Advertisement

Aquatic Geochemistry

, Volume 19, Issue 2, pp 135–145 | Cite as

Carbon Sequestration and Release from Antarctic Lakes: Lake Vida and West Lake Bonney (McMurdo Dry Valleys)

  • G. M. Marion
  • A. E. Murray
  • B. Wagner
  • C. H. Fritsen
  • F. Kenig
  • P. T. Doran
Original Paper
  • 343 Downloads

Abstract

Perennial ice covers on many Antarctic lakes have resulted in high lake inorganic carbon contents. The objective of this paper was to evaluate and compare the brine and CO2 chemistries of Lake Vida (Victoria Valley) and West Lake Bonney (Taylor Valley), two lakes of the McMurdo Dry Valleys (East Antarctica), and their potential consequences during global warming. An existing geochemical model (FREZCHEM-15) was used to convert measured molarity into molality needed for the FREZCHEM model, and this model added a new algorithm that converts measured DIC into carbonate alkalinity needed for the FREZCHEM model. While quite extensive geochemical information exists for ice-covered Taylor Valley lakes, such as West Lake Bonney, only limited information exists for the recently sampled brine of >25 m ice-thick Lake Vida. Lake Vida brine had a model-calculated pCO2 = 0.60 bars at the field pH (6.20); West Lake Bonney had a model-calculated pCO2 = 5.23 bars at the field pH (5.46). Despite the high degree of atmospheric CO2 supersaturation in West Lake Bonney, it remains significantly undersaturated with the gas hydrate, CO2·6H2O, unless these gas hydrates are deep in the sediment layer or are metastable having formed under colder temperatures or greater pressures. Because of lower temperatures, Lake Vida could start forming CO2·6H2O at lower pCO2 values than West Lake Bonney; but both lakes are significantly undersaturated with the gas hydrate, CO2·6H2O. For both lakes, simulation of global warming from current subzero temperatures (−13.4 °C in Lake Vida and −4.7 °C in West Lake Bonney) to 10 °C has shown that a major loss of solution-phase carbon as CO2 gases and carbonate minerals occurred when the temperatures rose above 0 °C and perennial ice covers would disappear. How important these Antarctic CO2 sources will be for future global warming remains to be seen. But a recent paper has shown that methane increased in atmospheric concentration due to deglaciation about 10,000 years ago. So, CO2 release from ice lakes might contribute to atmospheric gases in the future.

Keywords

Gas hydrates Sediment carbon dioxide FREZCHEM simulation Global warming 

1 Introduction

There are abundant geochemical analyses of McMurdo Dry Valley ponds and lakes (e.g., Wharton et al. 1993a, b; Doran et al. 1994, 2003, 2008; Welch et al. 1996; Marion 1997; Priscu et al. 1999; Neumann et al. 2001; McKay et al. 2003; Lyons et al. 2005; Knoepfle et al. 2009; Welch et al. 2010). Most of these studies have focused on Taylor Valley lakes, especially with respect to Lake Bonney, Lake Hoare, and Lake Fryxell (e.g., Wharton et al. 1993a, b; Doran et al. 1994; Welch et al. 1996; Priscu et al. 1999; Lyons et al. 2005; Neumann et al. 2001; Knoepfle et al. 2009; Welch et al. 2010), as the Taylor Valley is a branch of the Antarctic Long-Term Ecological Research (LTER) site. Much less studied has been Lake Vida (Doran et al. 1994, 2003, 2008; Murray et al. 2012), which is the highest elevation lake at 390 m.a.s.l. in the McMurdo Dry Valleys (Doran et al. 1994), and the brine in the ice cover was only recently sampled (Doran et al. 2008; Murray et al. 2012). West Lake Bonney, for example, only has an elevation of 60 m.a.s.l. (Doran et al. 1994). A consequence of the higher elevation of Lake Vida is a much thicker ice cover that could slow CO2 releases as global climate warms.

The environmental properties of McMurdo Dry Valley lakes are largely controlled by perennial ice covers that minimize wind-generated currents, restrict exchange of gases between water columns and the atmosphere, reduce light penetration, and limit sediment deposition (Wharton et al. 1993a, b) (Fig. 1). As a consequence, many lakes have developed brines that are high in carbon compounds such as carbonate alkalinity, DIC, DOC, and POC (e.g., Priscu et al. 1999; Neumann et al. 2001; Knoepfle et al. 2009). Surface waters in ice-covered lakes in Antarctica are supersaturated with respect to atmospheric CO2 and at depth, can be highly supersaturated with respect to atmospheric CO2 (Neumann et al. 2001). Lake Fryxell, on the other hand, is highly supersaturated with respect to atmospheric CO2 throughout the entire water column (Neumann et al. 2001). Should Earth’s climate continue warming (Hattermann and Levermann 2010; Joughin and Alley 2011; Raisanen and Ylhaisi 2011), eventually Antarctic lake ice covers would melt, leading to release of CO2 into the atmosphere that could contribute to global warming.
Fig. 1

Cross-sectional view of West Lake Bonney

FREZCHEM is a geochemical model that deals with chemistries of cold waters in molal units (Marion and Kargel 2008) that will be used for brine and CO2 chemistries in Antarctic lakes. A recent paper (Marion 2007) developed a molarity to molality transformation that allows FREZCHEM to cope with measured data in molar units, such as the Antarctic lakes described below. But an existing limitation for FREZCHEM is that carbonate alkalinity (=HCO3  + 2CO3 2−) is required, but in many cases, measured data are based on DIC (=HCO3  + CO3 2− + CO2).

The specific objectives of this study were to (1) develop a new algorithm for converting DIC into carbonate alkalinity for the FREZCHEM model, (2) evaluate the brine and CO2 chemistries of Lake Vida using FREZCHEM, (3) evaluate similar chemistries of West Lake Bonney with FREZCHEM, and (4) compare the potential roles of CO2 chemistries in Lake Vida and West Lake Bonney as Earth continues to warm.

2 Methods

Lake Vida is located in Victoria Valley, the northern most of the McMurdo Dry Valleys of East Antarctica (Doran et al. 2003). During initial coring of the >25-m-thick ice over the lake, brine infiltrated the borehole 16.0 m below the surface of the ice and ~9.0 m above the estimated main body of the brine (Doran et al. 2008). The brine then rose in the borehole to 10.5 m below the ice surface, a hydrostatic depth lower than expected for a floating, >25 m thick, ice cover, indicating that the lake’s thick ice cover is partly supported by grounding closer to the lake edge. The brine consistently returned to the same level in the borehole following pumping, indicating connection with the main brine body of the lake (Doran et al. 2008). We collected Lake Vida brine (LVBr) on December 3, 2005, from within the borehole (~8.5 m above the lake body-ice cover interface, ~16.5 m below the lake surface) using sterilized submersible pumps and sterile polytetrafluoroethylene (PTFE) tubing (Doran et al. 2008). Samples for chemical analyses and activity assays were collected in a sterile anoxic chamber. LVBr is slightly acidic (pH 6.2), anoxic, and turns from light yellow to dark orange upon exposure to the atmosphere as a result of ferric iron precipitation.

The Lake Vida chemistries were published in a recent paper (Murray et al. 2012). The geochemical and physical data for West Lake Bonney were taken from Lyons et al. (2005) and Knoepfle et al. (2009). West Lake Bonney was selected because it contains especially high concentrations of CO2 (Neumann et al. 2001), which is a good comparison to Lake Vida. In this paper, Tables 1 and 2 contain the geochemical and physical data used from these sources.
Table 1

The geochemical and physical data from Lake Vida brines (Murray et al. 2012) used in FREZCHEM

Solution species

Molarity (mol/l)

Molality (mol/kg(H2O))

Na

1.914

2.036

K

0.0828

0.0881

NH4

0.00388

0.00413

Mg

0.6649

0.7073

Ca

0.0301

0.0320

Sr

0.00045

0.00048

Fe(II)

0.00031

0.00033

Al

0.000010

0.000011

Cl

3.318

[3.4360]a

F

0.0015

0.0016

NO3 (+NO2)

0.00093

0.00099

DIC (HCO3 + CO3 + CO2)

0.0610

0.0649

Carbonate alkalinity (HCO3 + 2CO3)

0.0431 (equil./l)b

0.0459 (equil./kg(H2O))

SO4

0.0584

0.0621

Salinity

187.98 (g/l)

199.96 (g/kg(H2O))

Physical properties

Experimental data

Model data

Temperature (°C)

−13.4

−13.4

pHc

−log[H+] = 6.51d

−log[H+] = 6.49d

−log(H+) = 6.20e

pCO2 (bars)

 

0.60

aCI was adjusted to lead to a perfect charge balance for molality

bModel estimated, not directly measured

cThe brackets in the pH row refer to concentrations, and the parentheses refer to activity

dCalculated

eMeasured

Table 2

The geochemical and physical data from West Lake Bonney brines (Lyons et al. 2005; Knoepfle et al. 2009) used in FREZCHEM

Solution species

Molarity (mol/l)

Molality (mol/kg(H2O))

Na

1.71

1.798

K

0.036

0.0379

Mg

0.369

0.3880

Ca

0.057

0.0599

Cl

2.47

[2.5659]a

DIC (HCO3 + CO3 + CO2)

0.078

0.0820

Carbonate alkalinity (HCO3 + 2CO3)

0.0576 (equil./l)b

0.0606 (equil./kg(H2O))

SO4

0.050

0.0526

Salinity

149.11 (g/l)

156.79 (g/kg(H2O))

Physical properties

Experimental data

Model data

Temperature (°C)

−4.69

−4.69

pHc

−log[H+] = 5.60d

−log[H+] = 5.58e

−log(H+) = 5.46e

pCO2 (bars)

 

5.23

aCI was adjusted to lead to a perfect charge balance for molality

bModel estimated, not directly measured

cThe brackets in the pH row refer to concentrations, and the parentheses refer to activity

dMeasured

eCalculated

FREZCHEM was used to access chemical equilibrium in Antarctic lakes. FREZCHEM is an equilibrium chemical thermodynamic model parameterized for concentrated electrolyte solutions (to ionic strengths = 20 m) using the Pitzer approach (Pitzer 1991) for the temperature range from −100 to 25 °C and the pressure range from 1 to 1,000 bars (Marion and Farren 1999; Marion 2001; Marion et al. 2005, 2006, 2011, 2012; Marion and Kargel 2008). The current version of the model (FREZCHEM-15) is parameterized for the Na–K–NH4–Mg–Ca–Fe(II)–Fe(III)–Al–H–Cl–ClO4–Br–SO4–NO3–OH–HCO3–CO3–CO2–O2–CH4–NH3–Si–H2O system and includes 108 solid phases, including ice, 16 chloride minerals, 36 sulfate minerals, 16 carbonate minerals, five solid-phase acids, four nitrate minerals, seven perchlorates, six acid salts, five iron oxide/hydroxides, four aluminum hydroxides, two silica minerals, two gas hydrates, two ammonias, and two bromide sinks (see above references, especially Marion and Kargel (2008), for model parameters and structure).

The measurements of the two brine samples examined in this work were in molar units (mol/l) (Tables 1, 2). The FREZCHEM model, on the other hand, requires molal units (mol/kg(H2O)). Fortunately, there is a FREZCHEM program that converts molarity into molality (Marion 2007) based on individual ions and salinity (Tables 1, 2). In this conversion, we arbitrarily assigned a temperature of 298 K and a pCO2 pressure of 1.01325 bars. But in the specific modeling of Lake Vida and West Lake Bonney, field temperatures and estimated field pCO2 pressures are used. All of the molar data in Tables 1 and 2, except for carbonate alkalinity, were measured data. All of the molal data in Tables 1 and 2 were calculated based on the FREZCHEM model (Marion 2007). In Table 1 for Lake Vida, there is a difference between molarity and molality of 6.4 %. In Table 2 for West Lake Bonney, there is a difference between molarity and molality of 5.2 %.

Another factor that is necessary for the FREZCHEM model is input of carbonate alkalinity (CA)
$$ {\text{CA}} = {\text{HCO}}_{{3{\text{T}}}}^{ - } + 2{\text{CO}}_{{3{\text{T}}}}^{2 - } $$
(1)
where “T” in bicarbonates and carbonates include free ions and ion-pairs (e.g., CaCO 3 0 ). But in the two sets of measured data (Tables 1, 2), dissolved inorganic carbon (DIC) is equal to
$$ {\text{DIC}} = {\text{HCO}}_{{3{\text{T}}}}^{ - } + {\text{CO}}_{{3{\text{T}}}}^{2 - } + {\text{CO}}_{{2{\text{T}}}} ({\text{aq}}) $$
(2)
that was measured instead of carbonate alkalinity. In this case, CO2T(aq) = CO2(aq) + H2CO3(aq), where CO2T(aq) is the unit used in FREZCHEM. As pointed out above, we arbitrarily assigned a temperature of 298 K and a pCO2 pressure of 1.01325 bars to calculate molality from molarity, and we used the same units to estimate CO2T in Eq. 2 based on the following equation:
$$ K_{{{\text{CO}}_{2} }} = \frac{{\gamma_{{{\text{CO}}_{2} }} \cdot [{\text{CO}}_{{2{\text{T}}}} ({\text{aq}})]}}{{f_{{{\text{CO}}_{2} }} \cdot [{\text{CO}}_{2} ({\text{g}})]}} $$
(3)
where γCO2 is an activity coefficient, fCO2 is a fugacity coefficient, and [CO2] refers to concentrations. So given CO2(g) = 1.01325 bars, allows an estimate of CO2T(aq) via Eq. 3. The following equation calculates HCO 3T and CO 3T 2−
$$ K_{{{\text{HCO}}_{3} ,{\text{CO}}_{3} }} = \frac{{({\text{H}}^{ + } ) \cdot \gamma_{{{\text{CO}}_{3} }} \cdot [{\text{CO}}_{3{\rm T}}^{2 - } ]}}{{\gamma_{{{\text{HCO}}_{3} }} \cdot [{\text{HCO}}_{{3{\text{T}}}}^{ - } ]}} $$
(4)
where T in the above equation includes ions and ion associates (e.g., CaCO 3 0 ). Replacing [HCO 3T ] in Eq. 2 with Eq. 4 and rearranging lead to
$$ [{\text{CO}}_{{3{\text{T}}}}^{2 - } ] = \frac{{{\text{DIC}} - {\text{CO}}_{{2{\text{T}}}} ({\text{aq}})}}{{\left[ {\frac{{({\text{H}}^{ + } ) \cdot \gamma_{{{\text{CO}}_{3}^{2 - } }} }}{{\gamma_{{{\text{HCO}}_{3}^{ - } }} \cdot K_{{{\text{HCO}}_{3} /{\text{CO}}_{3} }} }} + 1} \right]}} $$
(5)
So given the measurement of DIC, CO2T(aq) based on Eq. 3, replacing HCO 3T in Eq. 2 with Eq. 4, leads to an estimate of [CO 3T 2− ] (Eq. 5), which then can estimate [HCO 3T ] given DIC, CO 3T 2− , and CO2T(aq) in Eq. 2. Estimates of HCO 3T and CO 3T 2− were used to estimate carbonate alkalinity (Eq. 1) that is needed for the molar to molal conversion described above. Once all the above calculations were made, we brought the molality scale into perfect charge balance for the FREZCHEM model by adjusting the Cl concentrations (Tables 1, 2).

3 Results

3.1 Lake Vida

The temperature of the Lake Vida brine at collection was −13.4 °C at a pH of 6.20 (Table 1). To determine the in situ field pCO2 with the FREZCHEM model, it was necessary to use the molality data, temperature, and pH of Table 1. Using pCO2 in this model simulation does not imply that CO2(g) exists in the ice-covered lake, but does imply that a related CO2(aq) phase exists. But as we will show, when total pressure is released on these lake samples, CO2(aq) quickly changes into CO2(g), similar to beer bottles.

Assuming pCO2 = 3.9 × 10−4 bars, today’s atmospheric concentration, FREZCHEM predicted a pH of 8.91, which is far removed from the measured value of 6.20 (Table 1). Adjusting pCO2 to eventually = 0.60 bars led to a model-calculated pH = 6.20 in agreement with the field measurement (Table 1). What is clear in this case is that the lower-level brines (≈16.5 m) of ice-covered Lake Vida are much more soluble with respect to carbonate alkalinity (0.0459 equil./kg(H2O)) (Table 1) than seawater that has a carbonate alkalinity of 0.00228 equil./kg(H2O) (Millero et al. 2008), and a difference in pH ≈ 8.3 in seawater (assuming a (H+) activity scale, Marion et al. 2011) compared to pH = 6.20 in Lake Vida (Table 1). Another factor considered in our model calculations was ice formation, which did not occur at the brine temperature of −13.4 °C, albeit, surrounded by ice in the field, but did form at −13.7 °C using the FREZCHEM model, which was in reasonable agreement (−13.4 vs. −13.7 °C). In addition to allowing ice to form, we also allowed minerals to precipitate. In this case at the point where ice was forming (−13.7 °C), small amounts of magnesite (MgCO3) and siderite (FeCO3) were also precipitating at pH = 6.02. Running the same case without magnesite, siderite, or dolomite precipitation led to a pH of 6.20, the same pH of the original case (Table 1). Overall, the FREZCHEM model was able to simulate the geochemical data assuming pCO2 = 0.60 bars at pH = 6.20 (Table 1).

3.2 West Lake Bonney

Figure 1 depicts West Lake Bonney that is covered with a 5 m ice layer, 9 m of a relatively clean freshwater below the ice, 3 m of a chemocline zone, and about 23 m of relatively dense brine at the bottom. This lake is very different from Lake Vida, which is essentially a >25 m frozen lake. As was the case for Lake Vida, the measurements in West Lake Bonney were also in molar units (Lyons et al. 2005) that had to be converted into molality. Both molarity and molality are included in Table 2 for West Lake Bonney. The difference between molarity and molality is about 5.2 %, slightly lower than the 6.4 % difference in Lake Vida case (Table 1). While both Lake Vida and West Lake Bonney are brines (Tables 1, 2), most of the aqueous concentrations are somewhat lower in West Lake Bonney, except for Ca and carbonate alkalinity (compare Tables 1, 2).

For West Lake Bonney model simulations, it was necessary to estimate a pCO2 value that was consistent with the measured temperature (−4.69 °C) and pH = [5.60] that was taken from Knoepfle et al. (2009) (Table 2). Assuming pCO2 = 3.9 × 10−4 bars, today’s atmospheric concentration, led to a calculated pH of 9.04, which is far removed from the measured value of [5.60] (Table 2). Adjusting pCO2 to eventually = 5.23 bars led to a model-calculated pH = [5.58] or (5.46) (Table 2). This 5.23 bar of pCO2 in West Lake Bonney (Table 2) is significantly higher than in the case for Lake Vida, where pCO2 = 0.60 bars (Table 1.) These two Antarctic lakes are significantly different in chemistries (Tables 1, 2) that can lead to interesting pCO2 cases that we will discuss in the next section.

4 Discussions

In the discussion, we will examine three comparisons of these two Antarctic lakes: high pCO2 in West Lake Bonney, gas hydrate possibilities, and global warming of these Antarctic lakes.

There is independent evidence that West Lake Bonney has an exceptionally high CO2 level (Neumann et al. 2001). Also, on November 16, 2002, we sampled the sediments at the base of West Lake Bonney (38.7 m) (Fig. 1). Sediment cores were obtained using a 3-m-long piston corer (UWITEC, Austria). Although the corer was released directly at the sediment surface (controlled by video camera), we found about 40–60 cm of water above the sediment surface when the cores were recovered. After recovery, one of the cores was immediately split into two segments of 1.5 m length in a tent at a temperature around the freezing point. According to a first macroscopic description of the recovered sediments through the transparent PVC liner, the sediment was composed of brown to ocher silt and clay, with interspersed salty crusts that started to collapse immediately after recovery. Within several minutes (c. 5 min), when we tried to stabilize the soft surface sediment in the upper segment with foam, the top 1.5 m segment of the core reduced to 0.4 m, so a volume loss of ~73 % occurred. The sediment collapse was associated with apparent fizzing and release of CO2 and the release of liquid water, though the volume of water released was not measured in the field. The subsurface pCO2 = 5.23 bars (Table 2) in the brine layer (Fig. 1), which is highly supersaturated with the atmosphere above West Lake Bonney, is why the sediment cores quickly released CO2 when these samples were brought to the surface. Such possible CO2 releases from Lake Vida cannot be compared because sediments from Lake Vida have yet to be sampled, and the pCO2 = 0.60 bars for Lake Vida is much lower than the pCO2 = 5.23 bars for West Lake Bonney. This lower level of pCO2 in Lake Vida could be due to the lower temperature at −13.4 °C compared to a higher temperature of −4.69 °C in West Lake Bonney, which would favor more microbial activity.

For West Lake Bonney that has a relatively high pCO2 = 5.23 bars, we also considered the possibility that gas hydrate, CO2·6H2O, might also have formed in this lake. Using the chemical data from Table 2 at a temperature of −4.69 °C, we ran FREZCHEM with a range of pCO2 values from 1 to 15 bars. The CO2·6H2O gas hydrate started forming at 13.9 bars, which is much higher than the calculated pCO2 = 5.23 bars at the base of West Lake Bonney. If such a gas hydrate is stable today in this lake, it must exist deep within the sediment layer, ≈57 m in order to increase the pCO2 to 13.9 bars, or it could be present in a metastable phase having formed under colder temperatures or greater pressures. Buffett & Zatsepina (1999) have argued that gas hydrates can persist in metastable phases for periods up to 106 years. We also ran the possibility that Lake Vida might produce the gas hydrate, CO2·6H2O. In the Lake Vida case, CO2·6H2O would start forming at a pCO2 = 6.9 bars, which is substantially above the model-calculated pCO2 of 0.60 bars (Table 1). Compared to West Lake Bonney that could form gas hydrates at 13.9 bars of pCO2, the lower pCO2 of Lake Vida at 6.9 bars for gas hydrate formation is due to the much colder temperature of Lake Vida (−13.4 °C) compared to West Lake Bonney (−4.69 °C). The only Antarctic lake that has been suggested might contain gas hydrates is Lake Vostok (McKay et al. 2003). But this lake is 4 km beneath ice compared to Lake Vida and West Lake Bonney that are 0.025 and 0.005 km beneath ice. While Lake Vida and West Lake Bonney are unlikely to contain gas hydrates, Lake Vostok is possible for gas hydrate formation (McKay et al. 2003). But, nevertheless, despite these possibilities, gas hydrates in Antarctic lakes are currently speculative.

Were there any other interesting possibilities associated with these Lake Vida and West Lake Bonney chemical data? The high carbonate alkalinity data for Lake Vida (Table 1) are interesting because the ice cover prevents the loss of gases, liquids, and soluble species from the deep brines (Wharton et al. 1993a, b). But what would happen if the Earth eventually became warmer, as there is evidence today? We ran a FREZCHEM simulation for Lake Vida from 258.15 K (−15 °C) to 283.15 K (10 °C), which runs from near-current temperature to significant warming above freezing, to examine what would happen to carbonate alkalinity and associated components such as pH and solid-phase precipitations during a warming period. In this simulation, we retained all potential bicarbonate and carbonate species (16 minerals) within the FREZCHEM model. Figure 2 depicts how the molality values of Table 1 would affect carbonate alkalinity as the ice layers of Lake Vida melted. We assumed the calculated value of pCO2 = 0.60 bars with an initial pH = 6.20 for the early stages from 258 K to 273 K. But when the temperature exceeded the ice freezing point at 273 K, we assumed a pCO2 = 3.9 × 10−4 bars, today’s surface atmospheric CO2 partial pressure. Obviously, this is a timeless approximation, as it could take years, decades, or centuries for these eventually non-icy lakes to equilibrate with the atmosphere. The only carbonate minerals that precipitated were siderite (FeCO3) and magnesite (MgCO3) (Fig. 2). Because of the prevalence of Mg in this simulation (Table 1), magnesite was the dominant sink for carbonate alkalinity. The increasing precipitation of magnesite with increasing temperature led to decreasing levels of siderite (Fig. 2). Not surprisingly, at 273 K, when the ice layer was removed, this caused a major decrease in CO2, a major decrease in solution-phase carbonate alkalinity, a significant increase in magnesite precipitation, and pH at this point rose from 5.93 to 7.22 (Fig. 2).
Fig. 2

Lake Vida carbon simulations as temperature increases with time

FREZCHEM is an equilibrium geochemical model and does not consider kinetics. So how quickly such a warming process might occur is impossible to tell based on FREZCHEM. Next, we examine a similar simulation for West Lake Bonney with even higher carbonate alkalinity (Table 2) that is drastically different in many respects from Lake Vida.

What would happen in West Lake Bonney if the Earth became warmer? We ran a FREZCHEM simulation from 268.15 K (−5 °C) to 283.15 K (10 °C), which runs from near-current temperature to significant warming above freezing, to examine what would happen to carbonate alkalinity and associated components such as pH and solid-phase precipitation during a warming period. In this simulation, we assumed the calculated value of pCO2 = 5.23 bars with an initial pH = 5.46 for the early stages from 268 to 273 K (Fig. 3). But when the temperature exceeded the ice freezing point at 273 K, we assumed a pCO2 = 3.9 × 10−4 bars, today’s surface atmospheric CO2 partial pressure. The only carbonate mineral that precipitated in this case was dolomite (CaMg(CO3)2) (Fig. 3). It was probably the relatively high concentration of Ca in West Lake Bonney (Table 2) that led to dolomite, instead of magnesite in Lake Vida. At 273 K, when the ice layer was melted, this caused a major decrease in CO2, a major decrease in solution-phase carbonate alkalinity, a major increase in dolomite precipitation, and pH rose from 5.40 to 7.41 (Fig. 3).
Fig. 3

West Lake Bonney carbon simulations as temperature increases with time

If global warming continues, eventually ice-covered lakes in Antarctica will melt leading to the release of CO2 as either gases or precipitates, as illustrated in Figs. 2 and 3. While FREZCHEM cannot cope with time factors, West Lake Bonney will likely melt earlier than Lake Vida because of the colder and thicker ice layer at Lake Vida. How important these Antarctic lake CO2 sources will be for future global warming remains to be seen. But, a recent paper (Brosius et al. 2012) has argued that methane increased in atmospheric concentrations due to deglaciation about 10,000 years ago. So, CO2 releases from ice lakes might also contribute to atmospheric gases in the future.

Notes

Acknowledgments

We thank J. Kyne and B. Bergeron of Ice Coring and Drilling Services (ICDS), S. Sherif and M. Badescu of NASA, N. Bramall of UCB, and P. Glenday for field assistance. We thank C. Davis (DRI), M. Dieser (UM), and J.R. Henricksen (UGA) for laboratory assistance. R. Edwards, M. Potosnak, and C.S. Riesenfeld of DRI provided geochemical and programming skills. We thank Lisa Wable for developing Fig. 1. This research was supported by NASA-ASTEP NAG5-12889, and logistical support was provided by the National Science Foundation’s Office of Polar Programs through a cooperative agreement with NASA.

References

  1. Brosius LS, Walter Anthony KM, Grosse G, Chanton JP, Farquharson LM, Overduin PP, Meyer H (2012) Using the deuterium isotope composition of permafrost meltwater to constrain thermokarst lake contributions to atmospheric CH4 during the last deglaciation. J Geophys Res 117:G01022. doi: 10.1029/2011JG001810 CrossRefGoogle Scholar
  2. Buffett BA, Zatsepina OY (1999) Metastability of gas hydrates. Geophys Res Lett 26:2981–2984CrossRefGoogle Scholar
  3. Doran PT, Wharton RA Jr, Lyons WB (1994) Paleolimnology of the McMurdo Dry Valleys, Antarctica. J Paleolimol 10:85–114CrossRefGoogle Scholar
  4. Doran PT, Fritsen CH, McKay CP, Priscu JC, Adams EE (2003) Formation and character of an ancient 19-m ice cover and underlying trapped brine in an “ice-sealed” east Antarctic lake. Proc Natl Acad Sci USA 100:26–31CrossRefGoogle Scholar
  5. Doran PT, Fritsen CH, Murray AE, Kenig F, McKay CP, Kyne JD (2008) Entry approach into pristine ice-sealed lakes—Lake Vida, East Antarctica, a model ecosystem. Limnol Oceanogr Methods 6:542–547CrossRefGoogle Scholar
  6. Hattermann T, Levermann A (2010) Response of Southern Ocean circulation to global warming may enhance basal ice shelf melting around Antarctica. Clim Dyn 35:741–756CrossRefGoogle Scholar
  7. Joughin I, Alley RB (2011) Stability of the West Antarctic ice sheet in a warming world. Nat Geosci 4:506–513CrossRefGoogle Scholar
  8. Knoepfle JL, Doran PT, Kenig F, Lyons WB, Galchenko VF (2009) Particulate organic and dissolved inorganic carbon stable isotopic composition in Taylor Valley lakes, Antarctica: the effect of legacy. Hydrobiologia 632:139–156CrossRefGoogle Scholar
  9. Lyons WB, Welch KA, Snyder G, Olesik J, Graham EY, Marion GM, Poreda RJ (2005) Halogen geochemistry of the McMurdo dry valleys lakes, Antarctica: clues to the origin of solutes and lake evolution. Geochim Cosmochim Acta 69:305–323CrossRefGoogle Scholar
  10. Marion GM (1997) A theoretical evaluation of mineral stability in Don Juan Pond, Wright Valley, Victoria Land. Antarct Sci 9:92–99CrossRefGoogle Scholar
  11. Marion GM (2001) Carbonate mineral solubility at low temperatures in the Na–K–Mg–Ca–H–Cl–SO4–OH–HCO3–CO3–CO2–H2O system. Geochim Cosmochim Acta 65:1883–1896CrossRefGoogle Scholar
  12. Marion GM (2007) Adapting molar data (without density) for molal models. Comput Geosci 33:829–834CrossRefGoogle Scholar
  13. Marion GM, Farren RE (1999) Mineral solubilities in the Na–K–Mg–Ca–Cl–SO4–H2O system: a re-evaluation of the sulfate chemistry in the Spencer-Møller-Weare model. Geochim Cosmochim Acta 63:1305–1318CrossRefGoogle Scholar
  14. Marion GM, Kargel JS (2008) Cold aqueous planetary geochemistry with FREZCHEM: from modeling to the search for life at the limits. Springer, BerlinGoogle Scholar
  15. Marion GM, Kargel JS, Catling DC, Jakubowski SD (2005) Effects of pressure on aqueous chemical equilibria at subzero temperatures with applications to Europa. Geochim Cosmochim Acta 69:259–274CrossRefGoogle Scholar
  16. Marion GM, Catling DC, Kargel JS (2006) Modeling gas hydrate equilibria in electrolyte solutions. CALHAD 30:248–259CrossRefGoogle Scholar
  17. Marion GM, Millero FJ, Camões MF, Spitzer P, Feistel R, Chen C-TA (2011) pH of seawater. Marine Chem 126:89–96CrossRefGoogle Scholar
  18. Marion GM, Kargel JS, Catling DC, Lunine JI (2012) Modeling ammonia-ammonium chemistries in the Solar System’s icy bodies. Icarus 220:932–946CrossRefGoogle Scholar
  19. McKay CP, Hand KP, Doran PT, Andersen DT, Priscu JC (2003) Clathrate formation and the fate of noble and biologically useful gases in Lake Vostok, Antarctica. Geophys Res Lett 30(13):1702. doi: 101029/2003GL017490 CrossRefGoogle Scholar
  20. Millero FJ, Feistel R, Wright DG, McDougall TJ (2008) The composition of standard seawater and the definition of the reference-composition salinity scale. Deep Sea Res 55:50–72CrossRefGoogle Scholar
  21. Murray AE, Kenig F, Fritsen CH, McKay CP, Cawley KM, Edwards R, Kuhn E, McKnight DM, Ostrom NE, Peng V, Ponce A, Priscu JC, Samarkin V, Townsend AT, Wagn P, Young SA, Yung PT, Doran PT (2012) Microbial life at −13°C in the brine of an ice-sealed Antarctic lake. PNAS, CGI. doi: 10.1073/pnas.1208607109
  22. Neumann K, Lyons WG, Priscu JC, Donahoe RJ (2001) CO2 concentrations in perennially ice-covered lakes of Taylor Valley, Antarctica. Biogeochemistry 56:27–50CrossRefGoogle Scholar
  23. Pitzer KS (1991) Ion interaction approach: theory and data correlation. In: Pitzer KS (ed) Activity coefficients in electrolyte solutions, 2nd edn. CRC Press, Boca Raton, pp 75–153Google Scholar
  24. Priscu JC, Wolf CF, Takacs CD, Fritsen CH, Laybourn-Parry J, Roberts EC, Sattler B, Lyons WB (1999) Carbon transformations in a perennially ice-covered Antarctic lake. Bioscience 49:997–1008CrossRefGoogle Scholar
  25. Raisanen J, Ylhaisi JS (2011) Cold months in a warming climate. Geophys Res Lett 38:3L22704. doi: 10.1029/2011GL049758 CrossRefGoogle Scholar
  26. Welch KA, Lyons WB, Graham E, Neumann K, Thomas JM, Mikesell D (1996) Determination of major element chemistry in terrestrial waters from Antarctica by ion chromatography. J Chromatogr 739:257–263CrossRefGoogle Scholar
  27. Welch KA, Lyons WG, Whisner C, Gardner CB, Gooseff MN, McKnight DM, Priscu JC (2010) Spacial variations in the geochemistry of glacial meltwater streams in the Taylor Valley, Antarctica. Antarct Sci 22:662–672CrossRefGoogle Scholar
  28. Wharton RA, Lyons WB, Des Marais DJ (1993a) Stable isotopic biogeochemistry of carbon and nitrogen in a perennially ice-covered Antarctic lake. Chem Geol (Isotope Geoscience Section) 107:159–172Google Scholar
  29. Wharton RA Jr, McKay CP, Clow GD, Andersen DT (1993b) Perennial ice covers and their influence on Antarctic lake ecosystems. Physical and biogeochemistry processes in Antarctic Lakes. Antarct Res Ser 59:53–70CrossRefGoogle Scholar

Copyright information

© Springer Science+Business Media Dordrecht 2013

Authors and Affiliations

  • G. M. Marion
    • 1
  • A. E. Murray
    • 1
  • B. Wagner
    • 2
  • C. H. Fritsen
    • 1
  • F. Kenig
    • 3
  • P. T. Doran
    • 3
  1. 1.Desert Research InstituteRenoUSA
  2. 2.University of CologneCologneGermany
  3. 3.University of Illinois at ChicagoChicagoUSA

Personalised recommendations