The geochemistry of supraglacial streams of Canada Glacier, Taylor Valley (Antarctica), and their evolution into proglacial waters

The Geochemistry of Supraglacial Streams

of Canada Glacier, Taylor Valley (Antarctica),

and their Evolution into Proglacial Waters

SARAH K. FORTNER1,w, MARTYN TRANTER2, ANDREWFOUNTAIN3, W. BERRY LYONS1 and KATHLEEN A.WELCH1

1Department of Geological Sciences, Byrd Polar Research Center, The Ohio State University

Columbus, Ohio 43210-1002, USA; 2Bristol Glaciology Centre, School of Geographical Sciences,

University of Bristol, Bristol BS8 1SS, UK; 3Departments of Geography and Geology, Portland

State University, Portland, OR 97207-0751, USA.

(Received in final form: 8 December 2004)

Abstract. We have investigated the geochemistry of supraglacial streams on the Canada Glacier,

Taylor Valley, Antarctica during the 2001–2002 austral summer. Canada Glacier supraglacial

streams represent the link between primary precipitation (i.e. glacier snow) and proglacial Lake

Hoare. Canada Glacier supraglacial stream geochemistry is intermediate between glacier snow

and proglacial stream geochemistry with average concentrations of 49.1 leq L)1 Ca2+,

19.9 leq L)1 SO2�4 , and 34.3 leq L)1 HCO�3 .

Predominant west to east winds lead to a redistribution of readily soluble salts onto the glacier

surface, which is reflected in the geochemistry of the supraglacial streams. Western Canada

Glacier supraglacial streams have average SO2�4 :HCO�3 equivalent ratios of 1.0, while eastern

supraglacial streams average 0.5, suggesting more sulfate salts reach and dissolve in the western

supraglacial streams. A graph of HCO�3 versus Ca2+ for western and eastern supraglacial

streams had slopes of 0.87 and 0.72, respectively with R2 values of 0.84 and 0.83. Low con-

centrations of reactive silicate (>10 lmol L)1) in the supraglacial streams suggested that little to

no silicate weathering occurred on the glacier surface with the exception of cryoconite holes

(1000 lmol L)1). Therefore, the major geochemical weathering process occurring in the supra-

glacial streams is believed to be calcite dissolution.

Proglacial stream, Anderson Creek, contains higher concentrations of major ions than su-

praglacial streams containing 5 times the Ca2+ and 10 times the SO2�4 . Canada Glacier proglacial

streams also contain higher concentrations (16.6–30.6 leq L)1) of reactive silicate than supra-

glacial streams. This suggests that the controls on glacier meltwater geochemistry switch from

calcite and gypsum dissolution to both salt dissolution and silicate mineral weathering as the

glacier meltwater evolves.

Our chemical mass balance calculations indicate that of the total discharge into Lake Hoare,

the final recipient of Canada Glacier meltwater, 81.9% is from direct glacier runoff and 19.1% is

from proglacial Andersen Creek. Although during a typical, low melt ablation season Andersen

Creek contributes over 40% of the water added to Lake Hoare, its overall chemical importance is

diluted by the direct inputs from Canada Glacier during high flow years. Decadal warming

wAuthor for correspondence: E-mail: [email protected]

Aquatic Geochemistry (2005) 11:391–412 � Springer 2005

DOI 10.1007/s10498-004-7373-2

events, such as the 2001–2002 austral summer produce supraglacial streams that are a major

source of water to Lake Hoare.

1. Introduction

The geochemical relation between precipitation and runoff has been thesubject of in-depth investigation for at least 35 years (Garrels andMacKenzie,1967). Yet, only in the past decade has there been detailed work undertaken inglacier-dominated systems (Anderson et al., 1997, 2000; Hodgkins et al.,1997, 1998; Hodson et al., 1998, 2000, 2002; Tranter et al., 1993, 1994, 2002).The movement of meltwater through snowpack and/or glacier ice can modifiythe ‘primary’ geochemical signature of precipitation (Lyons et al., 1998;Tranter et al., 1993, 1994, 1996a, b). In addition to containing the primarysolutes derived directly from snow and ice melt, supraglacial streams acquiresolutes from external sources (Tranter et al., 1993). Solute concentrations insupraglacial streams increase when there is an abundance of fine-grained‘dust’ (Sharp et al., 1995; Tranter et al., 1993). As an example, supraglacialstreams on the Glacier Tsidjiore Nouve in Switzerland had initially low TDS(total dissolved solute) concentrations, but as the water flowed downstreamthrough a dust-covered surface it gained high concentrations of sodium andpotassium after only 30 m (Lemmens and Roger, 1978).

Changes in supraglacial stream chemistry may also be associated with theflushing of cryoconite holes. Cryoconite holes are melt holes that form on theice surface of glaciers by solar radiation heating and melting-in of sediments(Gribbon, 1979). Typically, cryoconite holes are open to the atmosphere,host microbes and algae, and, in some cases, invertebrates (Mueller et al.,2001; Takeuchi et al., 2000; Wharton et al., 1985). In the dry valleys ofAntarctica, the cryoconite holes have ice lids, which limit water and gasexchange (Mueller et al., 2001; Tranter et al., 2004). Various biogeochemicalprocesses such as photosynthesis and organic matter mineralization are en-hanced in closed cryoconite holes, which can greatly alter their waterchemistry (Tranter et al., 2004). Approximately half of these cryoconite holesare isolated; however, the other half are linked to the near surface hydrologyand they may openly connect with supraglacial streams during decadalwarming events (Fountain et al., in press). Cryoconite holes may contributeas much as 13% of the measured glacier runoff (Fountain et al., in press).

The vast majority of research done on the geochemistry of supraglacialstreams has been done on temperate glaciers. Only a few papers deal withsupraglacial streams on polar glaciers, and all of those have focused on theArctic (Hodgkins and Tranter, 1998; Hodgkins et al., 1997; Hodson et al.,1998). Previous work has demonstrated that the geochemistry of CanadaGlacier proglacial streams is very different from glacier surface geochemistry

392 SARAH K. FORTNER ET AL.

(Lyons et al., 2003). This study examines the chemical evolution of CanadaGlacier melt from the glacier surface to proglacial Lake Hoare.

The geochemistry of Lake Hoare has never been analyzed with respect tosupraglacial streams because previous investigations have occurred duringcooler years when these streams were not produced. During this unusual highmelt season, Lake Hoare received more water from supraglacial streams thanproglacial Andersen Creek. This study assesses how Canada Glacier supra-glacial streams relate chemically to snow, ice, and proglacial streams alongwith determining their influence on Lake Hoare. This study begins to explorethe influence of supraglacial streams on the ecosystems of all of theMcMurdo Dry Valleys proglacial lakes.

2. Field Area

Taylor Valley (77�000 S, 162�520 E) is part of the largest ice-free expanse inAntarctica, the McMurdo Dry Valleys. The valley trends east to west,spanning 34 km from the Ross Sea to the Taylor Glacier (Figure 1). TaylorValley is approximately 400 km2 in area and is bound at its northern marginby the Asgard Range and at its southern margin by the Kukri Hills (Fig-ure 1). Within the Taylor Valley there are 15 glaciers, 3 ice-covered lakes(Hoare, Fryxell and Bonney) and much soil and exposed bedrock (Fountainet al., 1999a). Glaciers cover 35% of Taylor Valley (Fountain et al., 1999a).Since 1993, Taylor Valley has been the primary field area for a U.S. NationalScience Foundation Long-Term Ecological Research (LTER) program.

The dry valleys are polar deserts. Mean annual temperatures in Taylorvalley are )17�C on the valley floor with summer months having tempera-tures close to freezing (Doran et al., 2002). Mass balance measurements ofTaylor Valley glaciers show average gains and losses of less than 5 cm waterequivalents per year (Fountain et al., in press). The strongest winds in thevalleys are katabatic winds blowing from the East Antarctic Ice Sheet downthe Taylor Glacier eastwards into Taylor Valley that reach daily highs of over30 m s)1 daily (Doran et al., 2002). Milder summer winds blow from theRoss Sea toward the west (Doran et al., 2002).

2.1. TAYLOR VALLEY GLACIERS AND MELT

The glaciers of Taylor Valley, Antarctica are cold-based alpine types, withthe exception of Taylor Glacier, an outlet of the East Antarctic Ice Sheet(Figure 1). These glaciers appear to be at their maximum extent since theLGM (Denton et al., 1989). This is anomalous compared to the overallglobal retreat of alpine glaciers (Dyurgerov and Meier, 1999). Recent massbalance measurements of Taylor Valley glaciers indicate that mass change inthe accumulation region does not exceed 10 cm water equivalents and mass

THE GEOCHEMISTRY OF SUPRAGLACIAL STREAMS OF CANADA GLACIER 393

loss in the ablation region is less than 35 cm water equivalents (Fountainet al., 1999a). Also, glacier position and equilibrium line altitude increasesinland from the Ross Sea because precipitation decreases and solar radiationincreases inland (Fountain et al., 1998). In the Asgard Range, the ELA risesfrom 375 m at its the eastern margin to 1100 m at its western margin(Fountain et al., 1999b). Glaciers actively melt between 4 and 12 weeksduring the austral summer and are the primary source of water to the lakesand streams in Taylor Valley (Fountain et al., 1999a).

2.2. CANADA GLACIER AND CANADA GLACIER WATERSHED

Canada Glacier extends from the Asgard Range (Figure 1) with an ablationarea of approximately 8.5 km2 (Lewis et al., 1999). The ablation zone ofCanada Glacier faces south with a 3� slope. It is 3-km long, and ranges inelevation from 100 m to the equilibrium line at 350 m (Lewis et al., 1999).

Figure 1. Taylor Valley glaciers, lakes, and streams (modified from Lyons et al. (1998b)).

394 SARAH K. FORTNER ET AL.

During most summers of the 1990s, little melt was produced over the surfaceof the ablation zone and terminus cliffs contributed up to 30% of the pro-glacial stream volume (Lewis et al., 1999). Supraglacial streams were neverobserved during the 10 melt seasons prior to the summer of 2001–2002.

In contrast, the austral summer of 2001–2002 was quite warm, producingatypically high volumes of meltwater on Canada Glacier. Lake Fryxellgained a volume of 4.75 · 106 m3 during this season, replacing about 85% ofthe water lost from the lake over the previous decade (Doran et al., 2002).Lake Hoare volume gains were observed to be of a similar magnitudealthough they have not been quantified at this time.

East of Canada Glacier, Lake Fryxell is fed by Canada Stream along withproglacial streams from other valley glaciers. To the west, Lake Hoare is feddirectly by ice melt from Canada Glacier and by the western proglacialstream, Andersen Creek (Figure 1). Lake Hoare also receives less than 20%of its total water from Suess Glacier via Lake Chad to the west.

3. Methodology

Supraglacial stream sampling on Canada Glacier took place between12/27/01 and 1/16/02. Samples collected included: 81 for major ion andreactive silicate analysis, and 10 for stable isotope analysis. The supraglacialstream channels that were sampled ranged from only centimeters in widthand depth to widths of 10 m and depths of 1 m. Larger, higher dischargesupraglacial channels were preferentially sampled to reflect the runoff fromthe glacier that produced proglacial waters. The two largest supraglacialstreams were also sampled along their course from the uppermost ablationregion until they drained off the glacier (Figure 2). One of these major su-praglacial streams drained west into Lake Hoare, and the other drained eastinto Lake Fryxell. Ten samples for major ions were also taken from pro-glacial Andersen Creek during the same time-period. Cryoconite holes weresampled in 1999–2000 in isolation from supraglacial streams as described inTranter et al. (2004). Snow samples were also collected during the 1999–2000austral summer and analyzed in the same manner as the supraglacial streams.This study focuses on the snow collected from the top meter of CanadaGlacier.

All samples for major ions and reactive silicate were collected in cleanpolyethylene bottles and later filtered into clean polyethylene bottles. Poly-ethylene bottles had been rinsed with 18-MW distilled-deionized water(DDW) three times, then filled with DDW and soaked for a minimum of24 h. After soaking, bottles were rinsed three times with DDW. NalgeneTM

polyethylene 1 L bottles for suspended sediment were also rinsed, soaked andrinsed in the sequence noted above.

THE GEOCHEMISTRY OF SUPRAGLACIAL STREAMS OF CANADA GLACIER 395

Major ion and reactive silicate samples were filtered through 0.4 lm poresize NucleporeTM filters in the field laboratory within 12 h of collection.Plastic filter towers were rinsed between samples. All bottled samples wereplaced in clean plastic bags for storage in a refrigerator until they could betransported to the Crary Laboratory in McMurdo Station, Antarctica. Thesesamples were stored for about 30–40 days in the dark at 5 �C in the CraryLaboratory until they were analyzed.

Major ions (Na+, K+, Mg2+, Ca2+, Cl), NO3), SO4

2–) were analyzedwith a DX-300 ion chromatograph (IC) using methods outlined in Welchet al. (1996). The greatest error, determined from duplicate analyses, was3.5%. Container blanks of distilled dionized water were analyzed in everysample run. Because of the high precision of these measurements and the lowconcentrations of HCO3

– in Holocene polar snow and ice (Legrand andMayewski, 1997) HCO3

– was determined by summing the cations (Ca2+,K+, Mg2+, and Na+) in equivalents and subtracting the sum of the anions(Cl–, NO�3 , SO

2�4 ) in equivalents.

Reactive silicate samples were analyzed at Byrd Polar Research Center,Columbus, Ohio no later than 90 days after collection. The concentrationof reactive silicate was determined colormetrically following the methods ofMullin and Riley (1955). Sample blanks were run to insure the precision ofreactive silicate measurements; the highest blank concentration found was<2 lM. Precision was better than 9%.

Figure 2. Canada Glacier supraglacial stream, and proglacial stream (Andersen Creek)

sample locations. W–E samples were collected from more than one supraglacial stream.

N–S samples contain: (1) Supraglacial waterfalls, (2) Lake Hoare supraglacial channel,

and Lake Fryxell supraglacial channel. The Lake Hoare and Lake Fryxell supraglacial

channels along with Andersen Creek were sampled throughout their course.

396 SARAH K. FORTNER ET AL.

Samples for dD determination were stored at room temperature andshipped to INSTAAR in Boulder, Colorado for analyses. dD values weremeasured with an automated uranium reduction method with a standarddeviation less than 0.5 ppm based on analysis of the standards (Vaughnet al., 1998).

4. Results

Solute concentrations varied between snow, ice, supraglacial streams,cryoconite holes on Canada Glacier and proglacial streams (Andersen Creekand Canada Stream) (Figure 3). Average Ca2+ and SO2�

4 (49.1 leq L–1,19.9 leq L–1) concentrations are higher in the supraglacial streams of Can-ada glacier compared to glacier snow (2.9 leq L–1, 4.9 leq L–1) and ice(36.0 leq L–1, 7.8 leq L–1) (Figure 3). Averages are used because repeatedsamples (after 2 weeks) of supraglacial streams were within 25% of theiroriginally measured concentrations in the upper and middle ablation area.

Table I (modified from Lyons et al., 2003) compares average TDS, Si,NO3

-and Ca:Cl values in water from/or neighboring Taylor Valley to thatfound in Canada Glacier supraglacial streams. The Ca:Cl ratio is useful toassess the presence of terrestrial dust (i.e. when Ca:Cl ratio is greater thanseawater) (Lyons et al., 2003). Ca:Cl equivalent ratios of 1.27 in CanadaGlacier supraglacial and 1.90 and 1.74 in proglacial streams indicate elevateddust concentrations compared to the Canada Glacier snow ratio of 0.20(Table I). The highest Ca:Cl equivalent ratio on Canada Glacier was found

Figure 3. Major ion concentrations for Canada Glacier snow, ice, supraglacial streams,

and cryoconite holes versus Anderson Creek concentrations. Mg2+, K+, and Na+ not

measured in ice and cryoconite holes.

THE GEOCHEMISTRY OF SUPRAGLACIAL STREAMS OF CANADA GLACIER 397

in cryoconite holes (2.17). By comparison the Ca:Cl equivalent ratio inseawater is 0.038.

With the exception of NO�3 , solute concentrations in Canada Glaciersnow and ice are much lower than concentrations in the supraglacialstreams (Table I). TDS in snow is lowest in the most inland, highest ele-vation sites, such as Taylor Dome and Newall Glacier (Lyons et al., 2003).These inland snows are most removed from local terrestrial sedimentsources. Commonwealth Glacier contains the highest glacier snow TDS(5 mg L–1) because the glacier is exposed to windblown particulate matterand closest to a marine aerosol source, i.e. McMurdo Sound/Ross Sea(Lyons et al., 2003).

Although temporal variations were low in the supraglacial streams, spatialvariations existed. Western supraglacial streams were more enriched in SO2

4,Ca2+, K+, and Mg2+ with respect to Cl) than eastern supraglacial streams(Table II). The sequence of major ion abundance in equivalents for thewestern proglacial stream, Andersen Creek was: Ca2+>SO2�

4 >Cl)>Na+>Mg2+>K+>NO�3 , whereas the relative concentrations of the easternlying Canada Stream was: Ca2+>Cl–>Na+>SO2�

4 >Mg2+>K+>NO�3(Lyons et al., 2003). Enrichments in SO2�

4 in the western draining proglacial

Table I. Distance to coast, TDS, reactive silicate, nitrate and Ca:Cl for glacier snow, supraglacial

streams cryoconite holes, and proglacial streams

Kilometers

to Coast

TDS

(mg L)1)

Si

lmol L)1

NO�3leq L)1

Ca:Cl

(equiv)

Glacier Snow

Canada 14 2a NA 0.39 0.20

Taylor Glaciera 40 1 NA 3.90 1.89

Taylor Domec 150 0.2 NA 0.76 0.07

Newall Glacierb 30 0.5 NA 1.10 0.23

Supraglacial streams

Canada Glacier 14 5 3.02 0.69 1.27

Cryoconite holes

Canadad <14 48 1000 24.00 2.17

Proglacial streams

Andersen Creeka <14 25 12.6 (1995–1996) 1.79 1.90

Canada Streama <14 8.4a 30.6 (1995–1996) 0.41 1.74

aLyons et al. (2003).bWelch et al. (1989).cMayewski (pers. communication).dTranter et al. (2004).

398 SARAH K. FORTNER ET AL.

meltwaters can be traced back to the enrichments in the western flowingsupraglacial streams (Table II). Figure 4 shows that eastern draining su-praglacial streams are more alkaline than supraglacial streams entering LakeHoare. Therefore, the SO2�

4 :HCO�3 ratio may provide insight to the spatialdistribution of salts on the Canada Glacier and in its meltwaters.

5. Discussion

5.1. SNOW SOLUTE VARIATIONS

Ternary diagrams of cations and anions show the evolution of meltwaterfrom the unmelted glacier snow and ice to proglacial water (Figures 5, 6,and 7). Snow from the upper reaches of the Newall Glacier located at anelevation of 1600–1700 m, well above the Canada Glacier accumulation zone(Welch, 1993) contains a greater relative cationic abundance of Ca2+

(�40%) than Canada Glacier snow (�20%) (Figure 5). Although there maybe higher solute concentrations on Canada Glacier than Newall Glacier(Table I), the higher snow accumulation on the lower elevation Canada

Figure 4. Lake Hoare and Lake Fryxell draining supraglacial streams: Total alkalinity

versus Ca2+.

Table II. Mean and range of major ions (equivalents) for eastern and western supraglacial

streams on Canada glacier

SO2�4 :Cl– NO�3 :Cl

– Ca2+:Cl– Na+:Cl– K+:Cl– Mg2+:Cl–

Western 0.55

(0.35–1.21)

0.02

(0.00–0.03)

1.65

(0.00–5.29)

0.76

(0.48–1.03)

0.12

(0.04–0.22)

0.32

(0.20–0.52)

Eastern 0.41

(0.17–0.58)

0.02

(0.00–0.04)

1.25

(0.68–1.98)

0.76

(0.37–0.80)

0.10

(0.05–0.15)

0.29

(0.14–0.48)

THE GEOCHEMISTRY OF SUPRAGLACIAL STREAMS OF CANADA GLACIER 399

Glacier ‘dilutes’ the aeolian dust (Lyons et al., 2003; Welch et al., 1993).Ca2+ concentrations in Canada Glacier snow are from non-marine sources,as indicated by their Ca2+:Cl– equivalent ratio (0.20) (Table I). Likewise, allmajor ions, excepting Mg2+ are enriched in Canada Glacier snow with re-spect to Cl– (Figure 8), suggesting significant terrestrial inputs of these ions tothe surface of the Canada Glacier. Accumulations of wind blown materialcan be observed on the surface of Lake Hoare and on the western flank (the

Figure 5. Cation equivalent percentages of Newall Glacier snow, Canada Glacier snow,

Canada Glacier supraglacial streams, Canada Stream, Andersen Creek, Lake Hoare, and

Lake Fryxell. Solid arrow shows evolution from unmelted glacier to proglacial streams.

Dashed arrows show contributions of direct glacial (bottom arrow) versus proglacial

sources (top arrow) to Lake Hoare.

400 SARAH K. FORTNER ET AL.

Lake Hoare side) of the Canada Glacier suggesting the primary transport ofthis ‘dust’ is from the west via katabatic winds.

5.2. EVOLUTION OF GLACIER SNOW AND ICE TO PROGLACIAL WATER

The relative cationic abundance of Ca2+ increases in Canada Glacier melt-water during its evolution from glacier snow and ice to proglacial water. The

Figure 6. Anion equivalent percentages of Canada Glacier ice, Newall Glacier snow,

Canada Glacier snow, Canada Glacier supraglacial streams, Canada Stream, and Andersen

Creek. (Lake Fryxell and Lake Hoare are not included because NO�3 fell below detection

limits.) Arrow indicates evolution from unmelted glacier ice to proglacial streams.

THE GEOCHEMISTRY OF SUPRAGLACIAL STREAMS OF CANADA GLACIER 401

unmelted glacier ice has the lowest percentage of Ca2+ (�15%) with respectto Na+ + K+ and Mg2+ and the relative abundance of Ca2+ increasessuccessively from supraglacial streams (�60%) to proglacial streams (�70%)(Figure 5). Mg2+ maintains the same relative cationic abundance throughoutthis transition from ice into supraglacial and proglacial water (Figure 5). Therelative amount of Na+ + K+ decreases as snow evolves into supraglacialstream water (Figure 5). The relative depletion of Na++ K+ with respect to

Figure 7. HCO�3 ternary diagram. Anion equivalent percentages for Newall Glacier snow,

Canada Glacier snow, Canada Glacier supraglacial streams, Canada Stream, Andersen

Creek, Lake Hoare, and Lake Fryxell. Solid arrow shows the evolution the unmelted

glacier to the eastern proglacial stream Canada Stream. Dashed arrow shows the evolution

from unmelted Glacier to the western proglacial stream Andersen Creek.

402 SARAH K. FORTNER ET AL.

Ca2+ and Mg2+ continues from supraglacial to proglacial streams, (AndersenCreek, Canada Stream) (Figure 5). Since supraglacial and proglacial streamshave similar Na++K+:Cl– ratios, the relative losses of Na++K+ reflectgains in Ca2+ as the meltwater evolves.

5.3. SPATIAL VARIATIONS IN MELTWATER EVOLUTION

The pathway from unmelted glacier ice to supraglacial and proglacialstreams occurs with gains in relative anionic abundance of both SO2�

4 andHCO�3 relative to Cl) (Figures 6 and 7). Gains in the relative abundance ofSO2�

4 and HCO�3 are likely to reflect the dissolution of gypsum (CaSO4)and calcite (CaCO3) as meltwater evolves (Figure 6). The supraglacialstreams draining to Lake Hoare and Lake Fryxell had very strong correla-tions (R2 ¼ 0.84 and 0.83, respectively) between HCO�3 and Ca2+ with slopesof 0.87 and 0.72 (equivalents) (Figure 4). Both slopes suggest the dominanceof carbonate dissolution. Silicate weathering provides insignificant HCO�3 tothe supraglacial streams because they contain very little reactive silicate(<10 lmol L–1). This is unlike proglacial Andersen Creek, where silicateweathering occurs (Nezat et al., 2001).

5.4. NO�3 DISTRIBUTION

All major ions except NO�3 have a west to east distribution in Canada Glaciersupraglacial streams (i.e. Table II). If concentrations of NO�3 were primarily

Figure 8. Major ions versus Cl– for Canada Glacier snow, ice, supraglacial stream, and

cryoconite holes versus Andersen Creek and seawater ratios. Mg2+, K+, and Na+ not

measured in ice and cryoconite holes.

THE GEOCHEMISTRY OF SUPRAGLACIAL STREAMS OF CANADA GLACIER 403

in Taylor Valley dusts, NO�3 would show a spatial distribution similar toSO2�

4 , since both anions increase in relative abundance in valley salts headinginland (Keys and Williams, 1981). Typical Antarctic snow and firn containrelatively high concentrations of NO�3 (Legrand and Mayewski et al., 1997).This is because at high latitudes, stratospheric aerosols are enriched in ionssuch as NO�3 and HCHO. NO�3 can be transported to glacier surfaces via drydeposition and lost to postdepositional diffusion (Legrand and Mayewski,1997). As a result, older glacier firn and ice gain NO�3 with respect to snowvia postdepositional additions and snow-firn redistribution (Legrand andMayewski, 1997). This may explain the high ratio of NO�3 :Cl

– in ice withrespect to seawater (Figure 8).

The ratio of NO�3 :Cl– is lower in supraglacial streams compared to snow

and ice (Figure 8). Algae and cyanobacteria occur in cryoconite holes andproglacial streams of the Canada Glacier (McKnight et al., 1998, 1999;Mueller et al., 2001; Tranter et al., 2004; Wharton et al., 1985). Cyanobac-terial mats in Taylor Valley proglacial streams are known to cause down-stream reductions in NO�3 concentration (McKnight et al., 1999). Therefore,downstream losses of NO�3 in supraglacial streams may not be the result ofsimple dilution, but rather biological uptake by autotrophic organisms livingon the glacier surface.

5.5 IMPORTANCE OF AEOLIAN SEDIMENT TRANSPORTATION TO CANADA

GLACIER SUPRAGLACIAL STREAM CHEMISTRY

Clearly, the aeolian input of dust onto the glacier surface plays a key role inthe geochemical evolution of supraglacial stream chemistry. Wind speedsrecorded at the Canada Glacier met station between September 1998 andDecember 2000 averaged 5 m s–1 (Doran et al., 2002; http://huey.colorado.edu/LTER). These winds came from both the Taylor Glacier and the RossSea (Figure 1). The strongest winds are over 35 m s)1 and flow from west toeast (http://huey.colorado.edu/LTER). Wind strength is the most importantparameter in determining the aeolian transport of sediment in the McMurdoDry Valleys region (Lancaster, 2002). This is the case for most desert systems.For instance, in the Mojave Desert, 80% of aeolian sands are transported bythe highest 6% wind speeds (Lancaster, 1985). In the Taylor Valley, theaeolian flux is a linear function of distance from the coast (R2 ¼ 0.92) andwind speed (R2 ¼ 0.74) (Lancaster, 2002). Winds blowing down TaylorValley lose momentum as they head east (Lancaster, 2002) and the undu-lating topography of the Canada Glacier is likely to decrease the wind,causing sediment to fall onto the glacier surface. The net deposition on thesurface of Canada Glacier is 0.43 gm m)2 yr)1 with dust size particles(<50 lm) comprising 84.1% of this total (Lancaster, 2002).

404 SARAH K. FORTNER ET AL.

Supraglacial streams capture windblown material preventing redistribu-tion of dust and sediments (Wang and Kraus, 1999). As wind blown particlesmove across the glacier they are preferentially ‘captured’ in western supra-glacial channels. The more reactive materials would rapidly dissolve in su-praglacial streams because of the dilute nature of snow and ice melt (Tranteret al., 1993).

Solutes in supraglacial streams show a non-random spatial distribution onthe surface of the Canada Glacier (Table II). Solute concentrations werehigher in the supraglacial stream samples taken from the western side of theablation zone. The solute distribution reflects the distribution of terrestrialmaterials being introduced by the wind to the glacier surface. This is the samewest to east solute concentration trend previously observed in the streamsand on the snow of Canada Glacier (Lyons et al., 2003). The correspondinglyhigher amount of solute in the western supraglacial streams are likely toreflect that more dust is deposited on the windward (i.e. western) side of theglacier during katabatic events.

Canada Glacier supraglacial and proglacial streams also show spatialvariations in their SO2�

4 :HCO�3 ratios. Canada Glacier supraglacial streamshave a broad range of SO2�

4 :HCO�3 equivalent ratios, spanning from 0.13 to2.5. The averages for western and eastern supraglacial streams were 1.0 and0.5, respectively. Andersen Creek SO2�

4 :HCO�3 ratios range from 0.5 to 2.5,and Canada Stream ratios range from 0.25 to 1.1. The ratios of SO2�

4 :HCO�3suggest that the western supraglacial and proglacial streams may undergomore gypsum dissolution relative to calcite dissolution than the easternstreams. The valley floor composition supports this hypothesis, becausegypsum on the valley floor increases in concentration to the west, whereascalcite decreases (Keys and Williams, 1981). Easterly winds off the Ross Seawould deposit more calcite with respect to gypsum on the eastern side ofCanada Glacier than the west side. Likewise, the westerly winds off theTaylor Glacier are likely to carry more gypsum to the west side of CanadaGlacier than the east (Figure 1). The evolution of Canada Glacier meltwaterfollows two distinct pathways, one enriched in SO2�

4 and one enriched inHCO�3 (Figure 7).

5.6. RELATIONS BETWEEN SUPRAGLACIAL STREAMS AND CRYOCONITE

HOLES

Meltwater dominated the ablation process during the extremely warm meltseason of 2001–2002 in contrast to the previous 10 years when sublimationdominated ablation (Fountain et al., 1998; Lewis et al., 1999). During the2001–2002 melt season, cryoconite holes were observed with open lids,directly linking to open supraglacial channels. These extreme melt conditionsin the dry valleys are similar to those ‘normally’ observed on the White

THE GEOCHEMISTRY OF SUPRAGLACIAL STREAMS OF CANADA GLACIER 405

Glacier, Axel Heiberg Island, Canada (Mueller et al., 2001). White Glaciercryoconite holes were smaller in diameter and depth, and had lower pHs,nutrient concentrations and numbers of organisms than Canada Glaciercryoconite holes (Mueller et al., 2001; Tranter, et al., 2004). The cryoconiteholes on the White Glacier are flushed as they coalesce with dilute supra-glacial streams during summer melt (Mueller et al., 2001).

In 1998 cryoconite holes were sampled on the Canada Glacier duringthe low melt year and had not mixed with the supraglacial streams. Theirmajor ion and other data are included here because the cryoconite holeswere observed draining into supraglacial streams during the 2001–2002melt season. When these cryoconite holes were still isolated, they con-tained higher TDS than both Andersen Creek and Canada Stream andreactive silicate concentrations of 1000 lmol L)1 (Table I). When cryoco-nite holes are ice encased, photosynthesis by cyanobacteria increase the pHof the solution (Tranter et al., 2004). This high pH water enhances thedissolution of alumnosilicate minerals present in the cryoconite holes(Drever, 1988).

Some of the lower pH cryoconite holes from 1998 had similar pCO2 values(10)3.37) (Tranter et al., 2004) to the calculated supraglacial stream values(10)2.95 to 10)3.30 atm). This suggests that there is an exchange of supra-glacial stream and cryoconite hole waters. The little reactive silicate measuredin Canada Glacier supraglacial streams (3.0 lmol L)1) is likely to be fromsilicate weathering in cryoconite holes.

5.7. RELATIONS OF SNOW/ICE, SUPRAGLACIAL AND PROGLACIAL STREAMS

TO LAKE HOARE CHEMISTRY

The ternary diagrams reveal that Lake Hoare surface waters are chemicallyintermediate between Canada Glacier snow, Canada Glacier supraglacialstreams, and Andersen Creek (Figure 5, 6, and 7). When supraglacial streamsexit the glacier, via western waterfalls, they dilute Andersen Creek as seen bydecreasing SO2�

4 and Ca2+ concentrations (Figures 9 and 10). This con-trasted with low melt seasons when SO2�

4 and Ca2+ concentrations increaseddownstream in Andersen Creek in the absence of supraglacial waterfalls(Lyons et al., 1998). Supraglacial streams have an even larger influence onproglacial Lake Hoare. An algebraic balance of cations shows that CanadaGlacier melt which flows directly into Lake Hoare contributes 81% of the topmeter of the lake’s chemistry, whereas the marginal stream, Andersen Creekcomprises 19%. Suess Glacier contributions via Lake Chad are negligible. Ofthe direct glacier contribution, supraglacial streams contribute 21% andice melt in the lake’s subsurface waters (i.e. the interface between CanadaGlacier and Lake Hoare) contribute 60%.

406 SARAH K. FORTNER ET AL.

Likewise, the deuterium measurements of the meltwaters revealed that themajor supraglacial streams flowing into Lake Hoare reached a downstreamdD value of )266& (Figure 11). The older ice exposed in the terminus hasdD values lower (lighter) than )300& (Lyons et al., 2003). The progressivelylighter dD values measured in supraglacial streams reflect mixing of meltfrom lighter dD ice. The lowest, most western dD supraglacial stream valuescorrespond well with the dD values of the top 5 m of Lake Hoare in 1993 that

Figure 9. SO2�4 concentration for Andersen Creek (top) and Canada Glacier supraglacial

waterfalls (bottom).

Figure 10. Ca2+ concentration for Andersen Creek (top) and Canada Glacier supraglacial

waterfalls (bottom).

THE GEOCHEMISTRY OF SUPRAGLACIAL STREAMS OF CANADA GLACIER 407

averaged around )260& (Lyons et al., 1998). Whereas, dD values averagedfrom proglacial streams are likely to be significantly heavier because they mixwith hyporheic water (Gooseff et al., 2003). Although our dD data arelimited (10 samples), we have used them to help assess the sources of water toLake Hoare. Calculations using dD suggest that Andersen Creek contributes24% of Lake Hoare surface water and Canada Glacier (snow, ice, and su-praglacial streams) contributes 76% during the high flow year. The similarityin cationic and deuterium balances show that the chemistry of the surface ofthe lake is dominated by snow, ice and supraglacial streams rather thanproglacial streams during high flow years.

6. Summary and Conclusions

As suggested from this study and previous glaciochemical and proglacialstream analyses, (Lyons et al., 1998, 2003), the chemistry of meltwater inTaylor Valley is greatly modified by supraglacial processes. Canada glaciersnow, ice, and proglacial streams are enriched in major ions with respect tochloride, suggesting an abundance of reactive dust/salt dissolution. Theexception is NO�3 , which is taken up by cyanobacteria in supraglacialstreams. The slope of HCO�3 versus Ca2+ for western and eastern supra-glacial streams (0.87 and 0.72, respectively) coupled with low reactive silicateconcentrations (3.0 lmol L)1) suggest that calcite dissolution is the dominantgeochemical process in supraglacial streams. However, cryoconite holes are alikely source of the observed silicate in supraglacial streams.

Figure 11. Deuterium fractionation in a Canada Glacier supraglacial stream heading west

toward Lake Hoare.

408 SARAH K. FORTNER ET AL.

Although calcite dissolution dominates the Canada Glacier supraglacialstream signal, sulfate salt dissolution occurs more in western than easternsupraglacial streams. Correspondingly, the soils of Taylor Valley containhigher gypsum and lower calcium carbonate concentrations from west to east(i.e. heading from Taylor Glacier to Ross Sea) (Keys and Williams, 1981).Katabatic winds from Taylor Glacier preferentially deposit gypsum on thewestern side of the Canada Glacier and summer gusts from the Ross Seadeposit calcite preferentially on the eastern side.

Previous estimates had demonstrated that during low melt years (i.e.1993–1994), 56% of Lake Hoare water comes from its direct contact with thewestern terminal face of the Canada Glacier and 44% comes from AndersenCreek (House et al., 1995). During 2001–2002, over three-fourths of thewater entering Lake Hoare came directly from the Canada Glacier. Sincehigh melt years contribute a large volume of water to Lake Hoare they likelydominate the geochemistry of Lake Hoare. It is important to research thefrequency of high melt events that produce supraglacial streams to assesstheir impact over time on the geochemistry and ecology of Lake Hoare.

Acknowledgements

We thank Thomas Nylen, Heather Pugh, and Dr. Virginia Butler for helpingin the collection of samples and Dr. Anne Carey who helped with the LakeHoare chemical balance calculations. This work was supported by NSFGrant OPP-9813061. We especially thank Dr. Christian H. Fritsen and ananonymous reviewer who greatly improved this manuscript.

References

Anderson S. P., Drever J. I., Frost C. D. and Holden P. (2000) Chemical weathering in the

foreland of a retreating glacier. Geochim. Cosmochim. Acta 64, 1173–1189.

Anderson S. P., Drever J. I. and Humphrey N. F. (1997) Chemical weathering in glacial envi-

ronments. Geol. 25, 399–402.

Denton G. H., Bockheim J. G., Wilson S. C. and Stuiver M. (1989) Late Wisconsin and early

Holocene glacial history inner Ross Embayment, Antarctica. Quat. Res. 31, 151–182.

Doran P. T., McKay C. P., Clow G. D., Dana G. L., Fountain A. G., Nylen T. H. and

Lyons, W. B. (2002) Climate observations from the McMurdo Dry Valleys, Antarctica,

1986–2000. J. Geo. Res.- Atmos 107, 1–12.

Drever, J. I. (1988) The Geochemistry of Natural Waters: Surface and Groundwater Environments,

3rd edn. Prentice Hall.

Dyurgerov M. B and Meier M. F. (1999) Analysis of winter and summer glacier mass balances.

Geograf. Ann. 81A, 541–554.

Fountain A. G., Tranter, M., Nylen, T. H., Lewis K. J. and Mueller, D. (in press). Evolution of

cryoconite holes and their contribution to melt-water runoff from glaciers in the McMurdo

dry valleys. J. Glac.

THE GEOCHEMISTRY OF SUPRAGLACIAL STREAMS OF CANADA GLACIER 409

Fountain A. G., Lyons W. B., Burkins M. B., Dana G. L., Doran P. T., Lewis K. J.,

McKnight D. M., Moorhead D. L., Parsons A. N., Priscu J. C., Wall D. H., Wharton Jr. R.

A. and Virginia R. A. (1999a) Physical controls on the Taylor Valley Ecosystem, Antarctica.

Bioscience 49, 961–971.

Fountain A. G., Lewis K. J. and Doran P. T. (1999b) Spatial climatic variation and its control on

glacier equilibrium line altitude in Taylor Valley, Antarctica. Glob. Plan. Change 22, 1–10.

Fountain A. G., Dana, G. L, Lewis K .J., Vaughn B. H. and McKnight D. M. (1998) Glaciers of

the McMurdo Dry Valleys, Southern Victoria Land, Antarctica. In Ecosystem Dynamics in a

Polar Desert: The McMurdoDry Valleys, Antarctica (ed. J. C. Priscu), Vol. 72, pp. 65–76.

American Geophysical Union, Antarctic Research Series.

Garrels R. M. and Mackenzie F. T. (1967) Origins of the chemical composition of some springs

and lakes. In Equilibrium Concepts in Natural Water Systems, Vol. 67, pp. 222–242. Advances

in Chemistry Series, American Chemical Society, Washington.

Gooseff, M. N., McKnight, D. M., Runkel, R. L. and Vaughn, B. H. (2003) Determining long

time-scale hyporheic zone flow paths in Antarctic streams. Hydro. Proc. 17, 1691–1710.

Gribbon P. W. (1979) Cryoconite holes on Sermikaysak, West Greenland. J. Glac. 22, 177–181.

Hodgkins R., Tranter M. and Dowdeswell J. A. (1997) Solute provenance, transport and

denudation in a high arctic glacierized catchment. Hydro. Proc. 11, 1813–1832.

Hodgkins R. and Tranter M. (1998) Solute in High Arctic glacier snow cover and its impact on

runoff chemistry. Ann. Glac. 26, 156–160.

Hodgkins R., Tranter M. and Dowdeswell J. A., (1998) The hydrochemistry of runoff from a

‘cold-based’ glacier in the High Arctic (Scott Turnerbreen, Svalbard). Hydro. Proc. 12,

87–103.

Hodson A., Gurnell A., Tranter M., Bogen J., Hagen J. O. and Clark M. (1998) Suspended

sediment yield and transfer processes in a small High-Arctic glacier basin, Svalbard. Hydro.

Proc. 12, 73–86.

Hodson A., Tranter M. and Vatne G. (2000) Contemporary rates of chemical denudation and

atmospheric CO2 sequestration in glacier basins: An arctic perspective. Earth Surf. Proc.

Land 25, 1447–1471.

Hodson A., Porter P., Lowe A. and Mumford P., (2002) Chemical denudation and silicate

weathering in Himalayan glacier basins: Batura Glacier, Pakistan. J. Hydro. 262, 193–208.

House H. R., McKnight D. M. and von Guerard P. (1995) The influence of stream channel

characteristics on streamflow and annual water budgets for lakes in Taylor Valley. Antarc. J.

Unit. States 10, 284–287.

Keys J. R. and Williams K. (1981) Origin of cold desert salts in the McMurdo region, Antarctica.

Geochem. Cosmochim. Acta 45, 2299–2309.

Lancaster N. (2002) Flux of eolian sediment in the McMurdo Dry Valleys, Antarctica: A Pre-

liminary Assessment. Arctic, Antarc. Alp. Res. 34, 318–323.

Lancaster N. (1985) Variations in wind velocity and sand transport on the windward flanks of

desert sand dunes. Sed. 32, 581–593.

Legrand M. and Mayewski P. A. (1997) Glaciochemistry of polar ice cores: A review. Rev.

Geophys. 35, 219–243.

Lemmens, M. M. and Rodgers M. (1978) Influence of ion exchange on dissolved load of Alpine

meltwaters. Earth Surf. Proc. 3, 179–187.

Lewis K. J., Fountain A. G. and Dana G. L. (1999) How important is terminus cliff melt? A

study of the Canada Glacier terminus, Taylor Valley, Antarctica. Glob. Plan. Change 22,

105–115.

Lyons W. B., Welch K. A., Fountain A. G., Dana G., Vaughn B. H. and McKnight D. M. (2003)

Surface glaciochemistry of Taylor Valley, Southern, Victoria Land, Antarctica, and its

relationship to stream chemistry. Hydro. Proc. 17, 115–130.

410 SARAH K. FORTNER ET AL.

Lyons W. B., Welch K. A., Neumann K., Toxey J. K., McArthur R., Williams C., McKnight D.

and Moorhead D. (1998) Geochemical linkages among glaciers, streams and lakes within the

Taylor Valley, Antarctica. In Ecosystem Dynamics in a Polar Desert: The McMurdo Dry

Valleys, Antarctica (ed. J. C. Priscu), Vol. 72, pp.77–92. American Geophysical Union,

Antarctic Research Series.

McKnight D. M., Niyogi D. K., Alger A. S., Bomblies A., Conovitz P. A. and Tate C. M. (1999)

Dry valley streams in Antarctica: Ecosystems waiting for water. Bioscience 49, 985–995.

McKnightD.M., Alger A., Tate C.M., ShupeG. and Spaulding S. (1998) Longitudinal patterns in

algal abundance and species distribution in meltwater streams in Taylor Valley, Southern

Victoria Land, Antarctica among glaciers, streams and lakes within the Taylor Valley,

Antarctica. In Ecosystem Dynamics in a Polar Desert: The McMurdo Dry Valleys, Antarctica

(ed. J. C. Priscu), Vol. 72, pp.109–127. American Geophysical Union, Antarctic Research

Series.

Mueller D. R., Vincent W. F., Pollard W. H. and Fritsen C. H. (2001) Glacial cryoconite eco-

systems: A bipolar comparison of algal communities and habitats. Nova Hed. 123, 173–197.

Mullin and Riley (1955) The estimation of silicate–silicon in natural waters. Anal. Chim. Acta 12,

162–176.

Nezat C. A., Lyons W. B. and Welch K. A. (2001) Chemical weathering in streams of a polar

desert (Taylor Valley, Antarctica). GSA Bull. 113, 1401–1408.

Sharp M., Skidmore M. and Nienow (2002) Seasonal and spatial variations in the chemistry of a

High Arctic supraglacial snow cover. J. Glac. 48, 149–158.

Sharp M., Tranter M., Brown G. H. and Skidmore M. (1995) Rates of chemical denudation and

CO2 drawdown in a glacier-covered alpine catchment. Geol. 23, 61–64.

Stumm W. and Morgan J. J. (1996) Aquatic Chemistry. Chemical Equilibria and Rates in Natural

Waters, 3rd edn. Wiley, New York.

Takeuchi N., Kohishima S., Yoshimura Y., Seko K. and Fujita K. (2000) Characteristics of

cryoconite holes on a Himalayan glacier, Yala Glacier Central Nepal. Bull. Glac. Res. 17,

51–59.

Tranter M., Fountain A. G., Fritsen C. H., Lyons W. B., Statham P. and Priscu, J. C. (2004)

Extreme hydrochemical conditions in natural microcosms entombed within Antarctic ice.

Hydro. Proc. 18, 379–387.

Tranter M., Sharp M., Lamb H. R., Brown G. H., Hubbard B.P. and Willis, I. C. (2002)

Geochemical weathering at the bed of Haut Glacier d’Arolla, Switzerland- a new model.

Hydro. Proc. 16, 959–993.

Tranter M., Sharp M. J., Brown G. H., Willis I. C., Hubbard B. P., Nielsen M. K., Smart C. C.,

Gordon S., Tulley M. and Lamb H. R. (1996a) Variability in the chemical composition of in

situ subglacial meltwaters. Hydro. Proc. 11, 59–77.

Tranter M, Brown G. H., Hodson A. J. and Gurnell A. M. (1996b) Hydrochemistry as an

indicator of subglacial drainage system structure: A comparison of alpine and sub-polar

environments. Hydro. Proc. 10, 541–556.

Tranter M., Brown G. H., Hodson A., Gurnell, A. M. and Sharp M. J. (1994) Variations in the

nitrate concentration of glacial runoff in alpine and sub-polar environments. Snow and Ice

Covers: Interactions with the Atmosphere and Ecosystems. No. 223.

Tranter M., Brown G. H., Raiswell R., Sharp M. J. and Gurnell A. M. (1993) A conceptual

model of solute acquisition by glacial meltwaters. J. Glac. 39, 573–581.

Vaughn, B. H., Wuite J. W. C., Delmotte M., Trolier, M., Cattani, O. and Stievenard M.

(1998) An automated system for hydrogen isotope analysis of water. Chem. Geol. 152(2–4),

309–318.

Welch K. A. (1993) Glaciochemical investigations of the Newall Glacier, Southern Victoria

Land, Antarctica. M.S. Thesis, University of New Hampshire. pp. 1–99.

THE GEOCHEMISTRY OF SUPRAGLACIAL STREAMS OF CANADA GLACIER 411

Welch K. A., Lyons W. B., Graham E., Neumann K., Thomas J .M. and Mikesell D. (1996)

Determination of major element chemistry from Antarctica by ion chromatography. J.

Chromatogr. A. 739, 257–263.

Wharton Jr. R. A., McKay C. P., Simmons G. M. and Parker B. C. (1985) Cryoconite holes on

glaciers. Bioscience 35, 499–503.

Wang P. and Kraus N. C. (1999) Horizontal water trap for measurement of aeolian sand

transport. Earth Surf. Proc. Land 24, 65–70.

412 SARAH K. FORTNER ET AL.