lable at ScienceDirect

Quaternary Science Reviews 30 (2011) 2676e2689

Contents lists avai

Quaternary Science Reviews

journal homepage: www.elsevier .com/locate/quascirev

Late Pleistocene glaciation of the Mt Giluwe volcano, Papua New Guinea

Timothy T. Barrows a,*, Geoffrey S. Hope b, Michael L. Prentice c, L. Keith Fifield d, Stephen G. Tims d

aGeography, College of Life and Environmental Sciences, University of Exeter, Exeter, Devon, EX4 4RJ, United KingdombDepartment of Archaeology and Natural History, College of Asia and the Pacific,, The Australian National University, Canberra, ACT 0200, Australiac Indiana Geological Survey and Department of Geological Sciences, Indiana University, 611 North Walnut Grove Avenue, Bloomington, IN 47405-2208, USAdDepartment of Nuclear Physics, Research School of Physics and Engineering, The Australian National University, Canberra, ACT 0200, Australia

a r t i c l e i n f o

Article history:Received 19 July 2010Received in revised form10 May 2011Accepted 31 May 2011Available online 18 July 2011

Keywords:Papua New GuineaPacific OceanExposure datingGlaciation

* Corresponding author. Tel.: þ44 1392 722494.E-mail address: T.Barrows@exeter.ac.uk (T.T. Barro

0277-3791/$ e see front matter � 2011 Elsevier Ltd.doi:10.1016/j.quascirev.2011.05.022

a b s t r a c t

The Mt Giluwe shield volcano was the largest area glaciated in Papua New Guinea during the Pleistocene.Despite minimal cooling of the sea surface during the last glacial maximum, glaciers reached elevationsas low as 3200 m. To investigate changes in the extent of ice through time we have re-mapped evidencefor glaciation on the southwest flank of Mt Giluwe. We find that an ice cap has formed on the flanks ofthe mountain on at least three, and probably four, separate occasions. To constrain the ages of theseglaciations we present 39 new cosmogenic 36Cl exposure ages complemented by new radiocarbon dates.Direct dating of the moraines identifies that the maximum extent of glaciation on the mountain was notduring the last glacial maximum as previously thought. In conjunction with existing potassium/argonand radiocarbon dating, we recognise four distinct glacial periods between 293e306 ka (Gogon Glaci-ation), 136e158 ka (Mengane Glaciation), centred at 62 ka (Komia Glaciation) and from >20.3e11.5 ka(Tongo Glaciation). The temperature difference relative to the present during the Tongo Glaciation islikely to be of the order of at least 5 �C which is a minimum difference for the previous glaciations.During the Tongo Glaciation, ice was briefly at its maximum for less than 1000 years, but stayed nearmaximum levels for nearly 4000 years, until about 15.4 ka. Over the next 4000 years there was morerapid retreat with ice free conditions by the early Holocene.

� 2011 Elsevier Ltd. All rights reserved.

1. Introduction

The tropics receive the greatest energy per unit area from thesun anywhere on Earth. Heat received in the tropics is distributedby wind, surface ocean currents and by latent heat to the rest of theplanet, controlling the distribution of climate. The mountains ofNew Guinea lie in the western Pacific Ocean tropics near the centreof the Indo-Pacific warm pool. This region is the largest body ofwarm ocean on the planet and spans almost half way around theglobe. Changes in the temperature of this water body not onlystrongly influence the climate of New Guinea but have hemisphericclimatic consequences.

Knowledge of the evolution of the warm pool through thePleistocene has grown rapidly over the last 30 years. The highlyinfluential CLIMAP project (Climap, 1976, CLIMAP Project Members,1981; Thompson, 1981) documented remarkably little change intropical sea surface temperature (SST) during the last glacialmaximum (LGM), an event when the lowest temperatures in thelate Pleistocene would be expected. Since this time, several new

ws).

All rights reserved.

techniques have been developed to estimate SST in deep sea sedi-ments, and a large number of new cores with good chronologiescovering the LGM have become available. Despite this, recentrevisions of SST during the LGM indicate that the CLIMAP conclu-sion of little change remains robust (MARGO Project Members,2009). Estimates of cooling in SST from three independentmethods around the perimeter of New Guinea in the WesternPacific Ocean range from 0 to 3 �C (Rosell-Melé et al., 2004; Barrowsand Juggins, 2005; Barker et al., 2005). This minimal coolingstrongly contrasts with the observation that the snowline high inthe mountains of New Guinea was as much as 1100 m lower at thesame time, requiring a temperature difference of 5e6 �C (Löffler,1972). This difference in surface cooling has been perceived as anapparent ‘paradox’ and has fuelled a long standing debateattempting to reconcile the two sets of estimates (Webster andStreten, 1978; Rind and Peteet, 1985; Peterson et al., 2002). Noconsensus has emerged as to an explanation that satisfactorilyexplains this phenomenon. Some steepening of tropical lapse ratesprobably occurred but this does not fully resolve the issue (e.g.,Blard et al., 2007).

Despite significant progress in refining SST estimates, lessprogress has been made in quantifying temperature changes in

T.T. Barrows et al. / Quaternary Science Reviews 30 (2011) 2676e2689 2677

tropical mountains. Glaciers that existed in the mountains of NewGuinea are ideally placed to record temperature changes in thewestern Pacific Ocean through time. However, the maximumlowering of the snowline is not directly dated, and hence it ispossible that cooling estimates may have been incorrectly attrib-uted to the LGM. In recent reviews, the lack of chronological controlwas highlighted (Mark et al., 2005). The island of New Guinearepresents an ideal area to reassess the tropical ‘paradox’, bothbecause it records extensive glaciation and because thesurrounding Indo-Pacific warm pool is central to the debate on themagnitude of maximum cooling in the tropics. Most moraines onNew Guinea are assigned to the LGM on the basis of limitingradiocarbon dates in only a few locations (Prentice et al., 2005), andbecause of a lack of evidence for more than one glaciation (Löffler,1972). Therefore the outermost are frequently assigned an LGM age.Clearly, the correct moraine sequences need to be identified beforetemperature estimates from snowline changes can be reliablyattributed to the LGM.

In this paper we present a new study of the moraine systems onthe southwest side of Mt Giluwe on the island of New Guinea. Wepresent maps of the glacial geology and review previous glacialstratigraphies. To constrain the age of glaciation we present expo-sure ages on boulders from representative moraine sequences.Together with radiocarbon dates from basal sediments in associ-ated basins, we construct the first glacial chronostratigraphy fora mountain in New Guinea. These ages are used as the basis fora regional review of the timing of glaciation in Southern Hemi-sphere sector of the western Pacific Ocean. Lastly, based on theheights of dated moraines, we review previous estimates oftemperature change attributed to changes in the equilibrium line-altitude (ELA).

2. Glacial geology

2.1. Background

The mountains on the island of New Guinea were the mostextensively glaciated area in the Asian tropics during the latePleistocene (Fig.1). On thewestern side of the island over 3400 km2

Fig. 1. Regional map of New Guinea showing the location of Mt

was glaciated, mostly in the form of an almost continuous ice fieldfromMt Jaya and Mt Idenburg to the Star Mountains Prentice et al.,(2011). On the eastern side of New Guinea, there is evidence forglacial activity on at least 20 mountains. The ice cap on Mt Giluwewas the most extensive area of glaciation (170e188 km2) followedby Mt Wilhelm (w107 km2) (Löffler, 1972; Bik, 1972). Löffler (1972)placed the LGM snowline at an altitude of 3500e3550 m in centralPapua New Guinea, rising to between 3600 and 3700 m in theSaruwaged and Owen Stanley Ranges. Today, there is no activeglaciation on the eastern side of the island and only three of thehighest peaks were recently glaciated on the western side of theisland (Mt Jaya, Mt Idenburg, Mt Trikora, and Mt Mandala). Inthe early 1970’s, the ELA lay near w4600 m (Löffler, 1972).

Mt Giluwe is an extinct shield volcano and the second highestpeak (4368 m) in Papua New Guinea, rising above the westernhighlands surface at 1500e1700 m (Fig. 2). The mountain is deeplyincised by radial drainage, and only the lower half still bears originalvolcanic geomorphology (Blake and Löffler, 1971). The upper half isextensively modified by glaciation (Fig. 2a). Löffler (1972) and Blakeand Löffler (1971) concluded that Mt Giluwe had been glaciated atleast twice, with an intervening interglacial period. The first eventwas associated with volcanic activity on the east of the mountain.The second event was responsible for the prominent morainesaround themountain. This most recent event (the ‘main’ glaciation)probably consisted of two stages separated by an unknown amountof time, and deposited five groups of recessional moraines (Blakeand Löffler, 1971; Löffler, 1972). Bik (1972) also observed twostages of glaciation based on the positions of the terminalmoraines,but concluded that there must be at least a stadial between theglaciations. Bik (1972) also suggested that there were 5 maingroupings of moraines on the southwest side. In contrast, earlierwork by Verstappen (1952) and Reiner (1960) concluded that therewas evidence for only one glaciation in New Guinea.

2.2. Previous dating

There are few limiting dates from New Guinea for the timing ofmaximum glaciation. On Mt Giluwe, the basal radiocarbon datesfrom peat behind terminal moraines provide minimum ages

Giluwe and deep sea cores ODP Hole 806B and MD97-2140.

Fig. 2. a) Shaded relief map of Mt Giluwe, b) Topographic map of Mt Giluwe showing sites of radiocarbon dating (sites 1e6 in Table 1) and dated palagonite (‘v’ symbols). The digitalterrain model used is the SRTM 90 m digital elevation data. Solid triangles indicate the east and west peaks of Mt Giluwe. Inset box indicates the location of Fig. 3.

T.T. Barrows et al. / Quaternary Science Reviews 30 (2011) 2676e26892678

(Table 1). Radiocarbon dates are calibrated using Calib 6.0 (Stuiverand Reimer, 1993) utilising the INTCAL09 calibration curve (Reimeret al., 2009). Uncalibrated radiocarbon dates are expressed inradiocarbon years (14C yr BP) and calibrated radiocarbon dates areexpressed as calibrated years (cal yr BP). Exposure ages are given inthousands of years as ‘ka’. Where referred to, oxygen isotopechronozones (OIC ¼ the informal ‘marine isotope stage’) are basedon the LR4 chronostratigraphy of Lisiecki and Raymo (2005).

Löffler (1972) limited the ages of the second youngest andyoungest of his five recessional groupings to at least 7890 þ 120/�170 cal yr BP and 3840 þ 130/�140 cal yr BP years based on dateson peat samples from pits in gullies incised through depositsbehind moraines (Fig. 1, Table 1). Radiocarbon dates limit theoutermost moraines to greater than 4170 þ 240/�180 cal yr BPyears old (Blake and Löffler,1971). On the basis of these radiocarbondates, Löffler (1972) concluded that Mt Giluwe only became ice free

Table 1Radiocarbon dates from Mt Giluwe.

Location Altitude (m) Site Sample ID Depth (cm) Lab. ID d13C Date (14C yr BP) Cal. Age (cal yr BP) 68% CIa Source

Alkena peat section 2200 1 240 Unknown 23,600 �. 1100 28,270 þ 1240/�1300 1SW flank, alluvial terrace 2 0e0.5 GaK-2523 1810 � 150 1740 þ 160/�190 2SW flank, alluvial terrace 2 60 GaK-2524 3780 � 140 4170 þ 240/�180 2SW flank, moraine ‘4’ 3550 3 GaK-2697 6480 � 130 7380 þ 120/�110 1SW flank, moraine ‘4’ 3550 3 GaK-2699 7070 � 150 7890 þ 120/�170 1SW flank, moraine ‘50 3670 4 GaK-2698 3550 � 100 3840 þ 130/�140 1Summit area fen 4170 5 GWS2 89e100 ANUe1338 3370 � 160 3630 þ 200/�180 3Summit area fen 4170 5 GWS1 275e285 ANUe1337 9980 � 280 11,570 þ 490/�400 3Nth Plateau 3680 6 GNP1 Fines 263e293 ANUe1342A 11,250 � 550 13,120 þ 670/�560 3SW flank, Tongo Valley 3500 7 GWA1 395e400 ANUe1339 13,050 � 700 15,620 þ 1170/�980 3SW flank, Tongo Valley 3590 8 GWB1 178e200 ANUe1340 8790 � 240 9870 þ 280/�310 3SW flank, Tongo Valley 3700 9 GWC1 270e293 ANUe1341 9850 � 240 11,340 þ 420/�540 3SW flank, Mengane Valley 3580 10 GMR160 155e165 OS-54747 �25.72 17,100 � 100 20,300 þ 130/�140 4SW flank, Camp Bog 3550 11 GCB240 240 OS-54747 �22.45 13,050 � 65 15,760 þ 320/�500 4

a Uncertainty is a 68% confidence interval around the median probability.Sources: (1) Löffler (1972), (2) Blake and Löffler (1971), (3) Hope and Peterson (1975), (4) Thispaper.

T.T. Barrows et al. / Quaternary Science Reviews 30 (2011) 2676e2689 2679

in the late Holocene. However, this dated sequence most probablyreflects erosion following human disturbance of the forest cover.

On the northeast side of the mountain in the Kaugel Plainwherethe glaciated catchments of the Gorgon and Tamal rivers drain themountain (Fig. 2), a 40 m section through the flood plain terracereveals stratified sediments intercalated with thick beds of peat(Löffler, 1972). The uppermost layer is underlain by peat dated at28,270 þ 1240/�1300 cal yr BP and deeper peat layers are beyondthe radiocarbon limit. Consequently, Löffler (1972) placed the finalstage of the last glaciation at 28,000 cal yr BP.

Subsequent radiocarbon dating casts doubt on how closely theabove dates limit the glaciation (Table 1; Figs. 2e3). A date of15,620 þ 1170/�980 cal yr BP was obtained on basal clayey gyttjain a small Carpha alpina quaking fen nestled between moraines inthe Tongo valley labelled ‘group 30 by Löffler (1972), confirmingthe pre-Holocene age of these moraines (Hope and Peterson,1975). Further up the valley, a further date of 11,340 þ 420/�540 cal yr BP was obtained on basal gyttja from a small basin in‘group 4’ moraines (Hope and Peterson, 1975). This is a similarstratigraphic position to the 7890 cal yr BP age of Löffler (1972).Finally, an age of 11,570 þ 490/�400 cal (ANU 1337) cal yr BP onbasal pond muds overlying sands was obtained within a smallcirque basin at 4170 m about 300 m southeast of the highest peak(site 5 in Fig. 2). The above ages strongly contrast with the inter-pretation of Löffler (1972) and suggest ice-free conditions on themountain by the beginning of the Holocene. The ages agree withthe chronology of deglaciation from nearby Mt Wilhelm (Hopeand Peterson, 1975).

At several locations there is evidence for much older glacialactivity closely timed with the volcanic activity of the mountain. Anundated cumulo dome on the northern part of the mountain isglacially eroded on its sides, but not its top (>3800 m) (Blake andLöffler, 1971). These authors suggested that it was extruded latein the last glaciation, and therefore represents the last knownvolcanic activity on the mountain. On the eastern side of thesummit area in the upper Tamal valley, at 3650 m, 2 m of rubble(inferred to be till) and 5 m of stratified sand and silt are inter-bedded between two subaerial lava flows, in turn overlain by tillfrom the most recent glaciation (Blake and Löffler, 1971; Löffleret al., 1980). The K/Ar ages on the upper and lower lava flows arestatistically indistinguishable (G31, 319 � 20 ka; G38, 292 � 20 ka)and have a mean age of 306 � 14 ka. The presence of palagonitebreccia on the north-eastern side of the Giluwe in the upper Gogonvalley indicates that volcanic activity on the mountain occurredunder a cover of ice, probably 100 m thick (Blake and Löffler, 1971).Palagonite results from the weathering of volcanic glass that must

have chilled rapidly in contact with ice or water. The breccia isdated at 293 � 28 ka (G9, 301 � 40 ka, 285 � 40 ka) (Löffler et al.,1980), within the error of the age of the inter-bedded till. We referhere to this early evidence of ice cover as the Gogon Glaciation, andinfer it occurred between 293 and 306 ka. The Gogon glaciationwasmore extensive than the last glaciation because the interbedded tillsite lies outside its limits.

Palagonite is also found on the east peak of Mt Giluwe (Blakeand Löffler, 1971). Here the volcanic activity is much older andthe eruptions took place under only thin ice. Lava near the pala-gonite dates to 701 � 41 ka (G42; 652 � 60 ka, 743 � 55 ka). Pal-agonitized lava from the main peak dates to 810 � 53 ka (G23;753� 70 ka, 885� 80 ka). These are some of the oldest ages for lavafrom Mt Giluwe. Only one other site near the main peak is older,dating to 1.21e1.28 Ma (Löffler et al., 1980) which signifies the firsttime themountain reached its present altitude. The younger ages at810 ka may therefore represent the first recorded time the regionalsnowline intersected the mountain and provide circumstantialevidence for ice cover at this time.

2.3. Glacial stratigraphy

In Fig. 3 we present a new map of the glacial features of thesouthwest quadrant of the mountain. Preservation of glacialfeatures is generally excellent, with virtually no modification of thedepositional form and even minor moraine ridges are preserved.Only the crests of the outermost moraines show rounding (Bik,1972). Bik (1972) attributed the preservation to the presence ofthick grass cover that protects the slopes from erosion. Weobserved that glacial till under boulders was compact and plasticdue to the high clay content, and therefore resistant to erosion fromsurface runoff.

The style of glaciation is strongly controlled by topography(Fig. 3). The low angle upper slopes of the mountain bore ice capglaciation,which terminatedaseitheroutlet glaciers in thevalleysoras a broad shield on the flanks of the mountain. On the southwestsideof themountain, iceoriginated fromtwomajorcentresoneitherside of an arêtewith headwalls up tow3800m. The northern cirque(N) supplied ice mainly into the Tongo River valley. The southerncirque (S) formed an ice cap over most of the southwest flank of themountain. The cirques are not over-deepened indicating that the icecoverwas not strongly erosive. The number and spacing ofmorainespreserved is a function of the thickness of the ice and the rate ofretreat. Where the ice cap spread out radially on the southwestand the ice was thinnest, small changes in thickness necessitatedconsiderable retreat of the ice front (Bik, 1972). Consequently, this

3400

3300m 3200m 3500m

3600m

3700m

3800m

3200m

3300m

3200m

3400

m

3500

m

3100

m

3600 3700m

3800m

6.06°S

Kettleholes

Kettleholes

Fluted till

Fluted till and kettleholes

Fluted till

Fluted till

S

N

lava flows

a

col

lava flows

9

10

12

13

Mengane River

Mengane River

Menken R

iver

Menke

n R

iver

Yamboro River

Yamboro River

Kerenda River

Kerenda River

Kom

ia R

iver

Kom

ia R

iver

Tongo R

iver

b

11

0 500 1000m 250

Tongo South Glaciation major/minor moraines

Tongo North Glaciation major/minor moraines

Komia Glaciation

Mengane Glaciation major/minor moraines

Watercourse

Kettlehole Sample site (Table 1)

Contours Lava flows

Arete

143.82°E

Fig. 3. Glacial geology of the southwest flank of Mt Giluwe. The map is based on unrectified aerial photography due to poor local height control. Contours are estimated from RoyalAustralian Survey Corps, Sheet 7685, Series T683, 1974. Inset boxes indicate locations of Fig. 4a and b. Numbers refer to sites of radiocarbon dating (sites 7e11 in Table 1).

T.T. Barrows et al. / Quaternary Science Reviews 30 (2011) 2676e26892680

area has the largest number of moraines. Conversely, the fewestmoraines are locatedwhere the icewas channelled into valleys, suchas the TongoValley in the north, and ice reached its lowest elevationat these sites. Post-glacial fluvial incision on the valley floor isminimal and in most cases has only dissectedmoraines and has notdeeply incised into the bedrock.

There are at least three distinct sequences of moraines markingindividual glaciations. The moraines are mainly separated on thebasis of stratigraphic cross-cutting relationships. The best stratig-raphy on this part of the mountain occurs on the west side of the

Tongo River and in the Mengane River catchment (Fig. 4a).Successive glaciations and fluvial incision have lowered the baselevel of the Tongo River valley floor, channelling more ice into thevalley and preventing it from flowing over the col into theMenganecatchment (Fig. 3). Consequently, the ice from the northern cirquewas progressively more confined to the Tongo valley through time,and older moraines, particularly thewestern lateral moraines, werepreserved instead of being over-ridden by subsequent ice cover.

The oldest moraines recognised to date occur in the MenganeRiver valley and we assign them to the Mengane Glaciation

Tongo R

iver

Kom

ia R

ive

r

3500m3600m3400m

x x

xxx xx

x xx

x xx

x xx x

xxx

xx

x

xx

xx

xGLW-02 GLW-01

GLW-13

GLW-04GLW-12

GLW-05

GLW-17

GLW-16

GLW-14GLW-15GLW-11

Col

0 250m

GLW-18

GLW-06

GLW-03Watercourse

Kettlehole

Komia Glaciation

Mengane Glaciationmajor/minor moraines

Tongo South Glaciationmajor/minor moraines

Tongo North Glaciationmajor/minor moraines

Sample site Contours

X

143.82E

6.04S

a

3200m

3300m 3400m

3500mMenken R

ive

r

Mengane River

xGLW-07

x

x

xx

xx

x

xx x x

x

xx

x x x

xx

x

x xxxx

x

x

GLW-08

GLW-09

GLW-33GLW-10

GLW-19 GLW-32

GLW-31

GLW-23/GIL-01

GLW-22GLW-21

GLW-20/GIL-02 GIL-03

GIL-04

GIL-12

GIL-06

GIL-11GIL-07

GIL-08

GIL-09

GIL-10

0 250m

143.82°E

6.06.06°S°S6.06°S

b

Bren Tarn

Fig. 4. a) Glacial geology of the Tongo River valley. b) Glacial geology of the Menken and Mengane River valleys.

T.T. Barrows et al. / Quaternary Science Reviews 30 (2011) 2676e2689 2681

(Fig. 4b). This glaciation was the most extensive and terminalmoraines occur as low as 3200 m. During this phase, glaciers fromthe northern and southern cirques coalesced through the MenganeRiver col to form an ice cap that reached almost as far down slope asthe confluence of theMengane andMenken Rivers. The presence ofrillen karren on the sides of basalt boulders and subaerial weath-ering rinds up to 1 cm thick indicate a considerable age for thisglaciation. Moraine crests are rounded, and rivers have cut into

bedrock in the valley floor upstream of the terminal moraines,particularly in the lower Menken River. Between the Yamboro andthe Mengane Rivers lies the oldest stratigraphic evidence for an icecap. These moraines seem to be at least in part conformable withthe terminal moraines in the Mengane valley, but the outermostmoraines may pre-date them.

The second youngest moraines lie between the Menganemoraines and the youngest glaciation in the upper Komia River

T.T. Barrows et al. / Quaternary Science Reviews 30 (2011) 2676e26892682

valley, as a western lateral moraine of the Tongo River (Fig. 4a). Weassign thesemoraines to the Komia Glaciation. Stratigraphically, wealso place the moraines between the Mengane moraines and theyoungest glaciation in the Menken river valley into this glaciation,but these are not contiguous with the moraines adjacent to theTongo River valley, making this assumption difficult to assess. It isalso difficult to separate moraines from the younger glaciationbetween the Mengane River and the Yamboro River on the basis ofpreservation alone.

The most recent glaciation, the Tongo glaciation, has verydetailed moraine stratigraphy preserved over much of the moun-tain, particularly in the Tongo River valley and on the southwestflank of the mountain where there are at least 52 moraines. TheTongo Glaciation was less extensive than the preceding glaciations.Moraines occur down to 3400e3300m in the southwest, extendingabout 2.5 km from the back wall of the cirque.

The lateral moraines of the Tongo Valley (Fig. 4a) are the largestdepositional landforms on this side of the mountain. From theheight difference between the valley floor and the top of the lateralmoraine, ice wasw100 m thick in the Tongo Valley (¼Komia Valleyof Bik (1972)) depositing a series of large lateral moraines thatextend down to w3100 m and 4.1 km from the headwall. Löffler(1972) attributed the lateral moraines of the Tongo River to thelast glaciation, but thought that it consisted of two stages. Thewestern lateral lies at the junction of the Tongo valley glacier anda lobe of ice from the northern cirque (Fig. 3).

Fivemoraines abut the largewestern lateralmoraine of the Tongovalley (Fig. 4a). Traced to the south these moraines progressivelysplit into as many as 20 narrower, lower moraines where the glacierwas thinner. The number of retreat moraines is mostly a function oftopography. Both Löffler (1972) and Bik (1972) placed the morainesin this glaciation into 5 groups or ‘bundles’ of moraines. Thisgrouping may be climatic, or a product of the topography of thesouthwest side of the mountain because it breaks down laterally tothe north and south of this area. At maximum ice extent, thegradient of the ice cap was lowest in the south. In the region of BrenTarn (Fig. 4b), themean gradient to the top of the backwall was of theorder of 9�. Bik (1972) noted that the distal slope of the terminalmoraines was 5e6�, probably indicating the slope near the ice front.

Ice was at least 100 m thick in each cirque, and split into twoindividual ice fields when it thinned to less thanw100m (Fig. 3). Asice retreated through the col at the top of the Mengane valley(Fig. 3), the Tongo glacier continued to occupy its lateral moraines.The Tongo glacier deposited a series of inset lateral moraines as itretreated and became an ice field, similar to that in the southerncirque. As ice retreated in the cirques it left multiple low moraines,numerous kettle holes and well-preserved subglacial topographysuch as fluted till. The spacing between the moraines is greatest inthe south where the bedrock slope was shallowest. From themoraine orientation the last ice persisted in a southwest facing cleftin the northern cirque and under the northernwall of the southerncirque (N and S on Fig. 3). There does not appear to have been anystrong aspect control on ice retreat.

As noted by Löffler (1972), the most recent ice cap draped tillover the older moraines without modifying them. An example ofthis is the northern side of the Menken River valley (marked bya bold line in (Fig. 4b)). This ridge is probably mostly bedrock, butis overlain by medial moraine deposited during the MenganeGlaciation when glaciers occupied both the Mengane and theMenken valleys. The ice during the Tongo Glaciation depositedlow terminal moraines over the ridge at right angles to themedial moraine. The till sheet is probably no more than 0.5e5 mthick (Löffler, 1972). Down valley from Bren Tarn, older terminalmoraines have been over-ridden and are still visible through thethin till sheet (Fig. 3).

3. Exposure dating

The exceptionally preserved moraines of Mt Giluwe providegood targets for exposure dating. We chose sampling sites to coverthe full range of glacial features from the outermost icecapmoraines to the retreat sequences. MP and GH collected the GILsamples in 2001 and TTB, MP and GH collected the GLW samplesin 2003. In addition, GH and MP collected peat cores in 2001 and2003 for radiocarbon dating. Because of poor regional elevationcontrol, we used the Shuttle Radar Topography Mission (SRTM)90 m digital elevation data for sample site heights in conjunctionwith GPS field measurements. Geographic coordinates for the2001 samples were estimated. We sampled only the largestboulders, but in some cases for the GIL samples only small boul-ders (<1 m in height) were available. Post-depositional weath-ering on the Tongo Glaciation boulders appears negligible.However, on some boulders on the Mengane Glaciation moraineswe observed significant surface weathering in the form of rillenkarren indicating potential removal of centimetres of rock andavoided those surfaces. Surface spalling from the boulders was notobserved because of the massive nature of the basaltic rock. Wefocussed sampling on moraine crests to reduce the risk ofsampling boulders exhumed by erosion. Snow cover as a potentialsource of shielding on Mt Giluwe is negligible, except duringglacial periods but this is difficult to quantify. The entire samplearea is virtually forest free today, and would have lain above thetreeline until w11,000 cal yr BP. A low subalpine forest probablyoccupied the well drained sites from 11,000- ca 3000 cal yr BP,possibly building up a layer of root mat and litter. The date ofwidespread clearance is currently unknown but the charcoal datesof Löffler (1972) indicate that clearing occurred soon after4000 cal yr BP. We assume any shielding had similar but negligibleeffects on all samples. Site data for the sampled boulders is pre-sented in Table 2.

The exposure ages on the thinmoraine belts are likely to providean accurate account of the retreat of ice. However, the larger lateralmoraines are thick and are sequentially constructed. Therefore ourexposure ages are inherently biased towards the most recent timeice stood at the moraine and provide a minimum age for the iceadvance. Because of the minimal erosive power of the ice on lowangle slopes, there is potential for survival of old surfaces. This,together with reworking of boulders from older moraines, acts toskew the age distribution towards older ages. Conversely on oldermoraines, there is a likely bias towards younger ages, becausethrough time erosion flattens the profile of moraines, exhumingpreviously buried boulders, and weathering lowers the surface ofboulders.

Because of the availability of basalt with high potassiumcontent, we chose 36Cl for exposure dating. Chlorine wasextracted from the whole rock because the fine-grained nature ofthe basalt made mineral separation impossible. The isotopicratios of 36Cl/Cl were measured by accelerator mass spectrometryon the 14UD accelerator at the Australian National University.Because of the very high production rates at high altitude, blankcorrections were negligible. The abundance of major targetelements in the dolerite samples was determined using X-rayfluorescence. The concentrations of trace elements with largeneutron capture cross sections (B, Gd, and Sm) and neutron-producing elements (U and Th) were measured by inductively-coupled plasma mass spectrometry. Chlorine content wasdetermined by isotope dilution. Chemical data are summarized inTable 3. Other major element and trace element geochemicaldata are listed in the Appendix which is also available from theWDC-A for Paleoclimatology, Boulder, USA or the AustralianQuaternary Data Archive.

Table 2Site data for exposure ages.

Sample Longitude (�E) Latitude (�S)a Altitude (m) Scaling factor(nucleons)

Scaling factor(muons)

Horizoncorrection

Thickness (cm) Thickness correctionb

Tongo Glaciation (Menken Valley)GIL-01 143.813 �6.0570 3444 5.802 2.497 1.0 1.3 0.9886/1.072/1.071GIL-02 143.818 �6.0554 3515 6.019 2.560 1.0 1.0 0.9912/1.056/1.056GIL-03 143.821 �6.0561 3534 6.077 2.577 1.0 0.9 0.9921/1.050/1.050GIL-04 143.820 �6.0576 3496 5.960 2.543 1.0 1.0 0.9912/1.055/1.056GIL-06 143.819 �6.0619 3519 6.044 2.567 1.0 1.4 0.9877/1.077/1.076GIL-07 143.820 �6.0615 3523 6.019 2.560 1.0 0.6 0.9947/1.033/1.035GIL-08 143.822 �6.0645 3515 6.019 2.560 1.0 0.9 0.9921/1.049/1.051GIL-09 143.822 �6.0645 3515 6.078 2.577 1.0 2.0 0.9826/1.108/1.104GIL-10 143.824 �6.0649 3534 6.166 2.603 1.0 1.0 0.9912/1.055/1.055GIL-11 143.826 �6.0645 3562 6.007 2.557 1.0 2.2 0.9808/1.118/1.112GIL-12 143.819 �6.0641 3511 5.982 2.550 1.0 1.0 0.9912/1.056/1.055GLW-20 143.8181 �6.0559 3520 6.034 2.565 1.0 2.0 0.9826/1.108/1.103GLW-21 143.8181 �6.0559 3520 6.034 2.565 1.0 0.9 0.9921/1.049/1.051GLW-22 143.8039 �6.0593 3266 5.283 2.344 1.0 1.9 0.9834/1.104/1.100GLW-23 143.8135 �6.0574 3459 5.847 2.511 1.0 1.4 0.9877/1.077/1.076

Tongo Glaciation (Tongo Valley)GLW-01 143.8178 �6.0436 3519 6.031 2.564 1.0 1.0 0.9912/1.055/1.055GLW-02A 143.8176 �6.0437 3518 6.028 2.563 1.0 3.9 0.9664/1.198/1.177GLW-02B 143.8176 �6.0437 3518 6.028 2.563 1.0 2.0 0.9826/1.107/1.103GLW-03 143.8237 �6.0438 3587 6.244 2.625 1.0 0.9 0.9921/1.050/1.050GLW-04 143.8238 �6.0439 3588 6.248 2.626 1.0 3.4 0.9706/1.176/1.160GLW-05 143.8229 �6.0429 3577 6.213 2.616 1.0 1.7 0.9851/1.092/1.091GLW-06 143.8202 �6.0398 3529 6.062 2.573 1.0 1.5 0.9869/1.082/1.080GLW-11 143.8188 �6.0432 3531 6.068 2.574 1.0 2.5 0.9783/1.133/1.125GLW-12 143.8224 �6.0435 3584 6.235 2.622 1.0 1.0 0.9912/1.055/1.056GLW-13 143.8218 �6.0446 3568 6.184 2.608 1.0 1.4 0.9877/1.077/1.076GLW-18 143.8177 �6.0365 3478 5.904 2.527 1.0 1.2 0.9895/1.066/1.066

Komia GlaciationGLW-14 143.8204 �6.0432 3556 6.146 2.597 1.0 2.0 0.9826/1.107/1.104GLW-15A 143.8198 �6.0430 3545 6.112 2.587 1.0 1.0 0.9912/1.055/1.057GLW-15B 143.8198 �6.0430 3545 6.112 2.587 1.0 1.2 0.9895/1.066/1.066GLW-16 143.8205 �6.0414 3542 6.102 2.584 1.0 1.2 0.9895/1.065/1.066

Mengane GlaciationGLW-07 143.8121 �6.0504 3451 5.823 2.503 1.0 2.4 0.9791/1.128/1.119GLW-08 143.8081 �6.0531 3402 5.676 2.461 1.0 1.8 0.9843/1.098/1.094GLW-09 143.8068 �6.0542 3372 5.587 2.435 1.0 1.2 0.9895/1.066/1.066GLW-10 143.8007 �6.0577 3211 5.129 2.299 1.0 1.1 0.9904/1.060/1.061GLW-17 143.8180 �6.0392 3503 5.981 2.549 1.0 1.0 0.9912/1.054/1.055GLW-19 143.7993 �6.0591 3190 5.071 2.281 1.0 2.0 0.9826/1.108/1.103GLW-31 143.8039 �6.0594 3266 5.283 2.344 1.0 1.2 0.9895/1.066/1.066GLW-32 143.8027 �6.0584 3236 5.198 2.319 1.0 1.5 0.9869/1.082/1.080GLW-33 143.8008 �6.0576 3213 5.135 2.300 1.0 2.0 0.9826/1.108/1.103

r ¼ 3.0 g cm�3; L ¼ 160 g cm.a Geomagnetic latitude.b Correction factors for spallation, thermal neutrons, and epithermal neutrons respectively.

T.T. Barrows et al. / Quaternary Science Reviews 30 (2011) 2676e2689 2683

Exposure ages are calculated as detailed in Barrows et al. (2002),with the modifications made by Barrows et al. (2004). We used theproduction rates of Stone et al. (1996a, 1996b, 1998), Evans (2001)and Masarik and Reedy (1995). We scaled the production ratesusing the procedures of Stone (2000) for consistency with regionaldata sets. We found excellent agreement between exposure agesand radiocarbon ages indicating that this scaling approach issatisfactory for high altitude at low latitude. For 36Cl production byneutron capture we followed the procedures of Liu et al. (1994),Phillips et al., (2001) and Stone et al. (1998).

Age calculations are presented as conventional exposure ages,calculated using geographic latitude and including 36Cl productionfrom titanium and iron. No corrections are made for geomagneticchanges through time due to uncertainties in scaling low latitudesites. Based on the exceptional hardness of the basalt and the lack ofevidence of weathering on the boulders we sampled, all ages arecalculated assuming that there has been no weathering since initialexposure. We cannot rule out at least some erosion on the boulderson the outermost moraines on the scale of centimetres. To bear in

mind the sensitivity of exposure ages to weathering, 1 cm ofweatheringwould lower the age of a boulder that has been exposedfor 150,000 years by less than 1%. All measurement errors,including production rate errors, are fully propagated and moraineages are weighted means. All discussion in the text refers to thecorrected ages and all ages and statistics are reported at one stan-dard deviation.

4. Results

Exposure ages calculated using the 36Cl data are presented inTable 4. The exposure ages are discussed below in order ofdecreasing age.

4.1. Mengane glaciation

All exposure ages on moraines from the Mengane Glaciation areolder than 100,000 years, and range from 109 to 192 ka. Of theseven ages on the outermost moraines, four ranging from 136 to

Table 3Chemical data.

Sample [CaO] (wt%) [K2O] (wt%) [TiO2] (wt%) [Fe2O3] (wt%) [Cl] (ppm) cross section(10�3 cm2 g�1)

Tongo Glaciation (Menken Valley)GIL-01 8.43 � 0.09 2.41 � 0.02 1.35 � 0.04 6.3 � 0.07 79.05 � 1.17 6.05 � 0.085GIL-02 7.72 � 0.09 2.42 � 0.02 1.58 � 0.05 8.75 � 0.1 47.97 � 0.895 6.50 � 0.074GIL-03 6.95 � 0.08 2.91 � 0.02 1.32 � 0.04 5.87 � 0.07 94.03 � 2.57 6.05 � 0.082GIL-04 8.38 � 0.09 2.74 � 0.02 2.02 � 0.06 9.93 � 0.11 20.66 � 0.267 7.11 � 0.088GIL-06 7.42 � 0.08 2.77 � 0.02 1.39 � 0.04 6.68 � 0.08 53.28 � 0.843 6.19 � 0.082GIL-07 8.12 � 0.09 2.76 � 0.02 1.9 � 0.06 9.69 � 0.11 88.91 � 1.14 7.11 � 0.088GIL-08 7.42 � 0.08 2.86 � 0.02 1.6 � 0.05 8.54 � 0.1 73.69 � 0.941 6.86 � 0.098GIL-09 7.18 � 0.08 2.88 � 0.02 1.37 � 0.04 6.36 � 0.07 148.9 � 2.2 6.31 � 0.100GIL-10 8.02 � 0.09 2.85 � 0.02 1.34 � 0.04 6.69 � 0.08 122.7 � 1.63 6.43 � 0.098GIL-11 6.81 � 0.08 2.95 � 0.02 1.33 � 0.04 5.67 � 0.06 174.9 � 2.62 6.20 � 0.099GIL-12 7.52 � 0.08 2.77 � 0.02 1.23 � 0.04 5.65 � 0.06 39.78 � 0.707 5.94 � 0.086GLW-20 8.71 � 0.1 2.65 � 0.02 1.88 � 0.06 10.49 � 0.12 22.8 � 0.324 7.05 � 0.077GLW-21 7.09 � 0.08 3.29 � 0.03 2.00 � 0.06 8.61 � 0.1 49.61 � 0.612 6.98 � 0.094GLW-22 9.05 � 0.1 3.13 � 0.03 1.56 � 0.05 7.58 � 0.09 10.9 � 0.154 6.13 � 0.059GLW-23 11.08 � 0.12 3.12 � 0.03 1.13 � 0.04 6.54 � 0.07 26.98 � 0.498 6.13 � 0.081

Tongo Glaciation (Tongo Valley)GLW-01 8.26 � 0.09 2.93 � 0.02 1.3 � 0.04 6.6 � 0.08 29.01 � 0.37 6.27 � 0.086GLW-02A 8.1 � 0.09 3.53 � 0.03 1.49 � 0.05 4.94 � 0.06 28.68 � 0.69 6.19 � 0.080GLW-02B 8.29 � 0.09 3.05 � 0.03 1.32 � 0.04 6.31 � 0.07 27.27 � 0.379 6.23 � 0.080GLW-03 7.98 � 0.09 3.08 � 0.03 1.34 � 0.04 6.82 � 0.08 14.93 � 0.35 6.30 � 0.066GLW-04 9.03 � 0.1 2.55 � 0.02 1.27 � 0.04 6.56 � 0.08 51.44 � 0.698 6.30 � 0.094GLW-05 9.02 � 0.1 3.08 � 0.03 1.66 � 0.05 7.49 � 0.09 7.883 � 0.385 6.55 � 0.071GLW-06 8.28 � 0.09 3.51 � 0.03 1.28 � 0.04 4.55 � 0.05 44.21 � 0.576 6.01 � 0.082GLW-11 7.9 � 0.09 2.93 � 0.02 1.26 � 0.04 6.37 � 0.07 12.61 � 0.29 6.15 � 0.073GLW-12 8.17 � 0.09 2.9 � 0.02 1.41 � 0.04 7.14 � 0.08 12.0 � 0.169 6.47 � 0.089GLW-13 6.59 � 0.07 2.89 � 0.02 1.1 � 0.03 7.45 � 0.09 12.07 � 0.199 6.31 � 0.086GLW-18 7.8 � 0.09 3.11 � 0.03 1.37 � 0.04 7.06 � 0.08 10.22 � 1.46 6.36 � 0.079

Komia GlaciationGLW-14 8.25 � 0.09 3.01 � 0.02 1.69 � 0.05 8.1 � 0.09 7.071 � 0.187 6.60 � 0.072GLW-15A 8.33 � 0.09 3.01 � 0.02 2.06 � 0.07 10.07 � 0.12 9.898 � 0.264 7.21 � 0.088GLW-15B 7.77 � 0.09 2.86 � 0.02 1.47 � 0.05 8.6 � 0.1 14.84 � 0.362 6.52 � 0.066GLW-16 8.33 � 0.09 3.28 � 0.03 1.49 � 0.05 6.84 � 0.08 15.18 � 0.219 6.59 � 0.092

Mengane GlaciationGLW-07 8.24 � 0.09 3.47 � 0.03 0.84 � 0.03 3.53 � 0.04 18.27 � 0.231 5.57 � 0.091GLW-08 9.86 � 0.11 2.47 � 0.02 0.86 � 0.03 5.67 � 0.06 12.18 � 0.179 5.88 � 0.120GLW-09 8.51 � 0.09 3.27 � 0.03 1.07 � 0.03 6.31 � 0.07 16.18 � 0.211 6.14 � 0.096GLW-10 8.26 � 0.09 2.75 � 0.02 1.3 � 0.04 6.84 � 0.08 4.638 � 0.164 6.22 � 0.080GLW-17 8.59 � 0.1 3.24 � 0.03 1.48 � 0.05 7.08 � 0.08 14.4 � 0.185 6.41 � 0.084GLW-19 7.73 � 0.09 3.65 � 0.03 1.02 � 0.03 5.31 � 0.06 224.2 � 4.19 6.34 � 0.130GLW-31 8.64 � 0.1 3.09 � 0.03 1.54 � 0.05 7.9 � 0.09 67.98 � 1.02 6.57 � 0.083GLW-32 8.51 � 0.09 3.1 � 0.03 0.97 � 0.03 5.12 � 0.06 56.14 � 0.693 6.04 � 0.130GLW-33 8.47 � 0.09 2.92 � 0.02 0.91 � 0.03 5.18 � 0.06 107.4 � 1.4 5.95 � 0.120

T.T. Barrows et al. / Quaternary Science Reviews 30 (2011) 2676e26892684

158 ka have a mean age of 146 � 10 ka but with scatter beyondwhat would be expected due to random error alone (c2/n ¼ 2.1).Due to their great age, it is likely that the boulders on the Menganemoraines have not all experienced a simple exposure history. Thethree younger ages younger than the group of four are likely to havebeen partially shielded by till early in their exposure history, andthe remaining age (192.3 ka; GLW-31) probably represents a case ofprevious exposure. Finally, an age of 208 � 12.1 ka (GLW-17) camefrom a boulder on the outer western lateral moraine of the TongoRiver. Although only a single age and older than expected, itprobably indicates that this moraine belongs to the MenganeGlaciation.

4.2. Komia glaciation

The only locations where we dated the Komia Glaciation wereon a terminal moraine mostly overridden by the Tongo Glaciationand on an adjacent western lateral moraine (Fig. 4a). Two ages ona single boulder (GLW-15) on the terminal moraine give an averageof 64.2 � 11.8 ka. The boulder on the lateral moraine has anexposure age of 61.8 � 3.1 ka (GLW-16). The two boulders give

a weighted mean age of 61.9 � 3.0 ka (c2/n ¼ 0.2) for the KomiaGlaciation. An additional boulder from the leading edge of theterminal moraine, where it was deeply embedded, was dated at38.4 � 1.9 ka (GLW-14), and the top of the boulder was probablyshielded by till during part of its history.

4.3. Tongo glaciation

Most of our dating was concentrated on the youngest and best-preserved glaciation to define the age of the maximum ice extentand timing of retreat during the last glacial maximum. Our expo-sure ages agree well with the general timing of ice retreat from theradiocarbon dating of Hope and Peterson (1975). The outermostterminal moraine abutting the large western lateral moraine of theTongo valley is dated at 19.4 � 1.0 ka (GLW-11). This is the older ofa series of six moraines adjoining the Tongo left lateral moraine.The age agrees well with a new radiocarbon date 20,300 þ 130/�140 cal yr BP we obtained from the base of a bog formed behinda branch of a lateral moraine on the Mengane River (Table 1). Thethird oldest moraine, which eventually merges with the first twointo a single crest, has an average age of 18.3 � 1.4 ka (GLW-01, 02).

Table 4Exposure ages.

Sample Lab code [36Cl]c(�105 g�1) [36Cl]r(�103 g�1) S Production rate (atoms g�1 yr�1) Exposure age (ka)

Tongo Glaciation (Menken Valley)GIL-01 ANU-C076-13 6.96 � 0.26 7.06 � 0.22 53.90 � 1.63 13.1 � 0.6GIL-02 ANU-C076-12 5.52 � 0.19 3.90 � 0.13 48.83 � 1.39 11.5 � 0.5GIL-03 ANU-C076-15 8.76 � 0.39 7.20 � 0.28 59.92 � 1.97 14.9 � 0.8GIL-04 ANU-C136-19 6.18 � 0.20 1.58 � 0.05 47.82 � 1.39 13.1 � 0.6GIL-06 ANU-C076-16 18.7 � 0.65 4.84 � 0.15 51.91 � 1.56 37.6 � 1.8GIL-07 ANU-C136-21 8.75 � 0.31 7.30 � 0.22 59.45 � 1.77 15.0 � 0.7GIL-08 ANU-C136-20 17.8 � 0.42 7.80 � 0.23 55.94 � 1.69 33.1 � 1.3GIL-09 ANU-C076-17 8.58 � 0.35 13.7 � 0.44 70.99 � 2.60 12.3 � 0.7GIL-10 ANU-C136-22 10.2 � 0.34 13.3 � 0.40 67.64 � 2.25 15.4 � 0.7GIL-11 ANU-C076-18 7.48 � 0.39 16.9 � 0.57 75.29 � 2.95 10.1 � 0.7GIL-12 ANU-C076-19 4.76 � 0.24 3.27 � 0.11 48.83 � 1.4 9.9 � 0.6GLW-20 ANU-C136-23 5.35 � 0.14 1.79 � 0.05 48.62 � 1.38 11.2 � 0.4GLW-21 ANU-C136-24 7.07 � 0.20 4.07 � 0.12 54.92 � 1.70 13.1 � 0.6GLW-22 ANU-C130-25 5.61 � 0.22 2.67 � 0.08 43.80 � 1.37 13.0 � 0.7GLW-23 ANU-C130-20 7.59 � 0.30 2.17 � 0.07 55.80 � 1.62 13.8 � 0.7

Tongo Glaciation (Tongo Valley)GLW-01 ANU-C130-16 9.47 � 0.37 2.54 � 0.08 50.28 � 1.50 19.3 � 1.0GLW-02A ANU-C118-27 8.90 � 0.19 2.61 � 0.09 53.97 � 1.72 16.8 � 0.7GLW-02B ANU-C130-17 9.03 � 0.35 2.47 � 0.08 50.79 � 1.54 18.2 � 0.9GLW-03 ANU-C130-04 7.77 � 0.32 1.61 � 0.06 50.06 � 1.60 15.8 � 0.8GLW-04 ANU-C130-07 8.02 � 0.33 5.23 � 0.17 54.89 � 1.58 14.9 � 0.8GLW-05 ANU-C118-25 20.2 � 0.36 0.66 � 0.04 50.83 � 1.58 41.6 � 1.6GLW-06 ANU-C130-09 8.94 � 0.34 4.65 � 0.14 57.83 � 1.81 15.7 � 0.8GLW-11 ANU-C130-23 8.79 � 0.33 1.19 � 0.04 46.23 � 1.44 19.4 � 1.0GLW-12 ANU-C130-24 7.00 � 0.27 1.29 � 0.04 48.35 � 1.50 14.7 � 0.8GLW-13 ANU-C130-22 7.17 � 0.28 1.86 � 0.06 44.27 � 1.43 16.5 � 0.9GLW-18 ANU-C130-14 8.59 � 0.31 1.02 � 0.15 46.35 � 1.50 19.0 � 0.9

Komia GlaciationGLW-14 ANU-C130-15 17.61 � 0.61 0.69 � 0.03 47.94 � 1.51 38.4 � 1.9GLW-15A ANU-C130-21 25.82 � 0.91 0.88 � 0.03 49.34 � 1.52 55.8 � 2.8GLW-15B ANU-C130-26 31.47 � 1.09 1.38 � 0.05 47.14 � 1.44 72.5 � 3.6GLW-16 ANU-C130-10 29.45 � 1.03 1.79 � 0.06 51.28 � 1.63 61.6 � 3.1

Mengane GlaciationGLW-07 ANU-C129-31 53.7 � 1.75 4.27 � 0.13 49.80 � 1.63 124.0 � 6.6GLW-08 ANU-C118-26 41.9 � 0.44 2.91 � 0.10 43.46 � 1.26 108.9 � 3.8GLW-09 ANU-C129-29 56.2 � 1.88 4.07 � 0.13 47.32 � 1.50 138.8 � 7.5GLW-10 ANU-C118-28 49.9 � 0.50 4.33 � 0.02 37.66 � 1.18 158.1 � 6.3GLW-17 ANU-C130-11 78.9 � 2.56 1.35 � 0.04 47.78 � 1.51 208.0 � 12.1GLW-19 ANU-C129-30 9.09 � 2.73 43.1 � 0.15 77.7 � 3.31 136.3 � 8.4GLW-31 ANU-C136-25 81.7 � 1.58 6.48 � 0.18 52.56 � 1.57 192.3 � 8.6GLW-32 ANU-C130-03 51.3 � 1.75 11.9 � 0.39 49.23 � 1.53 119.1 � 6.3GLW-33 ANU-C129-28 71.6 � 2.50 21.4 � 0.70 56.25 � 1.94 150.7 � 8.9

Data are normalised to the GEC standard (36Cl/Cl ¼ 444 � 10�15). Carrier 36Cl/Cl ¼ 1 � 10�15. 36Cl decay constant 2.3 � 10�6 yr�1 [36Cl]c ¼ cosmogenic component;[36Cl]r ¼ background nucleogenic component.

T.T. Barrows et al. / Quaternary Science Reviews 30 (2011) 2676e2689 2685

The second youngest moraine has a single age of 16.5 � 0.9 ka(GLW-13). Together, these ages indicate that the ice front occupiedthe 5 moraines near its maximum extent for 3000 years. Thevolume of the moraines is similar, and assuming debris supply wasconstant, indicates ice occupied each limit for only w600 years.

Shortly after deposition of the last of the 5 moraines, iceretreated through the col in the upper Mengane Valley (Fig. 3). TheTongo left lateral appears to have been occupied continuously from19.0 � 0.9 (GLW-18) until 15.5 � 0.5 ka (GLW-03, 04, 06) (c2/n ¼ 0.38). A fifth age on the moraine is an outlier (41.6 � 1.6 ka;GLW-05). Further south, 15 ages on 13 boulders constrain thesubsequent retreat of the ice cap into the southern cirque. A newradiocarbon date (Fig. 3; site 13) on this flank indicates that icestayed near its maximum limits until shortly before 15,760 þ 320/�500 cal yr BP. The 13 boulders range in age from 15.4 to 9.9 ka andoverlap in age to some extent between adjacent moraines. Twoboulders (GIL-01/GLW-23; GIL-02/GLW-20) were sampled on theseparate field trips and the exposure ages agree within error(13.1 � 0.6/13.8 � 0.7 ka; 11.5 � 0.5/11.2 � 0.4 ka respectively).Some boulders appear to record cases of small amounts of priorexposure (e.g., GIL-04, GIL-07, GIL-10) and two boulders (GIL 06 and

GIL 08) probably represent reworked boulders from the underlyingolder till. The oldest moraine has an average exposure age of13.4 � 0.5 ka (GIL-01/GLW-23).

The ice cap split into the northern and southern lobes shortlyafter 11.3 � 0.3 ka (GIL-02/GLW-20). The youngest moraine has anaverage age of 11.5 � 2.3 ka (GIL-04, GIL-12). No further boulderswere sampled up the valley. However, this final age of retreat isvery similar to the minimum limiting age of the glaciation fromthe summit bog (11,570 þ 490/�400), and the rate of retreatsuggests that the last few moraines were deposited in less than1000 years.

5. Discussion

Exposure dating of moraines has identified at least four distinctstages of glaciation on Mt Giluwe within the last 150,000 years.Thesemoraines are difficult to subdivide on the basis of weatheringand preservation alone and this has led to them being previouslyclassified as one (Blake and Löffler, 1971; Löffler, 1972) or twoglaciations (Bik, 1972).

Fig. 5. Regional sea surface temperature records in comparison to dated glacial maxima on Mt Giluwe. a) Distance ice reached from the headwall for each of the dated glacialmaxima. Error bar is for the weighted mean age for the Tongo, Komia and Mengane Glaciations, and is an age range for the Gogon Glaciation, b) SST record of ODP Hole 806B (Leaet al., 2000), c) SST record of MD97-2140 (de Garidel-Thoron et al., 2005), d) LR4 benthonic oxygen isotope record stack of Lisiecki and Raymo (2005). This stack is derived from d18Orecords from around the world and primarily shows eustatic sea level changes. The age models of ODP Hole 806B and MD97-2140 are tuned to the chronostratigraphy of the LR4stack based on tie points listed in the Appendix.

T.T. Barrows et al. / Quaternary Science Reviews 30 (2011) 2676e26892686

To interpret the glacial history of Mt Giluwe in a regionalcontext, we have assembled two key regional SST records from theequatorial western Pacific Ocean (Fig. 5). Core ODP Hole 806B (Leaet al., 2000) is located on the Ontong Java Plateau and MD97-2140(de Garidel-Thoron et al., 2005) is located north of New Guinea(Fig. 1). Both cores come fromwithin the bounds of the Indo-Pacificwarm pool of SST>28 �C and are derived usingMg/Ca in planktonicforaminifera tests. Comparison of the timing of glaciation withthese SST records allows us to interpret the extent to which glaci-ation over New Guinea is regionally representative and driven bytemperature changes.

Theoldest evidence forglaciationonMtGiluwe isdatedbyK/Arat293e306 ka early in OIC 8, associated with the last dated volcanicactivityon themountain (Blake andLöffler,1971; Löffler,1972; Löffleret al., 1980). No evidence of this glaciation was found on the southwest flank of the mountain. However, because of its age, it may have

little surface expression remaining. Core ODP Hole 806B recordswarm SST at 300 ka, but rapid cooling into OIC 8 at 289 ka whereasMD97-2140 (deGaridel-Thoron et al., 2005) records SST duringOIC 8at 293 ka as the coldest in the last 500 ka (Fig. 5). Considering thelikely errors on the conventional K/Ar ages, theGogonGlaciationwasprobably associated with cooling early in OIC 8.

The oldest directly dated glaciation in our study, the MenganeGlaciation, reached its maximum extent during OIC 6. MinimumSST in ODP Hole 806B occurs at 143 ka and is recorded earlier inMD97-2140 at 169 ka. Together, these ages bracket the exposuredated maximum of the Mengane Glaciation centred at 146 � 10 ka.The OIC 6 interval is one of strong cooling in the temperate latitudesof the western Pacific Ocean (Barrows et al., 2007a), but in contrastto New Guinea, no ice advance has yet been directly dated to thistime in Australia or New Zealand. In the central North Pacific Ocean,the oldest evidence for glaciation on Mauna Kea in Hawaii, the

T.T. Barrows et al. / Quaternary Science Reviews 30 (2011) 2676e2689 2687

Pohakula Glaciation, is poorly dated from about 100e150 ka to150e200 ka (Wolfe et al., 1997) and possibly records glaciationat the same time as on Mt Giluwe. Elsewhere, well preservedmoraines of OIC 6 age conducive to exposure dating are rare (Blardet al., 2009).

The Komia Glaciation reached its maximum extent on MtGiluwe at 62 � 3 ka. This timing is slightly later than the OIC 4 SSTminima in ODP Hole 806B at 66 ka and 65 ka in MD97-2140. Thisglaciation is only well preserved in some locations because it hada similar extent to the Tongo Glaciation, and so it is poorlypreserved and we know little about the timing of ice retreat.Glaciation of this age is not known from elsewhere in New Guinea,but Hope (2009) inferred cooling from pollen records around thistime in the Albert Edward Range. On Hawaii, the penultimateglaciation or the Waihu Glaciation, has a single exposure age of63� 2.3 ka, based on an early approximation of the 36Cl productionrates (Zreda et al., 1991). Limiting K/Ar ages place the glaciationbetween 100 and 150 ka, but it could be as young as 70 ka (Wolfeet al., 1997). There is no direct evidence for an ice advance at60 ka in New Zealand or on the island of Tasmania, but the greatestextent of ice in the Snowy Mountains of Australia is dated at 59 kaduring the Snowy River advance (Barrows et al., 2001). Mid latitudeSST records indicate that Snowy River advance occurred at the endof the coldest phase of OIC 4 which ended at 60 ka and began, withvariable timing between cores, at 75e67 ka (Barrows et al., 2007a).It is important to note that minimum SST in the western PacificOcean is similar during both OIC 4 and 6, despite large differencesin sea level and therefore global terrestrial ice volume.

Cooling of a magnitude similar to OIC 2 and 4 occurs intermit-tently during OIC 3 in both ODP Hole 806B and MD97-2140 cores,but no glacial record is known on Mt Giluwe from this time,presumably because ice did not advance to the same extent as theensuing LGM. However, Highland vegetation histories suggest lowtreelines from 35,000-28,000 cal yr BP (Prentice et al 2005). Incontrast, the temperate glaciations of Australia and New Zealandare characterised by large advances of ice late in OIC 3 (Barrowset al., 2001; Suggate and Almond, 2005; Mackintosh et al., 2006)when temperate latitude SST cools to levels similar to the LGM(Barrows et al., 2007a).

The youngest glaciation, the Tongo Glaciation, reached itsmaximum extent at 19.4 ka during OIC 2. The oldest limitingradiocarbon age of 20,300 þ 130/�140 cal yr BP (Table 1) placesmaximum ice advance shortly before this. Exposure ages are likelyto be minimum estimates for the age of a moraine because of thelag time between deposition and geomorphic stabilisation/coloni-sation by vegetation. Bearing this in mind, the volume of theterminal moraine at the site dated (GLW-11) is similar to that of theother four terminal moraines in that set. This indicates that the icewas only briefly at the maximum position, probably less than 1000years.

The maximum extent of the Tongo Glaciation occurs during themaximum phase of cooling in the equatorial Pacific Ocean between19.8 ka (ODP Hole 806B) and 16.8 ka (MD97-2140). In the centralPacific Ocean on Mauna Kea in Hawaii, maximum glaciation alsooccurs between 19e16 ka (Blard et al., 2007). This timing coincideswith the last glacial maximum, unlike some other locations in thetropics (e.g., Smith et al., 2005b). Differences in ages between Blardet al. (2007) and Pigati et al. (2008) from similar moraines may belargely due to production rate calibration and scaling schemes. Theage of maximum ice extent during the LGM in the Pacific Oceantropics is very similar in timing to the age of maximum ice extent inthe temperate mountains of the western Pacific Ocean. Barrowset al. (2001, 2002) found that maximum ice extent occurred inAustralia between 17e20 ka, after a period of periglacial activitycentred at 22 ka (Barrows et al., 2004). The last major West Coast

glacial advance culminated at 21.5 ka in New Zealand (Suggate andAlmond, 2005) with retreat well underway by 17 ka (Schaefer et al.,2006). Throughout the Indo-Pacific region, sea-surface tempera-ture (SST) was at aminimum at 21 ka (Barrows et al., 2000; Barrowsand Juggins, 2005).

Ice onMt Giluwe retreated only slowly from itsmaximumextentfor 4000 years, before a more rapid retreat from 15.4 ka over thesubsequent 4000 years. On the southwest flank, there is on averageone moraine per 150 years. Our radiocarbon ages indicate rapidestablishment of bogs behind the moraines as the ice retreated(Table 1). Ice was absent from the highest elevations after 11.5 kawith no late Holocene advances. Similarly on Hawaii, retreat of iceafter 16 kawasmarked by a brief pause at 15.5 ka, then rapid retreatleaving the mountain ice free by 15 ka (Blard et al., 2007). Thepattern of tropical ice retreat varies from that seen in the temperatewestern Pacific Ocean. In both Australia and New Zealand, thedeglaciation is marked by one major readvance of ice at 16e19 ka(Barrows et al., 2001, 2002; Suggate and Almond, 2005). However,across 45� of latitude there is very rapid retreat of ice after 15 ka. Inthe mid latitude SST records, warming is rapid after 18e19 ka, andreaches near Holocene temperatures by 15 ka (Barrows et al.,2007a). Similarly in the tropical deep-sea cores, warming wasvery rapid after 16.8e19.8 ka increasing to Holocene level temper-atures at 14.9 ka (ODP Hole 806B) and 12.3 ka (MD97-2140).

The strong relationship between glacial maxima and SSTminima across the central and western Pacific Ocean indicatesa common regional climatic forcing of glaciation through atmo-spheric cooling. The above comparison between the glacial recordof Mt Giluwe with records in the temperate latitudes revealsa distinctly Southern Hemisphere pattern (Barrows et al., 2007a) inthe western Pacific Ocean tropics. Blard et al. (2007) suggested thatthe warming pattern in the central Pacific followed the NorthAtlantic pattern, as recorded in the Greenland ice cores, but thereare few differences between Hawaii and the western Pacific Oceanglacial records. In both locations there is no evidence for a signifi-cant readvance of ice or North Atlantic Ocean style cooling duringthe Younger Dryas Chronozone (YD), which is actually a time ofwarming in the western Pacific Ocean (Barrows et al., 2007b;Williams et al., 2009). Suggested cooling during the YD nearby atNew Caledonia (Correge et al., 2004), has no support from inde-pendent estimates of SST, nor from the glacial records. There arealso no glacier advances associated with Dansgaard-Oeschgerevents which are classically North Atlantic Ocean phenomenon.At present we are not aware of advances of ice during OIC 3 in theTropical glacier record but these are prominent in the temperateglaciated areas. These latter advances appear to be events associ-ated with large changes in sensible heat content in the SouthernOcean (Barrows et al., 2007a).

Despite the Southern Hemisphere character of the timing ofdeglaciation in the western Pacific Ocean, the first order timing ofglacial maxima is of a global nature. There is a strong rhythmmatching glacier maximum timing on Mt Giluwe (20 ka, 62 ka,146 ka) and the periodicity of obliquity in the Earth’s orbit (40 ka).These times match minima in solar radiation during summer athigh latitudes in the Northern Hemisphere. These times also matchtroughs in carbon dioxide concentrations as recorded in ice cores(Barnola et al., 1987; Petit et al., 1999). This provides support thatsynchronisation of global cooling during glacial maxima was aidedby low greenhouse gas concentrations.

5.1. Climate change during the last glacial maximum from glacialevidence

Direct dating on themoraines of Mt Giluwe allows us to reassessthe Tropical LGM temperature paradox in the western Pacific

T.T. Barrows et al. / Quaternary Science Reviews 30 (2011) 2676e26892688

Ocean. We found the maximum extent of ice during the LGM lessthan previously mapped (Löffler, 1972). An estimate of thetemperature difference relative to present can be approximatedfrom the elevation of the ice cap on the southwest flank. Löffler(1972) used a THAR (toe to headwall altitude ratio) of 0.5 to esti-mate the equilibrium line altitude (ELA) on Mt Giluwe, and we usethe same approach here to provide a comparison. A more sophis-ticated glacier reconstruction will follow. Ice extended froma maximum of about 3800 m at the highest point of the headwalldown to 3300m. A THAR of 0.5 gives an ELA of 3550�100m. This isconsistent with the highest elevation of lateral moraine in theTongo River Valley of about 3600 m. This ELA estimate is at theupper limit of the original estimate of 3500e3550 m made byLöffler (1972).

There are presently no glaciers in Papua New Guinea (PNG). MtGiluwe reaches amaximum elevation of 4367m and is currently iceand permanent snow patch free. Löffler (1972) estimated that the1970’s equilibrium line at about 4600 m. The climate of montanePNG within the humid inner tropics is monsoonal, so the ELA willmost closely correspond to the summer 0 �C isotherm (Benn et al.,2005). Using 4600 m as a first estimation of the modern ELA, thedifference between the LGM and present ELA is about 1050 m.Combining this estimate with a lapse rate of 6�C/1000 m indicatesthat the LGM was approximately 6 �C colder than present. Evengiven possible errors in the position of the ELA (�100 m) and thelikelihood that the lapse rate would be a minimum for the driermountain air during the LGM, the temperature difference isunlikely to be less than 5 �C. This estimate of cooling is similar tothe estimate of 5e7 �C for Hawaii (Blard et al., 2007). Becauseprevious ice caps that formed on Mt Giluwe were larger inmagnitude, this temperature difference is likely to be a minimumfor these glaciations.

A 5e6 �C temperature difference contrasts with the most recentestimates of an SST difference of a maximum of 3 �C for the tropicalwestern Pacific Ocean (MARGO Project Members, 2009). Thissupports previous observations that greater cooling occurred athigh altitudes in New Guinea compared to the sea-surface. Thisdifferential is not unique to the tropics and is also observed in thetemperate latitudes of eastern Australia (Barrows et al., 2000).Given the improved dating and the refinement in SST methodolo-gies, a climatological explanation must be sought to explain thedifference. Although the magnitude of cooling in the terrestrialenvironment is magnified, the timing of cooling between theterrestrial and marine realms is very similar, implying a commonclimatic control, probably through atmospheric cooling.

Acknowledgements

This work was conducted as part of ARC Discovery grants awar-ded to TTB (DP0557143) and GH and an NSF grant to MP (0234546).We thank Ernst Löffler for assisting GH with earlier coring.

Appendix. Supplementary data

Supplementary data associated with this article can be found inthe online version, at doi:10.1016/j.quascirev.2011.05.022.

References

Barker, S., Cacho, I., Benway, H., Tachikawa, K., 2005. Planktonic foraminiferal Mg/Caas a proxy for past oceanic temperatures: a methodological overview and datacompilation for the last glacial maximum. Quaternary Science Reviews 24,821e834.

Barnola, J.M., Raynaud, D., Korotkevich, Y.S., Lorius, C., 1987. Vostok ice coreprovides 160,000-year record of atmospheric CO2. Nature 329, 408e414.

Barrows, T.T., Juggins, S., 2005. Sea-surface temperatures around the Australianmargin and Indian Ocean during the last glacial maximum. Quaternary ScienceReviews 24, 1017e1047.

Barrows, T.T., Juggins, S., De Deckker, P., Calvo, E., Pelejero, C., 2007a. Long-term sea-surface temperature and climate change in the Australian-New Zealand region.Paleoceanography 22, PA 2215. doi:2210.1029/2006PA001328.

Barrows, T.T., Juggins, S., De Deckker, P., Thiede, J., Martinez, J.I., 2000. Sea-surfacetemperatures of the southwest Pacific Ocean during the last glacial maximum.Paleoceanography 15, 95e109.

Barrows, T.T., Lehman, S.J., Fifield, L.K., De Deckker, P., 2007b. Absence of cooling inNew Zealand and the adjacent ocean during the younger Dryas Chronozone.Science 318, 86e89.

Barrows, T.T., Stone, J.O., Fifield, L.K., 2004. Exposure ages for Pleistocene periglacialdeposits in Australia. Quaternary Science Reviews 23, 697e708.

Barrows, T.T., Stone, J.O., Fifield, L.K., Cresswell, R.G., 2001. Late Pleistocene glacia-tion of the Kosciuszko Massif, Snowy Mountains, Australia. QuaternaryResearch 55, 179e189.

Barrows, T.T., Stone, J.O., Fifield, L.K., Cresswell, R.G., 2002. The timing of the lastglacial maximum in Australia. Quaternary Science Reviews 21, 159e173.

Benn, D.I., Owen, L.A., Osmaston, H.A., Seltzer, G.O., Porter, S.C., Mark, B., 2005.Reconstruction of equilibrium-line altitudes for tropical and sub-tropicalglaciers. Quaternary International 138e139, 8e21.

Bik, M.J.J., 1972. Pleistocene glacial and periglacial landforms on Mt. Giluwe and Mt.Hagen, western and southern highlands districts, Territory of Papua and NewGuinea. Zeitschrift für Geomorphologie 16, 1e15.

Blake, D.H., Löffler, E., 1971. Volcanic and glacial landforms on Mount Giluwe,Territory of Papua and New Guinea. Geological Society of America Bulletin 82,1605e1614.

Blard, P.H., Lave, J., Pik, R., Wagnon, P., Bourles, D., 2007. Persistence of fullglacial conditions in the central Pacific until 15,000 years ago. Nature 449,591e594.

Blard, P.H., Lave, J., Farley, K.A., Fornari, M., Jimenez, N., Ramirez, V., 2009. Late localglacial maximum in the central altiplano triggered by cold and locally-wetconditions during the paleolake Tauca episode (17e15 ka, Heinrich 1).Quaternary Science Reviews. doi:10.1016/j.quascirev.2009.09.025.

Climap, P.M., 1976. The surface of the ice-age Earth. Science 191, 1131e1137.CLIMAP Project Members, 1981. Seasonal reconstructions of the Earth’s surface at

the last glacial maximum. Geological Society of America Map and Chart Series,MC-36.

Correge, T., Gagan, M.K., Beck, J.W., Burr, G.S., Cabioch, G., Le Cornec, F., 2004.Interdecadal variation in the extent of South Pacific tropical waters during theYounger Dryas event. Nature 428, 927e929.

de Garidel-Thoron, T., Rosenthal, Y., Bassinot, F., Beaufort, L., 2005. Stable sea surfacetemperatures in the western Pacific warm pool over the past 1.75 million years.Nature 433, 294e298.

Evans, J.M., 2001. Calibration of the production rates of cosmogenic 36Cl frompotassium. PhD Thesis, The Australian National University, Canberra.

Hope, G., 2009. Environmental change and fire in the Owen Stanley ranges, PapuaNew Guinea. Quaternary Science Reviews 28, 2261e2276.

Hope, G.S., Peterson, J.A., 1975. Glaciation and vegetation in the high New Guineamountains. In: Suggate, R.P., Cresswell, M.M. (Eds.), Quaternary Studies. TheRoyal Society of New Zealand, Wellington, pp. 155e162.

Lea, D.W., Pak, D.K., Spero, H.J., 2000. Climate Impact of late Quaternary equatorialPacific Sea surface temperature variations. Science 289, 1719e1724.

Lisiecki, L.E., Raymo, M.E., 2005. A Pliocene-Pleistocene stack of 57 globallydistributed benthic d18O records. Paleoceanography 20, PA1003. doi:10.1029/2004PA001071.

Liu, B., Phillips, F.M., Fabryka-Martin, J.T., Fowler, M.M., Stone, W.D., 1994. Cosmo-genic 36Cl accumulation in unstable landforms 1. Effects of the thermal neutrondistribution. Water Resources Research 30, 3115e3125.

Löffler, E., 1972. Pleistocene glaciation in Papua and New Guinea. Zeitschrift fürGeomorphologie 13, 32e58. Supplementband.

Löffler, E., Mackenzie, D.E., Webb, A.W., 1980. Potassium-argon ages from some ofthe Papua New Guinea highlands volcanoes, and their relevance to Pleisto-cene geomorphic history. Journal of the Geological Society of Australia 26,387e397.

Mackintosh, A.N., Barrows, T.T., Colhoun, E.A., Fifield, L.K., 2006. Exposure datingand glacial reconstruction at Mt. Field, Tasmania, Australia, identifies MIS 3 andMIS 2 glacial advances and climatic variability. Journal of Quaternary Science 21,363e376.

MARGO Project Members, 2009. Constraints on the magnitude and patterns ofocean cooling at the last glacial maximum. Nature Geoscience 2, 127e132.

Mark, B.G., Harrison, S.P., Spessa, A., New, M., Evans, D.J.A., Helmens, K.F., 2005.Tropical snowline changes at the last glacial maximum: a global assessment.Quaternary International 138e139, 168e201.

Masarik, J., Reedy, R.C., 1995. Terrestrial cosmogenic-nuclide production systematicscalculated from numerical simulations. Earth and Planetary Science Letters 136,381e395.

Peterson, J.A., Hope, G.S., Prentice, M., Hantoro, W., 2002. Mountain environmentsin New Guinea and the late glacial maximum "warm seas/cold mountains"enigma in the West Pacific warm pool region. In: Kershaw, A.P., David, B.,Tapper, N., Penny, D., Brown, J. (Eds.), Bridging Wallace’s Line. 34. Catena Verlag,Reiskirchen, pp. 173e187.

Petit, J.R., Jouzel, J., Raynaud, D., Barkov, N.I., Barnola, J.M., Basile, I., Bender, M.,Chappellaz, J., Davis, M., Delaygue, G., Delmotte, M., Kotlyakov, V.M.,

T.T. Barrows et al. / Quaternary Science Reviews 30 (2011) 2676e2689 2689

Legrand, M., Lipenkov, V.Y., Lorius, C., Pépin, L., Ritz, C., Saltzman, E.,Stievenard, M., 1999. Climate and atmospheric history of the past 420,000 yearsfrom the Vostok ice core, Antarctica. Nature 399, 429e436.

Pigati, J.S., Zreda, M., Zweck, C., Almasi, P.F., Elmore, D., Sharp, W.D., 2008. Ages andinferred causes of late Pleistocene glaciations on Mauna Kea, Hawai’i. Journal ofQuaternary Science 23, 683e702.

Phillips, F.M., Stone, W.D., Fabryka-Martin, J.T., 2001. An improved approach tocalculating low-energy cosmic-ray neutron fluxes near the land/atmosphereinterface. Chemical Geology 175, 689e701.

Prentice, M.L., Hope, G.S., Maryunani, K., Peterson, J.A., 2005. An evaluation ofsnowline data across New Guinea during the last major glaciation, and area-based glacier snowlines in the Mt. Jaya region of Papua, Indonesia, during thelast glacial maximum. Quaternary International 138e139, 93e117.

Prentice, M.L., Hope, G.S., Peterson, J.A., Barrows, T.T., 2011. The glaciation ofthe Southeast Asian equatorial region. In: Ehlers, J.J., Gibbard, P.L. (Eds.),Quaternary Glaciations e Extent and Chronology, Part IV e Acloser look.Elsevier, Amsterdam.

Reimer, P.J., Baillie, M.G.L., Bard, E., Bayliss, A., Beck, J.W., Blackwell, P.G., Ramsey, C.B.,Buck, C.E., Burr, G.S., Edwards, R.L., Friedrich, M., Grootes, P.M., Guilderson, T.P.,Hajdas, I.,Heaton,T.J.,Hogg,A.G.,Hughen,K.A.,Kaiser,K.F.,Kromer,B.,McCormac,F.G.,Manning, S.W., Reimer, R.W., Richards, D.A., Southon, J.R., Talamo, S., Turney, C.S.M.,Plicht, J.v.d., Weyhenmeyer, C.E., 2009. IntCal09 and Marine09 radiocarbon age cali-bration curves, 0e50,000 years cal BP. Radiocarbon 51, 1111e1150.

Reiner, E., 1960. The glaciation of Mount Wilhelm, Australian New Guinea.Geographical Review 50, 491e503.

Rind, D., Peteet, D., 1985 Jan-22. Terrestrial conditions at the last glacial maximumand CLIMAP sea-surface temperature estimates: are they consistent? Quater-nary Research 24, 1e22.

Rosell-Melé, A., Bard, E., Emeis, K.-C., Grieger, B., Hewitt, C., Müller, P.J.,Schneider, R.R., 2004. Sea surface temperature anomalies in the oceans at theLGM estimated from the alkenone-UK0

37 index: comparison with GCMs.Geophysical Research Letters 31. doi:10.1029/2003GL018151.

Schaefer, J.M., Denton, G.H., Barrell, D.J.A., Ivy-Ochs, S., Kubik, P.W., Andersen, B.G.,Phillips, F.M., Lowell, T.V., Schluchter, C., 2006. Near-synchronous interhemi-spheric termination of the last glacial maximum in mid-latitudes. Science 312,1510e1513.

Smith, J.A., Seltzer, G.O., Farber, D.L., Rodbell, D.T., Finkel, R.C., 2005b. Early local lastglacial maximum in the tropical Andes. Science 308, 678e681.

Stone, J.O., 2000. Air pressure and cosmogenic isotope production. Journal ofGeophysical Research 105, 23,753e23,759.

Stone, J.O., Allan, G.L., Fifield, L.K., Cresswell, R.G., 1996a. Cosmogenic chlorine-36from calcium spallation. Geochimica et Cosmochimica Acta 60, 679e692.

Stone, J.O., Evans, J.M., Fifield, L.K., Cresswell, R.G., Allan, G.L., 1996b. Cosmogenicchlorine-36 production rates from potassium and calcium. Radiocarbon 38,170e171.

Stone, J.O.H., Evans, J.M., Fifield, L.K., Allan, G.L., Cresswell, R.G., 1998. Cosmogenicchlorine-36 production in calcite by muons. Geochimica et Cosmochimica Acta62, 433e454.

Suggate, R.P., Almond, P.C., 2005. The last glacial maximum (LGM) in western southisland, New Zealand: implications for the global LGM and MIS 2. QuaternaryScience Reviews 24, 1923e1940.

Stuiver, M., Reimer, P.J., 1993. Extended 14C data base and revised CALIB 3.0 14C agecalibration program. Radiocarbon 35, 215e230.

Thompson, P.R., 1981. Planktonic foraminifera in the Western North Pacific duringthe past 150 000 years: comparison of modern and fossil assemblages. Palae-ogeography, Palaeoclimatology, Palaeoecology 35, 241e279.

Verstappen, H.T., 1952. Luchtfotostudies over het Centrale Bergland van NederlandsNieuw-Guinea. Tijdschrift Van Het Koninklijk Nederlandsch AardrijkskundigGenootschap Series 2 (69), 336e363. 425e431.

Webster, P.J., Streten, N.A., 1978. Late Quaternary ice age climates of tropicalAustralasia: interpretations and reconstructions. Quaternary Research 10,279e309.

Williams, M., Cook, E., van der Kaars, S., Barrows, T., Shulmeister, J., Kershaw, P.,2009. Glacial and deglacial climatic patterns in Australia and surroundingregions from 35 000 to 10 000 years ago reconstructed from terrestrial andnear-shore proxy data. Quaternary Science Reviews 28, 2398e2419.

Wolfe, E.W., Wise, W.S. & Dalrymple, G.B. (1997) The geology and Petrology ofMauna Kea volcano, Hawaii: a study of Postshield Volcanism. In: US GeologicalSurvey Professional Paper pp. 129.

Zreda, M.G., Phillips, F.M., Elmore, D., Kubik, P.W., Sharma, P., Dorn, R.I., 1991.Cosmogenic chlorine-36 production rates in terrestrial rocks. Earth and Plan-etary Science Letters 105, 94e109.