Glaciological processes under ice sheets provide sustainable ecosystems for microbes, forming an aquatic environment through basal melting, and providing nutrients and energy from bedrock. Microbes facilitate solute production in most Earth surface environments, but the balance of biotic and abiotic weathering in subglacial environment is presently unknown. This study demonstrates an up to eightfold increase in dissolved major cations in biotic relative to abiotic weathering experiments using glacial sediments and meltwater. This conclusion greatly expands our view of Earth’s biogeochemically active weathering zone by incorporating the large wet-based portions of glaciated continents, both at present and during Earth’s history. The profound environmental significance is that microbial processes have the ability to maintain terrestrial chemical weathering rates in cooling climates during glacial advance.


The role of bacteria in mineral weathering in glacial environments on Earth remains poorly understood. In wet-based zones of alpine glaciers and larger contemporary ice sheets, interactions between meltwater and sediments derived from comminuted bedrock liberate significant solute (Anderson et al., 1997; Sharp et al., 1995; Skidmore et al., 2010). This contrasts with geochemical models that assume that negligible chemical weathering occurs at the base of ice sheets primarily due to a limited supply of CO2 from melting basal ice (Kump and Alley, 1994). Microbes have been detected in glaciated environments (Sharp et al., 1999) and ice sheets (Karl et al., 1999), and there is growing evidence that they may be active in bedrock biogeochemical weathering processes (Skidmore et al., 2000, 2005; Wadham et al., 2004, 2010). Short-term (one-week) laboratory weathering experiments at 3 °C suggested that microbial activity may enhance mineral weathering at these low temperatures, but there were no parallel abiotic controls (Sharp et al., 1999). Thus, the importance of biotic versus abiotic weathering processes at low temperatures and the potential for long-term enhanced microbial weathering of bedrock remains an open question. We addressed this issue by conducting long-term (up to 300 days) laboratory weathering experiments specifically designed to simulate natural subglacial conditions of low temperature (2 °C), dark, and realistic water:rock ratios. Parallel experiments with and without microorganisms were conducted to separate biotic and abiotic components of the weathering process.


Glacial waters and sediments were sampled aseptically from Bodalsbreen, an outlet glacier of the Jostedalsbreen ice cap (Norway), and from Haut Glacier d’Arolla (Switzerland). Bodalsbreen is underlain by 1.6–1.8 Ga granitic gneisses (Oftedahl, 1980), and Haut Glacier d’Arolla (hereafter Arolla) is underlain by gneisses, diorites, and gabbros of the Arolla Series (Mazurek, 1986) that contain trace amounts of geochemically reactive minerals (e.g., calcite and pyrite [Tranter et al., 1997]). Sediment slurries at concentrations of 0.6–1 g mL−1 were prepared using only the sediments and glacial waters to simulate natural weathering processes in subglacial environments (Kivimäki, 2005; Montross, 2007) (see a full methods description in the GSA Data Repository1). Indigenous microbes were present in the sediment and meltwater from Arolla, and sterilized sediment from Bodalsbreen was inoculated with indigenous microbes from Bodalsbreen meltwaters. No additional nutrients were added. Live and killed control slurries were both prepared in triplicate and incubated in the dark at 2 °C under oxic and anoxic conditions for 100 days for the Arolla experiments, and oxic conditions only for the Bodalsbreen experiments for 300 days. At eight time points (Bodalsbreen) and nine time points (Arolla), sets of triplicate live and killed control bottles were sacrificed and the chemical composition of the water was analyzed following Kivimäki (2005) and Montross (2007). Bacterial cells in the sediment slurries were enumerated at the start and the end of incubation experiments. Cells in the Bodalsbreen slurries were stained with acridine orange and counted as described by Sharp et al. (1999). Cells in the Arolla slurries were extracted from sediments following Foght et al. (2004), stained with 4’,6-diamidino-2-phenylindole (DAPI), and viewed using epifluorescence microscopy. Bacterial cells were enumerated following the methods of Porter and Feig (1980) and Kirchman et al. (1982).


The cell biomass in the Arolla incubations ranged from 104 to 105 cells g−1 dry wt sediment, and in the Bodalsbreen incubations from 106 to 107 cells g−1 dry wt sediment. These values are similar to those reported for subglacial sediments in the New Zealand Alps (Foght et al., 2004), Svalbard (Kastovska et al., 2007), the Canadian high Arctic (Cheng and Foght, 2007), and the Kamb Ice Stream, West Antarctica (Lanoil et al., 2009). Enhanced release of major base cations (e.g., calcium, potassium, and sodium) occurred in the biotic incubations under oxic conditions at 2 °C relative to the sterile controls in both Bodalsbreen and Arolla sediment slurries (Figs. 1 and 2; Fig. DR1 in the Data Repository; Tables 1 and 2). No significant release of base cations was evident in the biotic relative to the abiotic incubations under anoxic conditions in the Arolla sediment slurries (Table DR3). Nitrate reduction occurred in both oxic and anoxic biotic incubations of Arolla sediments over the first 21 days, with transitory nitrite accumulation (Figs. 3A and 3B), which is characteristic for microbial respiration of nitrate, as nitrite is a metabolic intermediate (Kelso et al., 1999). Release of phosphate into solution occurs contemporaneously with nitrate reduction and may indicate selective targeting of phosphate-bearing minerals (e.g., apatite by microbes during the period of intense nitrate-reducing activity and subsequent assimilation into cellular material). Nitrate, nitrite, and phosphate concentrations remained largely unchanged in the abiotic controls over the 100-day incubation period (Figs. 3C and 3D). Increases in both ferrous iron (Fe2+) and the organic acid anion, acetate, occurred in the biotic incubations of Arolla sediments under anoxic conditions following the depletion of nitrate (Figs. 3B and 4). This likely reflects a change from oxidative metabolism linked to nitrate respiration, which directly produces CO2 from complex organic matter hydrolysis products, to ferric iron reduction to ferrous, which involves fermentation of hydrolysis products as an intermediary reaction producing short-chain fatty acids (e.g., acetate [Lovley and Chapelle, 1995]). No significant increase in acetate was observed in either the biotic or abiotic oxic incubations.

Enhanced release of sodium and potassium in biotic Bodalsbreen and Arolla sediment incubations is primarily from silicate (albite, microcline, biotite) weathering. Only minor (1–2 μM) release of silica was observed in the oxic Arolla incubations (Fig. DR1), indicating that the silicate weathering occurring is incongruent. This is consistent with subglacial water chemistry from Arolla where similar excess concentrations of sodium and potassium over silica have been observed and attributed to incongruent silicate dissolution (Tranter et al., 2002). Calcium and magnesium in Bodalsbreen slurries may also arise from silicate weathering, though it is more likely from weathering of minor carbonates found in the catchment bedrock (0.016–0.19 wt% carbonate), because the dissolution kinetics of carbonates are several orders of magnitude greater than those of silicates (Tranter et al., 1997). Release of ∼30 μM phosphate (Fig. 3A) would produce 50 μM of calcium if the phosphate is solely from the stoichiometric dissolution of apatite; however, this is the maximum contribution of calcium from an apatite source because the phosphate may also be from Fe- or Al-oxyhydroxide bound phosphate (Chen and Wu, 2007). Thus, the majority of the 1105 μeq L−1 calcium released over the 100-day oxic incubation of Arolla sediment slurries is from carbonate weathering.


Microbial respiration produces CO2 that dissolves, forming carbonic acid that drives weathering of carbonate and silicate minerals in both Arolla and Bodalsbreen sediments (Table 1). Sulfide oxidation liberates additional protons for mineral weathering in both Bodalsbreen and Arolla under oxic conditions. However, the contribution is lower relative to that from carbonic acid (31% and 2%, respectively) based on anion concentrations (Table 1). Organic acids may also be a proton source for weathering, but under oxic conditions, only low concentrations of acetate were present (<5 μeq L−1; Fig. 4), suggesting that if acetate is produced, it is rapidly utilized by microbes, ultimately resulting in CO2 production. Sulfide oxidation is microbially mediated (Table 1) and is typically a chemolithautotrophic process requiring a CO2 source. However, we are unable to determine proportions of biotic/abiotic CO2 utilized in this reaction in our experiments.

The weathering experiments demonstrate the presence of a range of different active microbial populations that modify their surrounding environment and drive biogeochemical processes under both oxic and anoxic conditions. The presence of anaerobic nitrate reduction even in the oxic Arolla incubations, probably within reduced microenvironments, and despite physical mixing, is consistent with field measurements that show a gradation of oxic and anoxic zones in the subglacial environment at Arolla (Tranter et al., 2002). Microbial nitrate reduction is followed by both fermentative processes and iron reduction in the anoxic Arolla incubations. Thus, it is likely that anoxic zones in subglacial sediments would also produce reduced iron and acetate, which, when transported to oxic zones, would provide additional energy and carbon sources for aerobic microbes. Some of the microbial processes that are demonstrated and quantified in the laboratory weathering experiments using Arolla subglacial waters and sediment have been previously inferred from field measurements of aqueous geochemistry; however, prior to the work herein the microbial component was unknown and unquantified. The microbial processes include nitrate reduction (Tranter et al., 1994), microbial sulfide oxidation (Tranter et al., 2002), and production of carbonic acid via microbial oxidation of organic carbon to drive the subglacial weathering of carbonates and silicates (Tranter et al., 2005). Thus, this research shows a direct link between the results of laboratory weathering experiments and field measurements, confirming that subglacial microbial activity is a significant driver of chemical weathering in natural, glacial environments.

Estimates of the microbial respiration rates to generate the biotic CO2 for the carbonation reactions for the Arolla and Bodalsbreen incubations (Table 1) are 0.2 and 0.002 pmol CO2 cell−1 day−1, respectively (Table DR4). Calculated respiration rates using data on glacial sediments from Svalbard (Bekku et al., 2004) are within the same range, at 0.0035 pmol CO2 cell−1 day−1. These respiration rates are all lower than those for the microbial community from melted samples of Vostok (Antarctica) ice core from 3603 m depth, incubated at 3 °C, with acetate as the sole amendment and carbon source at 24–30 pmol CO2 cell−1 day−1 (Karl et al., 1999). Only a small fraction (0.0006 wt% and 0.001 wt%) of organic carbon in the Arolla and Bodalsbreen sediments was utilized to produce the respired CO2. Thus the organisms in these sediments do not appear to be organic carbon limited. The results indicate that microbial activity at 2 °C under oxic conditions significantly enhances mineral weathering and solute release relative to the sterile abiotic controls over environmentally relevant time periods.

In abiotic mineral weathering experiments, large increases in solute production result from significant temperature increases, all other things being equal (e.g., grain size, water:rock ratio). Lasaga (1995) measured a sevenfold increase in sodium production in abiotic weathering experiments using the pure mineral phase albite. The increase in sodium production required a 20 °C temperature change from 5 °C to 25 °C in the incubation temperature (Lasaga, 1995). In the microbial weathering experiments presented here, a similar increase in sodium production was observed from the abiotic to the biotic experiments, but at a constant temperature of 2 °C. Although the biotic weathering in the 2 °C sediment incubations likely involves a mixture of minerals, rather than a single mineral, the magnitude of increased weathering compared to abiotic conditions is equivalent to that observed for a large (20 °C) temperature increase if chemical weathering is solely abiotic. Hence, this has profound environmental significance, because microbial processes in cold (glacial) systems have significant impact on terrestrial weathering rates, indicating that terrestrial chemical weathering rates would not necessarily decline in a cool or cooling climate.


Biogeochemical mineral weathering and microbial processes in subglacial environments are enhanced by the availability of water flowing through fine-grained, freshly comminuted debris at the glacier bed (Tranter et al., 2005). Recent work indicates the presence of subglacial water at the bed of larger contemporary ice masses (e.g., Greenland [Layberry and Bamber, 2001] and Antarctica [Gray et al., 2005]) and that wet-based conditions are significantly more widespread than previously thought beneath both West Antarctica (Fricker et al., 2007) and East Antarctica (Siegert et al., 2005; Wingham et al., 2006). Where the ice sheets are wet-based, one would expect microbial activity, under both oxic and anoxic conditions, based on the results of this study, and thus the deep cold biosphere and microbial weathering in these regions may impact on a much broader area of Earth than previously thought. Similarly, reconstructions of basal conditions of the Laurentide ice sheet (Marshall and Clark, 2002) show that >25% of the ice sheet sole would have been at the pressure melting point, thus producing basal melt, from 110 ka to 20 ka, with this proportion increasing to a maximum of 75% for ∼5 k.y. during deglaciation. Thus it seems reasonable to assume that microbial weathering was more extensive beneath the midlatitude ice sheets during the past glacial cycle than suggested in earlier chemical weathering models (Kump and Alley, 1994). Further, it was recently suggested (Falkowski et al., 2008) that only small microbial habitat patches persisted during the late Proterozoic glaciations, when large parts of Earth’s surface may have been covered by ice. We contend that our results demonstrate that where ice sheets are wet-based, microbial populations in these subglacial systems could be active under both oxic and anoxic conditions. Thus a large proportion of the glaciated continents during the late Proterozoic could have been microbial habitat.

Glaciological processes beneath ice masses provide subglacial microbial communities with sustainable ecosystems. Basal melting provides an aqueous habitat and supplies dissolved gases, and mechanical comminution continually provides fresh mineral surfaces, which supplies nutrients (e.g., phosphorus) and energy sources in the form of sulfides. Hence, ice sheet beds provide refugia for microbial life on a timescale of millions of years.

We thank A. Ridgwell for comments on the manuscript. This research was supported by a Natural Environment Research Council grant to Tranter; a University of Bristol studentship and Finnish Environment Institute (SYKE), Kone Foundation (Koneen Säätiö), and Finnish Cultural Foundation (Suomen Kulttuurirahasto) grants to Kivimäki; a Montana State University VP Research grant to Skidmore; and Geological Society of America grant to Montross. Montross was partially supported by the National Science Foundation DUE grant to D. Mogk. J. Mills conducted IC and FIA analyses at University of Bristol; S. Lamoureux, Queen’s University, Canada, performed grain size analysis; and M. Bennes provided logistical field support in Norway.

*Current address: Department of Geography, Queen’s University, Kingston, Ontario K7L 3N6, Canada.
Current address: Association for Water and Environment of Western Uusimaa, P.O. Box 51, 08101 Lohja, Finland.
1GSA Data Repository item 2013052, supplemental methods, Figures DR1 and DR2, and Tables DR1 and DR2, is available online at www.geosociety.org/pubs/ft2013.htm, or on request from editing@geosociety.org or Documents Secretary, GSA, P.O. Box 9140, Boulder, CO 80301, USA.