Quick viewing(Text Mode)

Fossil Mammals of Rivensleigh, Northwestern Q,Ueensland

Fossil Mammals of Rivensleigh, Northwestern Q,Ueensland

FossilMammals of Rivensleigh, NorthwesternQ,ueensland: PreliminaryOverview of , Correlationand EnvironmentalChange

MichaelArcherr, Henk Godthelpl, SuzanneJ. Handl and Dirk Megirian2

Abstract

Aspects of the results of studies of the -rich Cainozoic deposits of Riversleigh, nofthwestern Queensland, are reviewed. A summary of five selected Riversleigh faunas representing the primary cerioCs of the region's Cainozoic history is provided. Faunal and environmental changes over the last 25 000 004 vears n the Riversleigh region are identified and changes in s rainforest comlnunities over the same period are discussed. Evidence for the origin of Australia s modern mammal qroups front ancestors now known to have lived in the l'ertiary rainforests of northern Australia is reviewed. fhe geological record for Riversleighs more than 100 local faunas is considered. At least three primary inteNals ol Oligo-Niocene deposition, one of and ntany of and Holocene depc.ssttbn are identified. An appendix is provided in which the princioal faunal assemblages from Riversleigh are allocated to these depositional intervals. The evidence for ccrrelating Riversleigh local faunas with faunal assemblages in the rest of Australia and the world is reviewed. Oligo-Niocene, Pliocene and Pleistocene anci monotreme correlate Riversleigh's local faunas with others from central and eastern Australia; correlate them with faunal assemblages in Europe: rodertts correlate them with Pliocene assemblages in eastern Australia.

Ke.v words: Monotremata, Marsupiaira.Chiroptera. Muridae. Pisces,Amphibia, Reptilia.Aves. A4am- malia. Evolution,. Rainforest. Freshwater Limestone. Tertiary, . . Pliocene. Pleistocene,Holocene. Correlation,Stratigraphy. Palaeoecoiogy.

INTRODUCTION summary of the currentbiostratigraphic data with a focus on Riversleigh'smammai record and its interpretationin Despitelon-cl-term interest in the distinctivemammais a local as well as general context. ManV students and of modern Australia.understanding about their origin colleaguesinvolved in analysesof Riversleighrnaterials and diversificationhas laggedfar behind that of all other have contributedto the ideas presentedhere and tc the conrinenis except Antarctica(Alcher anci Bartholomai growing awarenessof the significanceof the Riversieigh i 978, Arcner 1984, Hand i 984. Woodburne ef al 1985). discoveries. The reasons for this are manifold but primary among them has been the apparent scarcityo[ rich Cainozct.. Although Riversieighs fossil marnmal deposits wert: fossil deposits.Until 1983 rnost investigationsinto the: onginallythought to representone distincrassemblage early diversificationof Australia'sterrestrial of medial Mioceneanimals (the RiversleighLocal Fauna focussecion the Oligo-Miocene deposits of the Lakt of Tedford 1967.from his SitesA-D). we nou' recognise Eyre Basrnand Frome Embayment of South Australia about 100 local faunasfrom this area representrng:l. at and isolated early Miocene faunal assemblagesfrorn least thre--periods of Oligo-Miocenetime (about 25 to year richly fossiliferous Tasmania. In that the extent of l5 miliion vears BP) which we define as Svstems A-C Tertiary depositsat Riversleigh(Fig. I ), some of whrc[ witn System A beinq between late C)ligoceneto earli' nad beeri systematicallystuciied as eariv a:' 1967 Mrocerre in arte, Systern B possibly early to nriddle (Tedford (Archer, 1967), began to become apparent Mioceneand Systen':C oossiblymiddle Mioceneto early Hand and Codthelpl986). lateMiocene: 2, at leastone Plioceneassemblage (about Examination of these Riversleigh deposits, which 3-5 millionyears BP1. Rackham's Roost Local Fauna; 3, appear to span the last 25OOO000 years, has led to manv Pleistoceneassemblages (between 2 million and many publications (e.9., Archer, Hand and Codthelp l0 000 yearsBP) from caveand ancientriverine deposits 1988b, Anon. I 9B9a,b)and on-going researchfocussed such as TerraceSite; and 4 many Holocene(r.e., Recent) on individualtaxonomic groups. Here we presenta broad assemblages(less than 10 000 yearsBP) some ol which appear to span the arrivalof introduced mamnrals and 'Nodhern Universi\_ol f'lew South Wales lrrrlor\'Museuni Europeans.

June 1989 Australian Zoologist. Vol. 25(2,) 29 This overview includes: 1, a faunal list from Riversleigh'sUpper Site Local Fauna to introduce the biostratigraphic issues involved; 2, a summary of speciesJevel diversity (for mammals) in six representa- 'Site tive, different-aged local faunas (Riversleigh LF from D'; Upper Site L.F; Bitesantennary LF; Dwornamor LF from Cag Site; Rackham's Roost LF; and the modern (stillJiving) fauna of RiversleighStation); 3, a considera- tion of apparent trends in Australia'srainforest mammal communities through time; 4, a summary of what we at present interpret to be the stratigraphic relationships of representativeRiversleigh local faunas; 5, some of the bases for approximating the ages of Riversleigh'slocal faunas; and 6, an appendix listing the prir.cipal local faunas under study with a tentative assessment of their relative ages. Systematic nomenclature used for marsupial groups follows Aplin and Archer (1987); that for bats follows : Guttof Z, Hand (1984); that for non-Riversleigh faunal Carpentaria|1 assemblagesand stratigraphyfollows Archer (1984) and Woodburne et a/. (1985). In the list, the author- ship of a taxonomic name is followed by the author's last CE a comma and the date of publication. [f no o name, = o Normant0n comma separatesan author from the date of publication E (the format also used here for a conventional text refer- G, Burketown a work in which the taxon is l*J ence), the reference is to O Riversleigh discussed or perhaps re-defined. = & The term rainforest covers a multitude of vegetation t! l. types (e.g., Webb, Tracey and Williams 1984; Webb, I Camooweal Tracey and Jessup 1986). Our interpretation of the &, (3 o Oligo-Miocene rainforests of Riversleigh,based on the z, faunas they contained, are that they were dense, species- Mountlsa 9zg persist -l rich gallery rainforests probably similar to those that km today in mid-montane New Guinea. Fossil collecting localities in Riversleigh are called 'sites' (e.g., Upper Site, Cag Site, Outasite). Regionally Fig. 1. The location of Rivercleigh within the Gulf country of to be superpositionally-related northwesternQueensland. (After Archer, Hand and Godthelp clusteredsites that apPear 1986). (differing in age but not significantly in position) and/or space-related (spatially isolated but approximately con- temporaneous) comprise what we have come to call 'systems'. Considering the vast exPanse and complexity of the Our aim in using these concepts is to draw Riversleighsediments, the large number of local faunas, attention to the apparent age-related (sequential or the early state of analyses of both the geology and the comparable) nature of sites that occur as geographically defined clusters (the Systems) in the Riversleighregion. faunas and the fact that this summary does not include 'System' the preliminary results of the non-mammal data under Our restricted use of the term should not be study, the diagrams and summaries presented here confused with the stratigraphic analogue of the 'period' (e.g., must be treated as preliminary assessments'lt would be chronologic term the Tertiary System and naive of us to presume that after a combined total of Tertiary Period;Raup and Stanley 1971). Nor should our only 12 months spent in the field at Riversleigh,we would Systems be interpreted as regional faunas or be aware of all major features of the area. With each assemblages of contemporaneous faunas because (at successive expedition, more is learned about the inter- Systems B and C least)are clearlytime-transgressive. relationshipsof the various sediments within the region We regard each site in a Riversleigh System to be and about the faunal assemblagesthey contain. As more potentially unique in time as well as space. In contrast, of the various sediments are processed, the diversity, each Riversleigh System is interpreted to have accumu- relationships and non-Riversleighoccurrences of some lated over a fwed interval of time within a specific area. of the Riversleighanimals necessitatea continual revis- For example, Gag Site (which may be middle Miocene 'big' ing or refining of the picture. in age) occurs as an isolated bone-rich deposit within

30 Australian Zoologist, VoL25(2) June 1989 the Gag Plateau (a major topographic feature in the Upper Site Local Fauna Riversleighregion) and is part of System C which as a Mollusca whole appears to span middle Miocene to early late Gastropoda(snails; pers. comm. Coleman, Australian Miocene time. However, all of these preliminary strati- Museum) graphic concepts will require revision as understanding Family? about the nature and relationships of the sediments --Genusand sp. develops. Family? --Genusand sp. An assemblage of from a single site is called Family? a local fauna (often abbreviated LF; plural, lfs or LFs if --Genusand sp. specific sites are named; e.9., the Upper Site LF, the Arthropoda (pers. comm. P. Jell, Queensland Outasite and Upper Site LFs). A few local faunal names Museum) differ from their corresponding site names (e.9., Diplopoda Dwornamor LF from Cag Site) but most are the same Family? (no. of genera undet.) as the site name, commonly but not alwaysless the word --Genusand sp. 1 'site' (e.g., Henk's Hollow Site, Henk's Hollow LF; but --Genusand sp. 2 Upper Site, Upper Site LF). Each of the Riversleigh --Cenus and sp. 3 Systems contains many non-identical and different aged --Genusand sp. 4 local faunas.A glossary of additional terminology is pro- Insecta vided. Hymenoptera Formicidae (ants; no. of genera undet.) --Genusand sp. 1 --Genusand sp. 2 THE UPPER SITE LOCAL FA(NA: Hemiptera AN INTRODUCTION TO RIVERSLEIGH Cicadidae (cicadas) --Genusand sp. 1 Of the many faunal assemblagesfrom Riversleigh,we Coleoptera chose the Upper Site Local Fauna to introduce this Curculionidae (weevils) review of Riversleigh'sbiostratigraphic record. As will be --Qenusand sp. 1 discussed in more detail below, by any modern stan- Families? (no. of genera undet.) dards, this is a very diverse local fauna with almost twice --Genusand sp. 1 the number of mammal families and twice the number --Cenus and sp. 2 of marsupial species of any surviving Australian or New Crustacea Cuinean ecosystem. Isopoda (slaters) Family? Upper Site was discovered on the inclined slope of Cenus and sp. 1 Codthelp Hill by H. Godthelp in 1985. It stands proud of Pisces the surrounding slope as an erosional remnant of a Teleostii slightly higher level thus earning its name. We quarried Family? approximately 1 tonne of fossiliferous limestone from --Qenus and sp. this site in 1985 and another 0.5 tonne in 1986, almost Amphibia all of which is now processed. After dissolution of the Anura (frogs; pers. comm. M. Tyler, Universityof limestone with dilute acetic acid and recovery of the Adelaide) small to large specimens, the acid-insoluble residues Leptodactylidae (leptodactylid frogs) were sorted under a microscooe to recover isolatedteeth -Crinia sp. and other tiny items. -Kyarranus sp. 7 -Kyarranus sp.2 The anirnals recovered included a diverseassemblage --Lechriodus intergerivus Tyler, 1989 of vertebrates as well as, perhaps most surprisingly, -Limnodynastes sp. 1 uncrushed arthropods. Wrinkled sheets of what we --Limnodynasfes sp. 2 interpret to be algal mats were also recovered. Many of Hylidae (hylid frogs) the invertebrate remains (in particular, dozens of --Litoria sp. 1 'Cenus coleopteran larvae; and sp. f in the list below) --Litoria sp.2 occurred in vertical tube-like extensions of these mats --Litoria sp.3 which may have grown down into the bottom mud of --Litoria sp. 4 the pond or hung below the algal mats if they floated. --Litoria sp. 5 Although similarly preserved invertebrates are now --Litoria sp. 6 known from other sites, the Upper Site invertebrate fauna --Litoria sp.7 is so far the most diverse. --Litoria sp. 8

June 1989 Australian Zoologist, Vol. 25(2) 31 Reptilia (; pers. comm. J. Squamata Muirhead) Ophidia (snakes;pers. comm. J. Scanlon, University - dicksoni Muirhead and Archer, of New South Wales) 1989 Pythonidae (pythons) Peramelemorphia (bandicoots; pers. comm. J. - -cf. lvlontypythonoides riversleighensrs Smith Muirhead) 'V.D.' and Plane.1985 New family I (extinct bandicoots) Madtsoiidae (large extinct snakes) --Cenus and sp. new 1 --Cenus and sp. new --Genus? and sp. new 2 Typhlopidae (blind burrowing snakes) ?Peroryctidae (forest bandicoots) --?Ramphotyphlops sp. --Cenus and sp. new 1 Elapidae(poisonous snakes) --Genus?and sp. new 2 --New genus?and sP. --Cenus?and sp. new 3 Lacertilia --Cenus?and sp. new 4 Agamidae (dragons; Pers. comm. --Cenus? and sp. new 5 J. Covacevich and P. Couper, Queensland --Cenus?and sp. new 6 Museum) Notoryctemorphia --Physignathus (Cray, 1831) sp. c/i P. leseurii Notoryctidae (marsupial moles; pers. comm. --Physignathus sp. M. Cott, Universityof New South Wales) --Cenus and sp. new --Cenusand sp. new Gekkonidae (geckoes;pers. comm. M. Yalkaparidontia Hutchinson, La Trobe UniversitY) Yalkaparidontidae (extinct family) --[not yet determinedl --Yalkaparidon coheni Archer, Hand and Scincidae(skinks; cornm. M. Hutchinson) Pers. Codthelp, 19BB --[many taxa but not yet determined] Varanidae (monitor lizards; pers. comm. R. Molnar, Q.ueenslandMuseum) --[not yet determined] Phascolarctomorphia Crocodilia () Crocodylidae(crocodiles; pers. comm. P.Willis, :Litokoala n. sp. Universityof New South Wales) --Cenus and sp. indet. --Cenus? and sp. new; cf Q.uinkana ?New family 2 Aves(birds; pers. comm. W. Boles,Australian Museum) --Cenus and sp. new Ratites Vombatomorphia Dromomithidae (giant mihirung flightlessbirds) (extinct marsupial lions) --Baraweftornis tedfordi Rich, I 979 -- sp. cf. W. oldfieldi Clemens and --Bullockornis sp. cf B. planer Rich, 1979 Plane,I974 Casuariidae (emus and cassowaries) --Genus and sp. new cf Priscileo --Dromaius gidiuPatterson and Rich, l9B7 Wynyardiidae(extinct sheep-sizedbrowsers) Passeriformes(no. of genera not yet determined) --?Namilamadeta n. sp. lvlenuridae(lyrebirds) (giant extinct browsers) --Genus?and sp. new -- tirarensisStirton, 1967 Family(ies)? --Genusand sp. new --Genus and sp. new I --Cenus?and sp. new --Cenusand sp. new 2 New family 3 --Cenusand sp. new 3 --Genusand sp. new Mammalia Marsupialia Burramyidae (pygmy possums) -- n. sp. Dasyuridae (tiny to medium-sized carnivorous --Cercarfefusn. sp. ;pers. comm. J. Muirhead, Universityof New South Wales) (brushtail possums and ) --Cenus and sp. new 1 --Trichosurus sp. c/l T. dicksoni Flannery and --Cenus and sp. new 2 Archer,19B7a --Cenus and sp. new 3 --?Stngocuscus sp. --Cenus?and sp. new 4 Pilkipildridae (extinct small omnivorous ?Dasyuridae possums) --Genus new and sp. --?Djilgaringa sp.

32 Australian Zoologist, Vol-25(2) June 1989 Vespertilionoidea ? (gliders and relatives) Molossidae (free-tailbats) --Genus and sp. new --Cenus and sp. indet. New family 4 Family? --Cenus and sp. new 1 --Genusand sp. indet. --Cenus and sp. new 2 (ringtail possums) The questions we find ourselves askinq about --Paljara n. sp. Riversleighassemblages of this kind form the rJbrtun." --cf. Pildra sp. of the rest of this paper. Can we determine the nature of --cf. Pseudochlrops I (small) the palaeoenvironments that these assemblages rep- --cf. 2 (middle) resent (e.9., rainforest,open forest, woodlands etc.) and --cf. Pseudochlrops 3 (large) can we distinguish different microenvironments within ?New family 5 these broader categories?How did faunal diversrryvary --Cenus and sp. new with time in the Riversleighregion? What happened to (feather-tailpossums) the various groups present in the region during the time --New gen.? cf- Acrobates interval represented by the Riversleigh deposits? What --Genus?and sp. new subsequently happened throughout Australia to faunal Macropodoidea diversitywithin the ecosystems represented by the older (rat-; pers. comm. B. deposits of Riversleigh?How many different kinds and Cooke, Brisb. College of Adv. Ed. and ages of fossil deposits are represented in the Riversleigh u.N.s.w.) region? Can we correlate the kinds and aqes of the Hypsiprymnodontinae variousRiversleigh animals with those previoJslyrnown -- sp. new from Australiaand the rest of the world? Propleopinae -- sp. cf. E. r'maArcher and Flannery, CAINOZOIC CHANGES IN M4qry1qL6N DMERSITY: I 985 THE EVIDENCE FROM RMERSLEIGH Balungamayinae --cf. Wabularoo n. sp. One of the most striking attributes of the Oligo- --cf. Gumardee n. sp. Miocenelfs of Riversleighhas beentheir high family-level --Cenus and sp. new I and species-level diversity compared with previously --Cenus? and sp. new 2 known Australian fossil faunas. Here we will attempt to --Cenus? and sp. new 3 quantifyand comment on this feature. Potoroinae --Wakiewakie lawsoni Woodburne, 1984a Table 1 presentsa comparison of mammal specles- (Codthelp,Archer, Hand and Plane 1989) leveldiversity in six of the Riversleighlocal faunas, includ- ('ordinary' kangaroos; pers. ing the modern environment:Riversleigh LF 1:'p;u. tn comm. B. Cooke, Brisb.College of Adv. Ed. Table 1); Upper Site LF (-'U.'); BitesantennaryLF ('B. ); and U.N.S.W.) DwornamorLF (:'D.'); Rackhams RoostLF (:'R.R.'); Balbarinae and the modern local fauna of the Riversleighregron --?Nambaroo sp. new I (-'Mod.'). --?Nambaroo sp. new 2 Species-level ?Marsupialia diversityhas been calculated for family- level units. However, a few taxa cannot Yingabaianaridae (extinct family of probable be allocated to modern families marsupials) and, as a result, they are listed at what- ever le'relof relationshipseems most appropriate(e.g., --Yingabalanara richardsoni Archer et a/., 1989 'vombatomorphian Placentalia the taxon from the Upper Site LF is clearly Chiroptera not a vombatid but it appears to repiesent a taxon close Microchiroptera to this group). Rhinolophoidea The numbers of taxa shown in all of the extinctlfs are Megadermatidae(ghost bats) also undoubtedlyminimal estimates.In most cases --folacroderma these sp. numbers will rise once the groups are systematically Hipposideridae(leaf-nosed bats) studied. Further, the data for the recently discovered --Brachipposideros sp. I BitesantennaryLF are very incompletebecause -- to date Brac hipposideros sp. 2 we have only processedabout 0.25 tonnes (i.e., --Brachipposideros about sp. 3 3%) of the availablematerial. We have added it in here --Brachipposideros sp. 4 because it representsundoubted Oligo-Miocene cave -- B rach ipposid eros sp. 5 sediments, one of the major types of sedimenr rep- --Cenus and sp. indet. 1 resentedin the Riversleighregion. --Cenus and sp. indet.2 June 1989 Australian Zoologist. Vol. 25(2) 33 Ihe fossl/rsedskull and lower iaw of an extinct Riversleigh rat-kangarco, Bettongiamoyesi. This was a distant relative of living Eettongs.

Thejaw and teeth of the giant RiversleighPython known as Montypythonoidesriversleighensis. lt may commonlyhave reached lengths ot seven metes-

The skulland lower jaw of Yalkaparidon coheni, one of the unique Riversleigh specles now placed in its own order ol mam m als, the Yal kap aridonti a. Table l. Species diversity by family Local Fauna (millions Riv. U. of years) D. ?20-25 ?20 R.R. Mod.l ?15 Famitt' 3-5 0 Thylacinidaq3 Dasyuridaea 1 I z cf Peroryctidae I 6 Peramelidae l 6 ?3 7 Perameloidea5 Notoryctidae6 z 3 PhascolarctidaeT I ?Phascolarctoideas z Diprotodontidaee ; i Palorchestidaelo ? ; ?Wynyardiijsgrt 2 Vombatomorphial2 1 i ,1 Thylacoleonid6gr: 1 Burramfdaela 1 Phalangerida6l5 z z; Pilkipildrid6gto z J PseudocheiridaelT 1 q I Petauridaels q Petauroideaie 1 I I Petauroidea2o 1 I Acrobatidae2r i 2 I Potoroidae22 ?2 1 Macropodidae23 3 8 8 Yalkaparidontid6q2a I z I 2 Vngabalanari6lss2s 1 ; I Hipposideridae26 I Megadermati6lsq2T ?1 7 ; Molossidae2s 1 ; z Emballonuridae 1 z ?1 Vespertilionidae z I Pteropodidae ?1 ^ ?2 Muridae2s B Canidae3o ?3 Hominidae 'a 7 fachyglossidae 1 1 Totals (min. numbers): l I6 63 Additionalfamilies 5B 33 from other Riversleighlo;ilu"r,- 3B

A..Cainozoic changes in diversity and habitats within the Riverslelgh interpretthe region -,We olderRiversleigh lfs (i.e.,Systems L) to representrainforest A_ Allowing communitiesfor hve main for the artificiallylow reasons:1, high species recovered number of taxa so far aiuersity"in'"r"ri.i#a ro.ur from the newlydiscover"J faunas(e.g., 63 mammalspecie.s the low Al*"r,ifrary LF, in UpperSite LF) indi_ number of mammals k".*,f;;;lr," cate a rich, finely_divided.and water,facies o""p_ rt"biJi;.;,.;nstant) representedby the resourcebase;2, as a corre]ate, probably Riversl",g";'LF,""a,n" .;;;1"^,;:dii! guirds low diversitvof the largely invotvingup to 6 species collected ,i"gla"lrnutia_ of small ,ftil;; -;mmals assemblaoer.om. nuc[tj,-n,r'"n11-., within partrc.u.lar below).it is t."" groups:::..::9,"* .morphologicallysimitar evidentIhat species (e.9.,the perorvctid_likeuunji.o"G'ilr{. has dropped .aiveisit in't"t;-region J. .o.-. markedlyfrom a high in Muirhead).a situatic to early the lateOliqocene Miocenefaunls to a low in the pliocene/duater_ nary.The drop ::.'_o_y,."au"i,ity;;il',,'il.#;'j"15*#":"jT:i€i in endemic ramiry_1g;e]ai"irr'irv.oi,", of sympatricobtigate teaf_eate.s same period ,i," raunasIlyTb"l(e'g', lri rt"noi"ro"uf is evenmore striking,f--'"tl"r"ri":a 9 speciesof serenodontp.""j.j"#i). least7 in addition fut : phascolarctoid, to thoserepresented in the Upper .selenodont i" il;'d;;a=mir ""a LF; seeTabte I ptus Site Indrcatesthat r_n) footnotesii" ,r,"-cjrig;_'m"i5!"n",o manv differentspecies .i i.""rl..rr"a no more than 14 in the within relativelysmall areas, mooerncommunity.. a common featureof rain_ lorestsbut not sclerophylltorests; 4, the presenceof manv June1989 Australian Zoologist, Vot. 25(2) 35 taxa otherwise only known from rainforest communities rhinolophoids (specifically hipposiderids and mega- (e.g., IFlannery and Archer 1987a], dermatids) in the older deposits gives way to higher Hypsiprymnodon IFlannery and Archer 1987b], diverse diversityof vespertilionidsin the younger deposits (Hand Pseudochirops-type pseudocheirids, Cercaftetus 1987; see below). This inversion reflectsa similar change caudatus-like burramyids, diverse peroryctids and log- over the same period in Europe (Hand 1984, 1987) and runner birds lOrthony<; Boles 19881);and 5, while there probably reflects a world-wide late Cainozoic change are vast numbers of browsing marsupials, there is a from more tropical conditions in the early to middle complete absence of grazers (i.e., no macropodine Tertiary to more xeric conditions in the late Tertiary to macropodids or vombatids) suggesting that the forest Quaternary. canopy was closed except over pools where aquatic The paucity of marsupials known from the Pliocene grasses may have grown Rackham's Roost LF may be anomalous because it The last point deserves elaboration. High-crowned could be the result of limited prey selection by mega- macropodids do not make their first appearance until dermatid bats. This fauna is dominated by small the Bullock Creek LF (middle Miocene Icentral Northern mammals of a size able to be carried by these carnivorous Territoryl; see below) and even there they are extremely bats. However, when we compare species- and family- rare (Cooke et al. in prep.). Undoubted grazers do not level diversityin this site with that from Gotham City Site, become a significant feature of Australia's mammal another megadermatid-accumulated assemblage but communities until the early Pliocene (the Bluff Downs one from the Oligo-Miocene System C, the marked drop Inortheastern Queensland], Chinchilla Isoutheastern in diversity is still evident. For example, although Queenslandl and Bow Ieastern New South Wales] LFs; acrobatids, burramfds, petauroids, small phalangerids, Archer 1984, Woodburne et al. 1985). Early reports of peroryctids and a new family of tiny perameloids (all grass pollens in the Oligo-Miocene Etadunna Formation small enough to be megadermatid prey) are present in in South Australia (e.9., Harris cited in Woodburne ef a1. the Cotham City LF, they are missing from the 1985) may represent aquatic rather than emergent Rackham's Roost LF. grasses(H. Martin.pers. comm.). We consider that changes in Riversleigh's snake We are unclear about the precise nature of the Oligo- assemblagesover this same time period are a significant Miocene rainforests of Riversleighbecause we have no part of the whole picture because they may provide infor- direct record of the plant species themselves. However, mation about when the ancestral Cregory River began the high mammal diversity in single locations (particu- to flow into the Culf of Carpentaria.Scanlon (l988) notes larly for sl,rnpatric obligate folivores: at least 12 in the that while pythonid, madtsoiid, typhlopid and elapid Dwornamor LF from System C) and the presence of a snakes are well-documented from the Oligo-Miocene wide size range of (from mouse- to cow-size) Riversleigh deposits, the aquatic acrochordids (file suggests a comparably high diversity of plant species snakes) are not. File snakes are common in the modern and muiti-level distribution of these resourceswithin the Cregory River as they are in all of the river systems of forest.This suggests a complex lowland galleryrainforest northern Australia open to the sea, being aggressive with a partially open understorey. colonisers of any accessiblefish-rich freshwater bodies. They are also distributed widely around the northern We interpret the marked drop in species-level and areas of the Indian Ocean and almost certainly entered family-level diversity between the Oligo-Miocene and Australian freshwatersfrom the oceans to the north. The Pliocene/Quaternarylocal faunas of Riversleighto mirror oldest file snakes (Acrochordus dehmi) from Siwalik the regional collapse of the rainforest communities and sediments of medial Miocene age in southern Asia are habitats. their replacement with cpen foresVsavannah clearlyderived members of the group. Hence we suggest Looked at more carefully,it is evident that most of the that the absence of file snakes from all of Riversleigh's losses in species-leveldiversity have been among the more aquatic Oligo-Mioceneunits (e.g.,Site D, D-Site marsupialsand most of the gains by addition of placental equivalent units, Melody's Maze Site, Ringtail Site, groups (e.g., murids, vespertilionids,emballonurids, Quentin'sQuarry Site and Bob's BouldersSite), presuming rhinolophids and pteropodids). they were present at the time in the waters to the north of the continent, indicates that the Riversleighdepositional Considering marsupials over this period, while many basins were not at that time open to the sea. The ancestral groups suffered extinction or severe reduction, the northward flowing Cregory River probably did not estab- grazing macropodids markedly increased (see below). lish itself until late Miocene or earlv Pliocene time. Conversely the generally browsing to omnivorous potoroid kangaroos markedly declined during the same B. tvlid to late Cainozoic chanoes in Australian period. rainforest communities Within the placental groups over this same period, a Mid to late Cainozoic changes in habitats and diversity change is evident in diversity. High diversity of within the Riversleigh region appear to include loss of

36 AustralianZoologist, Vol 25(2) June 1989 rainforest probably by late Miocene to Pliocene time. many rainforest groups (e.g., dasyuroids, perameloids, What happened to Australia's rainforest mammals after hipposiderids and petauroids) are diverse on either side this time? Some lineagesbecame extinct;some persisted of the early Pliocene. Consequently, we presume that in rainforest remnants in eastern Australia; and some few of the processes responsible for collecting the sub- responded to the loss of rainforestby adaptationto more stance of the Upper Site LF were operating in the Hamil- mesic habitats (see below). ton area. lt is also possiblethat the lower diversityreflects a different kind of rainforest (i.e., probably cool (Fig. We have compared 2A-D) species diversitywithin temperate Nothofagus rainforest at Hamilton vs warm mammal families for four Cainozoic rainforest mammal tropical gallery rainforest at Riversleigh).Unfortunately, faunas: (System Opper Site LF B of Riversleigh'sOligo- because the Hamilton LF is Australia's only-known (an Miocene sequence);Hamilton LF early Pliocenerain- Pliocene rainforest mammal assemblage, we have no forest assemblage from western Victoria; Turnbull and other link between the Oligo-Miocene lfs of Riversleigh Lundelius I970 and various papers by Turnbull et al. in and those of the present. Archer 1987); the modern Atherton rainforest mammal assemblage of northeastern Queensland; and mid- In contrast, mammal data for the mid-montane rain- montane P.N.G. (data provided by T. Flannery for living forests of Papua New Guinea and the rainforests of the mammals in this habitatin Papua New Guinea). Atherton Tableland have been obtained by comprehen- sive surveysof large tracts of forest and hence are more We chose these four for the following reasons. The likely to approximate total regional diversity than the Upper Site LF is now one of the oldest-known undoubtecj fossil local faunas. rainforest mammal assemblages from Australia32.The Hamilton LF is the only known Pliocene rainforest Based on his experiencewith mid-montane rainforest mammal assemblage from this continent. Although faunasof P.N.G.,Ken Aplin (pers.comm.) suggeststhat Riversleigh's Rackham Roost is Pliocene, it appears to diversity in more restricted areas of this habitat (e.g., represent an open forest or woodland habitat.There are ranging altitudinallybetween I 000-1 500 metres) would no known early or mid Pleistocene rainforest mammal be significantlv less than that given in Fig. 2D. Specifi- assemblages.The Atherton rainforestis among the richest cally, he would expect to find in a region of this more of Australia'stropical rainforest refugia. The P.N.C.mid- restrictedsize no more than:25 murids; 7 vespertilionids; montane rainforest was chosen because it contains 7 pteropodids; 7 dasyurids; 6 hipposiderids; 5 many taxa that are at least distant relativesof taxa known phalangerids: 3 rhinolophids: 3 emballonurids; 4 from Riversleigh'sOligo-Miocene rainforestassemblages peroryctids; 4 macropodids; 4 pseudocheirids; 2 and becauseit containsNew Cuinea'srichest mammal petaurids; 2 molossids; 1 burramyid; 1 petaurid: 1 assemblages. acrobatid; and 1 tachyglossid. However, it would still be unlikely that these 83 mammal species would occur Each of these assemblages has its limitations and/or close enough together to be found in a single fossil biases, hence comparisons between them must be deposit like Upper Site. Hence we caution that the figures made with caution. Specifically, despite the Upper Site for diversity given in Fig. 2 are almost certainly higher LF's relativelyhigh diversity.a better indication of faunal than they would be if they had been determined by the diversitywithin the Riversleighregion at that time would same means that gatheredthe Upper Site LF. require a compilation of contemporary lfs. For example, it is evident that many other families (e.9., ilariids, lf we consider the probable historiesand similarities 'local' ektopodontids, miralinids) were probably representedin of the mammal communities of the four rainforest the Riversleigh region during Upper Site time because faunasselected for comparison,the Upper Site LF (aug- they are present in other adjacent System B local faunas. mented as noted above) diversity,if not taxa,would prob- Further, there is no evidence for transportation of any of ably have been typical for at least northern Australian the Upper Site material to the restricted, irregularly lowland rainforest of the Oligo-Miocene. Following shapeddeposit (about 2 m2,1from which the samplewas decline of this pan-continental rainforest,probably in the obtained so we presume that the Upper Site LF contains late Miocene, descendant ecosystems persisted as taxa that were actually sympatric within the immediate remnant rainforests in areas of eastern Australia. area. Consequently, we caution that the number of The Hamilton LF community probably represents an species and families listed for Upper Site are almost early Pliocene rainforest refugium on the southeastern certainly significant underestimations of biotic diversity edge of the continent. The Upper Site and Hamilton LFs in the region at that time. share severalmammalian genera such as Hypsiprymno- don, , Burramys, The same is true for the Hamilton specimens Strigocuscus, Irrchosurus and possibly . (collected from exposures less than 10 m2) none of which shows signs of transportation from a wider source The Atherton LF representsa similar although modern area. However, diversity for this assemblage is almost refugium for tropical rainforest in northeasternAustralia. certainly under-representedby the local fauna because It shares several mammalian genera with the Upper Site

June 1989 Australian Zoologist, Vol. 25(2) 37 >a A. Upper SiteLF, Oligo-Miocene rainforestfrom Riversleigh,N.W. Qld E(E o^ at,

o

o .o a1 z' eai'i!irigiE iFgEiE i; --E=.;s -E F;*iii;* rj ; : B. Hamilton LF. E. Pliocene rainforest, S.W. Victoria a aoo F 6a o. o. o 2 o C. Speciesdiversity in rainforests o .Ol of the AthertonTableland, E N.E. z Queensland 0 aa@a@O!!!C. gpppppE;i;p 3EF;EE:*Ei! EEsE-EFHX=+=*Y!;j=tYo Ee =o,*Fii ;* a--o o>

i E E ! € E E E ! ! € € € € i;:;9E b= eI E30 = F: =::* = 6 = s 9E r P H (E D. Speciesdiversity in =: gEeE"*e gE *tF;5 F=-FE 3 o. mid-montane rainforests E Es E5 E * a E-d= a e a oeiP'"Ei: € : o *E-a;€f E' -EE o of New Guinea 9 B-. 20 o o : l| tr 5 to z Fig. 2. Diagrams depicting species-level diversity within familiesof mammalsin four rainforestassemb/ages: A, the Oligo-Miocene Upper Site LF of Riversleigh; B, the early Pliocene Hamilton LF of western Victoria; C, the rainforests sss€€€ s{p EE€€!gg of the Athefton Tableland,northeastern Queensland; and D, the mid-montanerainforests of modern New Guinea(P.N'G. =:3PIiE$EI€gEE€g from K. Aplin). E a: data courtesy T. Flannery; with additions 5 5 * e* !; t € =E s -a Humansand dogs havenot been includedbecause they are fra=o;o E'Ed " E (espectively) late Pleistoceneand Holocene invaders.

38 Australian Zoologist, Vol. 25(2) June 1989 LF including ttacro!1y1, Hipposideros,Hypsiprymno_ don, Cercaftetus, ps.eudochir"p, Extinction of familyJevel groups uiJ" iri"nosurus. It share,s, Hyp s i p a m o on, st from rainforest communities possibly pseudocheirus7 ! d;;;;, Tri ch os u ru s and withlhe ifu-ifion LF. palyno_ logicat anarys", f::T, loss reflects Ly".h,; L;;. iX",rnu* rgzol ^.This"progressive the overall trend noted suggestthat during the pleistocene above for declinino diveoity tf,e itne.ton rain_ *itf,in'in#,"".rf region. forests underwent The Opper Site fF hu.r*2J episodes.f;;;";" iontraction, rurnir,"r'iwit;,",",gn least stressthat may have a species); the Hamitton.T. 63 redr."a .urnrnui di-versity t f ty,,n ;l f""r,' f g species); regton. in that theAtherton rainforest r a p.N.C. f*iir,'tjolpl.i"lj;.and.ll the mid_ montane rainforest15 (with The p.N.C.mid i'Ji"l,"rr. havebeen We interpretthis_ iso;i"lji#jI" i:'[.ff :j,# decline,in diversity since the early Miocelg TiliLx#,fl i:. pul presumed .family_level to lntannery iOeg; see below), :*:l chang6s ii' tr,e although ephimeral communitiessince ptant ;;riiir"r,'or dispersat the,origo-Mioceie and reactions routes to and from "plio;";; the arrivalof various to Australiau." inA.u1"j by shared pruc"itur ;;;;;; il.Tuaing, generaof kanqaroos(protem tatepleistocene, j; in tl-," nodin) una*aiiro,oaon,;a. humans..Ch;"A; T";#,i,-r,ol.., (Nototheriumfin the plant communities ruinror"rt r"urryerl"."#'ii,E ar may huu" ;;lt;i ,n.r"ur"a 1967,Woodburne tprun" seasonatityand/or deterioration r,ungi.oo,-dJ;;i"f;r_;:),'iBii.:?1f;Tif ;f ;rr;;;i; s rainfall :**:; lv1iocene. n"a *,i.. .i1he dtversity eastern Australia and New :;?ff:.I,S"tate Cuin"u Lu1 no,,n" Hamilton LF or the "ulj"-n,,y raria.is,;lilJ; old", Oli;o_;iJ"i=" .uinror"rt ;;';:'J.f T"."il:"!ilfl faunas of Riversleigh. decline in mammalian, ff H"?lH: dversity" rJi,i,ri' ^r.ii.","n forests because rain_ The ail of the..extinct apparent of.potoroids, representatives sr;;;; tiJ ,y.pu,r,. notoryctids "O::i:: phascolarctids, in the Oligo_Mioc"ene.coriiunities and mec Riversleigh. of ffiffi:ilTl h :;: Hfi:',i- 'i, ' "d:il::t :modern *.:;i rainforests :r*' srrdnccrc ,r.-r. , -,' lIH It is possible that the high diversitycharacteristic nothave r""" Oligo-Miocene of the irlffi .tiil;,1;,[:,-,i;:"A,[:l lfs of Ri Miocene, althouqh"," current orihat views are that "oNew Guinea *s;; ffi ;::i:'"ltj,:li:,ff?:::,:H,",f and Australia weie a Australianrainforests. tt "p"rffi [:J jnT t, ;,glifl"# thatthe th or;;;;;; i,n!"j j::i high hipposiderid diversrty(5 New" "', i^: J}::;tTil.nlil spp) characteristic of the Cuinea was",, isolated f;';#ii,u"by gh lfs is.best';i..h,ej' possiblyuntil the pleistocene ,ut", :lo:1.Ty"*l"i il ;"'riJi,i ur"u, lnranne.yiea6l; anabr "a theremay havebeen 2, *nr'e"'ppr *n"," nrt"i'u"iri"r, igliJn',il',, ;'*L'"TTx'J:J]i"T'J:,:t loin:" or biological)isolatinc climatic 'iddred*i;'c;;,';fiit"",*:"iT:'#;:ili; n",uc-u-in"l',i;,ilY"';[,l;',:TJ";l"f inthe :":'"-"fRiversleigh to the lateTertiarv dispeisal :ffi#T:xi,:? 4ru.ff,ilJi"":ent '"gioi;;.lnstl," of po,oioia. fi# nustralia); th9 presently -ir!i"g:'li.up"r"ri_,uy lld,.. ,3^r nuu" If we look rea c he d New Cu i ne a b ui su _i.'r"o1""".n at the families that declined ff e red ir" n.ti o n. Australia's to extinctlon in Submergenceof much or "*ti rainforests,most were represented Iowtarianei"iui,",."u in t|," one by only lateMiocene could ha or two species of sr anypotoroidsth"th#;'"X[t"T",1':ji:"":J'ff pirkipirdrras'e"ri"-,#4i,[::.,.?.-i;*flj; :"*".j i"r*1.,,1, because most Australi"" pj.tiJr';;"'T I,11J'J:lllqretc.) mammals. Losses tr,iJiini'-uy environments. lowland nave required very no, It is of < sionificant alterations"i io'the"rainforest these-sro"ups. ; ;;', p;;.:ffi T?::;_Ot onrythe :T:{notoryctids ;; il; ix'J,3*i[X,;.I.,n iff o"r::lgu.nt populations.rn ".i..v*ar'rrir,,,u", u, and meoadermatidsj ;;";."* xerrcenvironments (the Mar_ Guinea but n"* supialMole of thecentral have not iet U""n a"i".t"I AustralianJ"r"nrl,i"i"f .*1. temporal Otherlosses from rainforest .^.liu".uppur"nt changes evidenced by these profound communitieswere more four rainforestmammi suchas thoseof diprotodontias. 1,th" ;p;;,;; thylacoleonidsand phascolarctids iator-ctrestias, ;;;#J?TlTffi allof whlchwere mid_ groups; 2, the correloonding il""T""?ffi"?ldte-sizedto largemammals. increase in intrafamilial ro, ror*r'oriniJi,na, *" diversity of particular groups. that,more profound 1rnu.ropodij., murids, rarnrorestsf,:::T:. .nung", o;*.i"} to tr.," vespertilionids, emba althouqhthev may rn u".nlng", ilill.rrr"a pteiopodids ) ; 3,,n; ;;: j Quaternaryrather than prants :':Ji:,,"0,1; n,o',n"i? ?*,, 1," ,l: _the the Tertiary.AJI of these of rainforestmammals *,"ii.".ririiii, except phascotarctidsuunirt"l, placental tn" ,ir" .r enuronmentsl:::l: f-"-",-""i. diversjty and its.possibl" in the lateI .;;;;;;' ;:,ih" fi.rt in the Oligo_Miocene rainforest "il;.l5xS';:i,t"ffi or:fT::"".:, groups communities atr:.l?t:"1".-,i;;il"#":l'ff teast50 000 years ff: that survived in mesic aoc the timesof their ance from ;;#ilr;. disappear_ "nui.o*"nL'.. ; not clear but zygomiturine June 1989

Australian Zootogist, Vol. 2b(2) 39 diprotodontids, at least,persisted in upland rainforestsof Asia since at least the late (e.9., Audley- New Guinea until 20 000 years ago (Hulitherium; Charles 1987; Fig. 4). The northern portions of this Flanneryand Plane 1986). archipelago (e.g., Sumatra, Java and western Sulawesi) are Asian in origin while the southern portions (e.9., Increase with time in intrafamilial diversity of Timor, Seram and eastern Sulawesi) are Condwanan. mac ro po dids, m uri d s, vespe rti lion ids, emballon u ri d s By 40 million years ago (Fig. 4A), in the late , and pteropodids perhaps only 5 million years after Australia finally separated ln general, the decline with time of rainforest marsupials from Antarctica, the submerged shelf of Australia and rise of placentalsis apparent (Fig. 3). Placentalshave included the Gondwanan relicts of Timor, Tanimbar, come to represent more than 5O% of the modern rain- Seram, Banggai-Sula the eastern half of Sulawesi and forest mammal faunas of both Australiaand New Guinea the then small island of New Guinea. In turn, these were in contrast to the 16% of the Oligo-Miocene Opper Site adjacent to a string of possibly emergent Asian islands LF. However,the changes from what we interpret to be including Java and Sumatra which indirectly'linked' the 'archaic' the kind of Australian rainforest assemblage Australian continent to mainland southeastern Asia. (such as that of Upper Site) appear to be more profound By 30 million years ago (Fig. 4B), in the late in the New Guinean mid-montane communities than in Oligocene, New Cuinea may have been a small area of the Atherton rainforests of Australia. lowland but the archipelago linking Australia to south- Relevanthere are questions about the relationshipsof eastern Asia was less strung out and beginning to the drifting Australiato adjacent land masses as it moved deform as Australia ploughed northwards. At this time into lower latitudes after breaking free from Antarctica. New Guinea was probably covered in tropical lowland ln contrast to previous widespread views which depict a rainforests which might have permitted colonists from vast, vacant ocean barrier between the drifting Australia comparable habitats in northern Australia.The rapid rise and southeasternAsia, the northern edge of the Austra- of New Cuinea from 15 million years ago is evident from lian continental shelf has evidently been the end of an the fact that the central highlands consist of middle albeit ephemeral archipelago linking it to southeastern Miocene marine limestones. We have previously

0ligo-Miocene: Modern: Modern: 110 Riversleigh,Q'ld N.E.QId Nev Guinea o e'l 100 o a 90 E D Murids o v| 80 |/| Pteropodids r! 70 a t! E Emballonurids E 60 tr I Vespertilionids o l.4olossids E 50 E m Rhinolophids c 40 a Hipposiderids A) El 30 n Megadermatids l! I c At 20 T Marsupials o L '10 o 0. g Monotremes 0 UpperSite LF Atherton Mid-montane GodthelpHill Tableland rainforests

Fig. 3. Changes within rainforestcommunities in the relative abundance (specles percentages) of monotremes, marsupials, bats and from the Otigo-Miocene lJpper Site LF to modern Australian (Atherton) and New Guinean (mid-montane) rainforests.The HamiltonLF is not includedbecause, although it representsa Pliocenerainforest, it is evidentlybiased against volantand smaltterrestrial mammals and would misrcpresentthe transitionbetween the Oligo-Mioceneand recentcommunities.

40 Australian Zoologist, Vol. 25(2) June 1989 IND/AN OCEAN

J{ :€

t-] DURTNGIHT EOCENI f I fRAGMFNI5OtAUSIRAtANCONDWANAIANDASOWStA LIWLi*holyor . - AUSIRATIANCONTINENTAT MARCIN FLOODEO 8Y SEA ,. DOd] DUR]NG THELAIE MIO' tNT flrllm THROUGHOUT TH€EOCENT ^"r".,.' !u( pdt a\u N.wa\..^ND,.o^orD8r.ia Dr,prNc f-lT-l ',rr,U IfIFtATtMIOCFNI

Fig. 4 . Currentinterpretations of the relationshipsof the Aust- ralian continent to surrounding /and masses at 40 (A), 30 (B) pd__rt o!fqfl and 10 (C) million years ago. There is considerable con- troversyabout some aspectsof thesereconstructions and in particular about the time of fusion of Burma and Matava to southeasternAsia. Some authoritiessuggest that this fusion occurred prior to 40 mya. (Modified after Audlev-Chartes 1987)

suggested(e.g., Archer 1984) that the middle Miocene ,""1i uplift of the New Cuinean highlands preserved in rising altitudes part of the formerly widespread tropical low- lands rainforest of Meganesia (Australia plus New Cuinea; Filewood 1984) which otherwise almost disappeared from the Australian mainland. /i By 10 million yearsago (Fig. C), in the late Miocene, the position of Australia with respect to surrounding land !5 masses was much as it is at present. Late Miocene sub- mergence of the the New Cuineanlowlands (see above) need not have endangeredthe mid-montane biotas but it may have acted as an obstruction to dispersal of land mammals to and from the Australianland mass. Hand (1989b) and Hand, Archer and Codthelp (1989b) have considered the implications of this archipelago to an understanding of how bats may have dispersed r--l f RAGMINIS()tAUStRAt ANGONOWANATANDABOWSE LtUt DURING to and from Australia in the early to L-l rHEoilGoctNr middle Tertiary.Marsupials do not appear to have made AL]sIR^t^fr IoNr NtNIAt MARGIr!IrooDtD 8y 5rA DURTNG ['ffTiTl Or GOaENT effective use of this route to disperse from Australia to

June 1989 Australian Zoologist, Vol. 25(2) 41 A specles of Burramysunique to Riversleigh'solder deposits. Thesetiny animals were quite common in the lowland raintorestsof the day but their living descendants sutyive only in the alpine rcgion ot southeasternAustralia. (Drawing by Peter Murray).

42 Australian Zoologist, VoL 25(2) June 1989 southeastern Asia but it is clear that it was used by the Australia prior to Rackham's Roost time and possibly as invading rodents in the middle to late Tertiary and by early as the late Miocene although there are no rodents humans and dingoes in the Quaternary. known from the late Miocene (Northern Territory) or Beaumaris (Victoria) LFs. This absence from the Perhaps the higher diversity of placental mammals in Alcoota LF is probably significant and suggests that mid-montane New Quinea (Fig. 3) has resulted from rodents entered Australia sometime between Alcoota that area's relativelylonger period of exposure to placental and Rackham's Roost time. One of us (Godthelp), having invasions. However, considering the antiquity of the studied modern murids and the diverse Rackham's archipelago that links southeastern Asia to Australia as Roost assemblage,is convinced that rodents first entered closely as it does to New Cuinea, dispersing placentals Australia through the drier, non-rainforest corridors of may have been landing on the shores of Australia for at southern Indonesia (Timor, Sumba, Java etc.) rather least as long as they have on those of New Cuinea. than via New Cuinea. It is not clear how soon after their first arrival that murids invaded Australia's rainforesrs. Conversely, perhaps many of the Oligo-Miocene marsupial groups originating in Australiawere unable to The post-Oligo-Miocene decline in the diversity of disperse to New Cuinea. Flannery ( 1989) argues that by peroryctids (only one of which, the Rufous Spiny Ban- early Miocene time, mammals in the Meganesian region dicoot Echym ipera rufesceng survives in Australia's rain- had separated into two biogeographically isolated sub- 'V.D.' 'archaic' forests), bandicoots (which vanished before the regions: Australia (with a blend of and diverse 'modern' Pleistocene) and possibly the smaller omnivorous relatively marsupial groups) and New Cuinea 'modern' potoroids may be related to the late Tertiary murid radi- (with a subset of only the more groups). ation. All three groups overlapped in size and probably Placental invaders into New Cuinea may have simply dietary preference.However, the fact that peroryctids are insinuated into niches unoccupied by marsupials but largely confined to rainforests while murids appear to found the same niches occupied in Australia.In this case, 'modern have primarily radiated in drier habitats strains the argu- the New Guinean'mid-montane faunal balance ment for competitive exclusion. Certainly none of the summarised in Fig. 3 may fairly reflect that region's late old endemic murid genera (Zyzomys, Conilurus, Tertiary diversity. Pseudomys, etc.) occupy rainforest niches today. The only murid genera that are well-representedin Australian The only known Tertiary mammal fauna from New rainforests (e.9., species of Melomys, Uromys, Rattus Cuinea is the PlioceneAwe LF of P.N.C.(Plane 1967). It 'modern' etc.) have New Cuinean affinitiesand, if the fossil record contains only a subset of groups such as from easternAustralia is a reliable guide, did not appear macropodine macropodids (unknown in Australia prior in Australia before the Pleistocene. to the late Miocene), diprotodontids (groups similar to late Miocene or even Pliocene taxa in Australia). one The increase in vespertilionidbat diversityis almost as dasyurid (representing lvlyoictis, a derived dasyurine; striking as that of murids. From 1 or 2 taxa in the Oligo- Archer 1982)and an indeterminatenumber of murids. Miocene lfs of Riversleigh,to 4 in the Rackham's Roost LF, at least 8 in Riversleigh'smodern fauna, 11 in the In sum, in large part because of the depauperate late Atherton rainforest and 15 in the mid-montane rarn- Tertiary and non-existent middle Tertiary lfs from New forests of New Cuinea, the increase in diversity in and Cuinea, we are uncertain about the significance of the outside of rainforests is clear. differences between modern New Guinean and Austra- lian mammal diversity in rainforests. Emballonuridsshow a similar if more recent increase While most rainforest marsupial groups in Australia in diversity.None is known from the Oligo-Miocene rain- declined before the Pliocene,macropodids increased in forest faunas of Riversleigh;1 species is known from the diversity.By early Pliocenetime there were many macro- Pliocene Rackham's Roost LF; 2 coexist in the podid lineages in and outside of rainforests. Their Riversleighregion today; 4 occur today in the Atherton increase appears to correspond with the decline in rainforests;and 6 live in the mid-montane rainforest of balungamayine potoroids, a diverse group of Oligo- P.N.G. The higher diversity in modern rainforests 'rat-kangaroos' Miocene that had evolved lophodont suggests that during Rackham's Roost time they may teeth similar to those of macropodids (Flannery,Archer have been present in higher numbers in east coastal and Plane 1983). rainforestsand possibly enteredAustralia from the north, via rainforest corridors, sometime between the middle 'sudden' The seemingly appearance in the Pliocene Miocene and early Pliocene. Rackham's Roost LF of 13 spp of murids (Codthelp in prep.) and the subsequentlyrapid rise in diversityof this This same scenario may apply to rhinolophids. This group throughout the Australian region by late Quater- group of bats does not appear to be represented in any nary time are striking features of the late Cainozoic of the various aged Riversleigh fossil deposits; nor are record. It seems highly probable that rodents were in they found in the area today. However, two species are

June 1989 Australian Zoologist, Vol. 25(2) 43 known from modern Australia and both are restricted to What happened in these forests to cause the loss of the east coast. Four species occur in the mid-montane these large animals?As noted above, at least for diproto- rainforests of New Quinea and in any one habitat three dontids and probably palorchestids, thylacinids and may be sympatric. Their presence in the forests of eastern thylacoleonids, the arrival of humans in Australia may Australia evidently rePresents a relatively recent invasion well have been a significant factor. However,ilariids and from New Guinea probably via northeasternQueensland, wynyardiids did not survive beyond the middle Miocene perhaps in the Pliocene or Pleistocene. so their declines require a different explanation.

Pteropodids appear in the Riversleigh record, suddenly, Considering the phylogenetic relationshipsof the now in the late Quaternary - there being not a single extinct large rainforest herbivoresfrom Riversleigh,most to represent this distinctive group of herbivorous bats in were vombatiform (sens. Woodburne 1984b, Aplin and the Oligo-Miocene,Pliocene or Pleistocenelfs. Consider- Archer 1987) diprotodontian marsupials that were also ing modern diversityin Australia (e.g.,5 spp in the Ather- present in the Ditjimanka and Pinpa LFs of South Aust- ton rainforest)and abundance, particularlyof the species ralia.The early PlioceneHamilton LF retainsonly a single of Pteropus, the absence of this group from the vombatiform group, the palorchestids. The Oligo- Rackham's Roost LF puzzlesus. This apparent absence Miocene decline from the rainforests of these vom- could be an artifact of limited preparation of Rackham's batiform groups, most of which are only known from Roost material. However, because New Guinea as a medium-sized to large species, coincides with the whole has 25 species of pteropodids as opposed to Aust- apparent rise in diversityof the rainforest phalangeridans ralia's B, it is possible that Australia was only relatively (sens. Aplin and Archer 1987) which, although already recently colonised by this group of bats. diverse in the Upper Site LF, today are the only diproto- dontian marsupials in rainforests.Phascolarctids, which Hand (1989b) suggests that pteropodids may have were not uncommon in the Riversleighrainforests, had colonised New Cuinea prior to the establishment of evidentlyvanished from this habitat by the early Pliocene arboreal fruit-, blossom- and nectar-feeding Possums; (Archer and Hand 1987). lt would aPPearthat the Oligo- conversely,they may have had difficulty invading Austra- Miocene rainforests of Riversleigh rePresent a turning lian ecosystems because of the abundance there of a point in rainforest :nammal assemblages where the wide varietyof arboreal possums. The Quaternary estab- vestiges of an older vombatiform radiation overlapPed lishment of pteropodids in Australia's northern rain- with the rapidly diversifying phalangeridan marsupials. forests may have followed late Tertiary declines of many of Australia's phalangeridans such as the smaller It is relevant to point out here that although vom- phalangeroids and a wide range of petauroids.But with- batiform marsupials declined in rainforests during the out a pre-Pleistocene record of pteropodids in New Tertiary,some vombatiform groups (e.9.,diprotodontids, Guinea or Australia or of pre-Pleistocenepossums from vombatids, palorchestidsand thylacoleonids) underwent New Cuinea, evidence for or against possible late significant late Cainozoic adaptive radiations in the open Cainozoic competition between these groups does not forests and grasslands alongside similarly spectacular exist. radiations of mesic and arid kangaroos. However,by the end of the Pleistocene,all of the vombatiform groups in Overall, the increases in diversity of murids, vesPer- these drier habitats except for vombatids and phascolarctids tilionids, pteropodids and macropodids more than had vanished. accounts for the higher species diversitiesof the Atherton and mid-montane P.N.G.rainforest communities. ln the Among the smaller rainforest mammals that declined Atherton rainforests,50% of the mammal diversityis now or were lost after the Oligo-Miocene are several enig- made up by these four groups. ln contrast, in the Oligo- matic groups known only from the older Riversleighlocal Miocene Opper Site LF, these groups comPrise no more faunas including yalkaparidontids and yingabalanarids. than 5% of the mammal assemblage. Similarly, pilkipildrids, miralinids and several odd phalangeridan groups (e.g., an unnamed family of phalangeroids from the System B lfs) vanished after the The size, relationships and food-plants Other small groups, such of rainforest mammals that declined Oligo-Miocene. Phalangeridan as burramyids, acrobatids,petauroids and ektopodontids One of the most striking contrasts between the older oersisted in Australian rainforests until at least the early rainforest communities and those that survive today is Pliocen" and then declined or went extinct (e.g., the presence in the former of many medium-sized to ektopodontids) in that habitat. large mammals. In particular, the diprotodontids, palor- chestids,thylacinids, wynyardiids, ilariids and thylacolmnids Most of Riversleigh'slarger Oligo-Miocene rainforest were mostly dog- to cow-sized mammals. ln contrast, in mammals, apart from kangaroos, were quadrupedal, the modern P.N.G. and Atherton rainforests, although terrestrial browsers similar to the placental that dingoes and./orthylacines were present, the rest of the suMve today in the rainforestsof Malesia.Consequently endemic mammals are no larger than a tree-. we presume that they fed on an understorey of relatively

44 Australian Zoologist, Vol. 25(2) June 1989 soft-leafed plants within the forest. The same may have button-quails (Turnicidae). He interprets some of these been the case for the gigantic (up to 500 kg) (e.9., the casuariid) to represent possibly ancestral dromornithid birds that were abundant in the rainforests groups for more mesic descendants. of Riversleigh.lf the rainforest food resources of these possible giants vanished for whatever reason, it would explain It is that many of the modern mesic groups became pre-adapted their demise and that of the (e.g., within the Oligo-Miocene rain- forests thylacoleonids) that may have fed on them. However, for survival outside of that habitat. For example, ( other factors such as climatic changes may have caused Archer and Hand 1987) suggestedthat koalas may have their extinction without exterminating their food plants. specialised on ancestral eucalypts within rainforests which in The hypertrophied spines on many wattles (Acacia spp.), turn specialised on nutrient deficient soils. which seem unnecessarilylarge to deter living herbivorous Consequently, when the rainforest declined in central Australia leavingvast areasof nutrient-deficientsoils, this marsupials (R. Wells, pers. comm.), may be persistent 'symbiotic' pre-adapted, defence mechanisms against browsers that no longer pair would have been able to exist. Large-spined acacias in Africa are eaten by thrive. In this instance, we are, however, puzzled as to elephants and giraffes. why phascolarctidssubsequently disappeared from rain- forests.

The first appearances of many now mesic gtoups Another example we have previously cited (Archer, of mammals in the Oligo-l4iocene rainforests Hand and Qodthelp 19BBb) concerns the marsupial of Riversleigh moles (notoryctids). Although confined today to the The general hypothesisthat Australia'sOligo-Miocene central Australian deserts,this habitat is almost certainly 'Creen rainforests were the Cradle' for its xeric marsupials relativelyrecent. We were neverthelesssurprised to find abundant was advanced by Archer, Hand and Godthelp (19BBb) notoryctids in the System B units of (Cott and Archer and Hand (1987) so we will only summarise Riversleigh'sOligo-Miocene deposits 1989; Cott ef it here. aL in prep.). Their limb morphology indicates that they were burrowing mammals and their teeth indicate that In essence, it appears that one of three things they were in almost every respect ideal ancestors for the happened groups to each of the various of mammals living species. We have suggested that the Riversleigh present in the Oligo-Miocene lfs of Riversleigh:1, extinction notoryctids became pre-adapted for a burrowing lifestyle before Pliocene (e.9., yalkaparidontids, the ilariids, in mesic environments by specialising as burrowing wynyardiids, pilkipildrids, lngabalanarids, balun- insectivore/carnivoresin the possibly mossy floors of the gamayine potoroids); persistence 2, in refuge rainforest Oligo-Miocene rainforests. As with koalas, we again environments (e.g., hypsiprymnodontine potoroids, wonder why notoryctids do not appear to have survived peroryctid bandicoots, Strigocuscus, Pseudochirops); or in rainforests. 3, adaptation to the non-rainforest environments that began to dominate much of Australia by the Pliocene. Considering the Riversleighrecord as a whole, we are We have already considered examples of the first two convinced that the Oligo-Miocene rainforestsof at least fates. lt is the third that deserves further consideration northernAustralia were the cradlefor all of the mammalian taxa (except here. Among the mammal groups present in the Oligo- murids; see above) that have since adapted Miocene lfs of Riversleighthat subsequently adapted to to this continent'smesic and xeric environments.This in non-rainforest environments, there are two primary itself is a very strong argument for conservation of the categories: 1, those that persist(ed) in relict rainforests remnants of this procreative environment throughout as well as the relativelynew mesic and xeric habitats(e.g., Australia and in particular the tropical rainforest refuges ornithorhynchids, species of Trichosurus, Cercartetus, of northeastern Queensland. dasyurids, macropodids, potoroine potoroids, acrobatids, lnlacroderma, hipposiderids, vespertilionids, molossids; RIVERSLEIGH STRATIGRAPHY and among now extinct groups, diprotodontids, palor- AND PALAEOECOLOGY chestids, thylacoleonids, thylacinids); and 2, those that Riversleigh presents a vast complex of stratigraphic persist only in mesic and/or xeric habitats (e.9.,notoryctids, units and fossil faunas. Any simple model attempting to phascolarctids,vombatids, petaurids,Bettongia, Burramys). relatethe whole is doomed to an earlydemise. Although The same pattern we see for mammals has commonly our presentunderstanding (Figs 5-7) is the outcome of appeared for other groups of vertebrates known from 6 years'investigation, we are under no illusionsabout its the older Riversleigh deposits. For example, Boles probablelife span. ( l989) has identified among Riversleigh'srainforest birds (including logrunners, Orthonyx and possible 'drier' Challenges in defining Teftiary units lyrebirds, fulenura),taxa relatedto groups including emus (Casuariidae)and cockatoos (Cacatuidae),swifts While some of Riversleigh'ssediments, such as the (Apodidae),storks (Ciconiidae),hawks (Accipitridae)and ThorntoniaLimestone and'basal conqlomerate' (:'lower

June 1989 Australian Zoologist, Vol. 25(2) 45 7'' h1 aerial view of one of the fossil-rich plateaus at Riversleigh. The actual fossil s,les occur as rclatively small deposits sqead unevenly arcund the edges and tops of these structurcs.

After thrce weeks of lield work, upwards ol 15 tonnes of fossil-rich limestonewill have been gathercd in labelled hessian bags. Volunteerc help in all aspects of this wotk.

Fossl, bones are clearly visible on the ercded suiaces of grcy limestone that prcseNed them for 20 0O0000 years.

Site D, one ot the firct tossil-beailng sites discovered at Riversleigh.The massive grey limastone is approximately 20-25 000 00Oyearc old and contains the bones of giant flight ess birds (dromornithids), ctocodiles, turtles and rarc mammals such as marsuDiallions.

Frcm the air, the pale fossil-tichSystem C sedimerfs stand out lrcm the surrcunding plain ot rcd dust and the da*et grcy Cambilan limestones in the left foreground.

Thislower jaw ol an extinct balbarine kangaroo is one of many trcm the Oligo-Miocenelocal launas. lt appearcd in Last Minute Site which is among the lowest in the System C sequence on the Gag Plateau.

7. The upper jaw of Hypsiprymnodon barthofomaii,an extinct Rivercleigh ancestot fot the living Musky Rat-kangaroothattoday is conlined to the rainlorests ol northeastern Queensland.

d. The most common and widesprcad mammals in the Rivercleigh tossil depositsa/e bats. ln some deposits such as thrs one (Mictosite), their bones occur in the thousands. The most common species,n this affay of jaws is most closely rclated to bats pteviously known only from lossl/ deposfs in Europe.

9. Volunteerc,sfudents and prclessional palaeontologists wotk togethet to divide a larye rcck in White Hunter Site. The creaturcs lrom this rclatively old site arc rcptesented by many complete skulls.

10.A block and tackle ate used to lift laee blocks into the "Bonesmobile",a four-wheel truck provided by the Au strcli an Geogrcp hic Soci ety.

Helicopter transpottation enabled by Wang Australia is sometimes the only way we can rcach the more rcmote deposits and the only way we can rctileve large samples of tocks lrcm the many areas where ground-based vehicles cannot move,

12. Rackham's Roost, unique withirt the Rivercleigh'sarca because lt is so fal Rivercleigh'sonly Pliocene site. lt p/esewes the remnants ol an ancient cave (one wall is to the right) within which ancestral ghost bats ate theil prcy. The rccks in the lorcground arc solid with teeth and bones of small mammals including rcdents, bats and dasyurids. r ,_--

Riversleigh/Lawn Hill track, where exposures of ?late member' of Tedford 1967), appear relatively easy to Oligocene-early Miocene units (System A) abound on trace over wide areas, determination of the regional the" southern and eastern margins, ?early to middle extent and superpositional sequence of many of the Miocene units (System B) are exposed on the northern Cainozoic deposits has been aggravatedby the apparent edge (possibly including Neville's.9u.tdgl..Site) and lack of time-specific macroscopic sedimentary features' *"it"tn ur"ut j".g., the discrete Codthelp Hill and Hall's a problem noied by Tedford (1967) in regard to what we -"uu"' Hill sites) und deposits (e'g', Microsite and now call the SYstem A sediments' BitesantennarySite) of Oligo-Miocene age occur.on the Although many of the Tertiary carbonates are com- southeasternind northern edges respectively;2, the flat- "suspect plex, we that this complexity reflects cycles of lying mesas to the east of the Riversleigh/LawnHill track to be sedimentary'eventsthat may have repeated themselves whe"re additional exposures of what appear these at different times in isolated areas' Thus we often find Systems A and B occur (we have not y-etanalyse-d (e'g', and that different-aged bone accumulations Camel o,-,ttyingsites); 3, the slightly dipping Godthelp Hill Sput"- in Sysilm B and Henk's Hollow in System C) Hat;s Ai[ series west of Site D where at least 7 metres 4' display a similar lower silt-rich phase containing more of System B units lie on D-Site Equivalent units; and compiete bones capped by a similar relativelycrystalline tfre ilightly dipping Gag Plateau to the northwest of the (see phase containing more fragmented specimens D-Site'plaieauwhere approximately 12 metres of System in below). C units are superimposed on D-Site Equivalents' subregions such as Ray's Amphitheatre, Archie's efforts to maP contemporaneous units On-going Amphiheatre, Megabyte Hilt and Jeanette's Amphi- throug"houi the region are hampered by the isolated (see theatre. naturJ of *any oI the bone-bearing exposures below). For example, all known System B units occur in We have laser-surveyedmany of the sites on Godthelp a different region of Riversleighthan those-representing Hill and the Gag Plateau.Figure 4 is a graphic Presentatron north- System C without annectent Tertiary carbonates that of the relativJheights of selected sites on the Hill' could clarify the stratigraphic relationships of the two eastern end of the Gag Plateau and on GodthelP Systems. Each fossil-rich site, or isolated faunal assemblage For these reasons, our approach to determining the (thanatocoenosis),has been treated as a distinct local Tertiary bone- or identity and relationships of most of the iuunu potentially representing a -time-distinct bearini units has been to treat them as stratigraphers ecosystem-distinit facies despite the fact that some of - other- often i-rust treat cave deposits prepared to discover these assemblagesare less than 5 metres apart in that non-contemPoraneousunits may displaysimilar dis- wise non-fossiliierous limestone of a seemingly homo- tinctive lithologies. Accordingly, most of Riversleigh's geneous nature. Other local faunas are isolated by Cainozoic bioitratigraphic units have been defined f,undreds of metres from their nearest neighbour' primarily on the baiis of their faunas rather than their As seen in eroded or excavatedexposures (e'g', Upper petrologY. Oligo- 'lower Site and Henk's Hollow Site), many of these the For this reason, for example, we doubt that Miocene bone-rich deposits are well-defined laterally' by Tedford member', a basal conglomerate recognised commonly 3-5 metres in diameter, 0'5-1'0 metres in D.limestone' (1967) as a Tertiary unit underlying the Site thickness and roughly lens-shaped in cross-section' Our region' is contemporaneous throughout the Riversleigh subjective impressions are that in horizontal view these t and.resemble Such basalconglomerates are widespread deposits appear to be oval, suboval or arcuate in shape' that today s erosional lags of chert nodules of the kind They appear to grade laterally, albeit rapidly, into ru carpet the land surface wherever the non-fossiiiferous limestones. Macroscopic lithological (the genetic source of the Ei Thorntonia Limestone discontinuities between the bone-rich and adjacent eroded away We suspect that We are uncertain ia nodules) has completely depauperate limestones are not clear' in situ'by F these basal conglomerates have been formed of the precise shapes of any of the fossiliferous bodies in^the same different cycles of Tertiary sedimentatio-n. so far sanipled because we have left most of each for t are forming in local areas of the Riversleigh R way that thly future investigations. t4 ,"jion toduy. If so, they may have little biostratigraphic :-$1 sijnificance- beyond rePresenting post-Cambrian Often (e.g.,Henk's Hollow Site, Camel Sputum Site' F erosionalsurfaces. Cag Site) thl lower levels of these assemblages contain clasts or masses of yellowish-white 4i larie intraformational Oligo-l4iocene sites, faunal assemblages lim"estone(which experience tells us indicates relatively i: and palaeoenvironme nts laroe amounts of brown silt) cemented by whiter' more .ry"rtullin" carbonate' These lower levels grade upwards There are four maior regionswithin RiversleighStation E inio whiter more homogeneous although sometimes silty where most of the Oligo-Miocenesediments have been limestones.The lower levelsoften contain more complete * found: l, the flat-lying D-Site Plateau, to the west of the June 1989 # 48 AustratianZootogist, Vol 25(2) f; t80 et GagPlauau Sequence: o o ue) c 170 o 4l o D-Sib Equinalent vl c L o E Godthelp 160 Hilt c Sequence: +t S}l3Dm E B EN o D-SiE Equiv. t50 o.E.l -ef 3Ez EE€-[E3?.=;gEEr.or,T 6EE €seIE-- o3 -) r ifiti€t''. (D -h f u! aF Lia" .,++i EE-"* .,^EJ E- EdHrs Fig S' Laser survey data (perthe eftorts of M. Bannister,.H.Godthelp and A. white) for system B slfes on GodthetpHitt and Systemc slfeson the Gag Ptateau. Heighttot the /owestsife abovemean sealevel.was calculated using a barometer.Atthough the two systerns are plotted.here together, they are in fact separated geographicalty. Both have what we int?:pr:t to be- D-site Equiualent units at their base (e.g., Don bamitto on the edge of""q*ri""ll"o the Gag ptateau); henceit is unlikelythat system B unitsoccur below the ""rpo"t"a ,spring sysfem b units exposedii tii ^r"u of the Gag plateau. Mound, is a geological feature, not a local fauna (see text). The block symbol usedhere representsthe bas;l relativ,ei,tyiomogeneous, thick D-SiteEquivalent limestones; the horizontallines representihe superpositional'sediments ol System B and C; squaresrepresent the relative heights but not necessarityhorizontal proxinity olfossrTslfes within SvstemsB and C. skulls and bones while the bones in the hiqher more In a few sites (e.g., Last Minute Site) the more crystalline crystalline levels are often broken with 'clEan' the eoges phases contain thick sequences of what upp"ui to b" of the breaks lacking abrasion (which miqht otherwise shattered evaporite surface crusts that have periodicallv reflect post-depositionaltransport). settledto the bottom with the bones of the animals that may Although the precise nature of these accumulations is have caused the surface to break. While evaporites are unclear, for some of these sites (e.g.,Cag Site in System normally associatedwith salinewaters, calcium carbonate C) we hypothesiseisolated, possibly steep-sidedpools of evaporites are known to form on the surfaces of cave carbonate-rich water accumulating detritus within rain_ pools. We have also found evaporite crusts of calcium forest environments. Possibly silt-rich facies accumulatino carbonate on small isolated pools of the modem Cregory around the edges of deeper pools became periodicalli River system in open air situationsthereby demonstratnq unstable and slid en masse towards the centre of the that cave environments are not required ior the developf pools where the semi-consolidatedyellowish clasts were ment of animal traps of this kind. Small, still, super_ secondarily cemented by more crystalline carbonates saturated pools surrounded by warm gallery rainforest may precipitating, often with the bones of bats, in the spaces well have supported fragile evaporitecrusts. These crusts between the clasts.In the final stages of pond filling, the may have been augmented or initiated by floating mats of upper units may have become more crystalline as the calcium carbonate-secretingalgae of the kind which we shallowing water became more concentrated with appear to have recovered from Upper Site (see the dis_ dissolved calcium carbonate. cussion of the Upper Site fauna above). June 1989 Australian Zoologist, Vol. 25(2) 49 Relativelyincreased bone breakages in upper phase lake sediments to the east of the road mav have pro- materials might reflect more severe bioturbation in duced the presentblacksoil valley containing the isolated shallowing waters. Possibly some of these upper phase mesas as erosional remnants. materials actually accumulated on emergent surfaces where they suffered trampling before becoming However,not all of the System A and D-Site equivalent cemented onto the surfacesby travertinedeposits. Petro- deposits are of this kind; Burnt Offering Site appears to logical analyses under way may be able to test these be more similar to the lens-shapeddeposits common in hypotheses. System B and C sediments. Most of the System B and C deposits suggest that the Note should also be made of what we have tentatively pools were either shallow or short-lived because few identified as spring mounds and massive travertine contain more than rare remains of aquatic vertebrates deposits. Approximately halfway up the Cag Plateau (fish, System C sequence a shelf that appearsto be a travertine crocodiles and turtles) and of those that do most 'flow' contain bones and teeth of only small individuals.Excep- (Fig. 6) outcrops over a wide area. Several faunal tional sites (e.9.,Ringtail Site, Bob's Boulders Site,Crusty assemblages, such as the Two Trees LF, have been Meat Pie Site, Quentin's Quarry Site) are dominated by recovered from this layer which in areas may be as thick larger individuals of aquatic vertebratesand rnay reflect as 40 cm. It may represent an episode of widespread episodes of raised water tables (possibly because of outpourings of carbonate-rich spring water. Fractured tectonic lowering of the region or eustatic changes),con- surfacesof this material reveala highly porous carbonate sequently deeper pools and longer periods of groMh for with narrow, verticallyorientated columns (diagenetically fish, turtles and crocodiles. altered calcite crystals?).Near Jim's Jaw Site on the Gag Plateau there is a roughly circular, metre-wide cauli- A few (such 'flow'. Oligo-Miocene deposits as the Bitesanten- flowered structure contiguous with this We interpret nary and MP2 sites) are bounded by what appear to be this to represent a spring head and at least one of the thin, travertine laminations suggesting that the bone-rich sources for the carbonate rich waters that spread awav sediments filled travertine-linedcavities developed within from this point.A similar'flow'andspring mound system the parental limestone. In these situations we visualise has been recognised in the System B sequence on that phreatic caves may have developed in the older Godthelp Hill ('springmound' in Fig. 5). In this area we Oligo-Miocene units. Roof collapse may have have found fossils (e.g., pseudocheirid dentaries) in the subsequently allowed entrance (which of bats are rep- eroding surface of the flow. While we cannot exclude the resented by 'flows' millions of bones in the Bitesantennary possibilitythat these are intrusive,it seems much deposit: Hand, Archer and Codthelp 1989a), which more likely that they represent repeated, possibly roosted in the cave, and surface water that accumulated regional events that occurred within Systems B and C. as ponds containing ostracods, snails and possibly rafts of algae. Terrestrial reptiles, birds and mammals may In regard to the spring mounds and travertines, we have stumbled over the edge of the cave's mouth or have previously suggested (e.g., Archer 19BB) the been brought in by mammalian or avian predators to possibility that non-contemporaneous spring heads become rare elements in these particular deposits. formed part or all of some of the isolated hills capped with System B and C sediments (such as the Gag Plateau The only apparent exceptions to these generalised and Codthelp Hill). The model involved spring water palaeoecologicalmodels may be the faunal assemblages forced up through the underlying Cambrian and older in System A (e.g., Site D) and some of the D-Site Tertiary limestones to deposit calcite on the surface as equivalents (e.9., Sticky Beak Site and Bone Reef Site). travertinesheets and cauliflowersand then subsequently In these the bones are on average larger (almost exclusively re-erupting through these deposits to add progressively fragments of large crocodiles,dromornithids, turtles and younger carbonates until the accumulating mass diprotodontids) and rarer. The parent limestone in these became impenetrable. At that point, the spring water sites forms thicker units (in Tedford's 1967 Site D the might erupt in another area to start another younger fossiliferous unit is at least 3 metres thick) with some sequence. The faunal assemblages we find might then evidence of flysch-like sediments comprising coarser, be interpreted as the cemented contents of isolated rim sometimes better-sorted materials. A palaeoecological pools that had developed lateral to the spring head. This model that might explain these featureswould be a large might account for the oncolite-like concretions we lake that filled the central basin of the area (through sometimes find developed around bone fragments in which the Riversleigh/LawnHill road now runs) fed by some of the deposits (e.g., Grime's Site on the Gag springs, perennial streams or periodic rushes of surface Plateau). Against this model, however, is the fact that water from the surrounding forested terrain.The western while the Riversleigh limestones are clearly capable of margin of this lake's sediments would have extended at preserving surface features (such as the rippled travertine least as far as the thick units containing Tedford's ( 1967) rims and fragments of straws found in Neville's Carden Sites B-F and possibly further west as the underlying Site sediments in 1988), these features are unknown Tertiary sediments of the D-Site Plateau.Erosion of these from any of the other sites we have so far excavated.

50 Australian Zoolooist. Vol. 25(2) June 1989 It is clear to us that many different depositional Throughout the region there are many other presumably mechanisms were responsible for producing the variety Pleistocene (or late Tertiary) bone-bearing sediments. of Oligo-Miocene sediments at Riversleigh. The task One of the most widespread is a rust-coloured, possible now will be to test the competing models and to indurated soil horizon commonly found on subsoil contrast them with the palaeoecological interpretations surfaces or flanks of deeply fissured blocks of the Oligo- independently arising from study of the equally diverse Miocene limestones.When it occurs as a cement on the faunal assemblages. surfaces of bone-rich limestone (e.9., as it does at Helicopter Site), it sometimes contains broken pieces of Pliocene sediments, faunal assemblages teeth and bone of Oligo-Miocene taxa. We presume that and palaeoenvironments these had previously been lying free in the soil having been naturally etched from the receding surface of lime- The only Pliocene site recognised (Rackham's Roost) stone before being re-cemented as an overlying probably appears to be the floor of a long, narrow cave that at one Pleistocenesoil horizon. end opened onto a vertical face along the ancestral Cregory River (Godthelp 19BB). Another larger, lower opening may have been developed more than 50 metres The composite stratigraphic column for Riversleigh away in the opposite direction. Most of the millions of Figure 5 summarises what we now perceive to be the bones accumulated in this cave appear to have been major sedimentary deposits present in the immediate brought in by carnivorous megadermatid bats which Riversleigharea. Key aspects are: dropped uneaten bits and defecated the indigestible (e.9., teeth and bones to the floor below their roosts.There are 1. Iocal outcrops at the base of the Site D hill along rarer bones of snakes which may have preyed on the the Riversleigh/LawnHill Road, between the Cregory bats and very rare large kangaroos which may have fallen River crossing and Verdon Creek) of an early Protero- through holes in the roof. zoic sediment (PloughedMountain Beds) containing conspicuous stromatolite beds; Pleistocene sites, faunal assemblages 2. widespread basement of Cambrian carbonates and palaeoenvironments (Thorntonia Limestone characterised by chert The Pleistocene deposits are of two main kinds: nodules and, more rarely,invertebrate remains) which isolated cave deposits which have been treated in the is overlain or cut into by younger sediments; for units; 'basal same way described above the Oligo-Miocene 3. carbonate cemented conglomerates' mainly perched fluviatile The cave deposits rep- and terraces. containing eroded chert nodules which appear to resent fissure fills, megadermatid roosts and/or owl- have accumulated (possibly at different times; see pellet accumulations of the kind that now accumulate in above) as pavements throughout much of the region general region, for example in the caves around the and which probably represent in situ erosional Camooweal. remnants following dissolution of the Thorntonia Lst In some instances (e.9., at Two Trees Site), what may and/or older Cainozoic carbonates into which the be Pleistocene (or Pliocene?) fissure-fills have been nodules had previously been reworked; found as buff- or rust-coloured heterogeneous carbonates 4. At least five types of Oligo-Miocene bone-rich carbon- intruded into eroded spaces within the Oligo-Miocene ates: sediments. Some of these fissure materials contain 'Carl murids. - the Creek Limestone' [which may more appro- priately be called the Verdon Limestone - we are The fluviatileterraces are of undetermined extent but presently considering this probleml, a deep-water each is almost certainlydefined by subsequentchanneling. lacustrine unit that contains, e.g., the Riversleigh Terrace Site, for example, presents an approximately 3 Local Fauna from Site D Icollectively, these metre thick cross-section through stream bed deposits deposits are referred to here as System A]; comprising a basal poorly sorted conglomerate (also - 'D-Site containing bone, freshwatermussel shells and charcoal) the equivalents' which frequently overly the which grades upwards into finer sands and silts. The Thorntonia Lst but are not continuous with the bones in the fluviatile deposits may represent the dis- main Site D limestoneand which contain,e.g., the cards of crocodile meals, accidental drownings or Sticky Beak, Pancake, Hiatus 1, Hiatus ll, and seasonalflushing of the surrounding countrysideby rain- possibly White Hunter LFs; fed creeks. Some (but not most) of the bone fragments - an at least 7 m thick sequence of lacustrine carbo- are very worn which invites an interpretation of long- nates best exposed on Codthelp Hill and Hal's Hill distance transport and/or re-working from older which contain mainly non-aquatic LFs such as Cainozoic deposits. The Australian Ceographic Society Wayne's Wok, RSO, Outasite, Camel Sputum, has provided funds to have the charcoal and/or shells Upper Site and Boid Site LFs Icollectivelyreferred from the Terrace Site dated. to here as System Bl;

June 1989 Australian Zoologist, Vol. 25(2) 51 lThe sedimentsand their inter-relationships I IInterpretations of th9 qggryIlgjgr4im enrsl l- HolocerBcave dePosi8 (various) -lnrctocerrel

Perched sedimentsfrom Pleisbceneerrace d.eposits (e. g., g+qe.st_rgl.9tq g_otyBlye-r TerraceSiE LF & otherunnarned assernblages) Cavedeposits (e.g., Bottom SiE)-{pl€js,bcenel PliocerreRacklum's RoostLF ffi Ivfrr.y a)dirrtiolJ: a

?Rcgi,oul c>aiutiol of vyr.ytdins, i]rxiiiL etc.; ilrcliu of )duagrnryiars ctc TertiarySyeemB (e.g., LFs from Godthelp {U f I HilI (e.g.,Upper SiE LF, CamelSputurn LF, lo t"Itd"rtv I InabeyarrceSic LF, OutasiELF etc.) lttiocenel "D-site equivalents" "D-SiEequivalents" e.g., StickyBeakLF 4Ki< ',

?MiddleMiocene There are many SystemC faunal levels tuniGl & RepresentatrveLFs + olher local faunas in System exposedin Ray's Jaw Junction (terrest.) C. Thrs Amphitheatre of the Bob's Boulders (aquatic) subsetexposed on Gag Plateau. Henk'sHollow (terrest.) the flank of Ray's Grime'sSite (terrest.) Amphitheatre is a Main Site (temesrial) stack of superim- Skull Site (terrestrial) posedbiostrati- Gotham City Site (cave outwash?) graphic leve1sof Travertine la1rer the GagPlateau. Ringtail Site (mainly aquatic) The highest units Relative Iy unfoss ilife ro w level may be latest Last Minute Site (terrest.) middle to earliest Kangaroo Jaw Site (terrest.) late Miocene and contain forms thal 4i;agroenii or rargeU6n;;rtt "D-Site equivalent " (deep aquatic ) anticipate laxa in rhe N.T.Alcoota LE

Fig. 6. A composite stratigraphic column for the Riversleigh region depicting what we presently interpret to be the physical andlor temporal relationshipsol the main sedimentary deposlts producing fossrTsln the area.

52 Australian Zoologist, Vol. 25(2) June 1989 - an approx. 10 m thick sequence of lacustrine carbon_ Riversleighs stratigraphy and palaeoecology will be ates (see base of diagram) containing mainly: non_ examined. To this end our group is studlng conie-porary aquatic LFs such as Last Minute, Dwornamor from freshwater carbonate depositional situatiJns througnout Cag Site, Main Site, Henk's Hollow and Jaw Junc_ Australia and intends to extend this investigationto areas tion LFs; and largely aquatic LFs such as Ringtail of the Malaysia (e.g., Borneo) where therJ appear to be Site, Melody's Maze, euentin's euarry and Bob's contemporary environments possiblyanalogous to those Boulders LFs referred Icollectively to here as in the Riversleighregion in Oligo-Miocene ti.r,es. System Cl; and - cave deposits such as Microsite (source of the CORREI"{TION OF RIVERSLEIGH'S Nooraleeba LF) and Bitesantennary Site which SEDIMENTS appear to be pool deposits filling AND LOCAL FAUNAS BEYOND 'Carl caves developed in the Creek Lst' and/or D-Site equivaient THE RMERSLEIGH REGION units; Correlation of Australia's mammal-bearinq fossil sedi_ ' 5. Pliocene cave deposits, Rackham's Roost ,.inq the ments has been difficult because of a paucity of only one so far recognised,that formed from ceiren_ radiometrically datable materials in most of the older tation of organic debris on the relativelydry floors of sites. As a result, biocorrelation using shared taxa and caves cut into the Thorntonia Lst; interpreted stage of evolution arguments has been the most commonly used method for determininq the rela_ 6. Pleistocenefissure/cave deposits filling spaces in the tive and approximateabsolute ages of most of-theprevi_ Thorntonia Lst probably representinq severaldifferent ously known assemblages (Woodburne et al. 19g5). episodes of accumulation such as Bottom Site and Unfortunately,very littlewas known about the geographic Outer Limits Site: and/or temporal range of most of the taxa,,u.h ui dipro_ todontoids (Stirton, Woodburne plane 7. Pleistocenefluviatile silts, sands and conglomerates, and rc;67\. involved in studies of biocorrelation. e.g. Terrace Site and other unnamed unitsof possibly differing ages, exposed as perched and dissected The oldest mammal known from Australia is the 1i O channel deposits accumulated by the ancestral million year old monotreme Steropodon qalmani fron-, Cregory River system; ano lightning Ridge, New South Wales lArclier, Flannery, Ritchieand Molnar 1985).The oldest 8. Holocene (i.e., Recent) cave deposits, mostly marsupialsmay be those presently being retrieved unnamed, which carpet the floors and fissuresof rela- by our research group from the ?Eocene/Oligocene sediments tively recent fissures developed generally in the of Muroon. southeastern Queensland (Codthelp, Thorntonia Lst but sometimes in the Tertiary carDon_ Archer and H-and ates. 1989). Coming up the line, the next oldest marsupials. monotremes and placentals(bats) are probablvthe oldest In addition to these deposits presently being known from central Australia(e.g., the Ditjimanka and examined, we have encountered many others in the Pinpa LFs of South Australia;e.g., Woodburne l9g6) region that may prove to be significantwhen adequatelv which, on current understandingbased on illite dates assessed.Most commonly these are fissure deposits of (Norrish and Pickering 1983), foraminiferalbiocorrera_ unclear age some of which contain bones. Other tion (Lindsay1987) and palynology(Truswell, Sluiter and presumably Tertiary rock types in the region include Harris 1985), are probably late Oligocene in age (i.e., zones of secondary ferruginizationin isolated patches of about 25 0O0 000 years old; Archer. Every. CJdthelp, Cambrian (and possiblyTertiary) limestones. Wang Site Hand and Scally l9B9). itselfappears blackish-grey because of primary deposition ln our or diageneticconcentration of ferromagnesium minerals. efforts to establish the ages of the Riversleigh local faunas,we We have not seen undoubted e"posures of laterite are using radiometry, biocorrelation ano superposition to beneathSystem A units (althoughthey were reportedby clarify absolute and relative aoes. Radiometric dates for pleistocene Tedford i967) but we have seen a limited exposure of Riversleigh's units"will be obtainedusing what appearsto be lateriteimmediately north of Mega_ charcoaland shell common in rnanv of these byte Hill (the location of Wang Site) on the northwestern deposits. Efforts are also under wav (per C. McNamara) roBe edge of the primary Cag plateau. lts stratigraphic to determine if (now known to be present in the relationships to the fossiliferous deposits have not yet Riversleigh limestones) couid provrcle been determined. radiometric dates for the Tertiary deposits. Further, some of Riversleigh's Oligo-Miocene deposits are Work is presentiy under way to test and refine these producing invertebratesincluding ostracodswhich may preliminary concepts. One of us (Megirian) is concentratinq be useful for biocorrelation.However, most of our efforts on the geology while about 30 students, colleaques anJ to date the Riversleigh local faunas involve studies the rest of us are focusing on the faunas.All of ihe many of vertebrate biocorrelation within and outside of kinds of evidence that might clarify understanding aboui Australia.

June 1989 Australian Zoologist. Vol. 25(2) 53 A rcconstruction of the head and neck of Wakaleo vanderleuri,a teopard-sized marsupiattion that would have been one of the more elticient carnivoresof its day. A related specles proited the older rainforestsol Riversleighas did a tiny cat-sized cousin from ihe Upper Siie Locai Fauna.(Drawtng by Peter Murray). 54 Australian Zoologist, VoL25(2) June 1989 -

The potential of the Riversleighsequence to serye as lineages, previously only known (if at a basis for development of a biocorrelation all) from geo_ framework graphically isolated deposits, for late Oligocene such as,V.D.' perameloids, to at least early euaternary non_ thylacinids, hypsiprymnodontine and marine deposits of Australia is high. Riverslejqh,sabun_ .propleopine, potoroids, balbarine macropodids, dant Oligo-Miocene bats (at lea-st23 speci-ies; ?alungamayne Hand, Neohelos, Namilamadeta, Wakaleo, paljara, pseudo_ fucher and Godthelp 1g8gb) are enabling intercontinental chirgps, tulacroderma, correlation of Riversleigh's pilkipildrids, hipptsiderids and sediments ivith those from molossids. western ,Old Clarificationof slrmpatric evoiutionarytrends Europe and other areas of the World,. For in these lineages may increase the potential example, in terms of international correlations, bio_ the correlativevalue of these raxa. hipposiderid bat Hipposideros (Brachipposideros) Figure nooraleebus from the Nooraleeba (Micrositei, Bitesan_ 6 summarises some of the evidence we have (only the tennary and White Hunter LFs correlates with similar mammals are considered here) for interpreting material of Burdigalian age known from France (Sig6, the approximate age correlations of key i?iversleighloca-l Hand faunas to and Archer 1982, Hand 19g4, Hand, Archer and those of the rest of Australia and the world in general. C_odthelp 1989a). Similarly, a Riversleigh molossid Age determinations for the pre_Tertiary sedi_ (Petramops (j the proterozoic,palaeozoic creaseri) from Gotham City Sie resembles Jnelts 9,, early IThorntonia Eocene " Lstl and molossids from Europe although this Mesozoic [?Mullaman Beds] Eras) are based on previous resemblance is almost certainly symplesioinorphic studies of invertebrates.du, interest focuses (Hand 1989a). on the Tertiary and euaternary sediments of the Carnozoic. The oldest recognise.d In terms of intracontinentalcorrelations, the distinctive . 'Carl Tertiary units at Riversleighare potoroid the Creek Lst' (or Wakiewakie lawsoni is now known only from ?Verdon Lst), and System B units. the SVslgT A units may fall somewhere Upper Site LF of System B in Riversleiqh and the betweenihe Ditjimanka and Kutjamarpu Kutjamarpu LF of South Australia (Codtheip, Archer, and Tarkarooloo LFs of South Australia. Hand The as yet slim evidence and Plane 1989). The thylacoleonid Wakateo old_ for this hypothesis is based on fieldi.!om. the occunence the Kutjamarpu LF resembles possibly con_ of species of Ngapakatdia, Wakaleo, specific Neohelos tirarensis thylacoleonid material from Siie D and a and a wyryarJiia in tfre Riversleigh LF, taxa similar thylacoleonid from Upper Site appears to be a more to those in the Tarkarooloo and Kutjamarpu derlved relativeof the Ngapakaldi taxon priscileopitikan- LFs and in part to those in the Ditiimanka tensr's. LF. Further, the The pseudocheiri"d'gens eayiii i, ,i-iurry only System A sediments seem to be at least frgm marginally older than [1ow1 the Kutjamarpu LF and System B and C the Microsite and Bitesantennary Riversleigh lfs. LFs which are cut into A balbarine macropodid from some of them and contain a hipposiderid'in the taxon similar System B and C lfs appears to be Nambaroo, a to some of Burdigalian age France genus (Hand, otherwise only known from the Tarkarooloo and Archer and Codthelp 19891). Kutjamarpu LFs of South Australia (B. Cooke, We have in prep.). - tentativelyconcluded that the System B local diprotodontid zygomaturine fh" from one of tzunas compare best with the Tarkarooloo and Kutjamarpu Riversleigh'shighest Oligo-Miocene lfs, LFs the Jaw Junction of South Australia based in part on shared ,p".,"r- closely resembles (but is more plesiomorphic level fh, T9r! taxa such as Wakiewakiehwsoni and oenr.rs-level than) torus from the Alcoota LF of the southern taxa such as Namilamadeta, Nambaroo and it Northern Territory. least one . The ?wynyardiid genus lungfish (A. Kemp, pers. comm.). t\amttamadeta is now known only from Riversleigh,s D.-Site ,:y:l part of Sy:E- C appears to be younger Equivalent lfs, Ei lfs land porrlbry ^._t_n-" lystem thanlhe Kutjamarpu and Tarkarooloo LFs RiversleighLF) and the Tarkarooloo LF of Souih Aust- but olier than the Bullock Creek LF based in part ralia..The palorchestid genus Ngapakaldr.a,previously on some of the zygomaturines and thylacinids it contains only known from the Ngapakaldilnd Tarkarooloo LFs and its lack of the previously abundant species of Namilamadefz. of central Australia is now also known from the The upper.part of System C appears Riversleigh LF. The zygomaturine Neohe/os tirarensis to be close to but older rrom than the Alcoota LF because of the presence Riversleigh System C lfs (e.g., Dwornamor LF) of a pre_ Kolopsis-Iike zygomaturine appears to be intermediate in size and in the Jaw Junction LF of morphology Riversleigh. letugen specimens in the Kutjamarpu LF and ihose in the Bullock Creek LF of centril Northern Territory. The Rackham's Roost LF is interpreted to be pliocene (probably early Pliocene) _ Stratopheneticdata of this kind are providing the basis because it contains: abunoant for plesiomorphic rodents (at the development of the biocorrelativefrariewort for least 12 spp; Codthelp in the a population Riversleigh local faunas summarised below that Prep.); of llacroderma gigas that retains some features plioie-n encompasses otherwise isolated Cainozoic local faunas characteristicof the e llacroderma koppa from other areas of the continent. from the Wellington Big Sink Fauna of New South wales (Hand 1987, Hand, Dawson and Augee The sequence of sediments in 19gg); . Riversleighis also pro_ and. a macropodid similar plioceie protem- viding opportunities to the early to study temporai changes in nodon pliocene snewenifrom the early Bluff Downs LF. June 1989 Australian Zoologist, Vot. 25(2) 55 WidespreadHolocene sediments- Modern gpp; sorr with Mtrs

KatapiriSands; Easem DipfglgdQn--gptalgm arrdother DarlingDovns; et. Pleisoceraaxd Pleist. lHolocene spp 'ncerujn _t Moderngenen rePresentsd bY regiornlly extinct species ffi r^rrr' lPossibiY?roieiar rodoncf. P. Elsslbsle!3ee$- ?AllinChlmFormation -l snercni (also in Bluff Downs LF) Higherunrs: Pre-Koloprt! sP Lolrer units: pilkipildrids et. (Somegpups preseilin Oligo.- MuoceneLFs in centralAustelia)

Tertiary SystemB (e.g.,Upper Site)

li dc o) v7 pgnger age -lUnclear: - c) Unclear.but tlnn burbaa sqgesr Oi:go of units ino which these?cave lMr..* agebased on sirnilarig o depositshav= been cul Basal co leurylean hipposideridsof approx age ? (no fauna)?all conemporan.z lBrudigarian MullomanBeds'. Jurassic belemnies

Cambrian TrilobileJ et.

ProErozotc Stromabhtes

More Fig.7. Our presentunderstanding of the approximatecorrelation of the main fossl/iferousunits exposed at Riversleigh detaitabout correlationof the mammal-bearingsedlmenls is given in the text.

June 1989 56 Australian Zoologist, Vol. 25(2) The Terraces pleistocene Site LF is interpreted to be and in particular Mr and Mrs Selrmour, Mr and Mrs Stadler, because it contains Dipratodon optatum, a taxon other- Mr and Mrs Evetts, Dr and Mrs Robinson, Dr and Mrs wise unknown from pre-Pleistocene deposits in Australia. Fitzpatrick,the Mount lsa Shire, the Burke Shire, the NorthwestQueensland Tourism and Other Pleistocene deposits are concluded to be this DevelopmentBoard, the Culf Local Development Association age on more circumstantial evidence. Bottom Site, for and PROBE. example, contains taxa referable to modern genera but ln addition, colleagues, volunteers and financial most appear to represent species no longer surviving in supporters have provided key support for work at the Riversleighregion. Presumably they were present in Riversleighbetween 1978 and 1989 including (among the area under a different climatic regime, changes many others): Graham Anderson, Reg Angus, Ken Aplin, which characterised the glacial/interglacialfluctuations Kathy Atkinson, Ross Arnett, Dave and Virginia Austen, of the Pleistocene. Murray Bannister, Alan Bartholomai, Alex Baynes, Bob The Recent deposits contain modern taxa still living in Beale, Peter Beard, Ivor Beatty, Margaret Beavis, Susan the area although some of these sites appear to have Bergdolt,Michael Birt, Walter Boles,Peter Bridge, Barbara been accumulated prior to the introduction of u,- House Bridges, Peter Bridgewater, Catherine Brookes, Bill Mouse (/vlus musculus) possibly as early as 400 years Brown, Arthur Bruty, Ross Buchanan, Chris Buck, Ceoff ago. Burchfield, Lloyd Campbell, RussellCarrington, Bob and Laura Chatfield, Elaine Clark, Hal Cogger, Barry Cohen, Missing so far from the RiversleighOligo-Miocene to Phil Coleman, Bernie Cooke, Patrick Couper, John Holocene sequence appear to be any local faunas clearly Courtenay, Jeanette Covacevich, Sue Creagh, phil representing the late Miocene and possibly the middle Creaser, Archie Daniels, Lyn Dawson, Terry Dawson, and late Pliocene.Late Miocene local faunas from other Patrick DeDekker, Martin Dickson, Petrina Dorrington, areas of Australia include Alcoota (Northern Territory) Ken, Margo and Martin Dredge,Tim Flannery,Klim Gollan, and Beaumaris (Victoria). The diprotodontids in these Gill Goode, Miranda Cott, Des Griffiths, Ken Crimes, local faunasmost closelyresemble the zygomaturinesin Terry Gygar, Les Hall, lrma Havlicek, Georgina Hickey, Riversleigh'sJaw Junction Site and No-nime Site at the Clen Hutton, PeterJell, Barry Jones, Quentin Jones, Bob top of the System C sequence. These highest local Katter Snr, John and Murray Keir, Noel Kemp, Anne faunas on the Gag Plateau may have been the last to Kemp, Dietlind Knuth, Karl Koopman, Barry and Di accumulate before deposition on Riversleigh Station Kubala, Jim and Sue Lavarack, Dennis Lee Sye, Serge shifted to some other as yet undiscoveredarea. The lack Legendre, Bill Lockwood, Wayne Longmore, Barry of additional Pliocenesites is similarly probably the result Mackinnon, Brian Mackness, Andrew Marnie, Tony of imperfect knowledge of the area. If one isolated cave McGrady, Ralph Molnar, Alan Moyes,Jeanette Muirhead, accumulated Pliocene materials,surely there were others Frank Munis, Peter Murray, Ted Naughton, John and accumulating at different times in the late Tertiary. For Jan Nelson, Melody Nixon, Colin and Cheryl O'Keefe, the same reasons,we expect that continued exploration Mike Plane, Neville Pledge, Syp Praeseuthsouk, Alan of limestones adjacent to those presently being Pryke,Alan Rackham and family, Jim Reece, Neil Reid, examined will eventuallyturn up vertebratedeposits that Tom and Pat Rich, Craham Richardson,Alex Ritchie, predate the Oligo-Miocene System A sequence. These Karl Roth, Jim Ross, John Scanlon, peter Schouten, are challenges for future Riversleighexpeditions. Dave Schultz, Don and Sue Scott-Orr, Bernard Sig6, Craig Smith, Dick Smith, Jenny Taylor, Dick Tedfoid, ACKNOWLEDGEMENTS Cuy Thomas,John Thyne,Bill Turnbull,Steve Van Dyck, Paul Watkins, Shirley Webster, We wish to acknowledge the vital financial support the Rod Wells, Arthur White and family, Neville Whitworth RiversleighProject has had from: the AustralianResearch and family, Tony Wicken, Ray Williams and paul Grant Scheme (Grant PC A3 851506P); the National family, Stephan Williams, Willis, Janet and Keith Winsbury, Estate Crants Scheme (Queensland);the Department of Cary Witty, Mike Woodburne and Shelly Woodrow. the Arts, Sport, the Environment, Tourism and Territories; the Queensland Museum; the Australian The geological hypotheses presented here were Museum; WaDg Australia Pty Ltd; ICI Australia pty Ltd; discussed or refined in the field with Alex Bavnes. Ansett Wridgways Australia Pty Ltd; the Australian Ceo- Bernie Cooke, Phil Creaser,Lyn Dawson, Ken Crimes, graphic Society; Mount Isa Mines Pty Ltd; and Surrey Bob Jones, Noel Kemp, Jeanette Muirhead, peter Beatty t, Sons Pty Limited. Critical logistical support in Murray,Armstrong Osborne,Mike Plane,Sue Solomon, the field and laboratory has been received from the Dick Tedford,Bill Turnbull,Rod Wells and paul Willis. RiversleighSociety, the Friends of Riversleigh,the Royal Australian Air Force, the Australian Defence Force, the We thank Chris Von der Borch, Peter Murray, Tom Queensland National Parks and Wildlife Service, Rich, Neville Pledge, Dan Lunney, Jeanette Muirhead Queensland Helicopters,Hertz RentACar,the Riversleigh and Paul Willis for critically reading drafts of this Partnership (Riversleighbeing a privatelyowned station) manuscriot.

June 1989 Australian Zoologist, Vol. 25(2-) 57 the of the headof an extinctRiversteigh ringtait possum distantly rclated to .Grcen Reconstuction Riversleighlocal Ringtaittnat iuiiiii ioa"v in ihe rainforestsot frrtneisteri Queensland.some speciesof ringtails' faunas(e.g., theDwornamor LF) haue,ii"h"a up b .nine distinctsympatric (DrawingbY PetetMurraY)'

June 1989 58 AustralianZoologist, Vol' 25(2) REFERENCES Archer, M., Tedford, R. H., Rich, T. H., 1987. The pilkipildridae, a new family and four new species of Anon., 1.989a.Selected publications involving Riversleighfossils ?petauroid possums (Marsupialia: Phalangerida) frorn and their significance.Riversleigh lyofe;5: 3_6. the Australian Miocene. Pp. "possums 607-27 ln and opossums: studies in Anon., 1989b. Additional Riversleigh publications. evolution" Riversleigh ed M. Archer. Surrey Beatty t' Sons pty Ltd and lYofes 6: in press. The RoyalZoological Society oi New South Wal"", Sydn.y. Aplin, K.P., Archer, M., 1987. Recent advances in marsuoial Audley-Charles, M. G., 1987. Dispersal systematics of Condwanaland: with a new slncretic classification. Rp. xv-trc

June 1989 Australian Zoologist, Vot. 25(2) 59 dicksoni, a Flannery,T. F., Plane,M. D., 1986. A new late Pleistocenedipro- Muirhead. J.. Archer, M., 1989. Nimbacinus (Thylacinidae, from todontid (Marsupialia) from Pureni, Southern Highlands plesiomorphic thylacinid Marsupialia) Province, Papua New Guinea. Bl4R J. Aust. Geol. Ceophys. Tertiary sediments of Riversleigh, northwestern Queens- (in 7Oz65-76. land. Mem. Qd l4us.28 press). - 22. Pp.281' Codthelp, H., 1988. Riversleighscene 4: Rackham's Roost Norrish,K., Pickering,J. C., 1983. Clay minerals, "The "Soils the beginnings of the modern world. Pp. 81-83 rn 308 rn an Australian viewpoint". CSIRO/Academic Antipodean ark" ed S. J. Hand, M. Archer. Angus €' Press:Melbourne. Robertson Publishers:SYdneY. Patterson,C., Rich, P. V., 1987. The fossil history of the emus' Nus.27t 85-11. Godthelp, H., Archer, M., Hand, S. J., 1989. The Tingamarra Dromaius (Aves:Dromaiinae). Rec. S. Aust. Local Fauna from Murgon, southeastern Queensland: Plane, M. D., 1967. Stratigraphy and vertebrate fauna of the Australia s oldest marsuPials. Conf. Australasian Veft. Evol., Otibanda Formation, New Guinea. Bull. Bur. Miner. Resour- Palaeont. and System., Abstracts: p.6. Ceol. Ceophys. Aust 86: i-64. New Godthelp,H., Archer,M., Hand, S. J., Plane,M. D., 1989. Pledge,N. S., Archer,M., Hand, S. J., Codthelp' H., 1989-New N.T. potoroine from Tertiary Kangaroo Well Local Fauna, specimens of ektopodontid species including a new and description of upper dentition of potoroine Wakiewakie species of Ektopodon. Rec. S. Aust. Mus. (in press). lawsoni from Upper Site Local Fauna' Riversleigh. Contl The , a family of large, eKinct Australasian Veft. Evo!., Palaeont. and System, Abstracts: Rich, P. 1979. endemic to Australia.Bull. Bur. Miner. Resour. p. 6. ground birds Geol. Geophys. Aust. 184. Gott, M., 1989. Oligocene/Miocene notoryctids (Notoryctidae, The snakes of Riversleigh.Riversleigh Notes Marsupialia) from Riversleigh:first record for the family. Scanlon, J.. 1988. Conf. Australasian Veft. Evol., Palaeont and System' 3: 7-8. Abstracts: p.5. Sig6, B., Hand, S. J., Archer,M., 1982. An AustralianMiocene Brachipposideros (Mammalia, Chiroptera) related to Cray,J. E., 1831.A synopsisof the speciesof the classReptilia. representativesfrom France. Palaeovertebrata12: Pp. 1-110 rn "The animal kingdom arrangedin conformity Miocene with its organization by the Baron Cuvier." Whittaker: London. 149-72. M. J., Plane.M. D., 1985. Pythoninesnakes (Boidae) Hand, S. J., 1984. Bat beginnings and biogeography:a southern Smith, 'Vertebrate Australia. B/\R J. Aust. Ceol. Ceophys perspective.Pp.853-904 rn zoogeography and from the Miocene of 191-95. evolution in Australasia" ed M. Archer' C. Clay'ton 9: Hesperian Press:Perth. Stirton. R. A., 1967. The Diprotodontidae from the Ngapakaldi Australia. Bull. Bur. Miner. Resour. Ceol. Hand, S. J., 1985. New Miocene megadermatids(Megader- Fauna. South Aust. 85: 7-44. matidae, Chiroptera) from Australia with comments on Geophys. megadermatid Phylogenetics.Aust. Mammal. 8: 5-43. Tedford, R. H., 1967. Fossil mammals from the Carl Creek Limestone. northwestern Q.ueensland. Bull. Bur' Niner. Hand, S. J., 1987. Phylogeneticstudies of AustralianTertiary Geophys. Aust. 92:. 217 -36. bats.Summary of PhD thesis.Nacroderma 3:9-1?- Resour.Ceol. Truswell.E. M., Sluiter,l. R., Harrrs,W. K., 1985. Palynologyof Hand. S. J., 1989a. Australia'sfirst Tertiary molossid (Micro- the Oligo-Miocene sequence in the Oakvale 1 corehole. chiroptera: Molossidae): lts phylogenetic and bio- western Murray Basin, South Australia. BMR J. Aust. Geol. geographicimplications. Mem. Qd Nus.28 (in press). Geophys. 9: 267 -95. Hand. S. J., I989b. On the wings of fortune:the origin of Aust- Turnbull,W. D., LundeliusJr, E. L.. 1970.The HamiltonFauna ralia s bat fauna. Ausf. Nat. Hist.(in press). a late Pliocenemammalian fauna from the Crange Burn, Hand, S. J., Archer,M., Codthelp,H., 1989a.A fossrlbat-rich, Victoria,Australia. Fieldiana: Ceology 79. Oligo-Miocene cave-fill from Riversleigh Station norlh- Turnbull,W. D., Rich. T. H. V., LundeliusJr, E. L.. 1987. The western Queensland. Conf Australasian Ven. EvoL petaurids (Marsupialia:Petauridae) of the early Pltocene Palaeont.and Svstem.,Abstracts: p.7. Hamilton Local Fauna, southwestern Victor:a Pp. 629-38 Hand.S. J., Archer,M., Godthelp,H., 1989b.The Australianbat in "Possums and opossums: studies in evclution ed M. fauna: a 30 million year history? Aust. Mamm. Soc. Conf Archer. Surrey Beatty t- Sons Pty Ltd and The Royal April. 1989, Abstracts: P.26. Zoological Society of New South Wales: Sydney. Hand. S. J., Dawson, L., Augee, M., 1988. Macroderma koppa' Turnbull. W. D.. Rich. T. H. V., Lundelius Jr' E L., 1987 a new Pliocene species of false vampire bat (Micro- Burramyids (Marsupialia: Burramyidae) of the early chiroptera: Megadermatidae) from Wellington Caves, New Pliocene Hamilton Local Fauna, southwesternVictoria. Pp. "Possums South Wales. Rec. Aust. l4us. 4O2343-51- 729-39 in and opossums:siudies in evolution' ed M. Archer. Surrey Beatty & Sons Pty Ltd and The Royal Pleistoceneand Holocenepollen Kershaw,A. P., 1976. A late Zoological Society of New South Wales: Sydney diagram from Lynch s Crater, northeastern Queensland' 1984 A floristic New Phvtol. 77: 469-98. Webb, L. J., Tracey.J. C., Williams.W T. frameworkof Australianrainforests. Aust. J Ecoi-9: 169-98 Lindsay, J. M., 1987. Age and habitat of a monospecitic Recentevidence foraminiferal fauna from near-type Etadunna Formation Webb. L. J., Tracey,J. C.,Jessup, L. W.. 1986. and subtropical ratn- Lake Palankarinna,Lake Eyre Basin. Depaftment of Mines for autochthony of Australiantropical and Energy South Australia Rept Bk No.87193. forest floristic elements. Telopea 2: 575-89.

June 1989 60 Australian Zooloqist, Vol. 25(2) Woodburne, M. O., 1984a. Wakiewakie laWe lawsoni, a new genus recognise new species of Burramys (several LFs, and species of Potoroinae (Marsupialia: Macropodidae) of Nooraleeba LF and Systems B and C, several LFs) and Cer- medial Miocene age, South Australia.J- Paleont.5Bz 1062-73. caftetus (several LFs, System B). Woodburne, M. O., 1984b. Families of marsupials:relationships, 15Strigocuscus reidi and Trichosurus dicksoni (from several evolution and biogeography.Pp. 48-71 rn "Mammals: notes LFs, Systems B and C; Flannery and Archer l9B7a); in addition, for a short course" ed T. W. Broadhead. Univ. Tenn. Dept we recognise a new species of Wyulda (in prep.) and additional Ceol. Sci. Studies in Geology:8. new phalangerids. Woodburne, M. O., 1986. Biostratigraphy and biochronology. l6Djilgaringa gittespleae (from several lfs in Svstem C; Archer Univ. Calif. Publ. Ceol. Scri. 131: 87-93. ef al. 1987): in addition, we recognise another Lnnamed raxon. 'Cenozoic Woodburne, M. O., 1987. Definitions. Pp. xiii-xv rn 17We recognise new species of paljara, pildra, pseudochirops mammals of North America" ed M. O. Woodburne. Oniv. and new genera Calif. Press: Berkeley. and species representinga considerably wijer range of pseudocheirids than exist today in any single fauna Woodburne, M. O., Tedford, R. H., Archer, M., Turnbull, W. D., (many LFs, System B and C). ^roqy Plane, M. D., Lundelius, E. L., 1985. Biochrc- of the lqve recognise at least one new species (severalLFs, continental mammal record of Australia and hc* Guinea. System B). rlMe Spec. Publ. S. Aust. Dept lvlines and Energy 5:347-63. recognise new subfamilies,genera and speciesof enrgma- tic ?pseudocheirids(several LFs, Systems B and C). 2\Me NOTES PRIMARILY ABOUT ENTRIES IN TABLE I recognise a new genus and at least two new species of rThe 'modern' plesiomorphic petauroids (many LFs. Systems B and C). fauna of Riversleigh refers to the whole of the 2lWe station rather than a single habitat (i.e., from river margins, to recognise new relativelyplesiomorphic acrobatids (manv clay soil flats, limestone and quartzite ranges) and is thus not LFs. SystemsB and C). strictlycomparable with the fossil local faunas which 22Wabularoo come from naughtoni (balungamalne from D-Site. System single sites. A; Archer 1979); Gumardee pascuali (balungarnayine from D- zFootnotes plane below relate to mammalian diversity within the Site, System A; Flannery, Archer and l9B3l: Ekaltadeta Riversleighfossil deposits as a whole. lma (propleopine from many LFs, Systems B-C; Archer and JJ. Flannery 1985); Hypsiprymnodon baftholomari (hypsiprymno- Muirhead, in prep., identifies in the Riversleigh Oligo- dontine, many LFs, System C; Flannery and Archer 19g7b); Miocene sediments (Systems A-C) approx. 5 species in at least Bettongia moyesi (potoroine, Two Trees LF, System C; Flannery three genera ranging from very generalisedto highly specialised and Archer 1987c); Cooke et aL (in press) recognise many new (some in press: Muirhead and Archer 1989). genera and species of balungamayinesand new species of aJ. hyp_ Muirhead, in prep.. recognises several Riversleighgenera in and propleopines (several LFs, Systems A- addition to these.One of the Upper Site taxa may be a dasyuroid lf'Y*noaontines rather than a dasyurid.Van Dyck and Archer (in prep.) recognise aBalbaroo riversleighensis (balbarine from D-Site, System A; another new dasyurid (Panorama Site, System B) that resembles Flannery,Archer and Plane 1983); Cooke (1989) and Cooke species of Antechinus and Murexia. ef a/. recognise other new genera and species of balbarine macro_ sMuirhead (in prep.; noted in Bergdolt and Muirhead l9B9) podids (many LFs, Systems A-C). ln the pliocene deposits recognises another family of plesiomorphic perameloids (Rackham's Roost) there are species of and (several LFs. Systems A-C). . ln the Pleistocenedeposits (e.g.,Terrace SiteJ there 6Cott are species of Macropus. (1989) and Gott ef a/. (in prep.) recognise a new genus 2aYalkaparidon and species (several LFs, System B). coheni and Y. jonesi (yalkaparidontian marsupials, 7We Systems B and C; Archer, Hand and Godthelp lggga). recognise Litokoaia n. sp. and another new taxon (several zsYingabalanara LFs, Systems B and C). richardsoni (new family; Upper Site, System B; Archer, Every, Godthelp, Hand and Scally, 1989). Ve recognise an enigmatic higher level taxon of uncertain 26Hipposideros affinities (System B). (Brachipposideros) nooraleebus (Nooraleeba LF; Sig6, Hand and Archer 1982).Hand (1987), Hand,Archer 1Ve recognise new species of lYeohe/os (Systems A-C), Kotopsb and Godthelp (1989a,b)and Hand (in prep.) (System C) and new zygomaturine and diprotodontine recognisemany genera other hipposiderids (many (Systems A-C). LFs, Systems B-C and ?A). 2lMacroderma r0S. godthetpi (Dwornamor LF etc., System Creagh etal (in prep.) recognise new species of Ngapakatdia B; Hand 1985).Hand (1987 and in prep.)recognises at leastthree (System A) and perhaps Palorchestesas well as othetl qenera other megadermatids in various (several LFs. Systems A-C). LFs of Systems B and C and two in the PlioceneRackham's Roost LF (includinq rrN. one that resembles Pledge (in prep.) recognises at least 3 species of /4. gigas. the living species). Namilamadeta (several LFs. Systems A and B). 2sPetramops creaseri(Cotham City 12We LF, System C; Hand, 19g9; recognise at least three new genera of vombatomorphian at least two additional molossids are present in Svstem B lfs viz. marsupials (several LFs, System B) but no undoubted vom_ VD Site and PanoramaSite). batids. 2eCodthelp (in 13We prep.) recognises species of some modern recognise Wakaleo sp. cf. W. otdfietdi(Systems A-C) and genera as well as others representing previously unknown a genus and new species of cf. Priscileo from Upper Site. qenera.

June 1989 Australian Zoologist, Vol. 25(2) 61 30C an is fam il ia ris . Chronological: "Having to do with recognising or discriminating (Woodburne 31From other Riversleigh local faunas, we have representatives intervals of Earth history" 1987). of the: Ektopodontidae (Ektopodon sp.; e.9., Waynes Wok, Correlation: "Demonstration of the equivalence of two or more System B, Pledge ef ai. in press); Ornithorhynchidae (Obduro- geological phenomena in different areas. . .". The correlation don sp., e.g., Ringtail Site, System C; fucher ef a/., in prep.); may be lithologic or chronologic. However, correlated sedi- ?Miralinidae (e.g., Wayne's Wok, ?System B); (new ments or local faunas, in the sense used here, imply the con- genus and species;e.g., White Hunter, ?System B); a new family clusion or hypothesis that the objects are approimately the of vombatiform diprotodontians (e.g., Boid Site East, System same age. family (e.9., Hiatus Site 1, ?D- B); another new diprotodontian Diagenetic: Processes that have effected a sediment, fossil or phalangeridan family near to but dis- Site Equivalents); a new other geological object after it was initially deposited other (e.g., Wok, B). tinct from phalangeroids Wayne's System than weathering or metamorphism. 32The Ditjimanka LF of central Australia (e.9., Woodburne et a/. Eustatic: "Of or pertaining to worldwide changes of sea level". 1985) is probably slightly older than the Upper Site LF of Eustatic changes can bring about renewed deposition or Riversleigh (if only because of its relatively plesiomorphic erosion by changing the gradient over which freshwater flows. ektopodontids and pilkipildrids), but it is not clearly a rainforest Raised sea levels commonly reduce erosion and lead to local fauna. While it is true that its mammals display no evidence continentally peripheral lakes (e.g., such as we postulate for of grazing adaptations, there are proportionately fewer arboreal much of Riversleigh) and estuaries; lowered sea levels mammals in this fauna than in any of the Oligo-Miocene revitalise sluggish streams and increase erosion (such as RiversleighLFs or the Pliocene Hamilton LF. Further, while it is occurs in the Riversleighregion now). true that there is only doubtful evidence for believing that "The terrestrial grasses were present (contra pers. comm. from W. Facies: aspect, appearance, and characteristicsof a rock Harris to Woodburne et al. 1985), perhaps the Ditjimanka LF unit, usually reflecting the conditions of its origin; esp. as represents an open forest environment with a moss, fern and/or differentiating it from adjacent or associated units." As used shrub understorey. Pollen samples from the Etadunna Form- here it implies that contemporaneous but isolated local ation are being re-examined (H. Martin, pers. comm.). faunas may accumulate in different microenvirotiments (e.g., a forest floor and adjacent pool) and hence be different in composition. GLOSSARY OF PALAEONTOLOGICAL AI.{D CEOLOGICAL TERMS (SED Fauna: "... an assemblage of fossil vertebrates of specific taxonomic composition obtained from a number of geo- Definitions in double quotes, unless otherwise credited, are graphically diverse sites" (Woodburne 1987). We do not yet those of Bates and Jackson (1984). ln these definitions, 'paleontology','color' know which Riversleigh assemblages may be composite parts spellingsof the words and (where approp- 'ize' of more widespread, regional faunas hence all that are riate) the endings of words (American spelling) have been 'palaeontology','colour' 'ise'. identified at Riversleigh are called local faunas. altered by us to and Fissure: An eroded space within a deposit of limestone; Assemblage: ". . . a group of fossils that occur at the same 'faunal generally narrow if vertical or low if horizontal. Our interest in stratigraphic level .. .". As used here, we intend fill up with new 'a these natural artifacts is that they commonly assemblage' to mean group of animals that occur in the limestone that often contains bones of animals younger than same site'. In this sense, a faunal assemblage is the physical 'local the limestone that originally developed the fissures. basis for what we ultimately interpret to be the fauna'. Fluviatile: "Belonging to a river; produced by river action...". Biochronologt: ". . . the relative dating of geologic events by Flysch: "A. . . sedimentary facies characterised by a thick biostratigraphic or palaeontologic methods or evidence; i.e. sequence of poorly fossiliferous,thinly bedded, graded marls the study of the relationship between geological time and and sandy and calcareous shales and muds.. ." usually organic evolution". interpreted to represent a rapidly moving mud slurry that Biocorrelation: Demonstration that the same taxon (usually a rushes across a surface or into and through a body of water. species or genus) occurs in spatially isolated deposits thus Formation: "A body of rock strata that consists dominantly of leading to the hypothesis that the isolated deposits are probably a certain lithologic or combination of types. . . A litho- similar if not identical age. Wpe 'Beds' logically distinct, mappable body of . . . rock". (e.9., 'Limestone' Biostratigraphy: "Stratigraphy based on the palaeontologic Ploughed Mountain Beds) and (e.9., Carl Creek aspects of rocks; the differentiation of rock units through Limestone) used in this paper effectively mean the same 'Formation' study of the fossils they contain". thing as (e.9., Etadunna Formation). 'A Biostratigraphic unit: body of strata that is differentiated from Intraformational: "Formed within a geological formation, more adjacent strata by its fossil content or palaeontological or less contemporaneouslywith the enclosing sediments".As character". used here, it refers to a feature evident in some of the Bioturbation: "The churning and stirring of a sediment by Riversleigh limestones where apparent fracturing of semi- limey muds was followed by re-cementation of organisms." As used here we suggest that animals walking consolidated pieces by marginally younger carbonates. over previously accumulated bones might break these prior the fractured to final entombment by calcium carbonate. Lacustrine: "Pertaining to, produced by, or inhabiting a lake or lakes...". Carbonate: "A mineral compound characterised by.. . CO3-2. Calcite and aragonite, CaCO3, are examples of carbonates". Litholory: "The description of rocks.. . on the basis of such As used here it is generally intended as a synonym for lime- characteristicsas colour, mineralogic composition, and grain stone. size... The physical character of a rock."

62 Australian Zoologist, Vol. 25(2) June 1989 GEOLOGICALTIME SCALE (afterPalaeontology Group, Bureau of MineralResources).

Era Period Epoch Ageinyears Recent 0-10,000 Ouaternary Pleistocene 10,000-2,000,000 c) O C.{ Pliocene 2-5,000,000 O = c) Miocene 5-24,600,000 Tertiary Oligocene 24.6-38,000,000 Eocene& Paleocene 38-65,000,000

C) C) Cretaceous 65-144,000,000 O o C\l N O Jurassic 144-213.000.000 o c.r) & kl ztrl 213-248,000,000 rTl 248-286,000,000 A 286-360,000.000 s2 O c{ 360-408,000,000 O E Srlurian 408-438,000,000 a{a Ordovrcian 438-505,000,000 Cambrian 505-590,000,000

CJ O Adelaidean 590-1,400,000,000 D'-l z O & Carpentarian 1,400-1,800,000,000 ts{E-r fr O_ pa o{ lower Proterozoic 1,800-2,300,000,000

C) z Erl Frl & 2,300,000,000+ O{ (J E

June 1989 Australian Zoologist, Vol. 25(2) 63 Limestone: "A sedimentary rock consisting chiefly of the Apart from the collections made at Site D which have been mineral calcite (calcium carbonate, CaCO:), with or without obtained from boulders on a steep scree slope below the Site magnesium carbonate. . . the consolidated equivalentof limy D cliff, we have kept discrete faunal concentrations separate in mud...". order to maximise future ability to detect differences in age or palaeoecology.If and when it becomes apparent that some of Lst: Abbreviation for limestone. these assemblages are identical, we shall combine them as Local fauna: . . . assemblages, containing as few as a single regional faunas. taxon, from a single site or geographically adjacent and temporarily equivalent sites" (Tedford 1967). S.vstemA: ?late Oligocene to early Miocene Oncolite: "A concentrically laminated, calcareous sedimentary structure, resembling an oolith, formed by the accretion ot Sites include: Site A, Site B, Site C (type locality for successive layered masses of gelatinous sheaths of blue- Bematherium angulum Tedford, 1967), Site D, Site E. All of green algae." Normally these would form in water exposed to these and several unnamed taxa are described by Tedford sunlight. (1967). Following that work, we tentatively accept these local Palaeoecology:"The study of the relationship between ancient assemblages to be approximately contemporaneous and to organisms and their environnrent". representthe RiversieighLocal Fauna. Petrology: "That branch of geologv dealing with the origin, Probable System A local fauna (which may or mav not be occurrence,structure, and historyof rocks... . contemporaneouswith the RiversleighLocal Fauna):Alsite. Stratigraphy; "The science of rock strata. Ii is concerned witF ali characters and attributes oi rocks as strata; and thet: interpretation in terms of mode of origin and geologic D-Site Equivalents: ?late Oligocene to early lvliocene history". Limestonestentatively placed in this seriesare so determined Stratophenetics: An integrated apDroach to the study of the because of their resemblance to the sediments of Site D and/or stratigraphic and phylogenetic relationships of evolving their occurrenceas superpositionalsediments on the Cambrian organisms. lt is rlone by superimposingsequentially aged Thorntonia Lst and/or as sediments underlying undoubteC faunas and identifying probable ancestral and descendant System B or C units. Insofar as it has so far been possible tc, Irneaqespassing through the predeterminedttrne planes. determine. manv also share taxa otherwise rare or unknowit 'A, Tectonics: branch of geologt dealing witl' the broac from System B or C units. architecture of the outer part of the eartir. that is. the majo. The Azores Site: Bone Atoll Site; Bone Reef Site; Don structuialor deformationalteatures and their reiations,oriqir Camillo Site: Hiatus I Site; Hiatus ll Site; Jeanette's and historicalevolution . Neville's Pancake 'A Amphitheatre Site; Lockwood's Link Site; Thanatocoenosis: set of fossils brought together after death Site; Sticky Beak Site; White Hunter Site. by sedimentaryprocesses Ie.g.. stream flow etc.],rather than by virtue o{ havrng originally'lived there collectivell . In the sense used here, it describesthe assembiageof normally Cave depcsits: ?just younge: tttan Systern A assemblages disaggregatedorganic rernatnsin a vety restrictedsite (most of whicn we presume were small pools into whiclr antmals Tnesesites are interpretedto have formed as in-fillirrgrsol. or feli or into which their bones were washed out-washrnqsfrom cavesceveloped in surroundinq.older lime- stonc. Travertine: A finely crystallirre.rnassive deposit o1 calcium calbonate.ol white,tan. or cream colour.formed by chemicai BitesantennarySite (Hand. Archer and Codthelp i989a): precipitatronfrorn solutiorriri surtaceand ground waters.as Microsite (source of Nooraieeba Locai Fauna and type localiX, around the mouth of soringls.. lt alsc occurs in limestone for Hip;tosideros (Brachipposideros) nooraieebus Sige, Han

APPENDX: LIST OF THE PRINICPAL RTVERSLEIGH System B: 2earl.vlliocene' LOCAL FAONAS AND THEIR RELATME AGES These sitesappear to representassemblages accurnulated in Thereare in excessof 97 sitesand cottespondinE local taunai pools or shallou lakes Neville'sCarden Site was undor.rbtedly assemblaqesnow known from the Riversiei-qhregion. The iirst formei near caves n! limestone overhangs because oi tht: ol these may have been that which Tediord i196,7)oescribed presen:? oi broken straws (speleothems) and in -srfutravertine as the RiversleighLocal Fauna,a conrpositeassemblage frorr' ridgeson what appearto havebeen irregulaisurfaces. Because years I 979 and 1983-to the present. his Sites A-8. ln the 978-1 of this it is also ;:ossiblethat some of the vertebratesmay have expeditionsto Riversleighhave yielded manv more local faunas been washed in'.o the deposi'.from the cave. having been pit' the most importantol which are listedhere. trapped b.l holes in its rcof. ln what follows,a drstinctionmay be made betweenthe fcsstl (e.g.. site (e.q..Microsite) and its local fauna NooraleebaLocal Low assemblares Faunal as explainedin the Introductron.lf it is not, the local tauna name is that of the site name listedie.g.. Bitesantennary Bole's Bonanza; D.D.D.D. Site; Female Siie; Outasite: Site containsthe BitesantennaryLocal Fauna). Local fauna may Outasite A, B. C. D etc.: RSO Site: Stop Site; Trash Site; be abbreviateCas LF (LFs for plural) Mctor's Vacuum Site; Wayne's Wok Site: Wayne's Wok Il Site.

64 Australian Zoologist. Vol. 25(2) June 1989 Higher assemblages Jaw Function Site (source of Kolopsis-like zygomaturine); Neville's Riches Site; No name Site (contains large Boid Site; Boid Site East; CS Site 1:gu-"1 Sputum Site; zygomaturines); Site; Turtle Site; Two Trees Site type locality for Yalkaparidon coheni Archer, Hand and Godt- (type focality for Bettongia moyesl Flannery and Archer, 19B7c). help, l9BB); G Spot Site; Helicopter Site; Inabeyance Site; Mike's Menagerie Site; MP2 Site; Panorama Site; Paul Willis (type Site; R.V. Site; Ten Bag Site; Upper Site locality for Yrn- Pliocene Sites gabalanara ichardsoni Archer, Every, Godthelp, Hand and Scally, 1989); Mew Delightful Site. The only undoubted Pliocene site identified from the region is Rackham's Roost Site (m) which represents the indurated Probable System B equivalents floor of a cave develooed in Thorntonia Limestone.

Neville's Garden Site; Souvenir Site; Judy's Jumping Joint Site. Pleistocene Sites

There are many Quaternarydeposits in the Riversleighregion. System C: ?middle lvliocene to early late lvliocene However, few have been more than cursorily examined and The sites in this System are interpreted to represent one of none has been preciselydated. Only Terrace Site, developed in three basic environments: deeper pools dominated by aquatic perched fluviatile sediments adjacent to the present Gregory vertebrates (a); shallow pools or even emergent accreting River,has been excavated.The precise age of this site is unclear (see surfaces dominated by terrestrialvertebrates; and sites adjacent above under Correlation).Other fluviatiledeposits of prob- to cave entrances dominated by what appear to be out-washings able Pleistoceneage are known. (m). from beneath megadermatid roosts Cave deposits and fissures are also common in the area. Some contain bones but few have been examined. Their ages Lowest assemblages are likely to vary but have not yet been determined. An example is Atomic Fissure Site which is a vertical fissure fill in (source fuchie's Absence; Bernie's Bedford Site; Gag Site Thorntonia Limestone. lt contains rodents and a soecies of Pet- of Dwornamor LF and type locality for Ekaltadeta r'ma Archer rogale. and Flannery, 7985, Trichosurus dicksoniFlannery and Archer, 1987a, Yalkaparidon jonesi Archer, Hand and Godthelp, 1988, Some Quaternary sites are probably late Pleistoceneon the Macroderma godthelpi Hand, 1985, Hypsipryrnnodon grounds that they contain modern species but taxa that no baftholomaii Flannery and Archer, 19B7b); Kangaroo Jaw Site; longer survive in this region. Examples would be Bottom Site Last Minute Site (type locality for Djilgaringa gillespieae Archer, and Outer Limits Site developed as fissure fills in Thorntonia Tedford and Rich, 1987, Strigocuscus reidi Flannery and Limestone. tucher, 1987a); Wang Site.

Middle assemblages Recent Sifes Bat Smear Site; Bird Bone Site; Courtenay's Cache Site; There are also many sites that appear to represent Recent Gotham City Site (m) (rype localily of Petramops creaseriHand, episodes of accumulation. Most have neither been excavated 1989); Gotham East Site; Gotham North Site; Grime's Site; nor named. Some that have been cursorily examined contain Group Site; Harry's Hump Site; Incisor Site; Mac's Mound Site; modern taxa appropriate to the region today but lack the intro- Main Site; lvlelody's lvlaze Site (a); Quentin's Quarry (a); duced House Mouse (llus musculus), a marker which heralds Ringtail Site (a; source of Obdurodon sp. skull); Ringtail East the European invasion of Australia.Other sites have House Mice Site; Ringtail East Site 2; Scum Site; Skull Site; Sue's Dip- and are therefore presumably less than 400 years old. rotodontid Site. Sites and LFs of (as yet) unclear relationship Probable middle assemblage equivalents ABR Site; Biggles Site (between the Gag Plateaudominated fuchie's Parlour Site; Jim's Jaw Site. by System C lfs and Godthelp Hill dominated by System B lfs); Bat Patch Site; Burnt Offering Site; Chinatown Site; Fig Tree Highest assemblages Site; M6 Site; Murray B.'s Site; Neo Site; Rackham's Rock Pile Site; Roo Site; Sewer Site; Souvenir Heights Site; Sue's E Bob's Boulders Site (a); Crusty Meat Pie Site (a); Fireside Site; Sue's Selection Site; Syp's Siberia Site; Upper Burnt Favourites Site: Grimes' Site: Henk's Hollow Site: Offering Site.

Top View