingually from the tooth row, the Riversleigh fossil megadermatid lacks P2. However, although this tooth is present in M. spasma and M. lyra, few Tertiary Megaderma species preservethis region of the tooth row, and P2 is known to have been independentlylost in some megadermatid lineages (i.e. Macrodermd spp) and apparently from others. Australian Oligo-Miocene species of Macroderma (e.g. M. godthelpi and M. malugara) contain a P2 which has beenlost in the Pliocene-RecentM. gigas. The tooth

58 has also been lost, apparently independently, in the living African speciesInvia frons and Cardioderma cor, between which no close phylogenetic relationship has been established(Sig6 1976, Hand 1985, Griffiths er aI. 1992). The tooth was large and within the tooth row in the oldest megadermatid, the late Necromantis adichaster WeIrHorrR, 1887. In the Miocene Megaderma vireti MuN, 1964, it was evidently small though less extruded than in living Megaderma species. Loss of P2 within the Macroderma lineage has been correlated by Hand (submitted) with a general trend to shorten the face (thereby apparently increasing the power of the canines).This trend can be traced from the Oligo-Miocene M. godthelpi to the middle Miocene M. malugara, the early Pliocene M. koppa and Pliocene and Recentspecimens of M. gigas.Also possiblyassociated with this shorteningof the face is a gradualchange in the natureof the posterioraccessory cusp of Cr: from a discrete and well-separated,conical cusp in Miocene Macroderma taxa to a less separated(but still relatively short and conical) cusp in RecentMacroderma gigas (Hand submitted). The closeproximity of the blade-likeposterior accessory cusp to the paraconein the Ct of Megaderma richardsi has been interpreted here to be a feature shared symplesiomorphically with other species of Megaderma, and the absence of P2 interpretedto be autapomorphicloss. However, it is also possible that these features represent two of few apomorphies shared with some (but not all) species of Macroderma. A relatively tall, blade-like posterioraccessory cusp in Ct, little separatedfrom the paracone, is intelpreted here to be plesiomorphic within the Megadermatidae. Rhinolophidsand nycteridslack a posterioraccessory cusp in Cr, but a secondarycusp is variably present within the family . For example, species of Hipposideros (Brachipposideros), Syndesmotis, AseIIia, Cloeotis, Triaenops, Coelops andRhinonicteris have relatively tall posterioraccessory cusps on CI. Species of H. (Hipposideros) variably have a secondary posterior cusp, which in some taxa tends to be low and sometimes blunt. Species of H. (Pseudorhinolophus), Palaeophyllophora, Anthops and Aselliscus lack (or exhibit a very reduced or incipient) posterior accessorycusp. Within the Megadermatidae,a tall, blade-like (versus conical) posterior accessorycusp is present in M. richardsi, M. gaillardi, M. vireti, M. spasma and Cardioderma cor and variably in M. Iugdunensis. This condition appearsto occur in hipposideridsand megadermatidsirrespective of size, diet or presenceof P2. Although the three known upper canines of Megaderma richardsi show little variation in absolutesize or cusp morphology(tab. 1, pl.I-2), somemegadermatid taxa exhibit considerablevariation in Ct morphology,including in the degreeof separation and form of the posterior accessory cusp. A sample of seven upper canines of Megaderma lugdunensis from the early Miocene Vieux-Collonges locality (Gu6rin & Mein L97I) show striking differences in height and length, with smaller, more gracile teeth tending to havemore bladeJike posterioraccessory cusps. In the Vieux Collonges sample, however, the accessorycusp remains low with respectto the paracone(unlike in Megaderma richardsi in which the secondarycusp rises to more than half the height of the paracone).Less variation in C' morphology is displayedby a sample of six upper canines of Megaderma vireti from the late Miocene Lissieu locality (Mein 1964) and in

59 a sample of eight Macroderma gigas caninesfrom Rackham'sRoost (QM F23591-6, QM F23S5l-2). Sexual dimorphism, which has been documentedin the dentition of some Tertiary molossids (Engesser 1972, Rachl 1983), may explain the observed variation in the M. Iugdunenslscanine sample. Sexual segregationwithin or between roosts could enhanceor, equaliy,mask the extent of morphologicalvariation exhibited by small randomsamples of extinct megadermatids. The two upper molars from Rackham's Roost referred here to Megaderma richardsi exhibit features that in combination generally characterize megadermatids (i.e. reducedparacone, massive metacone, lingually displacedmesostyle and elongated postmetacrista) but their protocone morphology is quite different. In most megadermatids,the preprotocristaand postprotocristaextend towards (if not touch) the basesof the paraconeand metaconerespectively, thereby enclosing a relatively small protofossa. In M. richardsi, the short postprotocristadrops vertically so that the protocone basin or protofossa opens broadly posterolingually (in M. jaegeri the protofossaopens slightly posteriorly).The moderatelybroad buccal shelf and lingually displacedmesostyle of M I-2are similar to thoseseen in speciesof Megaderma. Assignmentof the small Australian megadermatidto the genusMegadermc has been basedprimarily on its lack of most of the apomorphiesthat appearto characterise Macroderma species.At the same time, it is recognisedthat the genus Megaderma probably representsa paraphyleticgroup (seee.g. Hand 1985). Becauseof its unique Mr-2 protoconemorphology, the Rackham'sRoost could conceivablybe placed in yei another monotypic megadermatid genus. However, the overall affinities of M. richardsi appear to lie with speciesin the Megaderma complex and its referral to that group probably most aptly reflects its relationships.

INTRAGENERICRELATIONSHIPS

Sig6 (1976) envisaged three possible lineages of Tertiary Megaderma: one essentially European and representedby taxa such as M. lugdunensis,M. vireti and M. mediterraneum; a second Euroafrican lineage representedby M. gaillardi; and a third, northwest African lineage representedby the small but already specialized M. jaegeri. It was suggested that the latter two lineages may have arisen from earlier stock morphologically similar to the very large and relatively unspecialized European early Mioiene M. brailloni. Which if any of theselineages gave rise to extant speciesis not clear (Sig6 1976). In 1990, Sevilla described the very small M. Iopezae from late Oligocene sedimentsof Carrascosadel Campo, central Spain, suggestingthat it may be closely related to M. jaegeri and that its many primitive features could place it close to the base of the megadermatidradiation proposedby Hand (1985). In that hypothesis, phylogenetic relationships within the Megaderma complex could not be confidently resolved but it was suggestedthat Megaderma spasma, Iugdunensis and jaegeri (at least) might be closely related.Ziegler (1993) consideredthe GermanM. franconica to be a possible ancestoror close relative of M. gaillardi. Megaderma richardsi is distinguished from most Megaderma species by its small size, with the exception of the comparably small Megaderma lopezae and M. jaegeri. The two tiny non-Australiantaxa sharea conspicuousfeature not present in

60 other Megaderma species:a strong inflexion in the hypocristid half way along its length (Sevilia 1990).This feature is not presentin M. richardsi nor doesthere appear to be any specialrelationship between the small Australianand non-Australiantaxa. Nevertheless, as noted above, M. lopezae and M. jaegeri are among taxa that seem to be most similiar to M. richardsi, along with the Miocene forms M. lugdunensi,sand M. vireti and living speciesM. spasma and Cardioderma cor. These are all small to medium-sized megadermatids and it is possible that size rather than phylogeneticrelationship might be reflectedin theseapparent similarities. Further, these taxa have been inteqpretedto representvarious megadermatidlineages (see Sig6 1976, Hand 1985) and it is likely that many similarities withM. richardsi, particularly in C' and Pa morphology and in M1 talonid and Mr-Z stylar shelf morphology, represent sharedplesiomorphies. Overall, the little Australian bat exhibits an enigmaticmixture of plesiomorphic and autapomorphicfeatures. Some points, however, can be made. Although referred here to the genus Megaderma, M. richardsi appeus to have no special relationship to the living M. spasma and M. lyra. lts autapomorphies(e.g. loss of P2 and unique protocone structure) exclude it from being ancestralto any living taxon, or to the AustralianMacrodermc lineage. Its retainedplesiomorphic features suggest it may be close to ancestral?Asian Megaderma stock that also gave rise to the late Oligocene Spanish M. lopezae, Miocene Moroccan M. jaegeri and European M. Iugdunensis-M. vireti lineage.

PALAEOBIOGEOGRAPHYAND PALAEOECOLOGY

Sig6 (1976) has suggestedthat becauseearly Tertiary megadermatidsappear sporadically in the Europeanfossil record, their centreof dispersalwas probably outside Europe. Species of Megaderrna appeared later in Europe, towards the end of the Oligocene, and radiated in an area including Europe during the rest of the Tertiary before disappearing by the Pleistocene. Concurrent Tertiary radiations evidently occurred in the Asian-Australian region to produce the Macroderma and Megaderma richardsi lineages,both apparently separatingfrom other megadermatidlineages by the late Oligocene. It is possible that Megaderma richardsi was a descendantof poorly represented and poorly understoodOligo-Miocene Riversleigh taxa such as the Dwornamor Variant (Hand 1985) or Henk's Hollow megadermatid(Hand submitted).Unfortunately, little evidence of such a lineage exists. The Australian fossil mammal record is very poor between the middle Miocene and early Pliocene, with only three mammal-bearing depositsof late Miocene age (Alcoota, Ongeva and BeaumarisLocal Faunas)being recognised(Woodbume et al. 1985, Murray & Megirian 1992), none of which has producedsmall mammmals. Altematively, M. richardsi could be the descendantof a late Miocene-early PlioceneAsian immigrant. Its occunencein the AustralianPliocene coincides with the first occurrence of other allochthronous taxa such as emballonurids and murids. The latter, which now comprise almost a quarter of Australia's native non-marine mammal

6L fauna, appear to have first colonised Australia from during the late Miocene or earliestPliocene (Archer et al. 1,99L),and it is possiblethat emballonurids and the ancestorsof M. richardsi also arrived at this time. Murids, at least, appear to have entered Australia via the dry woodland corridors of late Tertiary Indonesia and northwestern Australia, rather than the wetter route through New Guinea and eastern Australia (Godthelp in prep.). No megadermatids,fossil nor living, are known from Papua New Guinea, and no emballonurids are endemic to that landmass (Flannery 1990). During the late Miocene, Australian rainforestsdeclined and drier woodlands spreadin northern and centralAustralia (Archer et al. l99l). The PlioceneRackham's Roost Local Fauna is believed to have been culled from a dry sclerophyl forest or woodland, probably not unlike the dominantvegetation now presentin the Riversleigh area. In semi-arid and desert areas of northem Australia today, the most common mammalian remains in the feeding roosts of M. gigas have been found to be rodents, dasyuridsand small (Douglas 1967,Schulz 1986),abalance also characteristicof the Rackham'sRoost deposit. Among the Rackham'sRoost bat remains are hipposiderids,vespertilionids and two speciesof Taphozous,al| of which probably sharedthe cave with and/or were victims of Macroderma gigas. The small Tertiary megadermatidsM. Iopezae and M. jaegeri are believed to have lived in caves and Megaderma richardsi was probably a cave-dweller too; living Megaderma speciesusually roost in small groups in caves, pits, buildings and hollow trees(Brosset 1966, Nowak & Paradiso1983). All three tiny iossil megadermatidsare known from depositsinterpreted to have accumulatedin warm but probtbly dry habitatsin the Oligocene,Miocene and Pliocenerespectively (Sevilla 1990, Sig6 1976,Archer et aI. l99l). Megaderma richardsi is a very small megadermatid, being approximately one half to two-thirds the size (basedon tooth lengths) of the smallest living megadermatid M. spasma(av. weight23-25 g; Nowak & Paradiso1983). It occurs in the Rackham's Rooit deposit with the Macroderma gigas, the largest living megadermatid (weight tqO-ASg; Richards 1983) and the predator believed largely responsiblefor accumulation of the fragmentary bones comprising the deposit. Thus, two very differently sized megadermatidsoccur together in the deposit, making it difficult to determine what the diet of Megaderma richardsi might have been. Of living megadermatids, Megaderma spasma is closest in size and dental morphology to M., richardsi, if not phylogenetic relationship, and is probably its closest ecological analogue. It is less carnivorous than either Macroderma gigas or its living sympatric congener Megaderma lyra, eating small vertebratesbut favouring grasshoppersand moths for which it foragesirmong trees and undergrowth (Lekagul & McNeely 1911).

ACKNOWLEDGEMENTS

Work at Riversleigh has been supported by the Australian ResearchCouncil, the Department of the Environment, Sport and Tourism, National Estate Programme Grants (Queensland), the Australian Geographic Society, ICI, the QueenslandMuseum and the University of New South Wales. Access to comparative specimenswas generouslyprovided by L. Gibson, T. Flannery, B. Sig6, P. Mein, M.

62 Hugueney,B. Engesser,R. Rachl, G. Storchand H. Felten.Bernard Sig6 and two refereeskindly read and constructively criticized a draft of this paper. The SEM photographs were taken in the Microscopy Unit of the School of Biological Sciences,Macquarie University, Sydney. I am particularly grateful to my colleaguesMichael Archer, Henk Godthelp and all in the Vertebrate PalaeontologyUnit of the University of New South Wales's School of Biological Science, and to Alan Rackham who discovered the Rackham'sRoost depositand has helped excavateit over the past eight years.During this work, I was supportedby a SpecialResearch Grant from the University of New SouthWales.

REFERENCES

ARCHER,M., GODTHELP,H., HAND, S.J.& MEGIRIAN, D., 1989.- Preliminaryoverview of mammaliandiversity, biostratigraphy, correlation and environmental change evidenced by the fossil depositsof Riversleigh,northwestern Queenslan d. Aust. hol., 25: 35-69. ARCHER,M., HAND, S.J.& GODTHELP,H., 1991.- Riversleigh.Reed Books, Sydney' ARCHER,M. & WADE,M., 1976.- Resultsof theRay E. LemleyExpeditions, Part l. TheAllingham Formationand a newPliocene vertebrate fauna from northernQueensland. Mem. Qd Mus.,l'7:3'79- 3v t. BACHMAYER, F. & WLSON, R., 1970. - Small (Insectivora,Chiroptera, Lagomorpha, Rodentia)from the Kohfidisch Fissuresof Burgenland,Austria. Ann. Naturhist. Mu* Wien,74: 533- 587. BATE, D.M.A, 1937.- New Pleistocenemammals from Palestine.Ann. Mag. NaL Hist., (10) 20:. 397- 400. BROSSET,A., 1966.- La Biologie des Chiroptbres'Masson, Paris. BUTLER, P. & GREENWOOD, M., 1965. - Insectivora and Chiroptera. Pp. 13-15. /n: L.B.S. LEAKEY (Ed.), Olduvai Gorge l95l-1961. Volume 1: Faunaand Background.Cambridge University Press,Cambridge DOUGLAS, A.M., 1967. - The natural history of the Ghost Bat, Macroderma gigas (Microchiroptera, Megadermatidae),in WesternAustralia. W. Aust.Natur.,10: 125-138. ENGESSER, 8., 1972. - Die obermiozhne Saiigetierfaunavon Anwil @aselland). Tcitigkeitsberichte d er Nat urfo rsc h e nde n Ge s e ll schaft B ase llnnd, 28: 37-363. FLANNERY. T., 1990. - The Mammals of New Guinea. Robert Brown and Associates, Carina, Queensland. GODTHELP, H., 1988.- Riversleighscene 4: Rackham'sRoost--the beginnings of the modern world. Pp. 8l-83. /n: S.J. HAND & M. ARCHER, (Eds), The Antipodean Ark. Angus and Robertson, Sydney. GzumTHS, T.A., TRUCKENBROD, A. & SPONHOLTZ,PJ.,1992. - Systematicsof megadermatid bats (Chiroptera,Megadermatidae), based on hyoid morphology.Amer. Mus. Novit.,3Ml: l-21. GUERIN, C. & MEIN, P., 1971.- Les principauxgisements de mammiftres miocbneset pliocdnesdu domainerhodanien. Docum. Lab. Geol. Univ. Lyon, H.S., 1: l3l-170. HAND, S.J., 1985. - New Miocene megadermatids(Chiroptera: Megadermatidae)from Australia with commentson megadermatidphylogenetics. A&st Mammal.,8; 543' HAND, S.J., 1987. - Phylogeneticstudies of Australian Tertiary bats: a summary of PhD thesis. Macroderma,3:9-12. HAND, S.J., submitted. - Macroderma malugara, a new Miocene megadermatid (Mammalia:

63 Microchiroptera)from Australia,with commentson broader aspectsof megadermatidevolution. Geobios,Lyon. HAND, S.J.,DAWSON, L. & AUGEE,M., 1988.- Macrodermakoppa, a new Pliocenespecies of false vampirebat (Microchiroptera:Megadermatidae) from Wellington Caves,New South Wales. Rec.Aust. Mul. 40:.343-351. HILL, J.E.& SMITH,J.D. 1984.- Bats:A NaturalHistory. Rigby, Adelaide. HONACKI, J.H.,KINMAN, K.E. & KOEPPL,J.W., 1982.- MammalSpecies of the World. Allen Pressand Association of SystematicCollections, Lawrence. LEKAGUL, B. & McNEELY,J.A., 1977.- Mammalsof .Association for theConservation of Wildlife,Bangkok. MEIN, P., tg&.- Chiroptera(MiocBne) de Lissieu(Rh6ne). 8f CongresSoc. Sav.Lyon,237-253. MEURISSE,M., MICHAUX, J. & SIGE,B., 1969.- Un remplissagekarstique )r micromammifbres du Miocdneinffrieur )r Ia Serrede Vergds,prds St-Arnac (Pyr6n6es-Orienules). C.R. somm. Soc. 96ol. Fr,1969(5):166-167. MURRAY,P. & MEGIRIAN,D., 1992.- Continuityand contrast in middleand late Miocene vertebrate communitiesfrom the NorthernTenitory. TheBeagle, Records of the NorthernTerritory Museumof Artsand Sciences,9: 195-218. NOWAK, R.M. & PARADISO,J.L., 1983.- Walker'sMammals of theWorld. Volume I. JohnHopkins UniversityPress, Baltimore. RACHL,R., 1983.- Die Chiroptera(Mammalia) aus den mittelmiozdnen Kalken des Nordlinger Rieses (Siiddeutschland).PhD Thesis, Ludwig-Maximilians-Universifiit, Miinchen. REMY. J.A., CROCHET,J.-Y., SIGE, B., SUDRE,J., DE BONIS, L., VIANEY-LIAUD, M., - GODINET. M., HARTENBERGER,J.-L., LANGE-BADRE, B. & COMTE, 8., 1987. Biochronologiedes phosphorites du Quercy:mise ir jour deslistes fauniques et nouveauxgisements demammifbres fossiles. Miinchner geowiss. Abh,A' l0: 169-188. RICHARDS,G.C., 1983.- GhostBat, Macrodermagigas. Pp. 292-93'1n: R. STRAHAN (Ed.)' The CompleteBook of AustralianMammals. Angus and Robertson, Sydney. SCHULZ, M., 1986.- Vertebrateprey of the GhostBat, Macrodermagigas, at Pine Creek,Northern Tenitory. Macroderma, 2: 59-62. SESE,C., 1986.- Chiroptera(Mammalia) del yacimientodel Miocenomedio de Escobosade Calatanazor(Soria, Espafla). Estudios geoL, 42: 33-39. SEVILLA, P., 1990. - Rhinolophoidea(Chiroptera, Mammalia) from the Upper Oligocene of Carrascosadel Campo (central Spain). Geobios, Lyon, 23: 173- I 88. SIG6,B., 1967.- Les chiroptdresoligocdnes de Saint-Victor-la-Coste(Gard). Etude prdliminaire. C'R. somm.Soc. giol. Fr,1967(4):163-l&. SIGE, B., 1968.- Les chiroptdresdu Miocbneinfdrieur de Bouzigues.I. Etude syst6matique. Palaeov e rteb rata, Montryllier, I : 65-I 33. SIGE,B., 1974.- Prdsenced'un Megaderma(Manmalia, Chiroptera) dans le PldistocEneinf6rieur ir Sdte(H6rault). G6ol. mddit' Marseille,l:97-lW' SIGE, B., 1976.- Les Megadermatidae(Chiroptera, Mammalia) miocdnesde B6ni Mellal, Maroc. G6ol.midit.,Marseille, 3: 7l-86. TOPAL,G.,1974.- Thefirst record of Megadermain Hungary.Vertebr. Hung.,15: 95-104. TROUESSART,8.L.,1897-99.- CatalogusMammalium tam viventium quam fossilium. Berlin. WEITHOFER.A.. 1887.-Zur Kenntnissder fossilen Chiropteren der franzosischenPhosphorite. Sirz'- Ber.Akad. Wiss. Wien,96:341.

64 WOODBURNE,M.O., TEDFORD,R.H., ARCHER, M., TURNBULL, W.D., PLANE, M.D. & LUNDELruS,E.L., 1985.- Biochronologyof the continentalmammal record of Australiaand New Guinea.Spec. Publ., S. Aust.Dept Mines and Energy ,5:347-63. ZAPFE,H., 1950.- Die Faunader miozdnenSpaltenfullung von Neudorfan der March (C.S.R.): Chiroptera.Sitz.-Ber. Akad. Wiss. Wien, Abt I, 159:5l-64' ZIEGLER,R., 1993.- Die Chiroptera(Mammalia) aus dem Untermioziinvon Wintershof-Westbei Eichst2itt(Bayern). Mitt. Bayer.Staatsslg. Paldont. hist. Geol., Miinchen, 33:119-154. l

APPENDIX

Characterssurveyed for phylogeneticinformation were: CI anterior accessorycusp development,shape, height and separationof posterior accessorycusp with respectto the paracone;

O presenceor absenceofP2;

a paraconeposition in Pa, developmentof buccal cingulum, heel sizeand shape;

a Mr-2 postmetacristalength, reduction of postparacristaand premetacrista,mesostyle height and degree of lingual displacement,presence/absence of paraloph,heel size and direction; M r trigonid and talonid compression (transverse and anteroposterior),relative heights, lengths and widths of trigonid and talonid, pre-entocristiddevelopment, bucial displacementof hypoconulid,hypocristid inflexion (posteriordisplacement of hypoconulid),metaconid contribution to cristid obliqua; high crowns.

65 LEGENDS OF PLATES

PLATE 1

Megaderma richardsi n. sp. from Rackham's Roost Site, Riversleigh Station, northwesternQueensland, Australia. Holotype, QM F23619, right maxillary fragment with Cr and Pa: A-A', stereopairs, occiusalview. QM F23598,RMr: B-B', stereopairs,occlusal view; C-C', stereopairs,oblique-occlusal view. QM F23599, RM2: D-D', stereopairs,occlusal view; E-E', stereopairs,oblique-occlusal view. Scale= 1 mm.

PLATE 2

Megaderma richardsi n. sp. from Rackham's Roost Site, Riversleigh Station, northwesternQueensland, Australia. QMF23620, RCr: A, buccal view; B, lingual view; C-C', stereopairs,occlusal view. QM F23600,LM 1:D-D', stereopairs,occlusal view; E, buccal view; F,lingual view. Holotype, QM F23619, right maxillary fragmentwith Cr and Pa: G, lingual view; G', buccal view. Scale= I mm.

66 PLATE I

D' PLA]IE 2