519Journal of Paleolimnology 25: 519–529, 2001.© 2001 Kluwer Academic Publishers. Printed in the Netherlands.

Luminescence-dating zeroing tests in Lake Hoare, Taylor Valley,Antarctica

Glenn W. Berger1 & Peter T. Doran2

1Desert Research Institute, 2215 Raggio Parkway, Reno, NV 89512-1095, USA (E-mail: gwberger@dri.edu)2Department Earth & Environmental Sciences, University of Illinois at Chicago, 845 West Taylor St., Chicago,IL 60607-7059, USA

Received 21 October 1999; accepted 12 April 2000

Key words: Antarctica, geochronology, dating, lakes, sediments, luminescence

Abstract

In two of the perennially ice-covered lakes in the McMurdo Dry Valleys, lakes Hoare and Bonney in the TaylorValley, bottom water has 14C ages of ~2.7 ka and ~10 ka (respectively), rendering 14C ages of bottom sediments highlyproblematic. Consequently, we tested the effectiveness of thermoluminescence (TL) zeroing in polymineral fine siltmaterial from several depositional environments around and on the lake (stream suspensions, ice-surface sand dune,and silty sand from near the top of the more-than-3m-thick ice). We also conducted TL and infrared-stimulated-luminescence (IRSL) dating tests on material from three box cores recovered from the bottom of Lake Hoare, in atransect away from the abutting Canada Glacier. We observed effective zeroing of light-sensitive TL in suspendedsilt from one input stream and less effective zeroing from another stream. We observed effective zeroing of light-sensitive TL also in silt from a glacier-proximal eolian ice-surface dune and from sand from within the upper 5 cm ofice. In contrast, in box-core 1, the bottom sediment yielded minimum TL apparent ages of 1500–2600 yrs, with nodiscernable stratigraphic depth trend. IRSL dating applied to the same box-core samples produced significantlylower age estimates, ranging from ~600 ± 200 yrs to 1440 ± 270 yrs top-to-bottom, an improvement over the depth-constant ~2200 yrs TL ages. In two other cores closer to the Canada Glacier, IRSL ages from ~600 ± 200 yrs (top) to~2900 ± 300 yrs (at depth) were measured. Not only are the IRSL ages a significant improvement over the TL results,but the near-core-top IRSL ages are also a dramatic improvement over the 14C results (~2.7 ka). IRSL dating has ademonstrated potential to supplant 14C dating for such antarctic lacustrine deposits.

Introduction

The McMurdo Dry Valleys comprise the largest ice-freearea in Antarctica. Within these valleys are perenniallyfrozen lakes whose bottom deposits contain importantmodern and paleolimnological information (Doran et al.,1994). However, the ages of these deposits, and hencethe history of limnological changes in these lakes, havebeen very uncertain because 14C ages are affected by avariable carbon reservoir effect. In the marine andcoastal areas of Antarctica this reservoir effect hasmade 14C dating difficult, since the effect varies withlocation and sample type, but reasonably reliablecorrections can be obtained by systematic collectionand dating of modern carbon samples in a region (e.g.,

Domack et al., 1989; DeMaster et al., 1992; Gordon &Harness, 1992; Berkman & Forman, 1996). Correctionfor this effect in lacustrine sediments, however, is morecomplicated, if tractable at all (e.g., Zale & Karlén,1989; Melles et al., 1994; Gore, 1997).

Using accelerator-mass-spectrometric 14C analysis ofdissolved inorganic carbon compounds and particulateorganic carbon compounds from lake water (LakesBonney, Fryxell and Hoare), as well as 14C analysis ofstream and lake bottom organic material (microbial mat),and paleolake deposits, Doran (1996) and Doran et al.(1999) determined that although delta and lake-shoredeposits from such lakes have a small or negligiblereservoir effect, lake-bottom sediments have a variableand large effect. For example, bottom water in Lakes

520

Hoare and Bonney yielded 14C ages of ~2.7 ka and ~10ka, respectively. Thus if Lake Hoare drained today, its‘paleolake’ bottom sediments would have a relict 14C ageof ~2.7 ka. However, Doran et al. (1999) did derive alinear age-depth trend in their core, corresponding toan average sedimentation rate of ~0.015 cm/yr over asampling depth range of ~40 cm. Nevertheless, know-ledge of such a sedimentation rate is difficult toextrapolate to any 14C dating of subaerially exposed lake-bottom deposits, because unless the fossil sediment-water interface is preserved, older 14C ages cannot be‘corrected’ for any bottom-water effect.

Therefore, seeking a better dating method, we testedthe effectiveness of zeroing by daylight of light-sensitive luminescence in sediments from Lake Hoare,sampling sediments resulting from a representativevariety of modern depositional processes. The rough,jumbled topography of the ice surface traps wind-blownsand. By mass, most sedimentation occurs by graintransport through the meters-thick perennial ice cover(e.g., Nedell et al., 1987; Squyres et al., 1991; Simmonset al., 1993: Figure 17; Doran et al., 1994), with presumablya minor component coming from suspension rain outresulting from stream inflow around the lake edge (e.g.,Lawrence & Hendy, 1989). Sediment traps indicatespatially uniform accumulation of ‘mud’ (fine-silt orsmaller material), suggesting that most or all mud isintroduced by suspension, not through the ice coveras is coarse-grained material (Squyres et al., 1991;Andersen et al., 1993).

By a complicated, poorly understood process (e.g.,Squyres et al., 1991; Doran, 1996; Wand & Perlt, 1999),the thick ice restricts large clasts to the surface,allowing sand-sized and any ice-borne smaller particlesto migrate during austral spring-summer through smallcracks and fissures to the base of the ice, where theyare then released causing heterogeneous sedimentation.Linear sand ridges and conical sand mounds areobserved on the lake bottom. The sediment from thisspatially-restricted and rapid sedimentation is recolonizedby microbial mats, causing sediments in cores to consistof randomly alternating layers of microbial mat andorganic-poor sand. Indeed, ‘fine-grained sediment ismore abundant in mat layers than in sand layers,suggesting that the mat accumulates more slowly’(Andersen et al., 1993: 76). Analyses of sedimentstructure and stratigraphy in several Lake Hoare coresreveal no evidence of laterally consistent verticalsequences (Schackman, 1994). Sediments in the icecover and in the bottom deposits are mineralogicallyindistinguishable (Bishop et al., 1996), with quartz and

feldspar (including orthoclase and plagioclase) con-centrations of 30–40% (Nedell et al., 1987). Long-termsedimentation rates in Lake Hoare are estimated to be0.01 cm/yr (Squyres et al., 1991) to 0.015 cm/yr (Doran,1996: Doran et al., 1999), although local rates can varyby an order of magnitude over short periods (e.g., duringthe formation of mounds).

Most sand grains near the ice surface are expectedto be exposed to daylight prior to final release from theice and most silt grains in suspensions are likewiseexpected to be exposed to daylight, because grainmobility through ice and in suspensions can occur onlyduring the 4-month austral spring-summer. Interestingly,it is likely that even in the base of the thick ice grainswill be exposed to sufficient daylight to reduce somelight-sensitive luminescence. Although only 2–5% ofsurface irradiance penetrates the ice, a broader spectrum(~400–600 nm, peak ~480–580 nm) is passed throughthe ice (Palmisano & Simmons, 1987; Lizotte & Priscu,1992) than through some turbid suspensions in non-polar lakes (Berger, 1990). In summary, the depositionalconditions around Lake Hoare, at least, are expected topermit some zeroing of the luminescence sediment-dating clock.

Samples

During the austral summer 1993–1994 we collectedsamples from a variety of depositional settings in andaround Lake Hoare (Table 1 and Figure 1). Suspensionswere collected and transported in opaque, flexible 20-L containers. Here we report analyses of samplesrepresenting the variety of modern depositional pro-cesses and sedimentary end members. Limited results(without details) of thermoluminescence (TL) zeroingtests are reported by Doran (1996) and Doran et al.(1999). Here we present the details of these TL tests plusresults from infrared-stimulated-luminescence (IRSL)tests on several of the TL-tested samples, as well asothers. For box-core 1, the radiation dose rates and TLages reported by Doran et al. (1999) differ from thosereported here because they used only estimates ofpotassium values, derived from onshore sedimentsrather than the core itself. Furthermore, we use a largervalue for the estimated saturation water concentration.

Box cores 3 and 5 had been thawed, partly dried andcollapsed before we obtained funds to analyze them,thus preventing fine-scale subsampling. Subsamplingadequate amounts of material in dim amber laboratorylight precluded isolation of most microbial mats from

521

bracketing sand-rich zones. Thus, these samples, andto some extent those in core 1, provide at most a crudedepth profile.

Experimental procedures

The basics of TL and IRSL sediment dating have beenoutlined elsewhere (Aitken, 1985, 1998). Zeroing of light-sensitive TL requires daylight exposures from tens ofminutes to several hours, depending on several factors(Huntley, 1985; Berger, 1990). By its nature, IRSL lacksa light-insensitive component, and zeroing requires onlyminutes or tens of minutes of daylight. We employedboth TL and IRSL, and used fine-grain (4–11 µmdiameter), multi-aliquot methods (Aitken, 1985, 1998;Berger, 1990). The multi-aliquot approach assumes thatall grains have been exposed equally to daylight beforefinal burial. Carbonates and organic matter wereremoved by use of 1.0N HCl and 30% H

2O

2, respectively.

A luminescence age = equivalent-dose (DE)/effective

dose rate. For TL we used the partial-bleach (PB) andtotal-bleach (TB) variants to determine D

E values

(Aitken, 1985; Berger, 1988; Berger et al., 1994). For IRSLwe used the thermal-transfer, additive-dose protocol(Ollerhead et al., 1994), combined with the use of twopreheating experiments for some samples to isolatethermally stable signals. We applied no ‘short-shine’normalization (Aitken, 1998) to the pre-irradiated(natural) aliquots because the natural signals werelow. Furthermore, such normalization often has no

Table 1. Samples from Lake Hoare area

Sample Site code Deptha

(Fig. 1) (cm)

SuspensionPDR-70 6-Huey Creek –PDR-71 7-Anderson Creek –

Ice surfacePDR-16 8-East end, 0–4

~ 600 mb

Sand dunePDR-12 9-~ 50 mb ~ 36

Ekman box coresPDR-1A 1-333 mb 0–1, mat with siltPDR-1B 1-333 mb ~ 2–3, 1 cm silty sandPDR-1C 1-333 mb ~ 3.5–5.5, thin (~1 mm) silty matsPDR-1D 1-333 mb ~ 6–8.5, silty sand and matsPDR-3A 3-233 mb ~ 0–1, silty sand and thin matsPDR-3B 3-233 mb 0–1.5, silty sand and 2 thin matsPDR-5B 5-133 mb ~ 1.5–3.5, silty sand and 2 thin matsPDR-5C 5-133 mb ~ 3–5.5, silty sand and thin mat lenses

aDepth from top of deposit or depth interval within Ekmansample. Intervals within box cores 3 and 5 are approximate,because ‘bedding’ was not horizontal but concave up; bDistancefrom Canada Glacier.

Figure 1. Map showing: (a) the study area in Antarctica; (b) the Taylor Valley in the northern McMurdo Dry Valleys region; and(center) Lake Hoare (depth contours in meters) with samples sites (Table 1). The light stippled areas in Figure 1b represent glaciers.

522

beneficial effect for fine-silt IRSL dating. However,unlike Ollerhead et al., (1994) we used relatively narrowbandpass detection to isolate emissions centered about410 nm and 550 nm. Illustrative examples of thesemethods for determination of D

E values are given below.

Choice of laboratory added dose was guided byinferences from the theory of Berger et al. (1987).

The parameters needed to calculate the effective doserate are listed in Table 2 below, with methods used fordetermination of K

2O, U and Th stated in footnote a.

The dose rate was calculated from the equations anddose-rate conversion factors listed by Berger (1988). Inthese calculations, we assumed radioactive secularequilibrium, because there are little or no data ondisequilibrium in actual lake-core material. This as-sumption apparently is valid for luminescence datingin many sediments (Krbetschek et al., 1994) but not inall subaereally-exposed lake (Krbetschek et al., 1994),alluvial or fluvial sediments (Olley et al., 1996). Basedon the data of Olley et al. (1996), even in the unlikelysituation that some of the radioelements (mainly thosefrom the 238U decay series) in our samples exhibitedup to 50% disequilibrium, the calculated dose rate forlate-Holocene samples would be incorrect by at most~4% when thick-source-alpha-particle counting(TSAC) (Huntley et al., 1986) is used. Furthermore,

sealed-chamber TSAC counting indicated no sig-nificant radon loss from the dried sample powders inthis study.

We chose fine-silt-size grains for two reasons. Firstly,as outlined elsewhere (Berger, 1985a, b, 1988), fine-siltgrains in most waterlain deposits are likely to have hadlonger transport times (hence opportunity for daylightexposure) than sand-sized grains. In most lake sedimentsmany of the sand-sized grains could be deposited byhigh-energy processes (e.g., bottom flows from slumpingdelta fronts), precluding adequate exposure to light.However, in perennially ice covered lakes such as LakeHoare most sand-sized grains are expected to have beenadequately exposed to daylight while on and in the thickice cover, before final deposition. Nevertheless, thepossibility remains that a fraction of the bottom sandgrains in Lake Hoare was transported by slumping fromover-steepened shoreline deposits (e.g., a delta front),and the significance of this possibility for luminescencedating of lake-bottom sediments in Taylor Valley remainsa topic for future research. Secondly, we did not haveprotocols established (e.g., radiation-source calibrations,sample preparation procedures, etc.) for dating sandgrains, nor did we have funding to adapt and apply, inparallel with fine-grain protocols, such protocols to thesesamples. We consider that our experiments are only a first

Table 2. Dosimetry data

Sample K2Oa C t

b C t hb Th U b valuec Fγ

d Dose ratee

wt. % ks–1·cm–2 ks–1·cm–2 ppm ppm pGy·m2 Gy/kyr(± 0.05)

Ice surfacePDR-16 2.57 0.3286 ± 0.0068 0.157 ± 0.021 4.22 ± 0.56 1.34 ± 0.17 1.0 ± 0.2 0.6 ± 0.1 2.35 ± 0.37

Sand dunePDR-12 2.24 0.3255 ± 0.0066 0.169 ± 0.021 4.54 ± 0.57 1.23 ± 0.17 1.0 ± 0.2 1.0 ± 0.0 2.91 ± 0.17

Ekman box corePDR-1A 1.68 0.5649 ± 0.0093 0.276 ± 0.030 7.43 ± 0.81 2.25 ± 0.25 1.69 ± 0.22 0.6 ± 0.1 2.71 ± 0.19PDR-1B n m n m n m n m n m 0.84 ± 0.11 0.6 ± 0.1 2.25 ± 0.14PDR-1C n m n m n m n m n m 1.130 ± 0.089 0.7 ± 0.1 2.48 ± 0.15PDR-1D n m n m n m n m n m 0.89 ± 0.11 0.7 ± 0.1 2.35 ± 0.14PDR-3 2.16 0.524 ± 0.011 0.302 ± 0.037 8.13 ± 0.99 1.73 ± 0.30 1.0 ± 0.2 0.6 ± 0.1 2.53 ± 0.17PDR-5 2.06 0.3981 ± 0.0089 0.215 ± 0.029 5.76 ± 0.77 1.43 ± 0.23 1.0 ± 0.2 0.6 ± 0.1 2.20 ± 0.14

aFor box core 1, K2O and TSAC data were obtained only for sample 1A, whereas for cores 3 and 5 these data were obtained for asubsample representative of the entire core. nm – not measured (see text); bTotal and thorium count rates from finely powderedsamples for TSAC method (Huntley & Wintle, 1981; Huntley et al., 1986). Cu = Ct – Cth. Uncertainties here and elsewhere are ± 1σ;cThe alpha effectiveness factor (Berger, 1988; Huntley et al., 1988), measured only for box-core 1 and suspension-sample PDR-70.For the latter, b = 1.29 ± 0.14 pGy·m2. Values of 1.0 ± 0.2 are assumed; dThe fraction of the infinite-matrix γ dose rate, estimatedfrom (Aitken, 1985; Figure 4.3); eCalculated with the conversion factors and equations given by Berger (1988), and includes anestimated cosmic ray component of 0.13 ± 0.03 Gy/kyr. A water-weight/dry-sample-weight ratio of 0.30 ± 0.05 is assumed for allsamples (which are sandy), except for surface dune sample PDR-12, for which a value of 0.10 ± 0.05 is assumed.

523

step to proper luminescence analyses in the TaylorValley.

While this project was underway, Krause et al. (1997)reported their multi-aliquot IRSL results for lake depositsfrom the opposite side of Antarctica. Unlike here, theyused ~100–200 µm diameter plagioclase grains separatedfrom sandy horizons within three lake cores. Unlike here,their samples were older than the 14C dating limit, andso lacked independent age checks. Our oldest samplesare known to be younger than 1–2 ka. Furthermore,their IRSL experiments produced age reversals in twoof three cores. However, Krause et al. (1997) did isolateboth 410 nm and 550 nm emissions, and demonstratedthe rapid bleaching (to effective zero within ~100 s!) ofthe 550 nm plagioclase emission. Consequently, in ourproject we also tested the utility of 550 nm emissionsfor dating, but without IRSL bleaching-response tests.For the upper four samples in Table 1 we performed TLbleaching-response tests with relatively broadbanddetection centered at 410 nm emissions. For these testswe used equivalent subsets of aliquots and twodifferent (overlapping) spectral regions of visible lightfor bleaching (a double-wavelength bleaching test,Berger, 1990). These same spectral regions were notused for all samples in the determination of D

E values.

Further details of our TL and IRSL procedures are givenin footnotes to Tables 2 and 3 below.

Results and discussion

Dosimetry data and calculated dose rates are listed inTable 2, and TL and IRSL D

E values are listed in Table

3. Our K2O concentrations and Th/U ratios are similar

to those measured along and between cores by Bishopet al. (1996).

Suspensions

Following the results and arguments of Berger (1990),we think that double-wavelength bleaching-responseexperiments could help distinguish samples dominatedby grains carried in surface plumes or streams, fromsamples dominated by grains carried in turbid bottomor interlayer suspensions. Berger (1990: Figure 21)showed that unlike his suspension samples and onesurface sample of eolian silt, two samples from laminatedlake sediments evinced a clear distinction in bleachingresponse when blue-red bleaching was employedcompared to when orange-red bleaching was used.

Portions of these two laminated samples were likelydeposited from bottom and interlayer suspensions. Theambient light in bottom and interlayer suspensionsusually has most light with wavelengths shorter than~500 nm removed by adsorption and scattering ofphotons (Berger, 1990). Our suspension samplesPDR-70 and PDR-71 are from surface streams, so basedon the results from Berger (1990) we expected nodistinction in bleaching response when using the twodifferent (overlapping) bleaching spectra. Accordingly,we observed (Figures 2C & D) no resolvable distinctionsassociated with different bleaching spectra. However,the evident decrease in TL at high bleaching valuesfor sample PDR-71 demonstrates that light-sensitiveTL in sample PDR-71 was incompletely zeroed, where-as the results from sample PDR-70 (Figure 2C) suggesteffective zeroing of light-sensitive TL for that sample.

Sample PDR-71 is from Anderson Creek whichpasses along the edge of the Canada Glacier (Figure 1:site 7). Other luminescence studies of feldspar-dominant,polymineral, waterborne silt (Forman, 1990; Gemmell,1997, 1999) have shown that ice-proximal suspensionscarry a large fraction of grains not exposed to daylight,therefore yielding much light-sensitive luminescencein the laboratory. Large ‘relict’ light sensitive signalswere also observed in sand-sized glaciofluvial quartzgrains (Rhodes & Bailey, 1997). Unlike sample PDR-71, sample PDR-70 is from Huey Creek (Figure 1: site6) that flows for a few km distal to any glacier. Thissituation appears to provide ample opportunity foreach fine-silt-sized grain to be exposed to daylightbefore entering a lake (e.g., Berger, 1990). Thus thesebleaching-response tests suggest that silt from ice-distal streams entering the Taylor Valley lakes will havewell-zeroed light-sensitive TL (at 410 nm emissions)and silt from glacier-bordering streams will not.Equivalent-dose results (Table 3) for these two sus-pensions are not statistically different because of poorinter-aliquot reproducibility in the PDR-71 experiment(a different batch of aliquots was used for the bleaching-response tests).

Ice surface and sand dune

The fine-silt grains from the near surface of the icematrix (sample PDR-16, Table 1) preserve no signifi-cant light-sensitive TL, and the double-wavelengthbleaching-response test (Figure 2A) does not dis-tinguish differential bleaching, consistent with aneolian origin for these grains. Likewise, the partial-

524

Table 3. TL and IRSL equivalent doses and calculated ages

Sample Modea Bleachb Preheatc D Ed Temp./timee A g e f

( G y ) (y r s )

SuspensionPDR-70 T L PB/550 (238) 75 °C (7d) 0.4 ± 3.6 210–320 °C –PDR-71 T L PB/550 (328) 75 °C (7d) 5.4 ± 3.6 210–320 °C –

Ice surfacePDR-16 T L PB/550 (328) 75 °C (7d) –1.4 ± 2.6 210–320 °C –600 ± 1100

TB/430 (78) 75 °C (7d) –1.1 ± 1.4 230–370 °C –470 ± 600

Ekman box coresPDR-1A T L PB/550 (219) 75 °C (7d) 7.1 ± 3.5 240–320 °C 2600 ± 1300

IRSL/410 TB/780 (550) 150 °C (2d) 1.68 ± 0.82 1–10s 620 ± 310IRSL/550 TB/780 (470) 130 °C (2d) 1.45 ± 0.47 1–15s 540 ± 190

PDR-1B T L PB/550 (219) 75 °C (7d) 3.3 ± 1.8 190–290 °C 1470 ± 800PDR-1C T L PB/550 (219) 75 °C (7d) 5.1 ± 1.4 190–300 °C 2050 ± 570

TL TB/430 (117) 75 °C (7d) 5.27 ± 0.98 230–310 °C 2120 ± 400T L TB/Hg (1470) 75 °C (7d) 27.4 ± 2.1 260–340 °C 11000 ± 1000IRSL/410 TB/780 (550) 130 °C (2d) 3.64 ± 0.50 1–45s

TB/780 (550) 150 °C (2d) 4.56 ± 0.99 1–40sWeighted mean IRSL/410 DE = 3.83 ± 0.45 1540 ± 200

IRSL/550 TB/780 (470) 130 °C (2d) 2.69 ± 0.41 1–10s 1080 ± 180PDR-1D T L TB/430 (117) 75 °C (7d) 6.0 ± 1.5 270–360 °C 2550 ± 650

IRSL/410 TB/780 (550) 130 °C (2d) 3.40 ± 0.60 1–15s 1440 ± 270PDR-3A IRSL/410 TB/780 (550) 130 °C (2d) 1.72 ± 0.32 1–30s

TB/780 (550) 150 °C (2d) 2.03 ± 0.56 1–30sWeighted mean IRSL/410 DE = 1.80 ± 0.28 710 ± 120

IRSL/550 TB/780 (470) 130 °C (2d) 1.11 ± 0.91 1–25s 440 ± 360PDR-3B IRSL/410 TB/780 (550) 130 °C (2d) 0.91 ± 0.59 1–35s

IRSL/410 TB/780 (550) 150 °C (2d) 0.95 ± 0.66 1–20sWeighted mean IRSL/410 DE = 0.93 ± 0.44 370 ± 170

IRSL/550 TB/780 (470) 130 °C (2d) 0.60 ± 0.47 1–10s 240 ± 190PDR-5A IRSL/410 TB/780 (550) 130 °C (2d) 1.36 ± 0.34 1–15s

TB/780 (550) 150 °C (2d) 1.89 ± 0.92 1–20sWeighted mean IRSL/410 DE = 1.42 ± 0.32 650 ± 150

IRSL/550 TB/780 (470) 130 °C (2d) 0.75 ± 0.42 1–10s 340 ± 190PDR-5B IRSL/410 TB/780 (550) 130 °C (2d) 1.79 ± 0.68 1–20s 810 ± 320

IRSL/410 TB/780 (550) 150 °C (2d) 1.2 ± 1.1 1–15s –IRSL/550 TB/780 (550) 130 °C (2d) 1.36 ± 0.61 1–5s 620 ± 280

PDR-5C IRSL/410 TB/780 (550) 130 °C (2d) 5.99 ± 0.65 5–15sTB/780 (550) 150 °C (2d) 7.17 ± 0.93 5–20sWeighted mean IRSL/410 DE = 6.38 ± 0.55 2900 ± 300

IRSL/550 TB/780 (470) 130 °C (2d) 4.11 ± 0.67 1–15s 1870 ± 320

aFor all TL runs, the heating rate was 5C°/sec. TL was detected in the spectral region 350–500 nm at 1% cut (Kopp 5–60 optical glassfilter). For IRSL: 410 means detection of the emission band at ~ 410 nm using a filtered spectral region of 390–470 nm at 1% cut;550 means detection of the emission band at ~ 550 nm, using a filtered spectral region of 510–600 nm; bFormat xx/yyy (zzz) asfollows: ‘xx’ specifies PB (‘partial-bleach [R-β]’) or TB (‘total-bleach’) protocols; ‘yyy’ specifies the approximate lower wavelengthbound of the bleaching light (430 means 430–780 nm wavelengths passed at 10% cut; 550 means 550–800 nm wavelengths passedat 10% cut; 780 means > 780 nm solar spectrum passed; Hg means an unfiltered Hg-vapor ‘sunlamp’); ‘zzz’ specifies the measured(by radiometer) laboratory or solar bleaching fluence (J/cm2). For TL, these choices largely follow earlier practice (Berger et al.,1994); cThe chosen pre-readout heating for TL (see Berger, 1994) and preheating for IRSL (see Ollerhead et al., 1994); dWeightedmean equivalent dose plus average error over temperature or time interval in next column. Either of two weighted regressionmodels (straight line or quadratic; Berger et al., 1987) was employed, as appropriate for individual samples (see Figure 3 caption).Straight line regression was used for all IRSL experiments except the ‘410’ experiments on sample PDR-3A; eThe readouttemperature (TL) or readout time (IRSL) interval for which DE is calculated. For IRSL the typical chosen diode-current settingcorresponds to ~ 35 mW/cm2 at the sample; fFor all samples in box cores 3 and 5, the corresponding dose rates in Table 2 wereassumed for age calculation. A weighted mean DE was not calculated for sample PDR-5B because the precision of the DE value for the150 °C preheating experiment is relatively poor. The calculated error in Weighted mean the standard error of the mean.

525

bleach and total-bleach DE values (Table 3) are not

resolvable from zero. Clearly, such grains yield well-zeroed light-sensitive TL and, if they dominated thebottom sediment, would promise no relict TL age in lake-bottom sediment. The poor reproducibility of TL in thebleaching-response test for the sand dune sample PDR-12 (Figure 2B) precludes any clear statement from thisexperiment. We did not perform a D

E experiment for this

sand-dune sample because of insufficient fine-siltmaterial, but we would expect that because this sampleis eolian then all grain-size fractions would have producedeffectively-zero D

E values from TL experiments.

Box cores

Because we assumed that the dose rate calculated forthe uppermost core 1 sample (Table 1) is the same forall other core 1 samples, then the comments below onthe apparent age-depth trend for this core should beconsidered as tentative. On the other hand, the dose-rate data for cores 3 and 5 are more representative ofthe respective cores. In any case, in this dating-feasibility study the efficacy of our results dependsmainly on the relative D

E values for IRSL and TL for

each sample, so that the consequences of use of theselimited dose-rate data are not severe. Clearly, follow-up work will require more detailed IRSL and dose-ratemeasurements.

Samples from core 1 yielded TL apparent ages around2.2 ka with no depth trend (Table 3). Although this‘relict’ TL age is perhaps lower than the ~2.7 ka reservoireffect for 14C (Doran et al., 1999), the relatively lowprecision and lack of a depth trend discourages futureTL dating of such lake-bottom sediments younger than10–15 ka. An illustration of the PB and TB TL protocolsapplied to core sample PDR-1C is given in Figure 3.The low precision in D

E values arises from the need

to subtract a large TL signal (‘N+PB+75 °C’ or‘N+TB1+75 °C’) from another large signal (‘N+75 °C’)(Figure 3A). Although light-sensitive TL is a largefraction of natural TL (difference between N+75 °C andN+TB2+75 °C), only a small fraction of this light-sensitive TL in this sample is post-depositional (dif-ference between N+75 °C and N+PB+75 °C). Thisillustrates the perennial limitation of TL for dating late-Holocene-age waterlain sediment: the ubiquity of asubstantial light-sensitive TL within the pre-de-positional signal (Berger, 1985a, b, 1988, 1990). That

Figure 2. Plots of TL peak intensity vs. laboratory bleaching fluence for two ice-surface samples (A and B) and two fluvialsuspensions (C and D). Bleaching was conducted under two spectral conditions: wavelengths 435–750 nm (‘435’), and wavelengths550–800 nm (‘550’) (see Berger, 1990). For reference, 103 J/cm2 on the X-axis has about the same effect as ~4.3 h of direct sunlightat 65 mW/cm2 (e.g., Berger et al., 1994: Figure 5).

526

is, light-sensitive TL is not readily zeroed in waterlaindeposits, and restricted laboratory bleaching is requiredto tease out the post-depositional, light-sensitive TL.The trade-off is a poor precision in D

E values.

Partly to improve the precision of DE measurements

and particularly to avoid measuring anything but light-sensitive luminescence, we applied IRSL dating pro-cedures to the samples from core 1 and also to additionalsamples from cores 3 and 5 (Table 1, Figure. 1). Examplesof the IRSL ‘shine’ curves for sample PDR-1C are shownin Figure 4B. The count rates (after the 2-day 130 °Cpreheating) are low, but the resolution of a post-depositional signal is much higher (first 10 s of IRSL)than in the corresponding TL readout curves (Figure3A). For this sample, the resultant mean D

E value of 3.83

± 0.45 Gy is not statistically different from the DE value

of 5.27 ± 0.98 Gy from the TL experiment (Table 3), butthe precision is better. Interestingly, the D

E value of 2.69

± 0.41 Gy from the 550-nm experiment is apparentlylower (at 1σ) than that for the 410-nm IRSL experiment.

The IRSL age estimates in Table 3 for core 1 are lowerthan corresponding TL ages and also produce an

apparent depth trend (Figure 5, left). No depth trend isevident in core 3, but one appears again in core 5. Forillustrative purposes and assuming a relict IRSL ageof ~600 yrs for core-top feldspars (Table 3), we show(Figure 5) for core 1 a line (dashed) of possible IRSLages based on a prior estimate of sedimentation ratesin Lake Hoare. Since the average upper-core minimumIRSL age of ~600 yrs represents the top 1–1.5 cm ofsediment (Table 1) and not the top mm or so, the truerelict IRSL age of any ‘core-top’ material could bemuch lower. For example, we cannot know (withoutadditional, more sensitive experiments) if the grainsat the sediment-water interface were deposited withinthe last 50 yrs or only about 500 yrs ago. This am-biguity exists because sedimentation in this lakeoccurs mainly by episodic release through the icecover. Consequently, we would not expect in everycore a linear age-depth trend over such a short depthscale. It is important to recall that the approximatelylinear depth trend in 14C ages reported by Doran et al.(1994) represents more widely spaced core-samplingintervals over ~40 cm of deposition at only one site.

Figure 3. A) – Representative TL ‘glow curves’ for core-1 sample PDR-1C. Note that the dramatic difference between theN+TB1+75 °C and N+TB2+75 °C glow curves is not entirely due to the large differences in bleaching fluence (Table 3). Rather, asshown by Berger (1985a) and summarized in Figure 7 of Berger (1988), the inclusion of high-energy blue-ultraviolet photons in theHg lamp spectrum (used for TB2 bleaching) is the critical variable. B) – Dose-response curves for partial-bleach (PB) (intersectionof two curves) and total-bleach (TB) (intersection of upper curve with N+TB1 intensity line) protocols. Bleaching conditions aregiven in footnote b of Table 3. Weighted quadratic regression (Berger et al., 1987) was used for these data because the dose-responsecurves are significantly supralinear (e.g., Berger, 1990: Figure 15). The inset in (B) shows an expanded view of the region of curveintersection. C) – plot of DE vs. readout temperature. Error bars are ± 1σ. For clarity of viewing, the lower half of the symmetricalerror bars for PB data are omitted, and the PB data points are shifted +5C °. Bleaching conditions TB1 and TB2 correspond to ‘TB/430’ and ‘TB/Hg’, respectively, in Table 3.

527

Without additional experiments we can speculate nofurther on the chronological significance of anycomparison between these IRSL ages and such linearage-depth trends.

Finally, there appears to be a small but systematicdifference between the 410-nm and 550-nm resultsfrom some samples (Figure 5). Future, careful laboratory‘fading’ tests (e.g., Aitken, 1998) could elucidatewhether this apparent difference manifests the ex-istence of dissimilar thermal stabilities (hence signallifetimes) between these two emissions, or whether thisapparent difference reflects a distinction in theeffectiveness of signal zeroing during deposition. Forsand-sized plagioclase separates, Krause et al. (1997)stated that the 550-nm emissions were both thermallystable (in their antarctic setting) and much moresensitive to zeroing by daylight than are 410-nmemissions. It has not escaped our attention that if thedifferences in Figure 5 manifest mainly differencesin zeroing efficiency, then the ages from the 550-nmemissions would be more accurate and the relativedivergence of the two age estimates may then indicatewhich samples contain a significant fraction of grainshaving poorly zeroed light-sensitive IRSL, that is,which samples have been deposited most rapidly.

Conclusions

Considering the limited dose-rate data for the boxcores, the luminescence results suggest that in thesecores a relict IRSL age of ~600 yrs or less exists, andthat for core 1 the IRSL multi-aliquot approach canresolve an age-depth trend, unlike the less sensitiveTL method. Further experiments are needed on smaller(1–3 mm thick) core-top samples to determine moreaccurate relict IRSL ages. The relict IRSL ages (<~ 0.6ka) are much smaller than the core-top TL and 14C ages(~2.2 ka and ~2.7 ka respectively). Additional ex-periments on longer cores from more than one site areneeded to determine if IRSL age-depth trends can beresolved elsewhere in Lake Hoare (and similar lakes).Additional tests of the 410-nm and 550-nm emissionsare needed to ascertain the causes of the apparentdifferences in Figure 5.

The relatively small observed relict IRSL DE values

from modern day core tops could be subtracted from DE

values for deeper core samples to yield (in future IRSLapplications) accurate ages at depth. Furthermore, IRSLwould seem to be well suited to dating subaeriallyexposed (>∼ 5 ka) paleolake-bottom sediments becausesuch small relict D

E values would then be statistically

Figure 4. A) – Weighted-line-regression IRSL dose-response curves for core-1 sample PDR-1C. B) – Representative IRSL ‘shine’curves for the same. C) – DE vs. time plots for emissions near 410 nm. For clarity of viewing, the 150 °C data have been shifted+5C °. D) – DE vs. time plot for emissions near 550 nm. The most precise DE values represent the earliest signal release, where thesignal is greatest.

528

negligible. Finally, refined application of IRSL datingmight resolve non-uniform deposition rates (non-lineardepth trends) in present lake-bottom deposits, as ex-pected from the nature of sedimentation in Lake Hoare.Such future application might also answer the questionof whether Lake Hoare dried out ~1200 yrs ago (Spauldinget al., 1997; Lyons et al., 1998, 1999; Tyler et al., 1998;Doran et al., 1999 ).

Acknowledgments

Desert Research Institute’s former Quaternary SciencesCenter funded the TL work, while the U.S. NationalScience Foundation’s Office of Polar Programs fundedthe IRSL work through a subcontract to GWB from theUniversity of Alabama. We thank D. Huntley and ananonymous reviewer for their very helpful comments.

References

Aitken, M. J., 1985. Thermoluminescence Dating. AcademicPress, N.Y., 351 pp.

Aitken, M. J., 1998. An Introduction to Optical Dating, OxfordUniversity Press, Oxford, 262 pp.

Andersen, D. W., R. A. Wharton Jr. & S. W. Squyres, 1993.Terrigenous clastic sedimentation in antarctic Dry Valleylakes. In Green, W. J. & E. I. Freidmann (eds), Physicaland Biogeochemical Processes in Antarctic Lakes. Ant.Res. Ser., Am. Geophys. Un., Wash. D.C. 59: 71–81.

Berger, G. W., 1985a. Thermoluminescence dating studies ofrapidly deposited silts from south-central British Columbia.Can. J. Earth Sci. 22: 704–710.

Berger, G. W., 1985b. Thermoluminescence dating applied to athin winter varve of the Late Glacial South ThompsonSilt, south-central British Columbia. Can. J. Earth Sci. 22:1736–1739.

Berger, G. W., 1988. Dating Quaternary events by luminescence.In Easterbrook, D. J. (ed.), Dating Quaternary sediments.Geological Society of America, Special Paper 227: 13–50.

Berger, G. W., 1990. Effectiveness of natural zeroing of thethermoluminescence in sediments. J. Geophys. Res. 95:12,375–12,397.

Berger, G. W., 1994. Thermoluminescence dating of sedimentsolder than ~100 ka. Quat. Sci. Rev. 13: 445–455.

Berger, G. W., R. A. Lockhart & J. Kuo, 1987. Regression anderror analysis applied to the dose response curves inthermoluminescence dating. Nucl. Tracks Radiat. Meas.13: 177–184.

Berger, G. W., B. J. Pillans & A. S. Palmer, 1994. Test ofthermoluminescence dating of loess from New Zealandand Alaska. Quat. Sci. Rev. 13: 309–333.

Berkman, P. A. & S. L. Forman, 1996. Pre-bomb radiocarbonand the reservoir correction for calcareous marine speciesin the Southern Ocean. Geophys. Res. Lett. 23: 363–366.

Bishop, J. L., C. Koeberl, C. Kralick, H. Fröschl, P. A. J. Englert,D. W. Andersen, C. M. Pieters & R. A. Wharton Jr., 1996.Reflectance spectroscopy and geochemical analyses of LakeHoare sediments, Antarctica: Implications for remote sensingof the Earth and Mars. J. Geophys. Res. 60: 765–785.

DeMaster, D. J., R. B. Dunbar, L. I. Gordon, A. R. Leventer, J.M. Morrison, D. M. Nelson, C. A. Nittrouer & W. O. SmithJr., 1992. Cycling and accumulation of biogenic silica andorganic matter in high-latitude environments: the RossSea. Oceanography 5: 146–153.

Figure 5. Plots of estimated TL and IRSL ages for each of the three box cores, with distance from the face of the Canada Glacierin parentheses. Also shown in the leftmost plot is a hypothetical age-depth dashed line, based on an estimated average sedimentationrate of 0.01 cm/yr (Squyres et al., 1991). The base of core 1 is the base of that figure segment. Each depth point is the midpoint ofeach 1–2 cm thick sample (Table 1). Data for each depth point are offset vertically, for clarity of viewing.

529

Domack, E. W., A. J. T. Jull, J. B. Anderson, T. W. Linick & C.R. Williams, 1989. Application of tandem accelerator mass-spectrometer dating to Late Pleistocene-Holocene sed-iments of the East Antarctic continental shelf. Quat. Res.31: 277–287.

Doran, P. T., 1996. Paleolimnology of Perennially Ice-coveredAntarctic Oasis Lakes. Ph.D. Dissertation, University ofNevada, Reno.

Doran, P. T., R. A. Wharton Jr. & W. B. Lyons, 1994.Paleolimnology of the McMurdo Dry Valleys, Antarctica.J. Paleolim. 10: 85–114.

Doran, P. T., G. W. Berger, W. B. Lyons, R. A. Wharton Jr., M.L. Davisson, J. Southon & J. E. Dibb, 1999. DatingQuaternary lacustrine sediments in the McMurdo DryValleys, Antarctica. Palaeogeogr. Palaeoclimatol. Palaeo-ecol. 147: 223–239.

Forman, S. L., 1990. Thermoluminescence properties of fiordsediments from Engelskbukta, western Spitsbergen,Svalbard: A new tool for deciphering depositionalenvironment?. Sedimentology 37: 377–384.

Gemmell, A. M. D., 1997. Fluctuations in the thermoluminescencesignal of suspended sediment in an Alpine glacial meltwaterstream. Quat. Sci. Rev. 16: 281–290.

Gemmell, A. M. D., 1999. IRSL from fine-grained glaciofluvialsediment. Quat. Sci. Rev. 18: 207–215.

Gordon, J. E. & D. D. Harkness, 1992. Magnitude and geographicvariation of the radiocarbon content in Antarctic marinelife: Implications for reservoir corrections in radiocarbondating. Quat. Sci. Rev. 11: 697–708.

Gore, D. B., 1997. Blanketing snow and ice: constraints onradiocarbon dating deglaciation in East Antarctic oases.Antarctic Sci. 9: 336–346.

Huntley, D. J., 1985. On the zeroing of the thermoluminescenceof sediments. Phys. Chem. Min. 12: 122–127.

Huntley, D. J. & A. G. Wintle, 1981. The use of alpha scintillationcounting for measuring Th-230 and Pa-231 contents ofocean sediments. Can. J. Earth Sci. 18: 419–432.

Huntley, D. J., M. K. Nissen, J. Thompson & S. E. Calvert, 1986.An improved alpha scintillation counting method fordetermination of Th, U, Ra-226, Th-230 excess and Pa-231excess in marine sediments. Can. J. Earth Sci. 23: 959–969.

Huntley, D. J., G. W. Berger & S. G. E. Bowman, 1988.Thermoluminescence responses to alpha and betairradiations, and age determination when the high doseresponse is non-linear. Radiat. Effects 105: 279–284.

Krause, W. E., M. R. Krbetschek & W. Stolz, 1997. Dating ofQuaternary lake sediments from the Shirmacher Oasis (eastAntarctica) by infra-red stimulated luminescence (IRSL)detected at the wavelength of 560 nm. Quat. Sci. Rev. 16:387–392.

Krbetschek, M. R., U. Rieser, L. Zöller & J. Heinicke, 1994.Radioactive disequilibria in palaeodosimetric dating ofsediments. Radiat. Meas. 23: 485–489.

Lawrence, M. J. F. & C. H. Hendy, 1989. Carbonate depositionand the Ross Sea ice advance, Fryxell basin, Taylor Valley,Antarctica. N. Z. J. Geol. Geophys. 32: 267–277.

Lizotte, M. P. & J. C. Priscu, 1992. Spectral irradiance and bio-optical properties in perennially ice-covered lakes of thedry valleys (McMurdo Sound, Antarctica). In Elliot, D.E. (ed.), Contributions to Antarctic Research III. Ant. Res.

Ser., Am. Geophys. Un., Wash. D.C. 57: 1–14.Lyons, W. B., S. K. Frape & K. A. Welch, 1999. History of

McMurdo Dry Valley lakes, Antarctica, from stable chlorineisotope data. Geology 27: 527–530.

Lyons, W. B., S. W. Tyler, R. A. Wharton Jr., D. M. McKnight& B. H. Vaughn, 1998. A Late Holocene dessication oflake Hoare and Lake Fryxell, McMurdo Dry Valleys,Antarctica. Antarctic Sci. 10: 247–256.

Melles, M., S. R. Verkulich & W. -D. Hermichen, 1994.Radiocarbon dating of lacustrine and marine sedimentsfrom the Bunger Hills, East Antarctica. Antarctic Sci. 6:375–378.

Nedell, S. S., D. W. Andersen, S. W. Squyres & F. G. Love,1987. Sedimentation in ice-covered Lake Hoare, Ant-arctica. Sedimentology 34: 1093–1106.

Ollerhead, J., D. J. Huntley & G. W. Berger, 1994. Luminescencedating of the Buctouche Spit, New Brunswick. Can. J. EarthSci. 31: 523–531.

Olley, J. M., A. Murray & R. Roberts, 1996. The effects ofdisequilibria in the uranium and thorium decay chains onburial dose rates in fluvial sediments. Quat. Sci. Rev. 15:751–760.

Palmisano, A. C. & G. M. Simmons Jr., 1987. Spectraldownwelling irradiance in an Antarctic lake. Polar Biol.7:145–151.

Rhodes, E. J. & R. M. Bailey, 1997. The effect of thermaltransfer on the zeroing of the luminescence of quartz fromrecent glaciofluvial sediments. Quat. Sc. Rev. 16: 291–298.

Schackman, M., 1994. Analysis of sediment and sedimentaryprocesses in perennially ice-covered Lake Hoare, Ant-arctica. unpublished M.S. thesis. San José State University.

Simmons Jr., G. M., J. R. Vestal & R. A. Wharton Jr., 1993.Environmental regulators of microbial activity incontinental antarctic lakes. In Green, W. J. & E. I.Freidmann (eds), Physical and Biogeochemical Processesin Antarctic Lakes. Ant. Res. Ser., Am. Geophys. Un.,Wash. D.C. 59: 165–195.

Spaulding, S. A., D. M. McKnight, E. F. Stoermer & P. T.Doran, 1997. Diatoms in sediments of perennially ice-covered lake Hoare, and implications for interpreting lakehistory in the McMurdo Dry Valleys of Antarctica. J.Paleolim. 17: 403–420.

Stuiver, M., G. H. Denton, T. J. Hughes & J. L. Fastook, 1981.History of the marine ice sheet in west Antarctica duringthe last glaciation, a working hypothesis. In Denton, G. H.& T. H. Hughes (eds), The Last Great Ice Sheets. Wiley-Interscience, N.Y.: 319–436.

Squyres, S. W., D. W. Andersen, S. S. Nedell & R. A. WhartonJr., 1991. Lake Hoare, Antarctica: Sedimentation through athick perennial ice cover. Sedimentology 38: 363–379.

Tyler, S. W., P. G. Cook, A. Z. Butt, J. M. Thomas, P. T. Doran& W. B. Lyons, 1998. Evidence of deep circulation intwo perennially ice-covered Antarctic lakes. Limnol.&Oceanog. 43: 625–635.

Wand, U. & J. Perlt, 1999. Glacial boulders ‘floating’ on theice cover of Lake Untersee, East Antarctica. Antarct. Sci.11: 256–260.

Zale, R. & W. Karlén, 1989. Lake sediment cores from theAntarctic Peninsula and surrounding islands. GeografiskaAnnal. 71A: 3–4.

530