Perennial Antarctic Lake Ice: An Oasis for Life in a Polar Desert

See allHide authors and affiliations

Science  26 Jun 1998:
Vol. 280, Issue 5372, pp. 2095-2098
DOI: 10.1126/science.280.5372.2095


The permanent ice covers of Antarctic lakes in the McMurdo Dry Valleys develop liquid water inclusions in response to solar heating of internal aeolian-derived sediments. The ice sediment particles serve as nutrient (inorganic and organic)–enriched microzones for the establishment of a physiologically and ecologically complex microbial consortium capable of contemporaneous photosynthesis, nitrogen fixation, and decomposition. The consortium is capable of physically and chemically establishing and modifying a relatively nutrient- and organic matter–enriched microbial “oasis” embedded in the lake ice cover.

The McMurdo Dry Valleys, Antarctica, is one of the coldest and driest deserts on Earth. Lakes in this region are permanently ice covered (1). The ice is typically 3 to 6 m thick and contains a layer of sand and organic matter of aeolian origin below the surface. This layer represents a dynamic equilibrium between downward movement of sediments as a result of melting during the summer and upward movement of ice from ablation at the surface and freezing at the bottom. Liquid water inclusions are present in this layer for about 150 days during the summer when solar radiation is continuous; up to 40% of the total ice cover volume during this period can be liquid water (2, 3). We discovered that the ice meltwater supports a viable microbial assemblage associated with the sediment layer. Here, we describe the ecosystem.

We collected ice samples from six lakes (Bonney, Hoare, Fryxell, Miers, Vanda, and Vida) between August and October 1993 and 1995 using 10-cm-diameter coring devices. Cores were sectioned, melted, and analyzed for photoautotrophic and heterotrophic activity; biomass, sediment, and nitrous oxide content; and chemistry (4). Most of our studies were on Lake Bonney. The ice cover of Lake Bonney was 4 m thick, and the sediment layer was 2 m beneath the ice surface (Fig. 1A). N2O and dissolved inorganic nitrogen (DIN) (NH4 + + NO2 + NO3 = DIN) reached a maximum at this depth, indicating that the region was a source for nitrogen (Fig. 1, A and B). N2O is a known intermediate of nitrification in the liquid water column of this lake (5). Soluble reactive phosphorus (SRP) showed no clear trend with depth. The DIN:SRP ratio averaged 31.3 (range of 20.3 to 107.7), indicative of phosphorus deficiency (6). N2fixation (7) ranged from 9.4 to 91.3 nmol of N2per milligram of chlorophyll a per hour in lake ice aggregates. Primary productivity, bacterial activity, chlorophylla concentration, and bacterial cell number also were highest in or near this sediment layer (8) (Fig. 1, C and D). Dissolved organic carbon (DOC) was highest near the peaks in primary and bacterial productivity. Primary productivity ranged up to 7.8 μg of C per liter per day, and bacterial productivity ranged up to 0.40 nM thymidine per day in the combined data set from all six lakes. High photoautotrophic biomass was associated with high bacterial biomass in the combined data set.

Figure 1

(A to D) Vertical profiles of selected constituents in the lake ice cover of Lake Bonney (8). TdR, incorporation of 3H-thymidine.

A complex microbial assemblage was physically associated with the sediment. The sediment is composed of aggregated sand and gravel, distributed in centimeter-scale inclusions within the ice when no liquid water is present (Fig. 2A). Bubbles present above the sediment layer likely formed from gas exsolution as the liquid water lens froze from the top downward during early winter (2, 3). Confocal microscopy (9) revealed bacterial and cyanobacterial cells attached to the sediment particles (Fig. 2B). Most of the cyanobacteria are filamentous species of the genus Phormidium (Fig. 2B, top right). More rarely, we observed cells of the cyanobacterial genusChamaesiphon (Fig. 2B, top left), the nitrogen-fixing genusNostoc, unidentified coccoidal cyanobacteria, and diatom algae. Phylogenetic analysis (10) of the oxygenic phototrophs was performed by cloning bacterial 16S ribosomal RNA genes directly from the ice assemblage DNA (Fig. 3). Thirty-one of the 198 bacterial genes sequenced were of cyanobacterial origin; among these were four clades. Three clades were allied with the genera Leptolyngbya,Chamaesiphon, and Phormidium. The fourth could not be associated with any of the cyanobacterial genes in public sequence databases. These observations establish that representatives of widely distributed cyanobacterial groups have successfully colonized the ice habitat. The lake ice organisms do not resemble the predominantly eukaryotic plankton within the water column of Lake Bonney.

Figure 2

(A) An aggregate from 2 m beneath the surface of the permanent ice cover of Lake Bonney. The photograph was taken from within a 3.5-m-deep trench cut into the ice in early September before the formation of liquid water. Scale bar, 2 cm. (B) Confocal laser photomicrographs showing microorganisms associated with a sediment particle, with enlarged views of two species of cyanobacteria (blue, DAPI-stained bacteria; red, chlorophyll autofluorescence; gray, sediment particle). Scale bars, 10 μm. (C) Microautoradiograph of sediment particles bound by cyanobacterial filaments (dark regions denote sites of active 14CO2 accumulation, indicative of photosynthetic activity). Scale bar, 100 μm.

Figure 3

Phylogenetic tree illustrating relations among selected 16S ribosomal DNA clones from the ice cover on Lake Bonney (clones from this study are designated LB3) and representative cyanobacterial sequences. The number of bootstrap replicates that supported each branch, from a total of 100 replicates, is shown above (neighbor-joining) and below (parsimony) each node. Bootstrap values below 50% are not shown. The 16Sribosomal DNA sequence of Escherichia coli was used to root the tree.

Our data imply that ice sediment particles serve as nutrient (inorganic and organic)-enriched microzones for the establishment of a physiologically and ecologically complex microbial consortium capable of contemporaneous photosynthesis, N2 fixation, and decomposition. All of these processes are needed to complete essential nutrient cycles. CO2- and N2-fixing capabilities ensure access to biologically available carbon and nitrogen, key growth-limiting nutrients in aquatic and terrestrial habitats. Microautoradiography revealed that filamentous cyanobacteria actively fix CO2 upon exposure to meltwater (Fig. 2C). Microautoradiographs also show that bacteria are capable of metabolizing a range of organic substrates closely associated with cyanobacteria (11). These activities imply a consortial arrangement within the assemblage where photoautotrophs are supplying fixed carbon (and nitrogen in the case of N2-fixing species) and heterotrophs are cycling CO2 and inorganic nutrients back to the photoautotrophs. Highest dissolved inorganic carbon (DIC) concentration (1.2 mg of C liter−1) corresponded to the region of greatest metabolic activity. Substrate kinetics experiments showed that DIC in this region does not limit photosynthetic activity. Microscopy also revealed that cyanobacterial filaments bind sediment particles, forming a cohesive fabric that produces a relatively nonfriable aggregate (Fig. 2C). Our data support the contention that this assemblage is capable of physically and chemically establishing and modifying a relatively nutrient- and organic matter–enriched microbial “oasis” embedded in the lake ice cover.

A model of sediment and associated particulate organic carbon (POC) dynamics within the ice cover of Lake Bonney illustrates the dominant physical and biological processes relevant to the transport and production of POC in this system. Aeolian transport is the primary allochthonous source of sediment and POC to the lake ice. The sediments and associated POC migrate to the depth in the ice in dynamic equilibrium with radiation-induced downward melting caused by sensible heating and with the net upward movement of the ice (3, 12). Data from thermocouples embedded in the ice and observations from ice-core data indicate that meltwater in association with the sedimentary layers is present from November to February (3) and supports photosynthetic carbon production during this period of continuous sunlight (13,14). Daily average photosynthetically available radiation (PAR) reaching the photosynthetic layer ranged from 50 to 200 μmol of photons m−2 s−1 during the period of liquid water (14). We estimated that the median growth rate of photoautotrophs is 0.5 year−1 (range = 0.08 to 1.6 year−1) using rates of photoautotrophic protein synthesis (14).

POC is lost from the ice cover through cracks and meltwater conduits that develop during the austral summers. We evaluated the POC balance within the ice cover of Lake Bonney using the following numerical relation in concert with data collected on the system:Embedded Imagewhere C is the POC concentration per area,t is time, μ is the photoautotrophic growth rate (0.5 year−1), C i is the standing stock of microalgal POC in the ice (375 mg of C m−2) (15), Q a is the aeolian flux of POC onto the lake ice (43 mg of C m−2 year−1) (15), and Q w is the sinking flux of POC from the ice (2.3 mg of C m−2 year−1) (15). This model represents net fluxes of POC only, neglecting losses to DOC and CO2.

Our data imply that 229 mg of POC m−2year−1 accumulated in the Lake Bonney ice cover during our study. The model predicts an unsteady-state condition where both POC and sediment accumulate over the year. These dynamics are expected to vary annually as local climate conditions affect PAR, aeolian deposition, and sediment loss from the ice cover. The model indicates that >80% of the computed POC accumulation results from carbon produced by cyanobacterial photosynthesis despite low growth rates and shows that the permanent ice covers provide viable habitats for the microscale proliferation of life in what would appear to be an otherwise inhospitable macroscale environment. The physical and biological dynamics of the permanent ice covers we describe differ from that in marine pack ice (16) and winter ice of temperate lakes (17). Both of these systems are characterized by freeze-flood cycles driven by snow deposition and are inhabited by organisms originating from the underlying liquid water column. Conversely, the microbial habitats in the ice covers of the dry valley lakes arise from internal melting associated with aeolian-deposited sediments, which also provides the biological seed.

The habitat in the Antarctic lake ice may serve as a model for life on Mars and Europa. Although Mars may have had extensive liquid water at one time, it rapidly cooled, and ice would have become, as it is today, the dominant form of water on the surface (18,19). On Europa, surface ice appears to exist in contact with subsurface liquid water (20–22). Solar heating of the subsurface could result in melt layers similar to those we describe here.


View Abstract

Stay Connected to Science

Editor's Blog

Navigate This Article