Records of the Western Australian Museum Supplement No. 57: 341-350 (1999).

The Pleistocene mammal fauna of Kelangurr Cave,central montane Irian Jaya, Indonesia

Timothy F. Flannery

Mammalogy, Australian Museum, 6 College St, Sydney, NSW 2000;email: timf@amsg.austmus.gov.au

Abstract - Sixteen mammal taxa have been identified on craniodentalmaterial from a rich deposit of bones in the first chamber of Kelangurr Cave.The remains of all species except the largest (Maokopia ronaldi andProtemnodon hopei) and the smallest (Miniopterus sp. cf. M. macrocneme) andpossibly a single tooth cap of Anisomys imitator, are considered to have beenaccumulated by Pleistocene Sooty Owls (Tyto tenebricosa). About a dozenbird bones representing species ranging in size from finch to medium-sizedhoneyeater, and currently unidentifiable, are the only non-mammalianremains from the deposit.

Kelangurr Cave is at an elevation of 2,950 m, and presently located in tallupper montane forest. Analysis of the habitat requirements of the fossilmammal fauna indicates that at the time of deposition (25-20,000 BP) the cavewas surrounded by alpine tussock grassland and scrub similar to thatoccurring today at 4,000-4,200 m on Mt Carstensz and Mt Wilhelmina.

The age of faunal remains not accumulated by owls is uncertain. There isno evidence of non-analogue mammal communities, and no evidence ofextinction among species under 2 kg in weight in the fauna.

INTRODUCTIONThe Kwiyawagi area (Figure 1) is the only part of

Irian Jaya that has proved to be rich in Pleistocenevertebrate fossils (Flannery 1994). Remains of giantextinct marsupials were first discovered there byPastor Ooug Hayward of the Unevangelised Fields

Mission during or before the 1980s. In January 1990Or Geoffrey Hope walked in to the West Baliemvalley from Wamena, and found bones to beabundant both in Kelangurr Cave and in sedimentsexposed in the banks of the West Baliem River. Ireturned to the area with Or Hope in June 1990 and

Settlement

x Fossil sites

o Pollen analysis sitesTiom 0

IRIAN JAYA

\

Mt Trikora, 4760m

\--L---------c'--------J

4"$

Figure 1 Location diagram, upper Baliem River catchment, Irian Jaya (inset).

Table 1 The fossil mammal fauna from Chamber 1, Kelangurr Cave: ,. possibly or certainly not accumulated byPleistocene owls. MNI = minimum number of individuals; % = per cent of fauna; Wt = mean or estimatedadult weight; E. r. =present elevational range (m a.s.l.); (weight and elevational range data for living speciesmainly from Flannery 1995); Survival: p =still present, np =not present in the area, ex =extinct. ? =no data.

Taxon MNI % Wt E. r. Survival

DasyuromorphiaNeophascogale lorentzii 1 0.2 200 g 1500-3300 PPeramelemorphiaMicroperoryctes longicallda 48 8.7 540 g 1000-3950 PDiprotodontiaMaokopia ronaldi ,. 1 0.2 100 kg ? exProtemnodon hopei ,. 1 0.2 40 kg ? exThylogale christenseni 2 0.4 2 kg ? exThylogale sp. 1 0.2 6 kg ? exCercartetlls calldatus 46 8.3 20 g 1500-3450 PPseudochirlllus mayeri 290 52.6 150 g 1500-4200 PMicrochiropteraMinioptenls sp. cf. M. macrocneme ,. 1 0.2 9g 0-3200 ?RodentiaAnisomys imitator ,. 1 0.2 500 g c. 0 - 3500 PCoccymys riimmleri 38 6.9 30 g 1900-3600 ?Coccymys sp. 2 0.4 c. 40 g ? ?Hydromys habbema 1 0.2 c. 80 g 2800-3600 PMallomys gllnung 6 1.1 2 kg 3500-4430 np5tenomys richardsoni 101 18.3 70 g 3225-4500 np5tenomys omlichodes 11 2.0 75 g 2950-3950 P

TOTALS 551 100.1%

Figure 2 Plan of Kelangurr Cave showing location of Chambers and bone specimens.

T.F. Flannery

--+-.~ N

described by Hope et al. (1993) and the sites areshown in Figure 1.

The Cave lies about 10 km west of the settlementof Kwiyawagi in a side valley leading to the WestBaliem River. It consists of a passage about 350 m

Bedrock

Talus core

Scattered black bone

~ Palaeo-f1ow direction

o 1001 I

metres

made additional collections. The faunal remainsincluded those of two large, extinct marsupialsMaokopia ronaldi Flannery, 1992b and Protemnodonhopei Flannery, 1992b. The geology, palynology anddates obtained from the deposits have been

342

Mammal fauna of Kelangurr Cave, Irian Jaya

long averaging 3 m high, which slopes gentlydownward from a cliffed entrance (Figure 2). It issurrounded by diverse, tall upper montane forest,which includes species of Podocarplls, Pltyllocladllsand other gymnosperms. An area of swampy,valley floor vegetation is present approximately 1km away.

During a visit to Kelangurr Cave in 1994 by theauthor and Alexandra Szalay, additional materialwas collected, including approximately 18 kg ofmatrix and lag deposit, rich in fossil bones, fromthe first chamber (Chamber 1). This material, whichconsists largely of the bones of small mammals,forms the focus of the present study.

MATERIALS AND METHODS

About 4 kg of primary matrix and 14 kg of lagdeposit (consisting principally of mud and fossilbones) were removed from the walls and floor ofChamber 1 and taken to Kwiyawagi, where the lag

343

material was washed and sieved through flywire(mesh approximately 1 mm). The bones and rockfragments were then dried, packed and transportedto the Australian Museum, where the loose boneswere sorted and the craniodental fragmentsseparated from the rest. An attempt was made toacid prepare a small sample of matrix(approximately 2 kg), but with little success. Thesmall sample of bones recovered from this attemptcomprises only taxa recorded from the lagmaterial.

The craniodental material was identified using theAustralian Museum mammal collection as areference. This collection includes examples of allextant species identified except Coccymys sp. Detailsof identification criteria are provided in the MammalSystematics subsection of the Discussion, below. Themegafaunal specimens (Maokopia ronaldi AM Fl00743and Protenmodon /lOpei AM F83413) and four 5tenomysspecimens (5. ricltardsoni AM FI00744 and 5.omlicltodes AM FlO0745) have already been accessed

Table 2 The present mammal fauna of the Kwiyawagi area, including the vicinity of Kelangurr Cave, as determinedfrom faunal survey 4-10 June 1994. Data on body weight range (Wt), present elevational range in metresa.s.!. (E. r.), and primary habitat (Habitat) mainly from Flannery (1995). Dendrolagus mbaiso Flannery, Boeadiand Szalay, 1995 and Mallomys gUnlmg were collected only high in the Prinz Wilhelm Range to the south ofKwiyawagi.

Taxon Wt E. r. Habitat

DasyuromorphiaDasyurus albapunctatus 580-710 g 0-3500 forestNeophascogale lorentzii 212 g 1500-3300 forestPeramelemorphiaMicroperoryctes longicauda 350-670 g 1000-3950 forest/ alpinePeroryctes raffrayana 650-1000 g 60-3900 forestDiprotodontiaPhalanger carmelitae 1700-2600 g 1400-3660 forestPllalanger sericeus 1750-2405 g 1500-3900 forestDendrolagus dorianus 6.5-18 kg 600-3300 forestDendrolagus mbaiso 8.5-9 kg 3250-4200 alpineDorcopsulus vanheurni 1500-2340 g 800-3100 forestDactylopsila palpator 260-550 g 1200-2950 forestPetaurus breviceps 69-114 g 0-3000 forestPseudochirops corinnae 925-1300 g 1200-2900 forestPseudochirops cuprells 1300-2250 g 1700-3996 forest/ alpinePselldochirullls mayeri 105-206 g 1500-4200 forest / alpine

MicrochiropteraPipistrellus collinus 5.1-6.9 g 1700-2950 ?Tadarida kuboriensis 23-27.5 g 1900-2950 ?

RodentiaAnisomys imitator 388-580 g 0-3500 forestHydromys sp. cf. H. lzabbema c. 80 g 2800-3600 rivers / lakesMallomys gllnlmg 2000 g 3500-4430 alpineMallomys rothschildi 925-1500 g 1550-3700 forestParamelomys rubex 31-57 g 900-3000 forest/ alpineRattus steini 110-220 g 20-2800 gardensStellomys niobe 36-55.5 g 760-4050 forest/ alpineStenomys omlichodes 69-78 g 2950-3950 swamps/alpineUromys anak 450-1020 g 850-2950 forest

344

into the Australian Museum palaeontologicalcollection. The other material will be added to thatcollection in the near future.

RESULTS

For the modem mammal species, classificationgenerally follows Flannery (1995), includingrecognizing Stenomys as a genus separate fromRattus. An exception is members of the Melomyscomplex, where Menzies (1996) is followed. Theextinct Thylogale, previously identified as T. brunii,is here referred to as Thylogale sp. followingFlannery (1992a). Authorities are only given forspecies not included in Flannery (1995) or Walton(1988). Higher category taxa of marsupials followAplin and Archer (1987).

Fossil Fauna from Kelangurr CaveSixteen species of mammals, ranging in

estimated mean body weight from 9 g to 100 kg,have been identified in the fossil fauna fromChamber 1 of Kelangurr Cave (Table 1, overleaf).Minimum numbers of individuals weredetermined from the most abundant identifiedskeletal element of each species (Table 1 and Table4, below). The only other vertebrate remainscollected consist of about a dozen currentlyunidentifiable bones of birds, ranging in size froma finch to a medium-sized honeyeater.

Extant Mammal Fauna of the Kwiyawagi AreaA survey of the extant mammal fauna of the

Kwiyawagi area was undertaken at the time of the1994 visit. Results are given in Table 2, andindicate that the area today supports a faunawhich is primarily forest dwelling. The alpinespecies listed in Table 2 were collected in themountains which rise on the southern side of theWest Baliem Valley.

Dating of the DepositNew optically-stimulated luminescence (OSL)

and uranium series dates, the details of which willbe published elsewhere in conjunction with RG.Roberts and J.M. Olley, indicate that the deposit inChamber 1 accumulated between 25,000 and 20,000BP. The OSL dating was undertaken on tiny sandgrains which had probably made their way in tothe deposit in the fur of prey items. These wouldpresumably have been picked up from the surfaceof the ground where they would have beenthoroughly bleached, so it seems reasonable toassume that the faunal remains derived from owlpellets are accurately dated by this technique. Thebackground radiation dose rate, used to calibratethe OSL dates, was measured in the laboratory onthe bulk sample of deposit.

As will be discussed below, a small proportion of

T.F. F1annery

the fossil material from Chamber 1 is probablyreworked from an earlier fossil accumulation andmay thus be considerably older than 25,000 BP.This material includes the remains of the extinctgiant marsupials.

DISCUSSION

Mammal SystematicsThe species identified from Kelangurr Cave can

be divided into three categories according to theirelevational and temporal distributions. A) Speciesunknown from the immediate area of Kwiyawagitoday, being restricted to higher elevations, B)Species still occurring in the Kwiyawagi area, orknown from similar elevations elsewhere in IrianJaya, and C) extinct species.

A) Taxa unknown from the immediate area ofKwiyawagi today, being restricted to higherelevations (Mallomys gunung, Stenomys richardsoni).

The Kelangurr Cave sample of giant rats wasidentified as Mallomys gunung (rather than one ofthe other three species of Mallomys known toinhabit the Maokop, Flannery 1995), primarily onthe basis of white incisor enamel, a characteristicunique to M. gunung in its genus.

The molars of the fossil Mallomys are comparablein size to those of M. gunung, and larger than thoseof M. rothschildi (the only species known to inhabitthe area of the cave today; Table 3). They are alsolarger than those of M. aroaensis, but overlapconsiderably with M. istapantap (Table 3). None ofthe palatal remains from the Kelangurr sample hasan ovoid choana (unique to M. istapantap in thegenus), so despite the similarity in molar size,identification of palatal remains with this specieswas excluded.

The identification of the Stenomys sample provedmore difficult. Three species of Stenomys are knownfrom subalpine grassland on the Maokop (Flannery1995). Their taxonomy is still poorly understood.Stenomys omlichodes has only recently been re­elevated to full specific status, distinct from S.richardsoni (Flannery 1995). Adequate craniodentaldiagnosis of these taxa based on examination of alarge sample is still unavailable.

I was able to satisfy myself that S. niobe is absentfrom the Kelangurr Cave sample. Its molars aremarkedly smaller, more slender and more cuspidatethat its congeners, and none resembling it was presentin the fossil sample. I had difficulty, however, indistinguishing S. richardsoni from S. omlichodes.

One useful character for distinguishing thesespecies is the colour of the incisor enamel; that ofS. richardsoni is white, while that of S. omlichodes isyellow. Unfortunately, many fossil specimens donot include an incisor, so other criteria had to beused as well.

Mammal fauna of Kelangurr Cave, Irian Jaya 345

Table 3 Dental measurements of species of Mallomys. (a) fossils identified as M. gHllIlIlg from Chamber 1 deposit,Kelangurr Cave; (b) = single fossil identified as M. glllllillg from deeper inside Kelangurr Cave; (c) = modernsample M. gllllllllg from all New Guinea; (d) remains of recently dead individuals of M. rotlzsclzildi fromChambers 2-4, Kelangurr Cave; (e) modern sample M. rothsclzildi wcylalldi from all New Guinea; (f) modernsample M. aroaensis aroaellS1S from all New Guinea; (g) modern sample M. istapalltap from all New Guinea.Upper molar measurements of (c), (e), (f) and (g) from Flannery et al. (1989). All measurements in mm.

(a) M. gl11l111lg

(b) M. gWllIllg

(c) M. gll1l11ng

(d) M. rothschildi

(e) M. rothsclzildi

(f) M. aroaensis

(g) M. istapalltap

M' 1 length MJ width M"length MJwidth

x 18.8 58 18.1 5.0R 17.9-20.0 5.4-6.4 17.6-185 4.8-5.111 3 3 4 8s 1.33 0.51 0.39 0.14

19.2 6011 1 1 0 0

x 18.8 5.8 17.9 4.9R 17.4-20.0 5.6-6.0 17.7-18.0 4.8-5.011 5 4 2 2s 0.96 0.18-

16.8 5.5 16.2 4.5xR 16.6-17.0 5.4-5.5 16.1-16.2 4.511 2 2 2 2

x 16.5 5.2 16.3 4.5R 15.7-17.5 5.1-5.4 16.0-16.6 4.3-4.6n 11 9 2 2s 0.21 0.11- 17.0 5.2xR 16.3-18.0 4.9-5.4n 10 10s 0.67 0.13

x 18.4 5.4R 17.4-19.1 5.0-6.0n 8 8s 0.57 0.35

My small sample of modem specimens of bothspecies (about six of each) indicate that some S.richardsoni possess accessory buccal cusps on thelower molars (see Figure 3), but these are notinvariably present. In addition, the molars of S.richardsoni are broader and more robust than thoseof S. omlichodes (see Figure 3 for examples).Although these differences did not always allow for

unequivocal identification of each fragment, thefossil sample was divided into specimens moreresembling S. richardsoni in molar robustness andmorphology, and those resembling S. omlichodes.

The results of these comparisons (Table 4) suggestthat whichever criteria are used, specimens referredto S. richardsoni are more abundant than thosereferred to S. omlichodes (in total about 74% of all

Table 4 The Stellomys dentary and dentary fragment sample (N = 123). Groups: (1) = identified as S. richardsolli onthe basis of both molars and white incisors; (2) = identified as S. richardsolli on the basis of molars (incisorsabsent); (3) = identified as S. richardsolli on the basis of white incisors (molars absent); (4) = specimens withwhite incisors but S. omliclzodes-like molars; (5) = specimens with yellow incisors, but S. richardsolli-likemolars; (6) = identified as 5 omlicllOdes on the basis of molars (no incisors); (7) = identified as S. omlichodes onthe basis of yellow incisors (no molars); (8) = identified as 5 omlichodes on the basis of both molars andyellow incisors; (9) could be identified only to genus Stellomys. Per cent = percentage of sample in each ofthe four major categories.

S. richardsolli intermediate S. omlichodes Stellomys sp. indet.

Group (1 ) (2) (3) (4) (5) (6) (7) (8) (9)CountsLeft 44 35 20 3 4 0 7 1 14Right 48 30 23 4 1 4 0 7 19Per centLeft 77.3 5.5 6.3 10.9Right 74.3 3.7 81 14.0

346

mm

T.F. Flannery

mm

Figure 3 Photomicrographs of murid dentaries identified as Stenomys ric1ulrdsoni, AM FlOO744 (left) and S. omlichodes,AM FlOO745 (right).

remains v. about 8%: with about 18% unidentifiableto species).

B) Taxa still occurring in the Kwiyawagi area, orat similar elevations elsewhere in Irian Jaya(Neophascogale lorentzii, Microperoryctes longicauda,Cercartetus caudatus, Pseudochirulus mayeri,Miniopterus sp. cf. M. macrocneme, Anisomys imitator,Coccymys rummleri, Hydromys habbema, 5tenomysomlichodes).

These nine species are all known to occur ataround 2,900 m elevation in central Irian Jaya atpresent. Identification of all except C. rummleri,which is in need of systematic revision, wasstraight forward. The Kelangurr Cave sample of C.rilmmleri consisted of larger individuals than those(all from Papua New Guinea) which wereavailable for comparison. Hope (1976) also notedthat Holocene specimens of this species from theCarstensz area were larger than the modernspecimens with which she compared them. Thesignificance of this is presently unknown, but it

may indicate the presence of cryptic species withinwhat is currently referred to as C. rummleri. Thetype locality of C. rilmmleri is Lake Habbema, notfar to the east of Kwiyawagi (see Figure 1), hencethe fossil sample is likely to represent thenominotypical taxon. The fossil material referredto Coccymys sp. is extremely fragmentary. Itappears to represent a species slightly larger thaneven the fossil C. rilmmleri, but with less complexmolars. It may be the poorly-known C. albidens,only recorded from the alpine zone on MtWilhelmina (Trikora).

The present elevational distributions of the eighttaxa identified to species level vary, but all exceptAnisomys imitator (which is restricted to forest) areknown to inhabit areas of subalpine scrub and tooccur above the treeline; and all of the othersexcept Neophascogale lorentzii and Pseudochirulusmayeri (the latter the most common species in thesample, see Table 1) can be found in tussockgrassland. 5tenomys omlichodes is not known to

Mammal fauna of Kelangurr Cave, Irian Jaya

inhabit forest. The upper elevational limits of mostspecies are poorly known. There is some doubtabout the provenance of two of these taxa(Miniopterus sp. cf. M. macrocneme, Anisomysimitator). Both may represent moderncontamination of the lag deposit (see below).

C) Extinct taxa (Maokopia ronaldi, Protemnodonhopei, Thylogale sp., Thylogale christenseni Hope,1981).

The elevational ranges of all of these taxa arevery poorly known. Fossils of Maokopia ronaldi andProtemnodon JlOpei are only recorded from theKwiyawagi area (Flannery 1992b). Remains of bothspecies have been found in sediments exposedalong the West Baliem River. Fossil pollen andplant macrofossils from the same deposits suggestthat the area was above or near the treeline at thetime of deposition (Hope et al. 1993).

The dentitions of both M. ronaldi and P. hopei areconsistent with a diet relatively rich in silica andother abrasive material (Flannery 1992b). Bothspecies have higher crowns and more prominentlinks than related species whose fossils have beenfound at lower elevation. These data suggest thatboth consumed grasses and lived above thetreeline, where grass is most abundant. As theremains of these taxa were clearly not accumulatedby owls, their association with the rest of the faunain the deposit is uncertain (see below).

PaderneIons (genus Thylogale) are known fromthe Maokop only as fossils. Although their originalelevational distribution is poorly known, remainsof late Holocene age have been recorded fromrockshelters in tussock grassland at 3,990 and 3,700m elevation (Hope 1976; Hope et al. 1993). Thisevidence, plus the nature of their dentitions(relatively high crowned with well-developedlinks) and the ecology of related populations(Flannery 1992a) all suggest that they inhabitedgrassland and herbfield communities above thetreeline.

The time of the extinction of these species ispoorly known. The remains of Maokopia ronaldi andProtemnodon JlOpei have been found in associationwith peats dated to 39,860 ± 770 BP in the WestBaliem River, and with wood, the cellulose fractionof which has been dated at 42,740 ± 1020 BP (Hopeet al. 1993).

Thylogale sp. and T. christenseni are known frommuch more recent deposits, including Billingeekrockshelter in the Prinz Wilhelrn V Range, wheretheir remains are associated with charcoal dated asyoung as 3,250 ± 80 BP (Hope et al. 1993).

Agents of AccumulationWith the exception of the remains of the very

large and very small taxa (see below), all of theother mammalian remains appear to have resultedfrom accumulation at an owl roost.

347

The bone accumulation present in Chamber 1 ofKelangurr Cave probably exceeds several cubicmetres in volume. It is plastered on the sides andfloor of the cave, and is most obvious about 5 m infrom the cave entrance. It is currently in thetwilight zone. It seems likely that the bulk of thedeposit has already eroded away.

As Baird (1991) noted, owls are "by far the mostfrequent accumulators of vertebrate bone in caves".The only other obvious possibility is derivationfrom acciptrid pellets, but the high frequency ofundamaged limb bones among the remains isinconsistent with this hypothesis.

Among the existing owl species of New Guinea,only the Sooty Owl (Tyto tenebricosa) occurs atelevations above 2,500 m (Beehler et al. 1986). Thisspecies also occurs in southeastern Australia,where its biology has been extensively studied.Australian Sooty Owls can reach 1 kg in weightand a total length of 51 cm (Schodde and Mason1980). They have large talons and are capablepredators, preying upon a wide variety ofmedium-sized to small mammals. Single or pairedSooty Owls forage over ranges of 200 to 800hectares around their roost (Schodde and Mason1980).

In Australia, the Sooty Owl has been found tofeed upon animals ranging in size from the HouseMouse (Mus musculus, mean weight 18 g) to theCommon Ringtail Possum (Pseudocheirus peregrinus,mean weight 900 g). The prey species mostcommonly recorded in a series of five sites in NewSouth Wales is the Sugar Glider (Petaurus breviceps,mean weight 128 g; Debus 1994). This general preyprofile fits well with that observed at KelangurrCave, where the estimated mean weight of thesmallest prey is 20 g, and the largest 1 kg Uuvenilesof Mallomys gunung and Thylogale sp.), with themost common prey species, Pseudochirulus mayeri,averaging 150 g.

New Guinean Sooty Owls average 36 cm inlength (Beehler et al. 1986); only slightly larger thanthe Barn Owl. The biology of the New Guineanpopulation is poorly understood, but it is known tohunt rodents and marsupials in forest and at theedge of alpine grassland (Beehler et al. 1986). Itoccurs from sea level to 4,000 m elevation. It istheoretically possible that another, as yet unknownowl species existed in New Guinea during thePleistocene, and this species was responsible forthe accumulation.

Ages and Provenance of the FaunasAs noted in Results, OSL and uranium series

dates indicate that the deposit in Chamber I,containing the fauna derived from owl pellets,accumulated between 25,000 and 20,000 BP. Fourspecies, however, represented by just five fossilfragments, were probably not accumulated by

348

Pleistocene owls. The age and provenance of theseremains needs to be separately assessed.

Two taxa (represented by just three fragments)may be recent remains that have been mixed withthe Pleistocene lag deposit. These are Miniopterussp. cf. M. macrocneme and Anisomys imitator.

The smallest species in the deposit is thecavemicolous bat Miniopterus sp. cf. M. macrocneme(weight about 9 g), which is represented by left andright dentary fragments, possibly from the sameindividual. This species has been recorded roostingin caves at 2,600 m elevation, and has beenobserved flying over alpine herbfield at 3,200 melevation (Flannery 1995). The two bones of thisspecies thus far identified may represent theremains of an individual which was roosting inKelangurr Cave in recent times. The fact that thebone is light and fragile and the tooth enamel white,is consistent with this interpretation.

Anisomys imitator is represented in the deposit bya single Ml enamel cap, which is white in colour.Many of the other specimens have variably stainedenamel. The Anisomys tooth cap may differ in ageand/or origin from the rest of the remains, althoughthere is no way to test this empirically at present. Itshould be noted, however, that the remains of otherlarge, recently deceased murids, have been founddeeper in the cave.

Only two mammal bones found in Chamber 1come from individuals estimated to have weighedmore than 1 kg (see Table 1). Neither could havebeen preyed on by owls. These are a dentary of anadult Protemnodon hopei (estimated weight 40-60kg) and a scapula of Maokopia ronaldi (estimatedweight 100 kg). (Although the estimated adultweights of the two Thylogale species were alsomore than 1 kg, the species are represented in thesample by juveniles.) The left dentary ofProtemnodon hopei (AM F83413), although exposedon one side, was actually embedded in the owlpellet deposit. The scapula of Maokopia ronaldi (AMFlO0743), was found lying loose on the chamberfloor, with fragments of matrix adhering to it.

Hope et al. (1993) visualized Kelangurr Cave asconsisting of six chambers (see Figure 2). Fossilremains were found in four of these, being mostcommon in Chambers 1 and 2. The cave slopesdownwards from its mouth, and Hope et al. (1993)hypothesized that the cave was once much larger,and that the majority of it (including the originalentrance) was carried away in a prehistoriclandslide. Some evidence suggests thatmegafaunal remains have been migratingdownslope in the cave for some time. Some bonesof M. ronaldi and P. hopei were found cementedinto a flows tone in Chamber 3. These bones do notappear to be in situ, but to have been derived fromsome other primary deposit (Hope et al. 1993).Depending upon the past morphology of the cave,

T.F. Flannery

it is possible that Kelangurr Cave may have actedas a pitfall trap for larger marsupials, ormegafauna may have used the cave as a shelterand died there.

Given the history of the cave and its deposition,and the fact that the megafaunal remains fromChamber 1 represent isolated, albeit well­preserved elements, I have no confidence that theyare in fact contemporaneous with the owl pelletremains. They may be much older, having beenwashed into Chamber 1 from the now lostproximal region of the cave, in an alreadyfossilized state.

The relationship between the owl pellet deposit inChamber I, and other deposits in the cave, remainsuncertain. Hope et al. (1993) analysed the chemicalcompositions of bones of Mallomys collected inChambers 2-4 of the Cave. Black-stained bonesdiffered substantially in composition from thelighter coloured bones, suggesting an agedifferential (see below). Hope et al. (1993) alsosuggested that the black bones werecontemporaneous with or slightly older than blackflowstones found in the same chambers andmeasured by ESR at 68 ± 14 and 80 ± 40 Grayunits. Unfortunately, in the absence of backgrounddose rates for these deposits, it is impossible toconvert these measurements into dates.Interestingly, the fluorine, organic nitrogen andapatite CO

2contents of the bones and flowstone

are similar. In the absence of obvious leaching, thisargues for a stable depositional environment inwhich the black materials represent a groundwaterchemistry different from that prevailing at present.

The strong correlation between the chemistry ofthe black flows tones and black bones fromChambers 2-4, suggests that both may be of similarage. The material from the deposit in Chamber 1 isvariably stained, some bones being black, whileothers are much paler. Fragments of black-stainedflows tone, apparently formed in situ, are alsopresent in the Chamber 1 deposit.

Heavily mineralized bones, blackish in colour, ofStenomys sp. and Mallomys gunung were foundlying loose in Chambers 2-3. Other fauna! remainsrecovered from Chambers 2-4 appear to have adifferent origin and possibly a different age. Theseinclude less mineralized bones of Microperorycteslongicauda, Phalanger sericeus, Thylogale sp., T.christenseni, Dactylopsila palpator, Mallomysrothschildi and Uromys anak. Of these taxa P.sericeus, D. palpator, M. rothschildi and U. anak areobligate forest dwellers. Some remains of the lasttwo species represented recently dead individuals.A living U. anak was found in the cave at the timeof the 1990 visit, and all except the two species ofThylogale (which became extinct some time after3,250 years ago) are still to be found around thecave (see Table 2).

Mammal fauna of Kelangurr Cave, Irian Jaya 349

Table 5 The fossil mammalian fauna from Mapalarockshelter. The left hand column lists taxarecognized by Hope (1976), and the right handcolumn the names used in this paper. 1

taxonomic change, 2 re-identification.

of deposition, forest communities lay at least thisfar and possibly even more distant from KelangurrCave.

The abundance of Pseudochirulus mayeri in thedeposit suggests the presence of significant areasof alpine scrub in the vicinity of the Cave. This wasmost likely confined to rocky slopes.

Hope (1976) gave an account of the mammalfauna of Mt Carstensz, as well as a subfossil faunarecovered from Mapala rockshelter, which islocated at 3,996 m elevation just above the treelineon the Kembau Plateau. The lower levels of theMapala rockshelter deposits have been dated to5,440 ± 130 BP (Hope 1976). 1 have re-examined thematerial from Mapala rockshelter and present inTable 5 a new faunal list for the site, based onmodem systematic concepts. All of the extant taxaexcept Dasyurus albopunctatus, Phalanger sericeusand Hyomys dammermani are known to inhabitalpine scrub above the treeline. As the depositresults from human accumulation, the threeexceptions may have been captured at lowerelevations and carried up to the shelter.

Pseudochirops cupreus is abundant in the Mapalarockshelter fauna and is the most commonlyencountered game item in alpine scrub in the areatoday (Flannery 1995). Yet, remarkably, it isentirely absent from both Kelangurr Cave andBillingeek rockshelter, dated at 3,250 ± 80 BP(Hope et al. 1993). It may be a recent arrival in thishabitat.

Hope (1976) also gave an infonnative account ofthe contemporary, indigenous mammal fauna ofthe highest parts of the Jaya mountains. Very fewspecies occur above 4,200 m elevation. Sherecorded only Mallomys gllnung (as M. rothscllildi),Stenomys richardsoni and S. omlichodes (the latter asS. niobe). To this list can probably be addedPseudoclzirulus mayeri (1 have found skulls of

Past Vegetation Fonnations Around KelangurrCave

A number of lines of evidence suggests that thefossil deposit preserved in Chamber 1 of KelangurrCave accumulated at a time when the surroundingarea was vegetated with tussock grassland andsome alpine scrub, the latter possibly confined torocky slopes. Forest was almost certainly absent, asthe remains of obligate forest dwelling mammalsare almost entirely absent from the deposit.

The presence of Mallomys gummg, and apparentlyof no other Mallomys species, in the deposit,provides an important insight into palaeoecology.Mallomys gummg is the only member of its genuswhich is an obligate inhabitant of plantcommunities occurring above the treeline. All otherMallomys inhabit forest, although M. istapantapextends into alpine grassland in Papua NewGuinea in the absence of M. gunung (Flannery1995). The fact that all identifiable remains ofMallomys in the deposit are M. gunung is consistentwith the hypothesis that at the time ofaccumulation, the treeline was below KelangurrCave.

Stenomys richardsoni has only been recorded fromalpine tussock grassland near and on MtWilhelmina and Mt Carstensz, where it occurs atelevations of up to 4,500 m (Hope 1976). Nomammal has been recorded at higher elevation inNew Guinea, although Mallomys glmung has beenrecorded at 4,430 m.

Limited trapping which 1 undertook at 3,950 melevation in the Meren Valley, revealed threespecies of Stenomys (niobe, omlichodes andrichardsoni) in approximately equal abundance,although their preferred habitats differed(unpublished observations). Stenomys richardsoniwas most commonly encountered in tussockgrassland growing in the valley floor; S. omlichodesfavoured the short scrub clinging to the rocky wallsof the valley; and S. niobe was associated with alarge, loose boulder slope and in particular withhuman camp sites in this area. At lower elevationsS. omlichodes is associated with heath and boggyscrub (down to 2,900 m), and S. niobe is common inforest.

The high relative abundance of S. richardsoni inthe Kelangurr Cave deposit provides furtherevidence that tussock grassland was an importantcomponent of the vegetation of the area at the timeof the accumulation.

The absence of forest-dwelling species from thefossil deposit is unlikely to have resulted from thefeeding preferences of the Sooty Owls, as they arewell-known to take a wide variety of forest­dwelling mammal species, both arboreal andterrestrial, in Australia. The present foraging areasof Sooty Owls in Australia vary from about 2 km 2 to8 km 2• On this basis, it seems likely that, at the time

Hope (1976)

ZnglosSlls bnlljllli (as bruijni)Satanelllls albopllnctatllsPeroryctes longicalldaPhalanger sp.Dendrolaglls dorianllsThylogale bnmii (as brltljni)Thylogale sp. novPselldocheinls cllprellsPselldocheirus mayeriMallomys rothschildiHyomys go!iathCanis fami!iaris

This paperZnglOSSllS brui;nii

Dasyllrus alboplmctatlls 1

Microperoryctes longicalldl1 1

Phalangcr scricells 2

Dendrolaglls Sp.2

71lylogale Sp2

71lylogale christenseni 1

Psclldochirops ClIprells 1

Psclldochintllls mayeri 1

Mallomys gllllllllg 1

Hyomys dammermani 1

Canis familiaris

350

freshly-killed individuals at about 4,200 m elevationin the Prinz Wilhelm V Range, unpublishedobservations) and Zaglossus bruijnii, which has beenrecorded at 4,150 m (Harrer 1964). The diversity offossil mammalian taxa recorded from KelangurrCave is considerably higher than that found above4,200 m today.

Overall, the closest match among contemporaryenvironments to that represented by the fossil faunain the Chamber 1 deposits from Kelangurr Cave, isthat occurring between 4,000 and 4,200 m elevationon the Maokop. At these elevations, valley floorssupport tussock grassland, while rocky slopessupport dense alpine scrub which grows to 1-2 m inheight. These habitats support a moderately diversemammal fauna. This interpretation is consistentwith other palaeoenvironmental data, for thetreeline in New Guinea dropped to as low as 2,100m elevation during the last glacial maximum(Hope and Hope 1976). Today tall upper montaneforest surrounds the site and the treeline lies atabout 3,900 m elevation on the mountains to thesouth.

ACKNOWLEDGEMENTS

I am indebted to Alexandra Szalay for literallygiving the shirt off her back to aid in the collectionof the sample analysed here. Mr Edward Seeto andMr Rui Lopes sorted and prepared the sample foranalysis. Mr Lopes also attempted the acidpreparation of the matrix sample, and carried outpreliminary taxonomic classifications. Dr GeoffHope kindly made available Figures 1 and 2 foruse in this publication.

REFERENCES

Aplin, K.P. and Archer, M. (1987). Recent advances inmarsupial systematics with a new syncreticclassification. In Archer, M. (ed.), Possums andopossums: studies in evolution: xv-lxxii, Surrey Beatty& Sons Pty Ltd, Chipping Norton, NSW.

Baird RF. (1991). The taphonomy of late Quaternarycave localities yielding vertebrate remains inAustralia. In Vickers-Rich, P., Monaghan, J.M., Baird,RF. and Rich, T.H. (eds), Vertebrate palaeontology ofAustralasia: 267-309, Pioneer Design Studios,Melbourne, Victoria.

Beehler, B., Pratt, T.K. and Zimmerman, D.A. (1986).Birds of New Guinea. Handbook No. 9, Wau EcologyInstitute: 1-293, Princeton University Press,Princeton, New Jersey, USA

T.F. Flannery

Debus, S.J.5. (1994). The Sooty Owl Tyto tenebricosa inNew South Wales. Australian Birds 28: 4-19.

Flannery, T.F. (l992a). Taxonomic revision of theThylogale bnmii complex (Macropodidae: Marsupialia)in Melanesia, with description of a new species.Australian Mammalogy 15: 7-23.

Flannery, T.F. (l992b). New Pleistocene marsupials(Macropodidae, Diprotodontidae) from subalpinehabitats in Irian Jaya, Indonesia. Alcheringa 16: 321-331.

Flannery, T.F. (1994). The fossil land mammal record ofNew Guinea: a review. Science in New Guinea 20: 39-48.

Flannery, T.F. (1995). Mammals of New Guinea, revisedand unpdated edition: 1-568, Reed Books, Sydney,NSW.

Flannery, T.F., Aplin, K.P., Groves, CP. and Adams, M.(1989). Revision of the New Guinea murid genusMallomys (Muridae: Rodentia), with descriptions oftwo new species from subalpine habitats. Records ofthe Australian Museum 41: 83-105.

FIannery, T.F., Boeadi and Szalay, AL. (1995). A newtree-kangaroo (Dendrolagus: Marsupialia) from IrianJaya, Indonesia, with notes on ethnography and theevolution of tree-kangaroos. Mammalia 59: 65-84.

Harrer, H. (1964). I come from the Stone Age, Rupert Hart­Davis, London, U.K.

Hope, G.5., Flannery, T.F. and Boeardi (1993). Apreliminary report of changing Quaternary mammalfaunas in subalpine New Guinea. Quaternary Research40: 117-126.

Hope, G.5. and Hope, J.H. (1976). PaIaeoenvironmentsfor man in New Guinea. In Kirk, RL. and Thorne,AG. (eds), The biological origins of the Australians: 29­53, Australian Institute of Aboriginal Studies,Canberra, ACT.

Hope, J.H. (1976). Fauna. In Hope, G.5., Peterson, J.A,Radok, U. and Allison, I. (eds), The equatorial glaciersof New Guinea. Results of the 1971-1973 AustralianUniversities' Expeditions to Irian Jaya: survey, glaciology,meteorology, biology and palaeoenvironments: 207-224,AA Balkema, Rotterdam, The Netherlands.

Hope, J.H. (1981). A new species of Thylogale(Marsupialia: Macropodidae) from MapaIa rockshelter, Jaya (Carstensz) Mountains, Irian Jaya(Western New Guinea), Indonesia. Records of theAustralian Museum 33: 369-387.

Menzies, J.I. (1996). A systematic revision of Melomys(Rodentia: Muridae) of New Guinea. AustralianJournal of Zoology 44: 367-427.

Schodde R and Mason, I.J. (1980). Nocturnal birds ofAustralia, (illustrated by J. Boot): 1-136, LansdownePress, Melbourne, Victoria.

Walton, D.W. (ed.) (1988). Zoological catalogue of Australia.5: Mammalia: i-x, 1-274, Australian GovernmentPublishing Service, Canberra, ACT.

Manuscript received 9 March 1998; accepted 19 August 1998.