Notice: Wiley Online Library will be unavailable on Saturday 30th July 2016 from 08:00-11:00 BST / 03:00-06:00 EST / 15:00-18:00 SGT for essential maintenance. Apologies for the inconvenience.
 Soluble salt accumulations in soils of Taylor Valley, Antarctica, provide a history of paleolakes and the advance of the Ross Sea Ice Sheet (RSIS). In western Taylor Valley, soluble salt accumulations are relatively high and are composed primarily of Na+, Ca2+, Cl–, and SO42–. In eastern Taylor Valley, soluble salt accumulations are much lower and are composed primarily of Na+ and HCO3–. Na-HCO3-rich compositions in eastern Taylor Valley are formed through leaching, calcite dissolution, and cation exchange reactions and appear to influence the chemistry of nearby streams and lakes. The data presented here support hypotheses that a lobe of the RSIS expanded into eastern Taylor Valley and dammed proglacial paleolakes. However, in contrast to previous studies, our findings indicate that the RSIS advanced deeper into Taylor Valley and that paleolakes were less extensive. By comparing soluble salt distributions across Taylor Valley, we conclude that a lobe of the RSIS filled all of eastern Taylor Valley and dammed paleolakes in western Taylor Valley up to approximately 300 m elevation. Following ice retreat, smaller paleolakes formed in both western and eastern Taylor Valley up to about 120 m elevation, with prominent still-stands controlled by the elevation of major valley thresholds. At higher elevations, soluble salt accumulations are consistent with older soils that have not been affected by the most recent RSIS advance.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
 Soluble salt accumulations are common in the hyperarid soils of the McMurdo Dry Valleys, Antarctica, and have been studied for over 40 years [Campbell and Claridge, 1987; Claridge and Campbell, 1968; Keys and Williams, 1981]. Soils accumulate salts over time from the input of marine sea-spray aerosols [Claridge and Campbell, 1968, 1977; Keys and Williams, 1981], oxidized nitrogen and sulfur compounds deposited on the East Antarctic Ice Sheet [Bao et al., 2000; Claridge and Campbell, 1968; Michalski et al., 2005], and chemical weathering [Claridge and Campbell, 1977; Keys and Williams, 1981]. These salts are reworked across the landscape by aeolian processes [Fortner et al., 2005; Lyons et al., 2003; Witherow et al., 2006] and leaching [Hagedorn et al., 2010; Keys, 1980; Wilson, 1979]. Salt accumulations have been used as an indicator of relative soil age [Bockheim, 1979, 1982, 1990; Everett, 1971; Pastor and Bockheim, 1980] because of hyperarid conditions that preserve salts in Dry Valley soils, potentially over million-year timescales [Bockheim, 1983; Hall et al., 1993; Marchant et al., 1994]. Soluble salt accumulations also can be modified during inundation by lakes and are useful for studying past lacustrine events [Barrett et al., 2010; Bockheim et al., 2008a; Field, 1975; Morikawa et al., 1975; Poage et al., 2008], such as those that are thought to have occurred in Taylor Valley.
 During Marine Isotope Stage (MIS) 2 (14.1 to 27.6 Ka), the Ross Sea Ice Sheet (RSIS) filled McMurdo Sound and abutted the Transantarctic Mountains near the McMurdo Dry Valleys [Denton et al., 1970; Stuiver et al., 1981]. In front of the RSIS, proglacial lakes are believed to have formed in Taylor Valley [Hall et al., 2000; Péwé, 1960; Stuiver et al., 1981], Wright Valley [Hall et al., 2001], Victoria Valley [Hall et al., 2002], Marshall Valley [Hendy, 2000], and Miers Valley [Péwé, 1960] up to about 450 m above present lake levels. In Taylor Valley, contrasting theories have been proposed about the extent of paleolakes and the RSIS. According to Stuiver et al. , proglacial paleolakes were dammed to a maximum elevation of about 310 m by a lobe of the RSIS that entered deep into Taylor Valley, so that much of eastern Taylor Valley was filled with ice. Prentice et al.  proposed that the RSIS entered farther into eastern Taylor Valley than Stuiver et al. . In contrast, Hall et al.  proposed that the RSIS was grounded at the valley mouth for about 10 Ka, forming a stable ice dam that resulted in the filling of all eastern and western Taylor Valley with proglacial paleolakes up to approximately 350 m elevation.
 These scenarios have different implications for the paleoclimate during and after MIS 2, the sensitivity of the RSIS to deglacial sea level rise and warming trends, and past hydrologic and climatic regimes in Taylor Valley. In the interpretation of Stuiver et al. , paleolake fluctuations primarily reflect the movement of the RSIS in Taylor Valley as it advanced or retreated in response to deglacial sea level rise and climate change. During advances, the RSIS would have displaced lake water, raising lake levels; during RSIS retreat from Taylor Valley, lake water would have filled in areas vacated by the retreating ice, lowering lake levels [Stuiver et al., 1981]. In contrast, Hall et al.  proposed that lake levels fluctuated on the basis of changing climatic regimes, which influenced melt water production from the RSIS and surrounding alpine glaciers [Hall et al., 2010].
 Salt distributions have been used to study the history of paleolakes in Taylor Valley by Foley et al. , Bockheim et al. [2008a], and Barrett et al. . Foley et al.  mapped the distribution of pedogenic carbonates in near-surface soil horizons and attributed an increase in carbonates below 336 m elevation to lacustrine carbonate deposition. Bockheim et al. [2008a] analyzed the distribution of soluble salts in soils, primarily in western Taylor Valley, and concluded that soluble salt accumulations are influenced primarily by soil age, not inundation by paleolakes. This conclusion was based on the interpretation that soils inundated by lakes are leached of soluble salts; however, Barrett et al.  showed that salts can accumulate near wetted lake margins as lake water moves along water potential gradients towards an evaporation zone near the soil surface. Barrett et al.  studied salt accumulations near Lake Fryxell, Lake Hoare, and Lake Bonney by measuring the electroconductivity of soil-water mixtures and found evidence of salt accumulation along paleolake margins in all three areas. The influence of paleolakes on these salt accumulations was interpreted as a function of lake stability, lake salinity, soil texture, and leaching [Barrett et al., 2010].
 These past studies on salt distribution in Taylor Valley are limited by either the method of analysis or the extent of soils studied. The pedogenic carbonates studied by Foley et al.  are sparingly soluble and are not as sensitive to the influence of paleolakes as more soluble salts. The electroconductivity measurements in Barrett et al.  cannot be used to determine the ionic composition of salts. Bockheim et al. [2008a] analyzed a large soluble salt dataset from soils in Taylor Valley; however, few soils were sampled in eastern Taylor Valley, and most soils were located at high elevations, above paleolake high-stands proposed by either Stuiver et al.  or Hall et al. . To use soluble salts as a tool to study the glacial and lacustrine history of Taylor Valley, a more detailed and comprehensive understanding of salt distribution is needed.
 In this paper, we present data on the water-soluble salt contents of 89 soils throughout Taylor Valley. By comparing salt distributions along elevational transects to modern salt accumulations and distributions along lake and ice margins, we present new evidence regarding the extent of the RSIS and paleolakes in Taylor Valley. Our results suggest that the RSIS advanced deep into Taylor Valley, and that proglacial paleolakes did not fill all of Taylor Valley, but formed within separate basins in eastern and western Taylor Valley.
2.1 Study Sites
 The McMurdo Dry Valleys are cold-dry deserts that are generally ice-free, with the exception of local alpine glaciers and ice-covered lakes. Minimal penetration by glacial ice is due to the Transantarctic Mountains, which divert ice from East Antarctica away from McMurdo Sound [Chinn, 1990]. Annual temperatures in the Dry Valleys average about –20°C [Doran et al., 2002], and annual precipitation rates are less than 100 mm in the form of snow [Bromley, 1985; Fountain et al., 2010]. Taylor Valley is the lowest elevation Dry Valley and is exposed to seasonally open seawater, making Taylor Valley the wettest and warmest of the McMurdo Dry Valleys [Doran et al., 2002]. Annual temperatures average –17°C, with average winter temperatures near –30°C and average summer temperatures near 0°C [Clow et al., 1988; Doran et al., 2002].
 Taylor Valley is bounded by Taylor Glacier, an outlet of the East Antarctic Ice Sheet, on the west, and by McMurdo Sound to the east (Figure 1A). Within Taylor Valley, cold-based alpine glaciers descend from the Asgard Range and Kukri Hills. During the austral summer, melt water from these glaciers feeds numerous ephemeral streams that drain into closed-basin lakes [McKnight et al., 1999]. The largest of these lakes are Lake Fryxell at 18 m elevation, Lake Hoare at 73 m elevation, and Lake Bonney at 58 m elevation [Spigel and Priscu, 1998]. These closed-basin lakes are sensitive to stream flow variations during warmer and cooler summers [Conovitz et al., 1998; Doran et al., 2008]. Since lake levels were first recorded in Taylor Valley by Scott , lake levels have risen because of greater melt water inputs from surrounding alpine glaciers [Bomblies et al., 2001; Chinn, 1981, 1993].
 Soil transects presented here were sampled in three distinct regions: soils near the valley mouth (Valley Mouth), soils within the closed drainage basin of Lake Fryxell (Fryxell Basin), and soils within the closed drainage basin of Lake Bonney (Bonney Basin) (Figure 1B). These regions are grouped into soils that occur in western Taylor Valley (Bonney Basin soils) and eastern Taylor Valley (Fryxell Basin and Valley Mouth soils). Fryxell Basin and Bonney Basin are the largest closed drainage basins within Taylor Valley and are distinctly different from each other. Bonney Basin is narrow (about 3 km wide) with steep valley walls, while Fryxell Basin is broad (about 10 km wide) with gentle north-facing slopes and steep south-facing slopes. Fryxell Basin and Bonney Basin are separated by a 116-m-high threshold below the terminus of the Suess Glacier (Figure 2) [Kellogg et al., 1980]. Near this threshold, much of Taylor Valley is occupied by a distinct bedrock hill called the Nussbaum Riegel, which constricts the valley bottom into a narrow defile. The Nussbaum Riegel reaches an elevation of about 700 m and limits the penetration of moist air masses from McMurdo Sound, making Bonney Basin drier than Fryxell Basin [Fountain et al., 1999]. As a result, average snow accumulation rates in Taylor Valley decrease from about 70 mm yr–1 near the Valley Mouth to about 10 mm yr–1 near Lake Bonney [Fountain et al., 2010]. Fryxell Basin is separated from McMurdo Sound by Coral Ridge (Figure 2), a large moraine complex near the valley mouth that has an average elevation of about 100 m and is cut by a major channel at its lowest point of 78 m [Kellogg et al., 1980].
 Soils in eastern Taylor Valley have formed on glacial, lacustrine, and fluvial sediments deposited during the MIS 2 advance of the RSIS into Taylor Valley [Denton et al., 1970; Hall et al., 2000; Stuiver et al., 1981]. Older glacial deposits from the RSIS and Taylor Glacier crop out near the Nussbaum Riegel and above approximately 350 m elevation [Bockheim et al., 2008b; Denton et al., 1970; Péwé, 1960]. Soils in western Taylor Valley have formed on glacial deposits from older advances of Taylor Glacier and alpine glaciations, with ages on the order of 0.1 to 4 Ma [Denton et al., 1970, 1989, 1993; Higgins et al., 2000; Krusic, 2009].
2.2 Sampling and Analysis
 Soils were sampled along five elevational transects: the Valley Mouth, Delta Stream 1, Delta Stream 2, Bent Stream, and Bonney Basin transects (Figure 1B). Additional soils were sampled from perched terraces in eastern Taylor Valley [Kellogg et al., 1980]. Samples were collected from soils at level sites on local topographic highs, such as paleoshorelines, moraine tops, and terrace apexes. Soils were sampled from each soil horizon to the depth of the hard ice-cemented soil; the top few cm of the ice-cemented soil were sampled using a chisel. Typically, soil pits were about 30 cm deep, except for several soils in Bonney Basin, where no ice-cement was present, and pits were excavated to 1 m depth. In general, if individual soil horizons were greater than 10 cm in thickness, the horizon was sampled at 10-cm intervals. For the Delta Stream 2 Transect, the top 10 cm of soil was bulk-sampled without regard to individual horizons. All samples were sieved in the field at 2 mm, and the weight of each fraction was measured. Laboratory tests were conducted on the <2 mm fraction.
 Past researchers have used a 1:5 soil-water extraction to measure soluble salts in Dry Valley soils [e.g., Bockheim, 1979; Claridge and Campbell, 1977; Gibson et al., 1983], but this may underestimate the amount of some ions [Bao et al., 2000]. To determine the most efficient and complete extraction method, soluble salts were extracted from sample no. 159 (a soil with a relatively high concentration of soluble salt) using four different soil-water ratios: 1:10, 1:25, 1:50, and 1:100. Soluble salts were extracted sequentially three times; for each extraction cycle, the soil-water mixture was shaken for 1 hour, centrifuged, and the supernatant decanted. Based on the results of these extractions (Figure 3, Table 1), soluble salts were measured using three sequential 1:25 soil-water extractions.
Table 1. Ion Concentrations (mmol kg–1) Measured in Individual Extractions of the Sequential Soil-Water Extractions
Ca2+ (mmol kg–1)
 Chemical analysis for major cations (Ca2+, Mg2+, Na+, K+) was conducted using inductively coupled plasma-optical emission spectrometry (Perkin-Elmer® Optima 3300DV) and major anions (Cl–, SO42–, NO3–) using ion chromatography (Dionex® ICS-2500). Total inorganic carbon was determined using an OI700 Carbon Analyzer, which acidifies the sample and measures evolved CO2. All inorganic carbon in the soil-water extractions is assumed to be bicarbonate (HCO3–), because the pH of extracting solutions is around pH 7. The accuracy of this analysis was checked by calculating the charge balance. In addition to the analysis of major soluble ions, soil pH was determined by shaking a 1:2 soil-water mixture for 1 hour, then allowing the mixture to rest for 1 hour before measuring the pH.
 To calculate the soluble salt content in soils, it is necessary to determine the soil bulk density. Past studies have assumed bulk densities of 1.5 or 1.6 g cm–3 to calculate soluble salt contents [e.g., Bockheim, 1979], but this can underestimate the bulk density in rocky soils. Following the method of Burnham and Sletten , we correct for the effect of gravel and cobbles on the bulk density by calculating the whole soil bulk density (BD) from the percent weight (%w) of the <2 mm soil fraction, the bulk density of the <2 mm soil fraction (ρ<2 mm), and the average grain density (ρg):
 The grain density was measured using volume as determined from water displacement for several samples, which yielded a constant grain density of 2.72 g cm–3. Because most soils in Taylor Valley are sandy, we use a bulk density of 1.55 g cm–3 for the <2 mm fraction, a value typical for sandy soils [Rawls, 1983]. To check the accuracy of whole soil bulk density calculations, 15 soils were measured for bulk density using a tin can with a diameter of 15 cm. The average measured bulk density for these soils was 1.76 ± 0.16 g cm–3, while the average calculated bulk density using equation ((1)) was 1.73 ± 0.10 g cm–3. The close agreement between measured and calculated values indicates that the bulk density corrections employed here are reasonable.
 Salt contents in the soil (equivalents m–2) are calculated by summing over all soil horizons the product of the salt concentration C (meq kg–1), the horizon thickness t (m), the whole soil bulk density BD (g cm–3), and the weight fraction of the <2 mm soil fraction for each soluble ion [Burnham and Sletten, 2010]:
 This equation assumes that all of the soluble salt in the soil is held in the <2 mm soil fraction. Total soluble salt contents are the equivalent sum of all anions and cations with depth.
 The results of this study are combined with soluble salt data compiled by Bockheim  (available at http://nsidc.org/data/ggd221.html) and data from I. B. Campbell and G. G. C. Claridge compiled by Gibb et al.  in the Antarctic Soil Database (available at http://soils.landcareresearch.co.nz/contents/index.aspx). Soluble salt contents in Bockheim  and Gibb et al.  are calculated from soluble salt and soil texture data using equation ((2)) and the whole soil bulk density in equation ((1)). Soil texture data was not available for some soil pits in Bockheim . For these soils, it was assumed that the bulk density was 1.75 g cm–3, the average in this study. Because inorganic carbon, and often nitrate, was not measured in either Bockheim  or Gibb et al. , total soluble salt contents are calculated as the equivalent sum of the cations multiplied by two. This will give the total equivalent sum of anions and cations, because charge balance requires that the equivalent sum of cations will equal the equivalent sum of anions.
3.1 Soluble Salt Extraction Procedure
 Tests on different soil-water extraction procedures show that measured soluble salts are strongly dependent on both the soil-water ratio and the number of sequential extractions performed (Figure 3, Table 1). Total equivalent ion concentrations (the equivalent sum of cations in the three sequential extractions multiplied by two) increase from 115 meq kg–1 in the 1:10 extraction procedure to 154 meq kg–1 in the 1:100 extraction procedure (Figure 3). Individually, all ion concentrations increase in higher soil-water extraction procedures except Cl–. Na+ concentrations are slightly lower in the 1:10 extraction procedure and plateau in the 1:25, 1:50, and 1:100 extractions. SO42−, Ca2+, Mg2+, and K+ concentrations do not plateau but increase at higher soil-water extraction procedures. In particular, Ca2+ concentrations increase by an order of magnitude from the 1:10 to the 1:100 soil-water extraction procedures.
 With respect to ion concentrations measured in each sequential soil-water extraction, nearly all Cl– and SO42– in the soil was extracted in the first sequential extraction, independent of the soil-water ratio used (Table 1). In contrast, significant concentrations of Ca2+, Mg2+, Na+, and K+ were liberated in the second and third sequential extractions. Na+ and K+ ions liberated in subsequent extractions are characterized by decreasing concentrations, while Ca2+ and Mg2+ increase in subsequent extractions. Based on total ion concentrations measured in the three sequential 1:100 soil-water extraction procedure (as shown in Figure 3), we calculate that, if only a single 1:10 soil-water extraction is performed, the total Na+ content in the soil will be underestimated by 31%, K+ content will be underestimated by 75%, Mg2+ content will be underestimated by 80%, and Ca2+ content will be underestimated by 97%. However, Ca2+, Mg2+, Na+, and K+ are high even in the third sequential 1:100 soil-water extraction, which suggests that the 1:100 extraction procedure also underestimates total ion concentrations in the soil.
 These tests on extraction methods show that it is difficult to completely extract all soluble ions from the soil, even using sequential extractions at high soil-water ratios. The sequential 1:25 soil-water extraction procedure was selected for this study because it is clear that sequential extractions remove more soluble Na+, K+, and Mg2+ ions from the soil, and total extracted Na+ plateaus after the 1:25 extraction procedure. The extraction of nearly all Cl– and possibly SO42− in the first sequential extraction indicates that these ions will be comparable between the sequential 1:25 soil-water method used in this study and the 1:5 soil-water method used by Bockheim  and Gibb et al. . In contrast, the extraction of significant concentrations of Na+, K+, Mg2+, and Ca2+ in the second and third sequential extractions indicates that these ion concentrations will vary depending on the extraction method.
3.2 Total Soluble Salt Content Distributions
 The distribution of total soluble salt contents in Taylor Valley is shown in Figure 4 using the datasets of Bockheim , Gibb et al. , and data from this study. Soils in this study are located primarily in the Fryxell Basin and Valley Mouth regions. Soils from Bockheim  and Gibb et al.  are located primarily in Bonney Basin. Some of the variability in soluble salt contents in Figure 4 may be due to greater dissolution of gypsum and calcite in the 1:25 soil-water extraction used in this study than in the 1:5 soil-water extraction used by Bockheim  and Gibb et al. . However, measured SO42− concentrations are typically low in this study, and carbonate dissolution can account only for about 20 eq m–2 of the total salt content (based on an average HCO3– content of 10 eq m–2 and assuming that all HCO3– is associated with equivalent concentrations of cations). Soluble salt contents in Figure 4 may also be influenced by the depth of soil pits. In Bonney Basin, the absence of ice-cemented soil allowed many soils to be excavated two to three times deeper than in eastern Taylor Valley, where ice-cemented soil occurs at about 30 cm depth. Greater pit depth could increase salt contents measured in Bonney Basin soils relative to eastern Taylor Valley. However, given the broad categories for salt contents in Figure 4, variability in total salt contents because of different extraction methods and pit depth is small relative to the value range of the categories.
 The dominant characteristic of soluble salt distributions in Taylor Valley is the occurrence of low salt contents in eastern Taylor Valley (averaging 50 eq m–2) and high salt contents in western Taylor Valley (averaging 900 eq m–2). It is believed that salt accumulations in Dry Valley soils are related directly to soil age [e.g., Bockheim, 1979] and, in general, salt distributions are consistent with differences in soil age between eastern and western Taylor Valley. In western Taylor Valley, soils are from MIS 5 or older advances of Taylor Glacier and alpine glaciations; while, in eastern Taylor Valley, soils are primarily from the MIS 2 advance of the RSIS [Bockheim et al., 2008b].
 There are a number of exceptions to this overall trend in salt content; some older soils have relatively low salt contents, and some younger soils developing on RSIS sediments have relatively high salt contents. In western Taylor Valley, older soils with low salt contents are typically ice-cemented near the surface, which suggests a higher local moisture regime that may have leached soluble salts from these soils. In eastern Taylor Valley, ice-cemented soil is ubiquitous, and salts have likely been leached from older soils due to the wetter climate or episodic snowmelt events [Hagedorn et al., 2010]. Relatively high salt contents in young RSIS sediments are primarily found below approximately 120 m elevation in eastern Taylor Valley. Barrett et al.  found similar high salt accumulations near Lake Fryxell and Lake Hoare and concluded that these salt accumulations were deposited at the margins of paleolakes.
3.3 The Chemical Composition of Salt Accumulations
 Average and median ion contents in Valley Mouth, Fryxell Basin, and Bonney Basin soils are given in Table 2. Only data from this study are considered in this section, because tests on different soil-water extraction procedures indicate that the different soil-water extraction procedure used in Bockheim  and Gibb et al.  will have a large effect on Ca2+, Mg2+, Na+, K+, and possibly SO42− concentrations. In eastern Taylor Valley, the concentration of all ionic species is low compared to western Taylor Valley, which reflects trends in total salt contents (Table 2). The greatest differences in ion contents between these two regions are found in Ca2+, Mg2+, Na+, SO42−, Cl–, and NO3–; while K+ and HCO3– contents are only slightly higher in western Taylor Valley.
Table 2. Average and Median Salt Contents of Major Cations and Anions and pH for Valley Mouth, Fryxell Basin, and Bonney Basin Soils
 The composition of salts and soil pH varies with distance from McMurdo Sound. Valley Mouth soils are composed almost entirely of Na+ and HCO3– ions, and the average soil pH is 9.7 (Table 2). Farther inland, in Fryxell Basin, Na+ and HCO3– are still the dominant ions, and the average soil pH is 9.6, but soils contain higher proportions of Ca2+ and Mg2+. In Bonney Basin, located farthest from McMurdo Sound, soils have high proportions of Cl– and SO42− relative to HCO3–, and the average soil pH decreases to 8.5. The trend in soil pH in Taylor Valley is similar to trends observed by Claridge and Campbell ; soils near the coast are alkaline with pH values around 9; while, near the East Antarctic Ice Sheet, pH decreases to as low as 6. Claridge and Campbell  suggested that this is due to the accumulation of slightly alkaline marine aerosols near the coast and slightly acidic aerosols near the East Antarctic Ice Sheet.
 Ionic ratios of soluble salts can be used to determine the source of salts to soils in Taylor Valley (Table 3). Near McMurdo Sound, average SO42−/Cl– and NO3–/Cl– ratios in soils are closer to ratios in seawater; while, farther from McMurdo Sound, SO42−/Cl– and NO3–/Cl– ratios increase. These trends in SO42−, NO3–, and Cl– with distance from McMurdo Sound are similar to findings by Keys and Williams  and are thought to be due to the influence of Cl–-rich marine sea-spray aerosols near McMurdo Sound and SO4–-NO3–-rich snow on the East Antarctic Ice Sheet [Claridge and Campbell, 1968]. SO42− and NO3– on the East Antarctic Ice Sheet are derived from nitrogen and sulfur compounds that have been oxidized in the upper atmosphere [Bao et al., 2000; Michalski et al., 2005]. These ions are precipitated with snow onto the East Antarctic Ice Sheet, which is carried down to the Dry Valleys by katabatic winds [Claridge and Campbell, 1968; Fountain et al., 2010]. SO42– also may accumulate in Taylor Valley soils from chemical weathering [Bao and Marchant, 2006; Claridge and Campbell, 1977; Kelly and Zumberge, 1961] and SO42–-rich aeolian dust from soils in western Taylor Valley [Fortner et al., 2005].
Table 3. Average Ionic Ratios in Taylor Valley Soils Compared to Ratios in Seawater
 With respect to other ionic ratios, soluble salts in Taylor Valley soils generally vary from seawater compositions. In Bonney Basin soils, Na+/Cl– ratios are similar to seawater but Ca2+/Cl–, Mg2+/Cl–, K+/Cl–, and HCO3–/Cl– ratios are higher than in seawater. In Fryxell Basin and Valley Mouth soils, Ca2+/Cl–, Mg2+/Cl–, Na+/Cl–, K+/Cl–, and HCO3–/Cl– ratios are 1 to 2 orders of magnitude higher than in seawater (Table 3). The large difference in soil Na+/Cl– ratios from the seawater ratio is surprising, because Na+/Cl– ratios in Taylor Valley glaciers, streams, and lakes are close to seawater ratios [Lyons et al., 1998]. Furthermore, Valley Mouth soils should be the most strongly influenced by marine sea-spray aerosols, but Na+/Cl– ratios in these soils deviate the most from the seawater ratio. As discussed in the sections below, ion ratios and compositions in these soils can be explained by calcite/gypsum dissolution and cation exchange reactions.
4.1 The Influence of Calcite/Gypsum Dissolution and Cation Exchange on Soil-Water Extractions
 Tests on different soil-water extraction procedures indicate that the measurement of soluble salts is strongly dependent on the soil-water ratio and the number of sequential extractions that are performed. The effect of different soil-water extraction procedures on soluble salts has been studied extensively by Eaton and Sokoloff , Kelley , and Reitemeier . These authors found that the measurement of soluble ions in soil-water extractions is complicated by calcite/gypsum dissolution and cation exchange reactions. In particular, the dissolution of sparingly soluble calcite or gypsum increases Ca2+, SO42–, and HCO3– concentrations in soil-water extractions and causes Ca2+ to exchange with Na+, K+, and Mg2+ on the exchange complex. By increasing the water content in soil-water extractions, greater calcite/gypsum dissolution occurs, which leads to even greater exchange of Ca2+ for Na+, K+, and Mg2+.
 Soil-water extractions in this study are likely affected by calcite dissolution and cation exchange, because calcite is ubiquitous in Taylor Valley soils [Foley et al., 2006; Keys, 1980; Nishiyama and Kurasawa, 1975] and soluble salts are measured in dilute soil-water extractions. Calcite dissolution and cation exchange can be described by the following generalized reaction:
where X is an exchange site associated with an adsorbed cation. The increase in Ca2+, Mg2+, Na+, K+, and HCO3– concentrations predicted by equation ((3)) is consistent with high ratios of Ca2+, Mg2+, Na+, K+, and HCO3– relative to Cl– in soil-water extractions of eastern Taylor Valley. Furthermore, the alkalinity produced by this reaction has been shown to raise soil pH [Cruz-Romero and Coleman, 1975; Gupta et al., 1981; Reitemeier, 1946] and is consistent with the high soil pH values measured in this study.
 To test the hypothesis that calcite/gypsum dissolution and cation exchange is affecting the composition of soil-water extractions, soluble salts in 33 samples were measured in a single 1:10 soil-water extraction (shaken for 1 hour) and then compared to cations measured in the three sequential 1:25 soil-water extractions (Figure 5). The results of this experiment show that all cation concentrations are higher in the sequential 1:25 soil-water extractions and that Mg1:25/Mg1:10, K1:25/K1:10, and Na1:25/Na1:10 ratios increase linearly with increasing Ca1:25/Ca1:10 ratios. On average, the increase in Ca2+, Mg2+, Na+, and K+ concentrations in the 1:25 soil-water extraction relative to the 1:10 soil-water extraction is 6.95, 1.90, 5.79, and 2.41 mmol kg–1, respectively. This is consistent with higher Ca2+ concentrations in the 1:25 extraction because of increased calcite/gypsum dissolution and the exchange of this Ca2+ with exchangeable Na+, K+, and Mg2+. Cation1:25/cation1:10 ratios shown in Figure 5 fall into distinct groups; in general, Mg1:25/Mg1:10 > K1:25/K1:10 > Na1:25/Na1:10. This relationship suggests the relative affinity of Mg2+, K+, and Na+ for exchange sites in the order Mg2+ > K+ > Na+. For example, the higher Mg1:25/Mg1:10 ratio indicates that a higher proportion of Mg2+ is removed from the soil in the 1:25 extraction relative to the 1:10 extraction and suggests that a higher proportion of Mg2+ remained on the exchange complex in the 1:10 extraction. Relative cation affinities for exchange sites are also consistent with how quickly and completely Mg2+, K+, and Na+ are extracted from soils in sequential water extractions (Table 1). Nearly all the Na+ is removed in the first extraction, while high proportions of Mg2+ and K+ remain in subsequent extractions.
 Although both calcite/gypsum dissolution and cation exchange will influence the chemical composition of soil-water extractions, only calcite/gypsum dissolution will affect total salt contents measured in equivalents because charge is conserved during exchange reactions. Furthermore, the effects of calcite/gypsum dissolution and cation exchange on soil-water extractions will be limited by the solubility of calcite/gypsum and the Cation Exchange Capacity (CEC). Since the CEC of Antarctic soils is generally low, averaging about 40 meq kg–1 for sandy soils [Cameron and Conrow, 1969; Cameron et al., 1970; Cameron et al., 1971], the addition of exchangeable Mg2+, Na+, and K+ to soil-water extractions will be limited to a maximum of about 40 meq kg–1. In soils with high soluble salt accumulations, such as in Bonney Basin, ions from calcite/gypsum dissolution and cation exchange will have a relatively small effect on the total ion content. This is consistent with ratios of Ca2+, Mg2+, Na+, K+, and HCO3– relative to Cl– in Bonney Basin that are closer to seawater ratios. In contrast, calcite/gypsum dissolution and cation exchange effects should dominate the chemistry of soil-water extractions in soils with low soluble salt accumulations, such as occur in eastern Taylor Valley.
4.2 Na-HCO3-rich Soils in Eastern Taylor Valley
 The influence of calcite/gypsum dissolution and cation exchange observed in soil-water extractions suggests that these reactions also occur naturally in Taylor Valley soils to produce Na-HCO3-rich soil solutions. The formation of Na-HCO3-rich soils has been studied extensively by authors investigating alkali (sodic) soils [e.g., de' Sigmond, 1927; Gedroiz, 1912; Kelley and Cummins, 1921] and is reviewed by Kelley . In calcareous soils, such as occur in eastern Taylor Valley, Na-HCO3-rich soils are thought to develop through a series of steps that include leaching with Na+-rich salt solution, calcite dissolution, and cation exchange. This process was first demonstrated over a century ago by Mondésir . By leaching a soil with Na+-rich salt solution, such as NaCl, the exchange complex becomes saturated in Na+. As Na+ adsorbs onto exchange sites, exchangeable Ca2+, Mg2+, and K+ is displaced into the soil solution and leached from the soil. When wetted, calcareous soils with high proportions of exchangeable Na+ follow the reaction in equation ((3)); Ca2+ from calcite dissolution exchanges with exchangeable Na+, producing a Na-HCO3-rich soil solution [Gedroiz, 1912]. Na-HCO3-rich soil solutions can also form enrichment can also occur in the absence of calcite from the hydrolysis of exchangeable Na+ [Gedroiz, 1912]:
 Although this hydrolysis reaction is important in some soils, Cummins and Kelley  demonstrated that Na-HCO3-enrichment formation in the presence of calcite is far greater than what can be produced by hydrolysis in the absence of calcite.
 In Taylor Valley, soils highly saturated with exchangeable Na+ likely form when Na-Cl-rich salts from marine aerosols leach down soil profiles following snowfall events [Hagedorn et al., 2010]. Because K+ ions are relatively minor in Dry Valley soils [Keys, 1980], downward-leaching Na+ will primarily displace exchangeable Ca2+ and Mg2+:
 The Ca-Mg-Cl-rich leachate predicted by this reaction may contribute to Ca-Mg-Cl-rich brines that have been found in groundwater seeps throughout Taylor Valley [Levy et al., 2011, 2012]. Furthermore, Na+ adsorbed onto the exchange complex in reaction (5) is no longer associated with Cl–, which is leached from the soil along with Ca2+ and Mg2+. When the soil is wetted, this exchangeable Na+ can exchange with Ca2+ from calcite dissolution to form Na-HCO3-rich soil solutions. The evaporation of Na-HCO3-rich solutions following wetting events could account for trona (NaHCO3·Na2CO3·2H2O) and thermonatrite (Na2CO3·H2O) salt encrustations that occur in soils of eastern Taylor Valley [Nishiyama and Kurasawa, 1975]. Similar leaching, calcite/gypsum dissolution, and exchange processes have been found to occur in Egyptian desert soils by Smettan and Blume . In these soils, downward-leaching Na+ from the soil surface has resulted in Na-HCO3-rich salts near the soil surface and the migration of Ca-Mg-Cl-NO3-rich brine to depth [Smettan and Blume, 1987].
 The process of Na-HCO3-enrichment outlined above is dependent on leaching and wetting of the soil; leaching with Na+-rich salt saturates exchange sites with Na+, and wetting soils highly saturated with exchangeable Na+ forms Na-HCO3-rich soil solutions. The dependence of Na-HCO3-enrichment on moisture suggests that Na-HCO3-rich compositions will occur in regions with precipitation rates high enough to cause salt leaching. In eastern Taylor Valley, precipitation rates are relatively high due to the influence of moisture from McMurdo Sound [Fountain et al., 2010], which is consistent with the prevalence of Na-HCO3-rich salts in eastern Taylor Valley. In addition, Na-HCO3-rich salts are often associated with the presence of calcite because of the combined effects of calcite dissolution and cation exchange [Kelley, 1951]. Although calcite is found in many soils of the Dry Valleys, it is particularly prevalent in eastern Taylor Valley because of the presence of marble in RSIS sediments, which dissolves and reprecipitates as calcite pendants beneath soil clasts [Campbell and Claridge, 1987]. Hence, the climate and mineralogy of soils in eastern Taylor Valley are well suited to the formation of soluble Na-HCO3-rich salts.
 Stream and lake waters in eastern Taylor Valley also have distinctive Na-HCO3-rich compositions [Green et al., 1988], which suggests that soils influence water chemistries in eastern Taylor Valley. Na-HCO3-rich soil solutions may be leached from soils into nearby streams and lakes during wetting events. In addition, wind could disperse Na2CO3-NaHCO3 salts and soil particles that are highly saturated with exchangeable Na+ into nearby stream and lake waters [Lancaster, 2002]. Alternatively, stream waters eroding into sediments or the reactivation of relict stream channels [McKnight et al., 2007] would enrich stream waters in Na+ and HCO3− because the introduction of dilute water to these soils is roughly analogous to a soil-water extraction. The flux of Na-HCO3-rich stream waters into closed-basin Lake Fryxell is thought to control the chemical evolution of Lake Fryxell, resulting in its Na-HCO3-rich composition [Green et al., 1988; Lyons et al., 1998]. Similarly, in western Taylor Valley, the Na-Ca-Cl-SO4-rich composition of soil-water extractions is reflected in stream [Welch et al., 2010] and lake waters [Angino et al., 1964]. These spatial relationships between soil chemistries, stream waters, and lake waters suggest that soils have a strong influence on solutes found in Dry Valley hydrologic systems.
4.3 Taylor Valley Paleolakes
 Soluble salts in Taylor Valley soils are thought to be sensitive indicators of past glacial and lacustrine events, such as the MIS 2 advance of the RSIS into Taylor and associated paleolakes [Barrett et al., 2010; Bockheim et al., 2008a]. To interpret relict soluble salt distributions from these events, we consider present-day salt distributions along lake and glacier margins.
 Lake water is thought to influence soluble salt accumulations in Dry Valley soils by leaching inundated soils during lake high-stands [Bockheim et al., 2008a; Keys, 1980] and accumulating soluble salts along wetted lake margins as lake levels lower [Barrett et al., 2009, 2010; Poage et al., 2008]. Salt accumulation along lake margins is driven by capillary forces that draw nearby lake water upwards against gravity, wetting soils up to 30 m horizontally and 1 m vertically from the lake surface [Gooseff et al., 2007; Northcott et al., 2009]. In a process that has been termed “evapoconcentration,” this lake water evaporates near the soil surface, concentrating salts dissolved in the lake water into the upper soil [Barrett et al., 2009]. Soluble salts in soils influenced by evapoconcentration have been sampled by Barrett et al.  to 10 cm depth. Cl– concentrations in these soils average about 100 mmol Cl– kg–1. Using this Cl– concentration and an average bulk density of about 1.65 g cm–3 as measured by Barrett et al. , the average Cl– content calculated from equation ((2)) is 16.5 mol Cl– m–2. This value is relatively high and is similar to Cl– contents measured in Bonney Basin soils, older soils near the Nussbaum Riegel, and soils below approximately 120 m elevation in eastern Taylor Valley.
 Soils adjacent to streams are also influenced by evapoconcentration, but these soils contain relatively low concentrations of salt compared to soils along lake margins [Barrett et al., 2009]. These lower salt concentrations are likely caused by the lower solute concentrations in streams relative to lakes [Lyons et al., 1998] and the instability of stream-wetted margins because of variable water fluxes [Barrett et al., 2009; Northcott et al., 2009]. Wetted stream margins sampled by Barrett et al.  have Cl– concentrations averaging about 5 mmol kg–1. The Cl– content of these soils calculated in equation ((2)) for a bulk density of 1.65 g cm–3, as above for wetted lake margins, is 0.5 mol Cl– m–2. This value is consistent with the low Cl– contents found in most soils of eastern Taylor Valley. Since melt water streams are commonly found along low-elevation glacier margins in the Dry Valleys [Atkins and Dickinson, 2008; Hambrey and Fitzsimons, 2010], we suggest that low salt concentrations associated with stream margins can be extrapolated to glacier margins.
 Using the distinct differences in salt content produced at lake, stream, and possibly glacier margins, we propose a delineation of past lacustrine and glacial margins in Taylor Valley; soils influenced by paleolake margins will have high salt accumulations relative to soils influenced by streams near glacial margins. The lake and ice margins determined this way will represent the last episode of lake or ice retreat, because any soluble salts deposited prior to the last event will have been redistributed. We expect that relict soluble salt accumulations from past glacial and lacustrine events will be somewhat altered by leaching, calcite/gypsum dissolution, and cation exchange reactions. This alteration will be more pronounced in the relatively low-salt-content soils of eastern Taylor Valley and less significant in the high-salt-content soils of Bonney Basin.
4.3.1 Bonney Basin
 In Bonney Basin, strandlines and lacustrine sediments provide evidence of paleolake margins with maximum high-stands between 306 and 350 m elevation [Hall et al., 2000; Stuiver et al., 1981]. Near the Rhone Glacier, lacustrine strandlines are etched into the valley walls from 116 to 306 m elevation [Stuiver et al., 1981]. Furthermore, terraces, which have been interpreted as deltas, are found throughout Bonney Basin up to 350 m elevation [Hall and Denton, 2000; Hall et al., 2010], and the elevation of several of these terraces coincides with lacustrine strandlines near the Rhone Glacier [Stuiver et al., 1981]. The age and elevation of these terraces is thought to record the timing and height of paleolake levels in Bonney Basin [Hall and Denton, 2000; Hall et al., 2010; Stuiver et al., 1981]. However, the interpretation that the terraces are deltas has been challenged by ground-penetrating radar (GPR) studies of terraces in eastern Taylor Valley [Arcone et al., 2008; Horsman, 2007; Horsman et al., manuscript in preparation]. Using GPR at frequencies of 50, 100, 400, and 900 MHz and interpreting reflectors as sedimentary bedding or unconformities, GPR shows that four out of five terraces between 20 and 220 m elevation lack the stratigraphic architecture characteristic of deltas. Although no terraces were studied by GPR in Bonney Basin, the results in eastern Taylor Valley cause uncertainty regarding whether terraces in Bonney Basin actually are deltas. Hall et al.  note that foreset beds are common in Bonney Basin terraces, but are rare in eastern Taylor Valley terraces. This suggests that terraces in Bonney Basin are more likely to be deltas than terrace, in eastern Taylor Valley, although this must be confirmed by further study of the internal structure of these deposits.
 The distribution of total soluble salt contents in Bonney Basin [this study; Bockheim, 2003; Gibb et al., 2002] is consistent with evidence of paleolakes up to approximately 300 m elevation (Figure 6). Soluble salts in Bonney Basin are characterized by relatively high salt contents above 300 m elevation (averaging 960 eq m–2), which sharply transition to relatively low salt contents below 300 m elevation (averaging 284 eq m–2). Soils above 300 m elevation in Bonney Basin have ages on the order of 0.1 to 4 Ma, and the high salt contents in these soils are thought to be derived from long-term accumulation of aerosols [Bockheim et al., 2008a]. Some of these older soils have relatively low salt contents and contain ice cement near the soil surface, suggesting salt leaching. Soils with ice cement are marked with a “+” symbol in Figure 6 and account for nearly all soils with low salt contents above 300 m elevation. Below 300 m elevation, soils in Bonney Basin are located on sediments deposited during the Taylor II glaciation between 113 and 120 Ka [Higgins et al., 2000]. The younger age of these soils suggests the possibility that the lower salt contents are due to differences in soil age; however, Taylor II soils also are found above 300 m elevation and have high salt contents similar to older soils [Bockheim et al., 2008a, 2008b]. Instead, the agreement between geomorphic evidence of paleolake high-stands and the sharp decrease in soluble salts near 300 m elevation strongly argues that soils below 300 m elevation in Bonney Basin were influenced by paleolakes.
 Paleolakes in Bonney Basin would have redistributed soluble salts in soils, leaching salts from inundated soils and accumulating salts by evapoconcentration along paleolake margins. The quantity of salt deposited by evapoconcentration is thought to be dependent primarily on paleolake stability [Barrett et al., 2009, 2010; Northcott et al., 2009], with greater salt accumulation from evapoconcentration occurring along more stable lake margins. Paleolakes in Bonney Basin above 116 m elevation would have been proglacial lakes dammed behind the RSIS as it advanced into eastern Taylor Valley. Proglacial lakes in such a system are dynamic and unstable, because lake levels would be sensitive to changes in glacial melt water inputs and the movement of the RSIS [Hall et al., 2010; Stuiver et al., 1981]. As a result, lake level lowering may have occurred rapidly at times, limiting evapoconcentration along paleolake margins. In contrast, paleolakes below 116 m elevation would have been closed-basin lakes controlled by the elevation of the Bonney-Fryxell threshold at 116 m elevation. Lowering of lake water in closed basins occurs slowly as lake water evaporates or lake ice sublimates [Clow et al., 1988], resulting in relatively stable lake levels and greater evapoconcentration in soils.
 Evapoconcentrated salt accumulations in Bonney Basin are characterized by a general trend of increasing salt contents at lower elevations (Figures 6 and 7), particularly below 116 m elevation, which is consistent with a transition from glacially dammed paleolakes above 116 m elevation to closed-basin paleolakes below 116 m elevation. The character of paleoshorelines in Bonney Basin provides additional evidence that the paleolakes below 116 m elevation were more stable. Shorelines above 116 m elevation are faintly etched onto the valley walls and can be found only near the Rhone Glacier [Stuiver et al., 1981]; in contrast, shorelines below 116 m elevation are broad and well defined. Soils at 86 and 116 m elevation (S046 and S047, respectively) were sampled on the largest paleoshorelines and contained the highest salt contents measured in this study, averaging 672 eq m–2 (Figure 7). The higher of these two soils at 116 m elevation corresponds to the elevation of the threshold separating Bonney Basin from Fryxell Basin [Kellogg et al., 1980]. This threshold makes 116 m elevation a natural limit for paleolake levels and the formation of a stable shoreline because an increase in lake levels above 116 m elevation would cause lake waters to spill over into Fryxell Basin.
 Because soluble salts in soils below approximately 300 m elevation in Bonney Basin are derived largely from paleolake waters, ionic ratios in these soils can be used to examine the chemical composition of paleolakes. In Table 4, ionic ratios in Bonney Basin soils (this study only) are compared to possible ionic compositions for paleolake waters, including: Blood Falls, surface waters of West Lake Bonney (WLB) above 6 m depth, highly concentrated hypolimnia in WLB below 15 m depth, and seawater. Blood Falls is a saline discharge from the snout of Taylor Glacier thought to derive from marine evaporite deposits beneath Taylor Glacier [Black et al., 1965]. This discharge has ionic ratios similar to seawater and is thought to influence the chemistry of waters in WLB [Lyons et al., 2005]. Near-surface lake waters in WLB are currently evapoconcentrating in wetted margins along Lake Bonney [Barrett et al., 2009] and represent a possible composition for evapoconcentrating lake waters in the past. WLB hypolimnia are thought to be derived from evaporative concentration of paleolake waters in Bonney Basin to below current levels [Hendy et al., 2000], which suggests that ionic ratios in WLB hypolimnia may reflect the composition of paleolake waters in Bonney Basin.
Table 4. Ratios of Salts in Bonney Basin Soils (this study only) Compared to Ratios in West Lake Bonney (WLB) Surface Water (above 6 m depth), WLB Hypolimnia (below 15 m depth), Blood Falls, and Seawatera
Ion ratios for WLB and Blood Falls are determined from the average of McMurdo Dry Valleys Long-Term Ecological Research data.
Soils between 116–300 m elevation
Soils below 116 m elevation
WLB surface water
 Soils between 116 and 300 m elevation have high Mg2+/Cl–, Na+/Cl–, and K+/Cl– ratios compared to most other soils and waters measured in Bonney Basin (Table 4). The HCO3–/Cl– ratio (0.27) is also high, which suggests that cation ratios relative to Cl– in these soils are influenced by calcite/gypsum dissolution and cation exchange, similar to soils in eastern Taylor Valley, but of a lesser magnitude. This makes it uncertain whether ionic ratios in soils between 116 and 300 m elevation are influenced by paleolake compositions or the soil-water extraction procedure. In soils below 116 m elevation, the HCO3–/Cl– ratio is low, and soluble salt concentrations are the highest measured in this study, indicating that the calcite/gypsum dissolution and cation exchange have a minimal effect on ionic ratios. Na+/Cl– ratios in soils below 116 m elevation (0.72) and WLB hypolimnia (0.71) are distinct from Na+/Cl– ratios in seawater (0.86), WLB surface water (0.83), and Blood Falls (1.05). K+/Cl– ratios are similar in all waters in Table 4 and in soils below 116 m elevation (0.02–0.03), as are Mg2+/Cl– ratios (0.1–0.16). Because Na+ and Cl– are conservative solutes during closed-basin lake evolution [Eugster and Jones, 1979; Hardie and Eugster, 1970], the unique similarity in Na+/Cl– ratios in soils below 116 m elevation and WLB hypolimnia suggests that WLB hypolimnia represent the original composition of paleolake waters in Bonney Basin, which have since evaporated to form a concentrated brine.
 SO42–/Cl–, Na+/Cl–, and Ca2+/Cl– ratios in Bonney Basin soils between 116 and 300 m elevation (0.58, 0.87, and 0.77, respectively) are higher than in soils below 116 m elevation (0.1, 0.72, and 0.12, respectively). Similarly, SO42–/Cl– and Ca2+/Cl– ratios in soils below 116 m elevation are higher than in WLB hypolimnia (0.03). The decrease in these ratios with decreasing elevation in Bonney Basin is consistent with proglacial paleolake waters dammed by the RSIS between 116 and 300 m elevation that transition into an evaporative closed-basin regime below 116 m elevation. In evaporating lake waters, gypsum, mirabilite, and calcite are often the first salts to precipitate, causing SO42–/Cl–, Na+/Cl–, and Ca2+/Cl– ratios to decrease in evaporating lake waters [Eugster and Jones, 1979].
 The distribution of evapoconcentrated salt accumulations in Bonney Basin soils with depth suggests that soluble salts have undergone significant postdepositional leaching (Figure 8). Salts accumulating from evapoconcentration are initially concentrated near the soil surface, as indicated by Cl– concentration profiles with depth in lake-marginal soils sampled by Barrett et al. . However, in Bonney Basin soils influenced by paleolakes, the highest Cl– concentrations are commonly found 10 to 50 cm below the soil surface. This indicates that evapoconcentrated Cl– has been partially leached down soil profiles following paleolake level lowering. The timescale for this Cl– leaching is between 3 and 10 Ka based on evidence of the last Lake Bonney low-stand sometime prior to 3 Ka [Poreda et al., 2004] and 14C dates of terraces that possibly date the last paleolake high-stand in Bonney Basin [Hall et al., 2010]. Leaching of near-surface Cl– accumulations down 10 to 50 cm over the Holocene represents a significant rate of leaching and is consistent with Cl– migration observed by Ugolini and Anderson  in Wright Valley soils. SO42– concentrations are highest directly beneath the soil surface, which indicates that SO42– has experienced much less leaching than Cl–. The close correlation between SO42– and Ca2+ in Bonney Basin suggests that SO42– has combined with Ca2+ to precipitate gypsum. Gypsum is a sparingly soluble salt that is not as easily leached from soils as more soluble chloride salts. This evidently preserves evapoconcentrated SO42– in its original position near the soil surface.
4.3.2 Eastern Taylor Valley: Fryxell Basin and Valley Mouth
 In eastern Taylor Valley, soluble salt accumulations are low and are strongly influenced by calcite/gypsum dissolution and cation exchange in soil-water extractions. This obscures salt distributions and compositions that may have been influenced by glacial and lacustrine paleoenvironments. To study the history of eastern Taylor Valley, we consider only total Cl– accumulations, because Cl– is a conservative ion in the soil that does not participate in exchange reactions and Cl– concentrations measured in different soil-water extraction procedures are comparable. The dominant distribution of Cl– contents in eastern Taylor Valley soils is that of higher Cl– accumulations near the Nussbaum Riegel, lower Cl– accumulations above approximately 120 m elevation, and higher Cl– accumulations below approximately 120 m elevation (Figure 9 and Figure 10).
 Soils older than MIS 2 in eastern Taylor Valley are found at higher elevations and generally have higher Cl– contents; however, Cl– contents in these older soils are not as high as in similarly aged Bonney Basin soils because of greater leaching of soluble salts in the wetter climate of eastern Taylor Valley. This leaching appears to be variable across the landscape, because soluble salt contents vary from high to low values. Heterogeneous leaching across the landscape may occur as a result of different snow accumulation patterns, soil texture, aspect, and slope [Keys, 1980]. Leaching was also found to vary within individual soils. For example, in a soil located at 384 m elevation on the Bent Stream transect (S016), several large, white, soluble salt precipitates were found beneath the desert pavement surface (Figure 11). Given this visible evidence of high salt content, it was expected that soluble salts extracted from the soil would be high, but measured soluble salts were fairly low (2.96 mol Cl– m–2). This suggests that salts in soil S016 preferentially accumulated beneath desert pavement clasts. The heterogeneous distribution of salts in soil S016 also suggests that bulk sampling of numerous soil pits on individual landscape elements is needed to accurately determine soil age from soluble salt contents.
 Below 320 m elevation in eastern Taylor Valley, all soils sampled in this study, except for higher elevation soils on the Bent Stream Transect located on pre-MIS 2 deposits, are located on deposits from the MIS 2 advance of the RSIS. Cl– contents in these MIS 2 sediments are primarily characterized by low Cl– contents above 120 to 140 m elevation, consistent with the young age of these sediments, and high Cl– contents below 120 to 140 m elevation, similar to older soils. In the Valley Mouth transect, two soils located at 120 m elevation on Coral Ridge have Cl– contents of 16.7 and 64.0 mol Cl– m–2, which is much higher than the Cl– contents averaging 0.9 mo Cl– m–2 in all other Valley Mouth soils. In the Delta Stream transects, Cl– contents in soils between 264 and 134 m elevation are uniformly low, averaging 0.3 mol Cl– m–2. At 128 m elevation, Cl– contents spike to 15.4 mol Cl– m–2 and are variable but generally high at lower elevations. Terraces in Fryxell Basin are characterized by two distinct spikes in Cl– content at 78 and 121 m elevation and low Cl– contents at all other elevations (Figure 10). In the Bent Stream Transect, fewer soils were excavated, and the boundary where Cl– contents increase is not clear, but the Cl– content in a soil at 143 m elevation is relatively high (4.9 mol Cl– m–2). High Cl– contents measured in RSIS sediments below 120 to 140 m elevation are much higher than Cl– contents found in RSIS sediments at higher elevations, which suggest that these soils were influenced by evapoconcentration from paleolakes that occupied Fryxell Basin up to 120 to 140 m elevation.
Barrett et al.  found similar increases in salts below approximately 120 m elevation near Lake Fryxell and Lake Hoare, in addition to increases in organic matter and fines (clay and silt), and attributed these trends to paleolakes. Although clay and silt content was not determined for soils in the current study, we observed that soils with the highest Cl– contents were commonly found in fine-grained soils below 120 m elevation. In particular, a soil sampled on a silt-rich moraine in front of Canada Glacier at 110 m elevation (soil S057) contained 43 mol Cl– m–2; and a soil composed of stratified, silty lacustrine sediment at 86 m elevation (soil S056) contained 47.5 mol Cl– m–2. Soils sampled along Coral Ridge at 120 m elevation (soils CR03 and CR04) were composed almost entirely of silt and had the highest Cl– contents in the Valley Mouth region. This association between fine-grained soils and soluble salts may be caused by increased deposition of clay and silt along stable lake margins or by lower rates of leaching in fine-grained soils [Barrett et al., 2010]. Fine-grained soils will slow down the rate of salt leaching, because fine-grained soils absorb water more strongly than coarse-grained soils [Hunt et al., 2007], reducing the downward percolation of snow melt. It is likely that both of these mechanisms influence the relationship between soil texture and salinity. If silt and clay deposition is associated with paleolakes, then increases in silt and clay contents may be useful as an additional indicator of paleolake levels, similar to soluble salts.
 Paleolakes filling eastern Taylor Valley were likely controlled by the elevation of the threshold that separated Fryxell Basin from McMurdo Sound at 78 m elevation. A large terrace is also found at 78 m elevation that has been confirmed by GPR as a delta having foreset stratigraphy [Horsman, 2007]. This delta indicates that paleolakes existed in eastern Taylor Valley up to at least 78 m elevation [Prentice et al., 2009]. Below 78 m elevation, paleolakes in eastern Taylor Valley would have been closed-basin lakes. Above 78 m elevation, paleolakes were likely proglacial lakes dammed by the RSIS; because, without an ice dam, lake waters would spill over into McMurdo Sound (although it is possible that the Coral Ridge threshold was higher in the past and has since eroded down from its maximum elevation). The upper limit of higher Cl– contents suggests that the maximum high-stand of these paleolakes was approximately 120 to 140 m elevation. The height of these paleolakes may have been limited by the Fryxell-Bonney threshold at 116 m elevation. To raise lake levels up to 116 m, water needs to fill Fryxell Basin; but, to raise lake levels higher than 116 m elevation requires the filling of both Fryxell and Bonney Basin. Alternatively, the elevation of paleolakes in Fryxell Basin may have been controlled by the elevation of the RSIS grounding line in eastern Taylor Valley.
4.3.3 Valley-Wide Glaciolacustrine History
Stuiver et al.  and Hall et al.  proposed that Taylor Valley was filled by valley-wide paleolakes (filling both eastern and western Taylor Valley) up to a maximum elevation of approximately 300 to 350 m. The minimum elevation of a valley-wide paleolake in Taylor Valley must have been at 116 m; because, below 116 m elevation, separate paleolakes would fill Fryxell Basin and Bonney Basin divided by the 116-m-high threshold near the Suess Glacier. As a result, soils that may have been affected by valley-wide paleolakes in Taylor Valley lie between approximately 116 and 350 m elevation. We expect that the formation of valley-wide paleolakes would result in similar soluble salt accumulations from evapoconcentration in soils across Taylor Valley, because the stability of lake margins and the concentration of solutes in lake water would be similar along the length of the paleolake. However, between 116 and 350 m elevation, salt contents in western Taylor Valley are much higher than salt contents in eastern Taylor Valley.
 The difference in soluble salt accumulations between eastern and western Taylor Valley could be explained as the result of higher leaching rates in eastern Taylor Valley. In eastern Taylor Valley, salt contents in soils influenced by evapoconcentration are generally lower than in Bonney Basin, which is consistent with leaching; however, leaching has not completely removed evapoconcentrated salt accumulations in eastern Taylor Valley, because high salt accumulations do occur locally. Similarly, although salts in older soils near the Nussbaum Riegel have been leached, relatively high salt accumulations remain in places. These two cases indicate that, while leaching occurs, it does not uniformly flush salts from the landscape. As a result, it is difficult to explain the uniformly low salt contents measured in MIS 2 sediments above 120 to 140 m elevation as the result of leaching. Below 140 m elevation, 36% of the soils sampled on MIS 2 sediments have significant Cl– contents above 5 mol Cl– m–2 (excluding terrace soils); while, above 140 m elevation, no soils have Cl– contents above 5 mol Cl– m–2, and most have Cl– contents less than 1 mol Cl– m–2.
 Because differences in soluble salt accumulations between eastern and western Taylor Valley are difficult to explain as a result of leaching, we conclude that soluble salt distributions in Taylor Valley are not consistent with valley-wide paleolake hypotheses. Instead, soluble salt distributions suggest that western Taylor Valley was filled with paleolakes up to approximately 300 m elevation, while eastern Taylor Valley was filled with paleolakes up to only 120 to 140 m elevation. To reconcile the occurrence of low soluble salt distributions in MIS 2 RSIS sediments above 120 to 140 m elevation in eastern Taylor Valley with evidence of evapoconcentration from paleolakes up to approximately 300 m elevation in Bonney Basin, paleolakes in Bonney Basin must have been dammed by a lobe of the RSIS that filled eastern Taylor Valley to at least 300 m elevation in Fryxell Basin. The margins of the RSIS in eastern Taylor Valley would have been characterized by ice-marginal streams that would leach soluble salts from soils. This RSIS lobe probably abutted against the Nussbaum Riegel, forming an ice dam in the narrow valley defile near the Suess Glacier. During the retreat of the RSIS from Taylor Valley, paleolake levels in Bonney Basin would have lowered from the high-stand as lake water filled space vacated by retreating ice [Stuiver et al., 1981], or paleolakes may have lowered in response to decreased melt water inputs from alpine glaciers and the RSIS [Hall et al., 2010]. Lowering of paleolakes during deglaciation could explain the preservation of low salt accumulations in RSIS sediments above 120 to 140 m elevation from the influence of paleolakes.
 The MIS 2 RSIS expansion into Taylor Valley proposed here is more extensive than in Stuiver et al.  and Hall et al. . Compared to the maximum position of the RSIS lobe in Stuiver et al. , the maximum elevation of ice proposed here is about 150 m higher near Canada Glacier. Hall et al.  proposed that the RSIS never extended beyond Coral Ridge near the Valley Mouth and that RSIS sediments throughout eastern Taylor Valley were deposited in a lake-ice conveyor system [Hendy et al., 2000]. Many of the sediments that have been interpreted as lake-ice conveyor deposits, including cross-valley ridges, sinuous ridges, and mounds, are found below 120 m elevation on the valley floor [Hall et al., 2000]. The primary evidence for high paleolakes above 120 m elevation in eastern Taylor Valley comes from radiocarbon dates of algae buried in terraces [Hall and Denton, 2000; Stuiver et al., 1981]. In Fryxell Basin, most terraces occur below 120 m elevation; out of a total of 78 terraces that have been radiocarbon dated, 70 are located below 120 m elevation, while only 8 are located above 120 m elevation. Hence, evidence for valley-wide paleolake levels above 120 m elevation rests on relatively few radiocarbon dates from terraces. Several terraces near 220 m elevation along Delta Stream were explored with GPR to determine whether the internal stratigraphy was deltaic, but deltaic clinoforms were not found [Horsman, 2007; Prentice et al., 2009; Horsman et al., manuscript in preparation]. This suggests that these features are not deltas deposited in large paleolakes, but may be stream sediments deposited at the margins of the RSIS [Prentice et al., 2009]. Even if some terraces above 120 m elevation are interpreted as deltas, this does not provide conclusive evidence for a valley-wide paleolake filling Fryxell Basin, because deltas could have been deposited in small lakes marginal to the RSIS lobe.
 This study uses soluble salt accumulations in Taylor Valley, Antarctica, to determine the history of paleolakes that are believed to have been dammed by the RSIS. Results from this study indicate that soluble salt distributions are controlled by soil age, distance from the sea (McMurdo Sound), leaching, soil texture, calcite/gypsum dissolution, cation exchange, and past glacial and lacustrine events. Soils in western Taylor Valley have relatively high soluble salt contents, and salt compositions are similar to those of seawater; in contrast, soils in eastern Taylor Valley have relatively low salt contents, and salt compositions are dominated by Na+ and HCO3–. The peculiar Na-HCO3-rich composition of salts in eastern Taylor Valley is consistent with calcite dissolution and cation exchange reactions in the presence of relatively high leaching rates. These reactions result in Cl– being leached from soils as Ca-Mg-Cl-rich brine and the association of Na+ with HCO3– from calcite dissolution.
 Soluble salt contents in soils are primarily determined by soil age. Soils that developed on glacial sediments from the MIS 2 advance of the RSIS have much lower salt contents than soils that developed on glacial sediments from older advances of Taylor Glacier, the RSIS, and alpine glaciations. This relationship between soil age and salt content is not uniform because of salt leaching (particularly in the wetter climate of eastern Taylor Valley) and paleolake events. Paleolakes have redistributed soluble salts in lower elevation soils through leaching and evapoconcentration along paleolake margins. Soils in western Taylor Valley are consistent with a history of paleolake margins up to approximately 300 m elevation, which agrees with paleolake high-stands inferred from lacustrine strandlines and terraces. In eastern Taylor Valley, salt distributions indicate maximum paleolake high-stands up to 120 to 140 m. Differences in soluble salt accumulations between eastern and western Taylor Valley are not consistent with the hypothesis that valley-wide lakes filled all of eastern and western Taylor Valley during the MIS 2 advance of the RSIS. On the basis of soluble salt distributions, we conclude that high paleolakes in Bonney Basin were dammed by a lobe of the RSIS that entered deep into eastern Taylor Valley, filling Fryxell Basin to at least 300 m elevation.
 This material is based upon work supported by the National Science Foundation under grants 0541054 and 0636998. Support was also provided by the Department of Earth and Space Sciences at the University of Washington. Funding was also provided by the Kenneth C. Robbins and Peter Misch Fellowship.