Global and Planetary Change 79 (2011) 61–72

Contents lists available at ScienceDirect

Global and Planetary Change

j ourna l homepage: www.e lsev ie r.com/ locate /g lop lacha

Elevated East Antarctic outlet glaciers during warmer-than-present climates insouthern Victoria Land

Kate M. Swanger a,⁎, David R. Marchant a, Joerg M. Schaefer b, Gisela Winckler b, James W. Head III c

a Department of Earth Sciences, Boston University, 675 Commonwealth Avenue, Boston, MA 02215, USAb Lamont-Doherty Earth Observatory, Route 9W, Palisades, NY 10964, USAc Department of Geological Sciences, Brown University, 324 Brook Street, Box 1846, Providence, RI 02912, USA

⁎ Corresponding author at: Department of EnvironmScience, University of Massachusetts, Lowell, MA 01854

E-mail address: Kate_Swanger@uml.edu (K.M. Swan

0921-8181/$ – see front matter © 2011 Elsevier B.V. Aldoi:10.1016/j.gloplacha.2011.07.012

a b s t r a c t

a r t i c l e i n f o

Article history:Received 10 December 2010Accepted 26 July 2011Available online xxxx

Keywords:McMurdo Dry ValleysTaylor DomeTaylor GlaciercosmogenicPliocenePleistocene

We document Plio-Pleistocene changes in the level of Taylor Glacier, an outlet glacier in southern VictoriaLand that drains Taylor Dome on the periphery of the East Antarctic Ice Sheet (EAIS). Chronologic controlcomes from 3He cosmogenic-nuclide analyses of 27 boulders sampled from drifts and moraines in KennarValley, a small hanging valley that opens onto a peripheral lobe of Taylor Glacier in the QuartermainMountains. Assuming a constant boulder-erosion rate of 10 cm Myr−1, our preferred age model spans the last3.1 Myr and calls for stepped ice recession from a local highstand ~200 m above the present base of TaylorGlacier at the mouth of Kennar Valley. The texture and sedimentology of all mapped moraines and driftsindicate deposition from cold-based ice, analogous with the modern Taylor Glacier at the mouth of KennarValley.The Kennar Valley glacial record shows an uncharacteristic relationship with average global temperatures,exhibiting higher-than-present ice levels during globally warm periods, including the Pliocene climaticoptimum (~3.1 Ma) and Marine Oxygen Isotope Stage (MIS) 31 (~1.07 Ma). The Kennar Valley record alsosuggests that the rate of ice-surface lowering accelerated after the mid-Pleistocene transition at ~0.9 Ma.Correlation of our moraine record with published reports for fluctuations of Taylor Glacier elsewhere in theQuartermain Mountains, and with a dated moraine record from Ferrar Glacier (a second outlet for TaylorDome), reveals similar ice-surface changes, highlighting minor, but widespread ice recession in southernVictoria Land since the mid- to late-Pliocene. Our record for minimal variability in the East Antarctic Ice Sheetcontrasts with recent data from nearby marine cores that call for dynamic fluctuations in the volume ofgrounded ice in the Ross Embayment, and significant reduction of theWest Antarctic Ice Sheet (WAIS) duringwarmer-than-present intervals. Taken together, these records from the Ross Embayment call for considerablevariation in the response of marine-based West Antarctic ice and terrestrial East Antarctic outlet glaciersduring Plio-Pleistocene time.

ental, Earth and Atmospheric, USA. Tel.: +1 978 934 2664.ger).

l rights reserved.

© 2011 Elsevier B.V. All rights reserved.

1. Introduction

One of the most important issues facing climate scientists todayconcerns the response of Antarctica's ice sheets to global climatechange. Recent interpretations of sediments recovered from the AND-1B marine core in the western Ross Embayment (78° S) (collectedunder the auspices of the multinational ANDRILL program, ANtarcticDRILLing) call for dynamic, obliquity-paced fluctuations in the volumeand areal extent of the West Antarctic Ice Sheet (WAIS) during Plio-Pleistocene time (Naish et al., 2009). The findings imply full to partialcollapse of the WAIS during warmer-than-present climate intervals,including marine oxygen isotope stage 31 (~1.07 Ma), and perhapseven during MIS 11 (~400 ka) and MIS 5.5 (~125 ka; Scherer et al.,

2008; Naish et al., 2009; Pollard and DeConto, 2009). Unknown,however, is whether nearby outlet glaciers draining the East AntarcticIce Sheet (EAIS) experienced similar fluctuations, or whether in theRoss Embayment only marine-based portions of Antarctica's icesheets underwent dynamic behavior during Plio-Pleistocene time(Denton and Hughes, 1981; see also Bentley et al., 2010). To begin toaddress this question, we mapped and dated glacial moraines anddrifts deposited from Taylor Glacier, an outlet glacier that drainsTaylor Dome on the periphery of the East Antarctic Ice Sheet. TaylorGlacier terminates on land, ~40 km from the coast, and about~150 km from the AND 1B core (Fig. 1). Consequently, the recordfrom Taylor Glacier affords an ideal opportunity to observe thephasing and dynamics between the marine-based West Antarctic IceSheet and the larger, terrestrial East Antarctic Ice Sheet. Ourchronologic control comes from 3He cosmogenic-nuclide analyses of27 boulders sampled from ninemoraines and drifts in Kennar Valley, asmall hanging valley that opens onto a peripheral lobe of Taylor

Are

na V

alle

y

Kennar Valley

Vernier Valley

Lash

ly M

ount

ains

Bea

con

Valle

yCas

sidy

Gla

cier

Taylor Dome

Ferrar G

laci

er

15 km

DepotNunatak

LashlyGlacier

Taylor Glacier

Fig. 2

Equilibrium Line

Friis Hills

N

0 10 km5

Ross Island

Ross Ice Shelf

Ross Sea

TaylorDome

Ferrar Glacier

Taylor GlacierMcMurdo

Dry Valleys

0 100 km50

E W

0 2000 km1000

TDDC

New Harbour

TAM

TurnaboutValley

1B

2A

FingerMountain

Taylo

r Glacier

TabularMountain

MountFeather

Kukri Hills

Fig. 1. Landsat 7 satellite image of upper Taylor Glacier and upper Ferrar Glacier, both sourced from Taylor Dome. The dotted black line shows the equilibrium line (here the general boundarybetweendry snowandwind-sweptblue ice) (Chinn,1980). Theblack rectangle corresponds to the regionofKennarValleydepicted inFig. 2a. Lower left inset:Antarcticawith the locationof theMcMurdo Dry Valleys indicated with the black rectangle. DC= Dome Circe (Dome C), TAM= Transantarctic Mountains, and TD= Taylor Dome. Lower right inset: Eastern Ross Sea regionshowing relative locations of Taylor Dome, Ross Ice Shelf, McMurdo Dry Valleys, outlet glaciers, and ANDRILL offshoremarine cores (1B and 2A) (see Naish et al., 2009). Black rectangle showslocation of the satellite image.

62 K.M. Swanger et al. / Global and Planetary Change 79 (2011) 61–72

Glacier in the Quartermain Mountains, McMurdo Dry Valleys (MDV)(Figs. 1 and 2).

2. Background and setting

2.1. Taylor Dome and Taylor Glacier

Taylor Dome is one of several peripheral domes along the marginsof the East Antarctic Ice Sheet. Taylor Dome (77°47′47″ S, 158°43′26″E) merges with a broad ice divide that extends inland to Dome Circe, amajor dome in interior East Antarctica (Fig. 1) (Drewry, 1982). Giventhis configuration, changes in the level of Taylor Dome reflect localchanges in precipitation (Steig et al., 2000; Grootes et al., 2001) aswell as major fluctuations in the level of interior East Antarctic ice(Chinn, 1980; Marchant et al., 1994). The 75-km long Taylor Glacierextends eastward from Taylor Dome and passes across a series ofhigh-level bedrock steps before terminating in central Taylor Valley~40 km from the coast (Fig. 1). In its upper reaches near KennarValley, Taylor Glacier is ~1000-m thick; on the basis of repeat GPSsurveys and synthetic aperture radar interferometry (InSAR), ice-surface velocities in this region are ~5–10 m yr−1 (Kavanaugh et al.,2009). East of Kennar Valley, Taylor Glacier is funneled throughnarrow bedrock constrictions and accelerates to a maximum velocity

of 15–20 m yr−1. Apart from strain-induced melting in these regionsof accelerated ice flow, Taylor Glacier is cold based, largely non-erosive, and frozen to its bed (Robinson, 1984; Staiger et al., 2006;Kavanaugh et al., 2009). Although we cannot preclude some level ofbasal entrainment beneath cold-based ice (e.g., Cuffey et al., 2000;Atkins et al., 2002) the absence of dirty basal ice at the margin ofTaylor Glacier alongside Kennar Valley suggests that basal plucking islikely insignificant in the upper reaches of the modern Taylor Glacier.The noted debris carried at the surface of Taylor Glacier today mostlikely arises from direct rock fall onto the ice surface and fromwindblown sands (Marchant et al., 1994; Staiger et al., 2006; see alsoSwanger et al., 2010).

Evidence for past changes in the elevation and areal extent ofTaylor Glacier comes from mapped moraines and drifts that crop outalongside Taylor Glacier in lower Kennar Valley, as well as in lowerArena and Beacon valleys (Fig. 1) (Brook et al. 1993; Marchant et al.,1994).

2.2. Kennar Valley

2.2.1. Physical settingKennar Valley is located along thewesternmargin of theQuartermain

Mountains, where Taylor Glacier first bends eastward toward the coast

K1

K2K2

K3

K4K5K6

K7

K8

0102

03

0405

0607

0809

2039

4546 47

50

61

6263

49

35

4342

4138

3736

55

0 500 mN

250

UD

UD 1500

1600

1700

1800 19

00

2000

1700

1400

1600

1700

1900

K2

K2

K3

K4K5K6K7

K8

Moraine without ice coreInferred moraine/ridgeContour (50-m interval)GlacierIce-cored drift N

Ice-cored moraine

01 km0.5

K1

a) b)

Fig. 3a

Fig. 3b

Fig. 2b

Fig. 2. (a) Sketch map of Kennar Valley on a topographic map base, showing the location of the Taylor Glacier lobe (light gray), moraines (black lines), and ice-cored drift (dark gray);contour interval=50 m. Enclosed box indicates area of coverage in panel b. (b) Aerial photograph (USGS TMA 3072 series) of lower Kennar Valley, showing the moraine sequence(white lines) and the locations of cosmogenic exposure samples (red/black circles with sample numbers). Locations for Fig. 3a and b are shown as numbered brackets that open inthe direction of view. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

63K.M. Swanger et al. / Global and Planetary Change 79 (2011) 61–72

(~77°45′ S and 161°24′ E). Kennar Valley is predominantly free of surfaceice and bounded to the west and south by a continuous bedrock cliffreaching ~1700-melevation (Fig. 2). The peripheral lobe of TaylorGlacierthat occupies the valley mouth displays a clean ice ramp and terminatesat ~1400 m elevation (Fig. 2). Adiabatically warmed katabatic winds arefunneled through Kennar Valley and produce a local blue-ice ablationzone on Taylor Glacier that drives continuous southward ice flow intoKennar Valley (Figs. 1 and 2).

As noted earlier, given that Taylor Glacier is cold-based in thisregion, the most obvious sources for debris entrainment are ice-marginal bedrock cliffs. Extensive bedrock cliffs of Ferrar Dolerite,150 m high, overlook Taylor Glacier just southwest of Kennar Valley(McElroy and Rose, 1987; Elliot and Fleming, 2004). At more distantlocations alongside Taylor Glacier, e.g., at Tabular Mountain, DepotNunatak, and the Lashly Mountains (all b25 km from Kennar Valley,Fig. 1), similar cliffs are incised in Feather Conglomerate, Weller CoalMeasures, and the Lashly Formation (fine-grained sandstones andgray shales with carbonaceous bands) (Barrett and Fitzgerald, 1986;McElroy and Rose, 1987); all such lithologies are present within thedrifts in lower Kennar Valley (see below). Beyond these regions ofcliffed bedrock, the East Antarctic Ice Sheet is sufficiently thick toovertop all bedrock topography and prevent debris entrainment fromdirect rockfall onto the ice surface. Assuming that all possible rockfallsources arewithin ~25 km, themaximumpotential time for transport ofsupraglacial-debris to Kennar Valley is ~5 kyr (assuming a minimumice-flow velocity of ~5 m yr−1 for Taylor Glacier and continuoussupraglacial exposure during transport).

2.2.2. ClimateEnvironmental conditions inKennarValley are among the coldest and

driest in the MDV. Mean annual atmospheric temperatures are b−20 °Cand summertime temperatures rarely, if ever, exceed 0 °C (Kowalewskiet al., 2006;Marchant andHead, 2007). Precipitation is b50 to 100 mmof

water equivalent per year (Clow et al., 1988; Beyer et al., 1999; Doranet al., 2002; Fountain et al., 2009). Such dry conditions limit soil moistureand active-layer processes; cryoturbation is largely restricted togravitational sliding along the margins of sublimation-type contraction-crackpolygons (e.g.,Marchant et al., 2002). Ingeneral, rocks at thegroundsurface tend to stay at the surface, since they arenot subjected to repeatedepisodes of burial and exposure as occurs in cold regions with saturatedactive layers (Hallet and Waddington, 1992; Marchant and Head, 2007;Morgan et al., 2010). Erosion is essentially restricted to salt weathering,wind abrasion, and thermal fracture (Marchant and Head, 2007; seeResults section). Estimates for erosion rates, via cosmogenic-nuclideanalyses, range from 5 to 10 cmMa−1 in high-elevation (2000–2500 m)nunataks (Summerfield et al., 1999) to ~20 cmMa−1 in nearby ArenaValley (Morgan et al., 2010) (see Fig. 1).

3. Methods

3.1. Mapping and sedimentological analyses

We employed geomorphic analyses of orthorectified aerial photo-graphs and detailed fieldwork in 2004 and 2006 tomap drifts depositedfrom Taylor Glacier in lower Kennar Valley (Figs. 2 and 3). Elevationcontrol was established using hand-held GPS units, a Garmin 3000witha reported accuracy of ±5 m horizontal and ±10m vertical and aTrimble GeoExplorer 3000 with ±1 m horizontal/vertical accuracy.During the course of fieldwork we collectedmultiple sediment samplesat 20- to 30-cm depth intervals (each 2 kg) for standard grain-sizeanalyses.We examined the 16- to 64-mm fraction (gravel) for lithologicconstituents and evidence for surfacemodification (such as glacial scourand/or weathering). At Boston University, we employed standard wetand dry sieving procedures to calculate weight percentages for finegravel, sand, and mud-sized components (Table 1).

2 m

K4

Distal

d)c)

a)

K5K4

b)

K6

K7

200 m200 m

Distal

2 m

K8

K2m

K3

K2d

Fig. 3. (a) Oblique aerial photograph looking north, showing K2 drift and bounding moraine. Note the high-centered sublimation polygons in the ice-cored drift (K2d) (locationshown in Fig. 2b). (b) Oblique aerial photograph of upper moraines K4–K7; view to the south (location shown in Fig. 2b). (c) Outer limit of K4 drift; view is to the south. (d) Outerlimit of K8 drift; view is to the northwest.

64 K.M. Swanger et al. / Global and Planetary Change 79 (2011) 61–72

3.2. Measuring weathering parameters

To measure progressive changes in the magnitude of surfaceweathering across Kennar drifts, we examined aminimum of 40 clastson each drift for morphologic evidence of salt weathering (solutionpits), wind abrasion (ventifacts and/or wind-abraded facets), andthermal fracture (fresh rock cores surrounded by thermally cleavedand spalled rock fragments (e.g., “puzzle rocks” of Marchant andHead, 2007)). We restricted analyses to clasts N104 cm3 (to ensuresufficient surface area for study), but otherwise sampleswere selectedrandomly. We also measured the a-axes of the largest clasts at thesurface of each drift (40 clasts per drift). Qualitative measures forwind abrasion and thermal fracture were noted in the field by thepresence of wind-polished facets and fractured clasts, whereas the

Table 1Physical characteristics of Kennar Valley drifts.

Drift Elev.a Height above Taylor Glacierb Reliefc dol:sst:cond

Grasize(m) (m) (m)

K2 1460 55 8 80:20:0 19:7K3 1475 70 1.5 82:18:0 32:6K4 1500 95 2.5 78:22:0 16:7K5 1490 85 1 90:0:10 58:3K6 1500 95 1.5 94:4:2 59:3K7 1495 90 1 68:25:7 11:8K8 1610 205 1 97:0:3 18:8UDi N/A N/A N/A 88:4:8 30:6

a Meters above sea level.b Maximum elevation of moraine compared to present-day average elevation of base ofc Maximum relief of moraine crests.d Ratio of lithologies in the N16-mm fraction of dolerite:sandstone:rounded quartz pebbe Ratio of gravel:sand:mud in the b16-mm fraction.f Ratio of equant:oblate:prolate:blade in the N16-mm fraction. dl = diameter of long axis

ds/diN0.67, Oblate=di/dlN0.67, ds/dib0.67. Prolate=di/dlb0.67, ds/diN0.67, Blade=di/dg Average maximum clast size determined by visual analyses in the field and of ground ph Letter indicates presence of: t: thermal fracture (puzzle rocks), h: weathered (oxidize

dolerite clasts, and p: development of solution pits N1 mm in diameter.i Undifferentiated drift that lies stratigraphically below moraines K4–K7.

effects of salt weathering (solution pits) were quantified throughdirect measurement of pit dimensions using digital calipers. For eachmeasured clast, we noted the width and depth of the largest solutionpits with a measurement precision of ±1 mm.

3.3. Cosmogenic 3He sample collection

We collected 27 surface clasts (Ferrar Dolerite) from nine mappedunits for cosmogenic-nuclide analyses (Figs. 2 and 3). Each cosmogenicsample was at least 103 cm3. To minimize the effects of potential rockdisplacement associated with the development of contraction-crackpolygons, we restricted sample collection to areas without polygons, orif necessary, to the center of the largest polygons (e.g., Marchant et al.,2002). We also collected samples along ridge crests, in order to reduce

ine

Zingg E:O:P:Bf

Average maximum clast size (cm)g Qualitativemeasures ofweatheringh

(cm)

6:5 16:49:15:20 200–300 t5:3 15:45:17:23 200 t, h, w4:10 13:62:5:20 100 t, h, w, v, p8:4 13:44:20:23 100 t, h, w, v, p4:7 22:51:11:16 100 t, h, w, v, p3:6 20:40:25:15 50 t, h, w, v, p0:2 15:44:12:29 50 t, h, w, v, p0:10 18:40:25:17 40 t, h, w, v, p

Taylor Glacier at Kennar Valley mouth (1405 masl).

les from conglomerates.

, di = diameter of intermediate axis, ds = diameter of short axis. Equant=di/dlN0.67,lb0.67, ds/dib0.67ictures.

d) soil horizon, w: wind faceting on surface clasts, v: varnish N1-mm thick on surface

Table 2aSample location, shielding and cosmogenic 3He data from the Kennar Valley drifts.

Drift Sample Longitudea Latitudea Altitudea Shieldingfactorb

3Hec

(masl) (108 at/g)

K1 DXP-04-03 160.142 −77.751 1405 0.912 0.088±0.003DXP-06-20 160.141 −77.751 1405 0.912 0.089±0.003KSX-06-39 160.434 −77.742 1395 0.883 0.094±0.003

K2md DXP-04-01 160.425 −77.752 1415 0.992 0.692±0.031KSX-06-35 160.416 −77.752 1415 0.993 0.522±0.021KSX-06-41 160.432 −77.751 1410 0.996 0.624±0.025KSX-06-42 160.433 −77.751 1410 0.996 0.917±0.037KSX-04-43 160.438 −77.752 1410 0.996 0.487±0.019KSX-06-47 160.422 −77.747 1450 0.995 1.220±0.037KSX-06-49 160.423 −77.746 1460 0.995 0.669±0.020

K2dd DXP-04-02 160.419 −77.752 1395 0.992 0.786±0.014KSX-06-36 160.421 −77.751 1400 0.993 1.388±0.056KSX-06-37 160.423 −77.750 1400 0.987 1.533±0.061KSX-06-38 160.428 −77.750 1400 0.991 0.429±0.017

K3 KSX-06-45 160.418 −77.747 1470 0.994 1.070±0.032KSX-06-46 160.417 −77.746 1475 0.995 1.770±0.053

K4 DXP-04-06 160.402 −77.751 1500 0.995 2.810±0.042DXP-04-07 160.403 −77.752 1490 0.995 3.290±0.058

K5 KSX-06-50 160.395 −77.751 1487 0.994 5.390±0.162KSX-06-63 160.395 −77.751 1482 0.995 2.410±0.072

K6 DXP-04-04 160.387 −77.751 1500 0.995 3.135±0.065DXP-04-05 160.387 −77.751 1503 0.995 2.370±0.066

K7 DXP-04-08 160.384 −77.750 1492 0.995 9.160±0.140KSX-06-62 160.384 −77.749 1494 0.995 8.890±0.267

K8 KSX 06-55 160.341 −77.751 1610 0.985 15.12±0.60UDe DXP-04-09 160.383 −77.750 1490 0.995 15.39±0.25

KSX-06-61 160.383 −77.749 1488 0.995 16.40±0.49

a Longitude, latitude and altitude (masl=meters above sea level) were measured ateach sample location using a Garmin 3000 (measurement error of ±5 m horizontal and±10 m vertical). Due to this relatively high vertical error, we compared these data totopographic maps andmeasured moraine ridge elevations using a Trimble GeoExplorer3000 (vertical error ±1 m).

b Shielding factors were calculated after Balco et al. (2008) from horizon geometrymeasurements recorded for each sample in the field.

c 1σ errors of 3He concentrations reflect propagated analytical uncertainties, basedon statistical errors and variability in the sensitivity of the mass spectrometer.

d K2 ice-cored moraines (K2m) and K2 ice-cored drift (K2d) (Figs. 2 and 3).e Undifferentiated drift distal to moraine K7 and stratigraphically below moraines

K4–K7.

65K.M. Swanger et al. / Global and Planetary Change 79 (2011) 61–72

the chances of burial beneath wind-blown snow (e.g., Margerison et al.,2005).

3.4. Mineral separation and gas extraction

Whole-rock, cosmogenic samples were cut at the Lamont-DohertyEarth Observatory (LDEO) and then crushed at Boston University (BU)using a Spex Certiprep 8515 Shatterbox; fragments were then sievedto isolate the N150 μm and b300 μm fraction. Typical sample weightswere about 30 mg. At LDEO, pyroxene grains were separated usingmagnetic and heavy liquid techniques, followed by handpicking.Separated pyroxenes were then analyzed for helium concentrationsand isotopic composition at LDEO on a MAP 215-50 noble gas massspectrometer calibrated with a known volume of a Yellowstonehelium standard (MM) with a 3He/4He ratio of 16.45Ra, where Ra=(3He/4He)air=1.384×10−6 (following protocols outlined inWinckleret al., 2005; Schaefer et al., 2006). Hot procedural blanks containedless than 5000 atoms of 3He with approximately atmospheric heliumisotopic composition. Blank corrections for 3He were smaller than 2%(for samples with low 3He concentrations, K1), and in most casessmaller than 1%.

3.5. Exposure age calculations and erosion rates

We use the sea level, high latitude production rate from spallationreactions of 120 at g−1 yr−1 for 3He in pyroxene (Goehring et al., 2010).At each sample site we recorded local and regional shielding bymeasuring horizon geometry. Shielding factors were calculated for eachsample after Balco et al. (2008) (Table 2a). Atmospheric pressures overAntarctica are anomalously low compared to typical pressure–elevationrelationships. Therefore, we employed the Antarctic-specific equationsin Stone (2000) to scale production rates to sample elevation. Choice ofthese production rates and scaling schemes relative to other reportedproduction rates (i.e. Lal, 1991; Licciardi et al., 1999) does not impact ourchronologyor ourmain conclusions.Nuclide-measurementdepths in allsurface clasts were b5 cm. Minimum, no-erosion, exposure ages werecalculated using the following equation:

N = Pt ð1Þ

where N is number of cosmogenic nuclides (at g−1), P is theproduction rate (at g−1 yr−1), and t is exposure time (yr).

By convention, exposure ages are typically reported assuming zeroerosion. This assumptionmay be valid for someHolocene and younger-aged samples, but for older samples, surface erosion typically removesthe outer crusts of rocks and reduces the total cosmogenic-nuclideinventory. As a consequence of this reduction, exposure ages notcorrected for erosion are typically viewed as minimum constraints. Ifone assumes a constant erosion rate (see below), the relationshipbetween exposure age and erosion rate is governed by the followingequation:

N =PLE

1− e−ELt

� �ð2Þ

where N is number of cosmogenic nuclides (at g−1), P is theproduction rate (at g−1 yr−1), L is the attenuation length (g cm−2),E is the erosion rate (g cm−2 yr−1), and t is time (yr). Solving Eq. (2)assuming an infinite exposure age (t) yields the maximum possibleerosion rate for a given sample (Table 2b). We used an attenuationlength of 155 g cm−2 (after Sarda et al., 1993) and an average rockdensity of 2.7 g cm−3.

4. Results

4.1. The spatial distribution and weathering characteristics of KennarValley drifts

The mapped pattern of drifts in lower Kennar Valley indicatesdeposition from southward advance(s) of Taylor Glacier. In mostcases, drifts are composed of scattered and isolated erratics whoseconcentration increases toward a single, major bounding moraineridge (Figs. 2 and 3); a similar pattern was also noted for driftsdeposited from the Ferrar outlet glacier in nearby Vernier Valley(Fig. 1) (Staiger et al., 2006). The Kennar Valley drifts are numberedsequentially from K1 (proximal to Taylor Glacier) to K8 (distal)(Fig. 2). As noted below, K2 is atypical in that it includes widespread,matrix-supported debris over stagnant, glacier ice. A second, matrix-supported drift of unknown origin (UD) underlies K4–K7 drifts andcrops out extensively on an upper-level bench beyond K7 (Fig. 2).

Kennar Valley drifts show an overall reduction in maximum clastsize with increasing distance from Taylor Glacier (Table 1). The a-axesof surface clasts decline from an averagemaximum of ~200 cm for K2,to ~100 cm for K5, and to ~50 cm for both K7 and K8 (Table 1). Someof the reduction in clast size likely reflects intermittent fracture(Fig. 4d–f) and pitting (Fig. 4a–c; Fig. 5). None of the clasts showevidence for transport beneath wet-based ice (e.g., striations, polish,molding, and/or faceting). For clarity we group Kennar drifts into twocategories, ice cored (K1 and K2) and non-ice cored (K3–K8).

Table 2bCosmogenic exposure ages and maximum erosion rates from the Kennar Valley drifts.

Drift Sample 3Hea Elevationscalingfactorb

Minimum 3He agec 3He-age 5 cm/Myrd 3He-age 10 cm/Myrd Max erosion ratee

(108 at/g) (ka) (ka) (ka) (cm/Myr)

K1 DXP-04-03 0.088±0.003 4.58 17.5±0.5 18 18 3600DXP-06-20 0.089±0.003 4.58 17.7±0.5 18 18 3600KSX-06-39 0.094±0.003 4.54 19.5±0.5 20 20 3300

K2mf DXP-04-01 0.692±0.031 4.62 126±6 127 127 460KSX-06-35 0.522±0.021 4.62 95±4 95 96 600KSX-06-41 0.624±0.025 4.60 114±5 114 115 500KSX-06-42 0.917±0.037 4.60 167±7 168 170 350KSX-04-43 0.487±0.019 4.60 89±4 89 89 650KSX-06-47 1.220±0.037 4.75 215±6 217 220 270KSX-06-49 0.669±0.020 4.78 117±4 118 120 500

K2df DXP-04-02 0.786±0.014 4.54 145±3 146 150 400KSX-06-36 1.388±0.056 4.56 255±10 258 260 220KSX-06-37 1.533±0.061 4.56 284±11 287 290 200KSX-06-38 0.429±0.017 4.56 79±3 79 80 730

K3 KSX-06-45 1.070±0.032 4.82 186±6 188 190 310KSX-06-46 1.770±0.053 4.84 306±9 310 320 190

K4 DXP-04-06 2.810±0.042 4.94 477±7 490 500 120DXP-04-07 3.290±0.058 4.90 562±10 580 590 100

K5 KSX-06-50 5.390±0.162 4.89 923±28 960 1000 60KSX-06-63 2.410±0.072 4.87 414±12 420 430 140

K6 DXP-04-04 3.135±0.065 4.94 532±11 550 560 110DXP-04-05 2.370±0.066 4.94 402±11 410 420 140

K7 DXP-04-08 9.160±0.140 4.91 1562±24 1700 1800 37KSX-06-62 8.890±0.267 4.92 1513±45 1600 1800 38

K8 KSX 06-55 15.12±0.60 5.38 2378±94 2700 3100 25UDg DXP-04-09 15.39±0.25 4.90 2630±43 3000 3500 22

KSX-06-61 16.40±0.49 4.90 2803±84 3200 3900 21

a 1σ errors of 3He concentrations reflect propagated analytical uncertainties, based on statistical errors and variability in the sensitivity of the mass spectrometer.b Cosmogenic production rates were scaled for elevation using equations from Stone (2000) for Antarctica.c We used a sea level, high-latitude cosmogenic 3He production rate of 120 at g−1 yr−1 (pyroxene) after Goehring et al. (2010). Minimum ages assume no erosion, accounting only

for production rates and shielding factors at each sample location.d Ages calculated with constant erosion rates of 5 and 10 cm Myr−1, with an attenuation length of 155 g cm−2 and an average rock density of 2.7 g cm−3.e Maximum erosion rates are calculated from the measured cosmogenic 3He assuming an infinite exposure time.f K2 ice-cored moraines (K2m) and K2 ice-cored drift (K2d) (Fig. 2 and 3).g Undifferentiated drift distal to moraine K7 and stratigraphically below moraines K4–K7.

66 K.M. Swanger et al. / Global and Planetary Change 79 (2011) 61–72

4.1.1. Ice cored drift: K1 and K2K1 drift includes the modern ice-cored moraine alongside Taylor

Glacier and all visible clasts embedded in the margin of Taylor Glacierat the mouth of Kennar Valley (Fig. 2). The modern ice-cored moraineis sharp crested, ~2 m wide, and ~3 m high. The only evidence forsurface alterations are thin iron-oxide stains that coat some rocksurfaces (Fig. 4); otherwise the rocks at the surface, and those partlyembedded in Taylor Glacier ice, are fresh, angular, and resemble thosefound in rockfall deposits elsewhere in the MDV (e.g., Swanger &Marchant, 2007).

K2 drift reaches a maximum elevation of ~1460 m, ~55 m abovethe base of nearby Taylor Glacier (Figs. 2 and 3). It includes anextensive sheet of matrix-supported, rocky debris that rests directlyon stagnant glacier ice (designated K2d); the drift is bound bymultiple, largely ice-cored, and partially cross-cuttingmoraines (K2m)(Figs. 2 and 3). K2 drift displays well-developed sublimation-typepolygons (Marchant et al., 2002; Marchant and Head, 2007), with arelief of ~3 m from elevated polygon centers to deep marginaltroughs. The contacts between drift (typically 25–50 cm thick) andunderlying glacier ice are sharp, dry, and planar. The largest ice-coredmoraine in K2 drift is 8 m high (with almost all of the relief arisingfrom the ice core itself); this moraine circumscribes a ~0.4 km2 regionof ice-cored drift (K2d) (Figs. 2 and 3a).

Clasts at the surface of K2 drift are composed of Ferrar Dolerite(80–90%), Feather Conglomerate, and undifferentiated sandstones,siltstones, and shales. The clasts of dolerite are typically ~1 to 2 m indiameter and exhibit weakly developed rock varnish (b1-mm thick)(Table 1).

4.1.2. Non-ice cored drift: K3–K8K3 drift reaches a maximum elevation of 1475 m, ~70 m above the

base of nearby Taylor Glacier. Its bounding moraine ridge is sharp-crested, 1–2 m high, and composed of dolerite (90%) and sandstonecobbles (Figs. 2 and 3). The surface dolerites exhibit minor wind-abraded facets and thin (b1-mm thick) rock varnish (Table 1).

K4–K7 drifts include a suite of closely spaced moraines and erraticcobbles that crop out between 1490 and 1500 m elevation, 85–95mabove the base Taylor Glacier (Fig. 2). The moraines and scatterederratics rest unconformably on an older, undifferentiated and matrix-supported drift sheet (UD). The boundingmoraine thatmarks the outeredge of K4 drift is 2- to 3-mhigh, whereasmoraines thatmark the outerlimits of drifts K5, K6 and K7 are relatively low and diffuse, reaching amaximumheight of ~1 m. Lithologieswithin K4–K7drifts are uniformlycomposed of ~80–90% dolerite and ~10–20% sandstone (with elevatednumbers of isolated quartz pebbles from the Feather Conglomerate).Clast size decreases from an average maximum of ~100 cm (a axis) onthe surface of K4 drift, to ~50 cm on K7 drift. Solution pits from saltweathering on the surface of clasts of Ferrar Dolerite increase from amaximum depth of ~15 mm on K4 drift to ~29 mm on K7 drift (Figs. 4and 5; Table 1). Ventifacts and puzzle rocks are also relatively common(Fig. 4). Salt-cementedhorizons bind iron-oxide stainedquartz grains inthe upper 20 cm of K4–K7 drifts (e.g., Bockheim, 2010).

K8 drift reaches a maximum elevation of 1610 m, ~205 m abovethe base Taylor Glacier at themouth of Kennar Valley (~2.5 km away).It terminates in a narrow moraine ridge ~400-m long and 1-m high(Fig. 3). As for the other Kennar drifts, the concentration of surfaceerratics increases up to the bounding moraine (Fig. 3). The clasts

a) d)

b) e)

c) f)

20 cm

Fig. 4. Relative weathering observed for clasts of Ferrar Dolerite on Kennar Valley drifts. Left column (a–c), shows the progressive increase in the development of solution pits: (a) nopits observed (fresh-appearing dolerite from K1 drift); (b) slightly pitted clast from K4 drift; (c) well developed solution pits on a clast from the surface of undifferentiated drift (UD)beyond K7 drift. Right column (d–f) shows the effects of thermal fracture: (d) fractured, but unbroken, boulder from K4 drift; (e and f) fractured and spalled clasts from K5 and K7drifts; arrows indicate direction of inferred movement away from central core cobble/boulder.

67K.M. Swanger et al. / Global and Planetary Change 79 (2011) 61–72

within K8 are N95% dolerite and display an average maximum size of~50 cm (Table 1); most clasts of dolerite exhibit thick rock varnish(≫1 mm thick) and large weathering pits (~20 mm deep). Sandgrains stained with iron oxides occur in the upper ~20 cm of themoraine and are cemented with visible salt encrustations.

4.1.3. Undifferentiated driftAs noted earlier, an undifferentiated drift (UD) crops out on the

upper-level bench between K7 and K8 drifts (Fig. 2). In hand-dugsections, this drift stratigraphically underlies erratics and morainesassociated with K4, K5, K6, and K7 drifts. Clasts at the surface aretypically b40 cm long (a axis) and exhibit solution pits as much as~45 mm deep, by far the largest for any mapped deposit in our studyarea (Figs. 4 and 5, Table 1).

4.2. Relative chronology

A relative chronology for Kennar moraines comes from notedchanges in surface and subsurfaceweatheringparameters, aswell as thepresence/absence of an underlying glacier ice. At the surface of K1 drift,

clasts lack evidence for weathering beyond slight iron-oxide stains.Clasts at the surface of K2 drift show slightly greater weathering, withrock varnish replacing iron-oxide stains (e.g., Staiger et al., 2006;Kowalewski et al., 2011). Thereafter, the development and progressivegrowth of solution pits, wind-polished facets, and puzzle rocks (thermalfracture) suggest that drift ages increase sequentially from K3 to K8(Fig. 4). Consistent with this assertion is the overall reduction inmaximum clast size from ~200 cm on the surface of K2 drift to ~50 cmon K8 drift. The reduction in clast size likely reflects the cumulativeeffects of surface weathering, especially episodic thermal fatigue androck fracture (Fig. 4). The undifferentiated drift (UD) is assigned theoldest relative age, with maximum surface clast sizes of b40 cm andsolution pits as much as 45 mm deep (Fig. 5, Table 1).

4.3. Numerical chronology

Our cosmogenic 3He analyses of 27 surface cobbles from Kennardrifts corroborate our relative chronology. In our preferred age model(see Section 5.1.3), we use the age of the oldest dated boulder fromeach drift to approximate drift age. As noted below, we adopt this

50

40

30

20

10

00 500 1000 1500 2000 2500 3000 3500

Uncorrected Drift Age (ka)

Wea

ther

ing

Pit

Dep

th (

mm

)

K4K5K6K7UD

b)

30 m

m M

yr-1

10 mm Myr-1

20 m

m M

yr-1

70

60

50

40

30

20

10

00 500 1000 1500 2000 2500 3000 3500

Uncorrected Drift Age (ka)

Wea

ther

ing

Pit

Dia

met

er (

mm

)

K4K5K6K7UD

a)

20 mm M

yr-140

mm

Myr

-1

60 m

m M

yr-1

Fig. 5. Plots showing pit evolution over time. Panel (a) shows the change in maximumpit diameter and panel (b) shows the change in maximum pit depth; all pits measuredon clasts of Ferrar Dolerite. Data are plotted as a function of minimum (no-erosion)exposure ages. Dotted lines indicate maximum pit dimensions over time given specificrates of pit deepening and widening. In general, maximum weathering pit diametersincrease by ~30–40 mm Myr−1 and maximum pit depths increase by ~15–20 mmMyr−1.

68 K.M. Swanger et al. / Global and Planetary Change 79 (2011) 61–72

procedure because erosion, especially over the multi-million-yeartimescales considered here, leads to a reduction in the overallinventory of cosmogenic nuclides and significant underestimation ofdeposit age; hence, we assume all cosmogenic ages are minimum ageestimates (see Section 5.1.2 for more information on the impact oferosion on exposure ages). In our discussion below, we highlight theoldest dated sample on each drift; see Tables 2a and 2b for all results.With the exception of K5 drift, all ages on clasts from individual driftsare internally consistent (Tables 2a and 2b).

4.3.1. Uncorrected agesThe oldest sample from K2 drift yielded an uncorrected age of 284±

11 ka. Likewise, the oldest sample from K3 drift yielded an uncorrectedageof 306±9 ka. Theupper-elevationdrifts, K4–K8, contain samples thatare considerably older, with the oldest samples yielding uncorrected agesof 562±10 ka (K4), 923±28 ka (K5), 532±11 ka (K6), 1562±24 ka

(K7), and 2378±94 ka (K8). The oldest clast on the undifferentiated driftjust beyond theK7 yielded anuncorrected age of 2803±84 ka (Table 2b).

5. Discussion and wider implications

5.1. Sources of scatter in cosmogenic data and calculating a preferredage model

The chronologypresentedhere ranges from20,000 years to ~3 millionyears.Withoneexception(K6), cosmogenic samples showageneral trendof greater exposure age with increasing distance from the modern TaylorGlacier, consistent with our relative chronology. However, there is scatterin the dataset. As noted in several prior cosmogenic studies (e.g., Brooket al., 1993; Schaefer et al., 1999; Gosse and Phillips, 2001; Ackert andKurz, 2004; Margerison et al., 2005; Staiger et al., 2006; Balco et al.,2008), the scatter in measured nuclide inventories is likely associatedwith some combination of (1) inheritance during prior exposure on cliffwalls and/or during transport to ice margins, (2) inclusion of non-cosmogenic 3He in analytical measurements, (3) diffusive loss of 3Hefrom minerals, (4) shielding by snow and ice, and (5) loss of thecosmogenic-nuclide inventory due tophysical erosion. Of these, thefirsttwo typically result in an overestimation of deposit ages, whereas thelatter three can causegreatunderestimationof ages. Potential impactsofeach of the above sources of error are discussed below.

Regarding the potential for diffusive loss of 3He from mineral grains,recent analytical studies have demonstrated that pyroxenes in FerrarDolerite quantitatively retain 3He (Schaefer et al., 2000); therefore,we donot consider diffusive loss to be a significant source of error. In addition,because precipitation in theMDV is extremely low (Fountain et al., 2009)episodic burial beneath snowfall is unlikely, however it cannot bedefinitively ruled out and could cause underestimation of exposure agesdue to occasional shielding of the samples (Margerison et al., 2005). Thisleaves the cumulative effects of cosmogenic-nuclide inheritance, non-cosmogenic sources, and erosional processes as the most likely factorsthat could cause significant scatter in our exposure ages.

5.1.1. Nuclide inheritance and non-cosmogenic 3HeEarlier cosmogenic noble-gas studies from Ferrar Dolerite in the Dry

Valleys have demonstrated consistent 3He and 21Ne exposure ages (e.g.Brunoet al., 1997, Schaeferet al., 1999, Staiger et al., 2006), indicatingonlyminor contributions by non-cosmogenic 3He. Here we expand thisargument by measuring the 3He concentrations in three samplesentrained in the terminus of Taylor Glacier in Kennar Valley (K1 drift)(Figs. 2 and 4). Because these clastswere entrained in the glacier, any 3Hein the pyroxenes must originate from prior exposure at the rock-fallsource area (inherited nuclides) and/or from a non-cosmogenic source.All three samples yielded very low concentrations of 3He, ~9×106 at g−1,which translates to an exposure age (if assumed to be entirelycosmogenic) of ~18 ka. Based on these data, we conclude that the non-cosmogenic and pre-exposure signal for the Kennar Valley samples isb20 kyr, which is minor relative to the timescales discussed here andtherefore, not a significant source of uncertainty.

5.1.2. Effects of erosion on cosmogenic exposure agesBy convention, exposure ages are typically reported assuming zero

erosion. This assumption may be valid for some late-Pleistocene andHolocene samples, but for older samples, surface erosion typicallyremoves the outer crusts of rocks (and/or causes exposure of entirelynew, fresh rock faces) and reduces the total cosmogenic-nuclide inventory(Fig. 4). As a consequence of this reduction, exposure ages that are notcorrected for erosion are typically viewed as minimum-age constraints.Published erosion rates for the MDV are typically ≤1 mMyr−1, and maybe as low as ~5 cmMyr−1 for high-elevation regions like Mt. Feather at2000–2500 m (Fig. 1) (Ivy-Ochs et al., 1995; Schaefer et al., 1999;Summerfield et al., 1999; Margerison et al., 2005).

69K.M. Swanger et al. / Global and Planetary Change 79 (2011) 61–72

As noted previously, clasts at the ground surface in Kennar Valleyshow evidence for surface erosion. Our quantitative analyses indicatethat the depth and diameter of solution pits increases with distancefrom Taylor Glacier, itself a proxy for increasing exposure age (Fig. 5).Similar changeswere observed in the size ofweatheringpits in doleritesalong a moraine sequence in Vernier Valley that spans ~4 Myr (Fig. 1)(Staiger et al., 2006). Although rates ofweathering-pit deepening do fallbelow the slowest reported erosion rates in the MDV of 5 cmMyr−1

(Summerfield et al., 1999), pit formation represents only one type ofweathering process, with others including wind abrasion and thermalfracture, for which our qualitative measures suggest increasing valuesfrom K2 to K8 (Table 1, Fig. 4). As shown in Fig. 6, thermal fractureproduces rock fragmentswith limited nuclide inventories and causes anincrease in age scatter with deposit age. For the oldest samples dated inthis study, on K8 drift and the undifferentiated drift (UD) themaximumpossible erosion rates are 21–25 cmMyr−1 (erosion rates calculatedfollowing equations in Gosse and Phillips, 2001). As noted below, whencalculating our preferred age model we assume erosion rates of10 cmMyr−1, a value that is consistent with other cosmogenic-nuclidestudies in this region (Ivy-Ochs et al., 1995; Schaefer et al., 1999).

5.1.3. The preferred age modelExcept for K5 drift, exposure ages within a single drift are internally

consistent. For K5, we received an age of ~430 ka for one clast and~1000 ka for another clast ~100 m distant (Table 2b). Our workinghypothesis is that the relatively young age of ~430 ka reflects nuclide lossdue to episodic erosion via thermal fracture/spallation (Fig. 6) (thoughthere were no obvious signs of recent erosion/spalling exceeding that ofnearby clasts). Although the old age could reflect some level of priorexposure, our findings of very low nuclide inheritance for clasts withinthe modern Taylor Glacier (K1 drift) suggest that the effects of prior

2000

1500

1000

500

00M

easu

red

Exp

osur

e A

ge (

ka)

500

400

300

100

00

200

Mea

sure

d E

xpos

ure

Age

(ka

)

After First Spalling Event

After Second Spalling Event

After Third Spalling Event

Initial Rock

Fracture

Fracture

Fracture

a) b)

c)

Fig. 6. Cartoon showing the effect of episodic thermal fracture on exposure histories and cothree spalling events over time. During each spalling event, a “buried-rock surface” is exposwithout other forms of erosion (e.g., no pitting or wind abrasion) its cosmogenic analyses wonew surface is exposed to cosmic rays; each newly exposed surface contains a reducedgreen/black slab, most recently exposed by thermal fracture, would contain the lowest nuclidfunction of spalling/thermal fracture; lines are color coded to spalling events/thermal fracturdeposit age through time. The lines for the other three fragments show exposure ages that in100-kyr intervals for a 500 ka deposit, whereas panel (c) assumes spallation/thermal fractureFig. 4; our assumption is that some of the scatter in exposure ages likely reflects the effects ofof the references to color in this figure legend, the reader is referred to the web version of

exposure may be minimal. In support of this assertion we note that theinternally consistent ages derived fromdated cobbles on drifts K4, K6, K7and the undifferentiated drift (UD) all indicate similar levels of exposure,an unexpected result if prior inheritance played a major role in alteringcosmogenic-nuclide inventories.

Therefore, our preferred age model for the cosmogenic data isgenerated by (1) applying a correction for constant erosion equivalent to10 cmMyr−1 (Table 2b), which is typical for the region (Ivy-Ochs et al.,1995; Schaefer et al., 1999) and (2) selecting the oldest cosmogenic agefor each mapped unit (Table 3). Assuming our preferred age model iscorrect, the ages are as follows: ~290 ka for the K2 drift, ~320 ka for K3drift, ~590 ka for K4 drift, ~1000 ka for K5 drift, ~560 ka for K6 drift,~1800 ka for K7 drift, and ~3100 ka for K8 drift (see Fig. 7 for all samplesages). The preferred age for undifferentiated drift (UD) that lies beyond(and stratigraphically below) the K7moraine is ~3900 ka (Table 3). K1 isthe modern ice-cored moraine.

5.2. Did Taylor Glacier fluctuate between periods of moraine formation?

Due to the episodic deposition of moraines alongside cold-basedoutlet glaciers, a key question is whether Taylor Glacier, and by inferenceTaylor Dome, could have experienced large-scale fluctuations duringintervals of non-deposition inKennarValley. To address this question,wecompare our dataset with other climate records in the MDV region. TheKennar Valley record, which indicates overall recession of Taylor Glaciersince a highstand ~3.1 Ma, is fully consistent with previously publishedchronologies for outlet glaciers in theMDV (see Fig. 1) (Brook et al., 1993;Marchant et al., 1994; Staiger et al., 2006). Indeed, cosmogenic analyses ofan extensive sequence of 39 moraines in nearby Arena Valley call foroverall ice recessionof TaylorGlacier since the late-Pliocene (Brownet al.,1991; Brook et al., 1993; Marchant et al., 1994). Likewise, cosmogenic

Second Spalling Event

First Spalling Event

Third Spalling Event

500 1000 1500 2000Deposit Age (ka)

Second Spalling Event

First Spalling Event

Third Spalling Event

200 300 400 500100Deposit Age (ka)

Age range:300 to500 ka

Age range:1000 to2000 ka

smogenic-nuclide inventories. Panel (a) shows an initial coherent rock that undergoesed. In the case of the blue/light gray fragment, its upper surface is continually exposed;uld yield the most accurate age. However, with each spalling/thermal-fracture event, a

cosmogenic inventory relative to that of the blue/light gray surface. In this case, thee inventory. Panels (b) and (c) show plots of “measured” exposure age vs. drift age as ae as noted in panel (a). Only the blue/light gray line shows equivalent exposure age andcreasingly underestimate deposit ages. Panel (b) assumes spallation/thermal fracture atevery 500 kyr for a 2 Ma deposit. Compare results with thermal fracture as observed in

intermittent thermal fracture (see text and exposure ages, Table 2b). (For interpretationthis article.)

Table 3Kennar Valley drift ages.

Deposit Heighta Mean exposure ageb Preferred age modelc

(m) (ka) (ka)

K1 0 19 20K2 55 160 290K3 70 260 320K4 95 550 590K5 85 720 1000K6 95 490 560K7 90 1800 1800K8 205 3100 3100UDd n/a 3700 3900

a Maximum elevation of deposit/moraine above current elevation of the base ofTaylor Glacier.

b Average of all clast exposure ages for each moraine/drift, in which a constanterosion rate of 10 cm Myr−1 is assumed for each sample.

c Preferred age model using oldest dated clast from each moraine/drift, assuming aconstant erosion rate of 10 cm Myr−1.

d Undifferentiated drift dated ~20 m distal to moraine K7. Drift is stratigraphicallybelow K7 drift.

70 K.M. Swanger et al. / Global and Planetary Change 79 (2011) 61–72

dates on a series of moraines in Vernier Valley (~25 km southeast ofKennar Valley) indicate gradual lowering of Ferrar Glacier (a secondoutlet glacier drainingTaylorDome) since themid-Pliocene (Staiger et al.,2006; Johnson and Staiger, 2007) (Fig. 1). Although these records sharesimilar first-order trends, they differ in the precise number and age ofindividualmoraines anddrifts; individual drifts cannot be correlatedwithcertainty from valley to valley. The variation likely arises from stochasticfactors associated with spatially variable and intermittent rockfall. Inaddition, temporal and spatial changes in rates of ice ablation, andresulting iceflow,might also influencemorainedeposition. Theunusuallylarge number of moraines in Arena Valley most likely reflects extensive

K1

K2K2

K3

K4K5K6

K7

K8

127150

18

560420

500590

18003500

1820

190320 220

1000

3900

1800430

120

96

89170

11580

290

260

3100

0 500 mN

250

UD

Fig. 7. Results from our preferred-age model (assuming 10 cm Myr−1 of erosion) for all27 exposure samples in Kennar Valley. Ages are listed in thousands of years (ka).

and persistent rockfall from exposed dolerite cliffs at the base of FingerMountain and/or potentially persistent higher-than-average ablationrates along the surface of Taylor Glacier ice at the mouth of Arena Valley(which may drive increased ice flow into the valley and result in morefrequent moraine formation).

5.3. Response of Taylor Dome to warmer-than-present conditions andthe mid-Pleistocene transition

The3.1 million-yearglacial record fromKennarValley (aswell as fromArena and Vernier valleys) implies overall ice recession for outlet glaciersdraining Taylor Dome since mid- to late-Pliocene time. In detail, theKennar Valley record indicates that the ice-surface elevation of TaylorGlacier, and hence of Taylor Dome, stood at higher-than-present levelsduring significant, globally warm intervals: the mid-Pliocene climaticoptimum(~3.0–3.1 Ma) andMIS 31 (~1.07 Ma). However, thesefindingscontrast with recent reports for significant reductions in the volume ofgrounded, marine-based ice from the WAIS in the Ross Embayment(Scherer et al., 2008; Naish et al., 2009; Pollard and DeConto, 2009; seealsoMiller et al., 2005). Thefindings call for considerable variability in theresponse of Antarctic ice to global climate change.

In addition, the combined glacier records from Kennar, Arena, andVernier valleys suggest that the rate of change accelerated at the mid-Pleistocene transition (or just shortly after), with both Taylor and Ferrarglaciers experiencing most (50–80%) of their total vertical recessionafter ~0.9 Ma (Table 3). One possible explanation calls on reduction insnowfall at Taylor Dome, which could reflect overall cooling ofatmospheric temperatures throughout the late Pleistocene (Lisieckiand Raymo, 2005) and/or northward displacement of openwater in theRoss Embayment associated with expanding sea ice and/or increasingfrequency ofWAIS expansion (Denton and Marchant, 2000; Steig et al.,2000; Grootes et al., 2001; Naish et al., 2009).

5.4. Implications for paleoclimate

The mapped drifts in Kennar Valley (as well as in nearby Arena andVernier valleys) were deposited from cold-based ice (Brook et al., 1993;Marchant et al., 1994; Staiger et al., 2006). Had wet-based conditionsoccurred, clasts within the Kennar Valley drifts would show evidence forglacial abrasion, including striations, polish, and faceting, which is not thecase (see alsoMarchant et al., 1994; Staiger et al., 2006). Also, had the icesurface experienced significant melting, outwash and/or stratifiedsediments would be commonplace (e.g., Denton et al., 1993). Instead,all moraines are texturally and morphologically identical to those foundtoday alongside cold-basedmargins of outlet and alpine glaciers that passacross the central and southern Transantarctic Mountains (Denton et al.,1989;Marchant et al., 1994; Staiger et al., 2006; Kowalewski et al., 2011).The implication is that climate conditions during moraine depositionwere similar to present-day conditions, andnot aswarmas those inferredfor the central Ross Embayment (Scherer et al., 2008; Naish et al., 2009).In addition, the presence of in-situ moraine ridges at the base of steepvalleywalls in allmapped valleys (K8 drift in Kennar Valley, for example)imply limited slope development for the last ~3.1 Myr. To be sure, arecord of morphologic change does exist, and is most notably expressedin the overall reduction in clast size on drifts K2 to K8 and a graduallowering of moraine heights with increasing exposure age (e.g., Morganet al., 2011).

6. Conclusions

The areal distribution of drifts in lower Kennar Valley, along with arelative and numerical chronology afforded by surface-weatheringcharacteristics and 3He exposure ages, call for overall thinning of upperTaylor Glacier over the last 3.1 Myr, although subtle readvances cannotbe precluded. At ~3.1 Ma, the margin of upper Taylor Glacier in KennarValley stood at ~1610 m elevation (~205 m higher than at present).

71K.M. Swanger et al. / Global and Planetary Change 79 (2011) 61–72

Between ~1.8 and ~0.56 Ma, a series of four closely spaced moraineswere deposited between ~1500- and 1490-m elevations (~85–95 mabove present Taylor Glacier). During the last ~0.32 Myr, the level ofupper TaylorGlacier in KennarValley retreated from~1475-melevation(~70 mabove present values) to its present value of ~1405 melevation.

Throughout all intervals of moraine deposition, the glacier marginremained cold-based (frozen to its bed) and lacked significant surfacemelting. There are no textural features within drifts to suggest clasttransport beneath wet-based ice (i.e., no striated, molded, or polishedclasts) and moraines lack associated outwash sediments. Theimplication is that climate conditions during drift deposition at thissite were essentially similar to modern conditions.

Comparison of our moraine record with published reports forfluctuations of Taylor Glacier elsewhere in the Quartermain Mountains,and with a dated moraine record from Ferrar Glacier (also sourced fromTaylor Dome), reveals consistent ice-surface changes, highlightingminor,but widespread ice recession in southern Victoria Land since the mid- tolate-Pliocene. The combined records show an atypical relationship withaverage global temperatures, with higher-than-present ice levels duringglobally warm periods, including the Pliocene climatic optimum (~3.0–3.1 Ma), MIS 31 (~1.0 Ma), and MIS 5.5 (~125 ka) (Brook et al., 1993;Marchant et al., 1994;Higgins et al., 2000; Staiger et al., 2006). TheKennarValley glacial record highlights the potentially complex and non-uniformresponse of Antarctic ice to climate change. The record suggests low-amplitude fluctuations of Taylor Dome during the last 3.1 Myr, whileoffshore sediment cores indicate considerable variability in the extent ofgrounded marine-based ice from the WAIS in the Ross Embayment(Fig. 1) (Naish et al., 2009).

Acknowledgements

We thank Douglas Kowalewski and David Shean for excellentassistance in the field and for insight regarding GPS data. We alsothank two anonymous reviewers whose comments helped improvethis manuscript greatly. Funding for this research was provided byNSF Polar Programs Grants ANT-1043706 and ANT-0944702 to DRMand ANT-1043724 to KMS.

References

Ackert Jr., R.P., Kurz, 2004. Age and uplift of Sirius Group sediments in the DominionRange, Antarctic, from surface exposure dating and geomorphology. Global andPlanetary Change 42, 207–225.

Atkins, C.B., Barrett, P.J., Hicock, S.R., 2002. Cold glaciers erode and deposit: evidencefrom Allan Hills, Antarctica. Geology 30, 659–662.

Balco, G., Stone, J.O., Lifton, N.A., Dunai, T., 2008. A complete and easily accessible meansof calculating surface exposure ages or erosion rates from 10Be and 26Almeasurements. Quaternary Geochronology 3, 174–195.

Barrett, P.J., Fitzgerald, 1986. Deposition of the lower Feather Conglomerate, a Permianbraided river deposit in southern Victoria Land, Antarctica, with notes on theregional paleogeography. Sedimentary Geology 45, 189–208.

Bentley, M.J., Fogwill, C.J., LeBrocq, A.M., Hubbard, A.L., Sugden, D.E., Dunai, T., Freeman,S.P.H.T., 2010. Deglacial history of the West Antarctic Ice Sheet in the Weddell Seaembayment: constraints on past ice volume change. Geology 38, 411–414.

Beyer, L., Bockheim, J.G., Campbell, I.B., Claridge, G.G.C., 1999. Review: genesis,properties and sensitivity of Antarctic gelisols. Antarctic Science 11 (4), 387–398.

Bockheim, J.G., 2010. Evolution of desert pavements and the vesicular layer in soils ofthe Transantarctic Mountains. Geomorphology 118, 433–443.

Brook, E.J., Kurz, M.D., Ackert Jr., R.P., Denton, G.H., Brown, E.T., Raisbeck, G.M., Yiou, F., 1993.Chronology of Taylor Glacier advance in Arena Valley, Antarctica, using in situcosmogenic 3He and 10Be. Quaternary Research 39, 11–23.

Brown, E.T., Edmond, J.M., Raisbeck, G.M., Yiou, F., Kurz, M., Brook, E.J., 1991.Examination of surface exposure ages of Antarctic moraines using in situ produced10Be and 26Al. Geochimica et Cosmochimica Acta 55, 2269–2283.

Bruno, L.A., Baur, H., Graf, T., Schluchter, C., Signer, P., Wieler, R., 1997. Dating of SiriusGroup tillites in the Antarctic Dry Valleys with cosmogenic 3He and 21Ne. Earth andPlanetary Science Letters 147, 37–54.

Chinn, T.J., 1980. Glacier balances in the Dry Valleys area, Victoria Land, Antarctica.Proceedings of the Riederalp workshop: AHS-AISH Publication, vol. 125, pp.237–247.

Clow, G.D., McKay, C.P., Simmons, G.M., Wharton Jr., R.A., 1988. Climatologicalobservations and predicted sublimation rates at Lake Hoare, Antarctica. Journalof Climate 1 (7), 715–728.

Cuffey, K.M., Conway, H., Gades, A.M., Hallet, B., Lorrain, R., Severinghaus, J.P., Steig,E.J., Vaughn, B., White, J.W.C., 2000. Entrainment at cold glacier beds. Geology 28,351–354.

Denton, G.H., Hughes, T.J., 1981. The Last Great Ice Sheets. Wiley, Chichester, New York,Brisbane, Toronto. 484 pp.

Denton,G.H., Bockheim, J.G.,Wilson,S.C., Stuiver,M., 1989. LateWisconsinandearlyHoloceneglacial history, inner Ross Embayment, Antarctica,. Quaternary Research 31, 151–182.

Denton, G.H., Sugden, D.E., Marchant, D.R., Hall, B.L., Wilch, T.I., 1993. East Antarctic IceSheet sensitivity to Pliocene climatic change from a Dry Valleys perspective.Geografiska Annaler 75A, 155–203.

Denton, G.H., Marchant, D.R., 2000. The geologic basis for a reconstruction of agrounded ice sheet in McMurdo Sound, Antarctica, at the last glacial maximum.Geografiska Annaler 82A, 167–211.

Doran, P.T., McKay, C.P., Clow, G.D., Dana, G.L., Fountain, A.G., Nylen, T., Lyons, W.B., 2002.Valley floor climate observations from the McMurdo dry valleys, Antarctica, 1986–2000. Journal of Geophysical Research 107, D244772. doi:10.1029/2001JD002045.

Drewry, D.J., 1982. Ice flow, bedrock, and geothermal studies from radio-echo soundinginland of McMurdo Sound, Antarctica. In: Cradock, C. (Ed.), Antarctic Geoscience.University of Wisconsin Press, Madison, pp. 977–983.

Elliot, D.H., Fleming, T.H., 2004. Occurrence and dispersal of magmas in the JurassicFerrar Large Igneous Province, Antarctica. Gondwana Research 7, 223–237.

Fountain, A.G., Nylen, T.H., Monaghan, A., Basagic, H.J., Bromwich, D., 2009. Snow in theMcMurdo Dry Valleys, Antarctica. International Journal of Climatology.doi:10.1002/joc.1933.

Goehring, B.M., Kurz,M.D., Balco, G., Schaefer, J.M., Licciardi, J., Lifton,N., 2010. A reevaluationof in situ cosmogenic 3He production rates. Quaternary Geochronology 5, 410–418.

Gosse, J.C., Phillips, F.M., 2001. Terrestrial in situ cosmogenic nuclides: theory andapplication. Quaternary Science Reviews 20, 1475–1560.

Grootes, P.M., Steig, E.J., Stuiver, M., Waddington, E.D., Morse, D.L., 2001. The TaylorDome Antarctic 18O record and globally synchronous changes in climate.Quaternary Research 56, 289–298.

Hallet, B.,Waddington, E.D., 1992. Buoyancy forces inducedby freeze–thaw in the active layer— implications for diapirism and soil circulation. In: Dixon, J.C., Abrahams, A.D. (Eds.),Periglacial Geomorphology: Proceedings of the 22ndAnnual Binghamton Symposium inGeomorphology. Wiley, New York, pp. 251–279.

Higgins, S.M., Hendy, C.H., Denton, G.H., 2000. Geochronology of Bonney Drift, TaylorValley, Antarctica: evidence for interglacial expansions of Taylor Glacier. GeografiskaAnnaler 82A, 391–409.

Ivy-Ochs, S., Schlüchter, C., Kubik, P.W., Dittrich-Hannen, B., Beer, J., 1995. Minimum10Be exposure ages of early Pliocene for the Table Mountain plateau and the SiriusGroup at Mount Fleming, Dry Valleys, Antarctica. Geology 23, 1007–1010.

Johnson, J.V., Staiger, J.W., 2007. Modeling long-term stability of the Ferrar Glacier, EastAntarctica: implications for interpreting cosmogenic nuclide inheritance. Journal ofGeophysical Research—Earth Surface 112. doi:10.1029/2006JF000599.

Kavanaugh, J.L., Cuffey, K.M., Morse, D.L., Conway, H., Rignot, E., 2009. Dynamics andmass balance of Taylor Glacier, Antarctica: 1. Geometry and surface velocities.Journal of Geophysical Research—Earth Surface 114. doi:10.1029/2009JF001309.

Kowalewski, D.E., Marchant, D.R., Levy, J.S., Head III, J.W., 2006. Quantifyingsummertime sublimation rates for buried glacier ice in Beacon Valleys, Antarctica.Antarctic Science 18, 421–428.

Kowalewski, D.E., Marchant, D.R., Swanger, K.M., Head III, J.W., 2011. Modeling vapordiffusion within cold and dry supraglacial tills of Antarctica: implication for thepreservation of ancient ice. Geomorphology. doi:10.1016/j.geomorph.2010.11.001.

Lal, 1991. Cosmic ray labeling of erosion surfaces: in situ nuclide production rates anderosion models. Earth and Planetary Science Letters 104, 424–439.

Lisiecki, L.E., Raymo, M.E., 2005. A Pliocene–Pleistocene stack of 57 globally distributedbenthic δ18O records. Paleoceanography 20, PA1003. doi:10.1029/2004PA001071.

Licciardi, J.M., Kurz, M.D., Clark, P.U., Brook, E.J., 1999. Calibration of cosmogenic 3Heproduction rates fromHolocene lava flows in Oregon, USA, and effects of the Earth'smagnetic field. Earth and Planetary Science Letters 172, 261–271.

Marchant, D.R., Denton, G.H., Bockheim, J.G., Wilson, S.C., Kerr, A.R., 1994. Quaternarychanges in level of the upper Taylor Glacier, Antarctica: implications forpaleoclimate and East Antarctic Ice Sheet dynamics. Boreas 23, 29–43.

Marchant, D.R., Lewis, A.R., Phillips, W.M., Moore, E.J., Souchez, R.A., Denton, G.H., Sugden,D.E., Potter Jr., N., Landis, G.P., 2002. Formations of patterned ground and sublimationtill over Miocene glacier ice in Beacon Valley, southern Victoria Land, Antarctica.Geological Society of America Bulletin 114, 718–730.

Marchant, D.R., Head III, J.W., 2007. Antarctic dry valleys: microclimate zonation,variable geomorphic processes, and implications for assessing climate change onMars. Icarus 192, 187–222.

Margerison, H.R., Phillips, W.M., Stuart, F.M., Sugden, D.E., 2005. Cosmogenic 3Heconcentrations in ancient flood deposits from the Coombs Hills, northern DryValleys, East Antarctica: interpreting exposure ages and erosion rates. Earth andPlanetary Science Letters 230, 163–175.

McElroy, C.T., Rose, G., 1987. Geology of the Beacon Heights Area Southern Victoria LandAntarctica, Miscellaneous Series Map 15, New Zealand Geological Survey. Dept. ofScientific and Industrial Research.

Miller, K.G., Kominz,M.A., Browning, J.V.,Wright, J.D.,Mountain, G.S., Katz,M.E., Sugarman,P.J., Cramer, B.S., Christie-Blcik, N., Pekar, S.F., 2005. The Phanerozoic record of globalsea-level change. Science 310, 1293. doi:10.1126/science.1116412.

Morgan, D., Putkonen, J., Balco, G., Stone, J., 2010. Quantifying regolith erosion rateswith cosmogenic nuclides 10Be and 26Al in the McMurdo Dry Valleys, Antarctica.Journal of Geophysical Research 115, F03037. doi:10.1029/2009JF001443.

Morgan, D.J., Putkonen, J., Balco, G., Stone, J., 2011. Degradation of glacial depositsquantified with cosmogenic nuclides, Quartermain Mountains, Antarctica. EarthSurface Processes and Landforms 36, 217–228.

72 K.M. Swanger et al. / Global and Planetary Change 79 (2011) 61–72

Naish, T., Powell, R., Levy, R., Wilson, G., Scherer, R., Talarico, F., Krissek, L., Niessen, F.,Pompilio, M., Wilson, T., Carter, L., DeConto, R., Huybers, P., McKay, R., Pollard, D.,Ross, J., Winter, D., Barrett, P., Browne, G., Cody, R., Cowan, E., Crampton, J., Dunbar,G., Dunbar, N., Florindo, F., Gebhardt, C., Graham, I., Hannah, M., Hansaraj, D.,Harwood, D., Helling, D., Henrys, S., Hinnov, L., Kuhn, G., Kyle, P., Laufer, A., Maffioli,P., Magens, D., Mandernack, K., McIntosh, W., Millan, C., Morin, R., Ohneiser, C.,Paulsen, T., Persico, D., Raine, I., Reed, J., Riesselman, C., Sagnotti, L., Schmitt, D.,Sjunnesko, C., Strong, P., Taviani, M., Vogel, S., Wilch, T., Williams, T., 2009.Obliquity-paced Pliocene West Antarctic ice sheet oscillations. Nature 458,322–329.

Pollard, D., DeConto, R.M., 2009. ModelingWest Antarctic ice sheet growth and collapsethrough the past five million years. Nature 458. doi:10.1038/nature07809.

Robinson, P.H., 1984. Ice dynamics and thermal regime of Taylor Glacier, South VictoriaLand, Antarctica. Journal of Glaciology 30, 153–160.

Sarda, P., Staudacher, T., Allegre, C.J., Lecomte, A., 1993. Cosmogenic neon and helium atReunion: measurement of erosion rate. Earth and Planetary Science Letters 119,405–417.

Schaefer, J.M., Ivy-Ochs, S., Weiler, R., Leya, I., Baur, H., Denton, G.H., Schlüchter, C., 1999.Cosmogenic noble gas studies in the oldest landscape on earth: surface exposure agesof the Dry Valleys, Antarctica. Earth and Planetary Science Letters 167, 215–226.

Schaefer, J.M., Baur, H., Denton,G.H., Ivy-Ochs, S.,Marchant, D.R., Schluchter, C.,Weisler, R.,2000. The oldest ice on Earth in Beacon Valley, Antarctica: new evidence from surfaceexposure dating. Earth and Planetary Science Letters 179, 91–99.

Schaefer, J.M., Faestermann, T., Herzong, G.F., Knie, K., Korschinek, G., Masarick, J.,Meier, A., Poutivtsev, M., Rugel, G., Schluchter, C., Serifiddin, F., Winckler, G., 2006.Terrestrial manganese-53 — a new monitor of Earth surface processes. Earth andPlanetary Science Letters 251, 334–345.

Scherer, R.P., Bohaty, S.M., Dunbar, R.B., Esper, O., Flores, J.A., Gersonde, R., Harwood,D.M., Roberts, A.P., Taviani, M., 2008. Antarctic record of precession-pacedinsolation-driven warming during early Pleistocene Marin Isotope Stage 31.Geophysical Research Letters 35, L03505. doi:10.1029/2007GL032254.

Staiger, J.W., Marchant, D.R., Schaefer, J.M., Oberholzer, P., Johnson, J.V., Lewis, A.R., Swanger,K.M., 2006. Plio-Pleistocene history of Ferrar Glacier, Antarctica: implications for climateand ice sheet stability. Earth and Planetary Science Letters 243, 489–503.

Steig, E.J., Morse, D.L., Waddington, E.D., Stuiver, M., Grootes, P.M., Mayewski, P.A., Twickler,M.S., Whitlow, S.I., 2000. Wisconsinan and Holocene climate history from an ice core atTaylorDome,WesternRoss Embayment, Antarctica. GeografiskaAnnaler 82A, 213–235.

Stone, J.O., 2000. Air pressure and cosmogenic isotope production. Journal ofGeophysical Research 105 (B10), 23,753–23,759.

Summerfield,M.A., Stuart, F.M., Cockburn,H.A.P., Sugden,D.E., Dunai, T.,Marchant, D.R., 1999.Long-term rates of denudation in the Dry Valleys, Transantarctic Mountains, southernVictoria Land, Antarctica based on in-situ-produced cosmogenic 21Ne. Geomorphology27, 113–129.

Swanger, K.M., Marchant, D.R., 2007. Sensitivity of ice-cemented Antarctic soils togreenhouse-induced thawing: are terrestrial archives at risk? Earth and PlanetaryScience Letters 259, 347–359.

Swanger, K.M., Marchant, D.R., Kowalewski, D.E., Head III, J.W., 2010. Viscous flow lobesin central Taylor Valley, Antarctica: origin as remnant buried glacial ice.Geomorphology. doi:10.1016/j.geomorph.2010.03.024.

Winckler, G., Anderson, R.F., Schlosser, P., 2005. Equatorial Pacific productivity and dustflux during theMid-Pleistocene climate transition, Paleoceanography, Vol. 20, No.4,PA4025, doi: 10.1029/2005PA001177.