Geomicrobiology of Subglacial Ice Above Lake Vostok, Antarctica

See allHide authors and affiliations

Science  10 Dec 1999:
Vol. 286, Issue 5447, pp. 2141-2144
DOI: 10.1126/science.286.5447.2141


Data from ice 3590 meters below Vostok Station indicate that the ice was accreted from liquid water associated with Lake Vostok. Microbes were observed at concentrations ranging from 2.8 × 103 to 3.6 × 104 cells per milliliter; no biological incorporation of selected organic substrates or bicarbonate was detected. Bacterial 16S ribosomal DNA genes revealed low diversity in the gene population. The phylotypes were closely related to extant members of the alpha- andbeta-Proteobacteria and the Actinomycetes. Extrapolation of the data from accretion ice to Lake Vostok implies that Lake Vostok may support a microbial population, despite more than 106 years of isolation from the atmosphere.

Lake Vostok is the largest (∼14,000 km2) and deepest (maximum depth ∼ 670 m) lake identified beneath Antarctic glacial ice (1, 2). The residence time of the water in the lake has been estimated to be about 10,000 years, and the mean age of water, since deposition as surface ice, is about 1 million years (2). The ice above the lake has been cored to 3623 m, stopping ∼120 m above the surface of the lake. The upper 3310 m is glacial ice that represents an environmental record covering four complete ice age climate cycles. Ice between 3310 and 3539 m is transitional between glacial and accretion ice; ice below 3539 m represents refrozen lake water accreted to the bottom of the glacial ice (3, 4). Here we describe the geomicrobiological environment within the accretion ice and use the information to predict conditions in Lake Vostok.

We studied a core from a depth range of 3588.995 to 3589.435 m (core 3590) (5). Cross-polarized light observations of the optical section revealed two distinct ice crystals (Fig. 1). The crystal boundaries extended beyond the edge of the core, making it impossible to estimate the exact grain size of either crystal. The C axes of the two crystals made a three-dimensional angle of 24.3° with each other (6). The small and large crystals had declinations of 62° and 43° from the vertical direction, respectively. The horizontal and vertical crystal misalignment could have arisen from seed crystals that nucleated in the lake water or along the margins of the lake before attaching to the bottom of the overlying ice. Alternatively, sheer stresses may have reoriented or recrystallized the ice after accretion.

Figure 1

Cross-polarized image of a 5-mm-thick core section taken from near the middle of core 3590. The horizontal core dimension is 9 cm.

Unfiltered Cl and SO4 2−concentrations in core meltwater fall between the Vostok modern and Vostok Last Glacial Maximum values, indicating that glacial and interglacial snow and ice have melted to produce Lake Vostok (Table 1). Elemental ratios for Al/Rb and Al/Ba in core 3590 were 714 and 192, which are similar to Earth crustal ratios of 704 and 116, respectively (7). NO3 in core 3590 was depleted relative to concentrations in ice from the Last Glacial Maximum and from the last interglacial period. It is not known whether the depletion of NO3 is related to its preferential retention in the lake or loss by biological incorporation or denitrification. Recent experiments (8) indicate little difference between liquid-solid water phase partitioning coefficients for Cland NO3 , implying that NO3 was depleted biologically. Using liquid-solid chemical partitioning coefficients obtained from another Antarctic lake (9), we predict that the upper water column of Lake Vostok contains Na+-SO4 2−waters, similar to many lakes in North America (10).

Table 1

Major ion and trace element concentrations in melted ice from core 3590. Predicted concentrations for water in Lake Vostok were derived from water-to-ice partitioning coefficients obtained from another Antarctic lake (9). Vostok modern and Vostok (LGM) represent concentrations in Vostok glacial ice from the present interglacial period and the Last Glacial Maximum, respectively (32). NA, not applicable. <, limits of detection.

View this table:

The δ18O (11) and δD values of core 3590 ice were −56.8‰ and −445‰, respectively, supporting the results of Jouzel et al. (4). If ice in this core was accreted from Lake Vostok water, as implied by our crystallography data, and the ice was in equilibrium with water at 0°C, the water in Lake Vostok should have isotopic values of −59‰ and −463‰ for δ18O and δD, respectively (12). The δD value from core 3590 is within the range reported for Vostok glacier ice (−420 to −485‰) (3, 4), again suggesting that the lake water is derived from a mix of melted ice from glacial and interglacial periods.

Mineral analysis showed that biotite (73%), quartz (13%), potassium feldspar (9%), plagioclase (2%), muscovite (2%), and iron oxide (1%) were the primary minerals (13). The distribution of mineral phases in these sediments does not reflect the expected proportions of minerals observed in common crustal granitoid rock types (biotite: <20%; quartz: 20 to 55%; potassium feldspar + plagioclase: 40 to 80%; and muscovite and iron oxide: trace amounts) (14). Whether through transport by air, glacier, or subglacial streams, a mechanical sorting process likely operated to concentrate biotite to relatively high levels in core 3590.

Microscopic analysis of melted ice revealed microbes in core 3590 (Fig. 2). Bacterial abundances determined by epifluorescence microscopy of DNA-stained samples and scanning electron microscopy (SEM) were 2.8 × 103 and 3.6 × 104cell ml−1 of melted ice, respectively. These concentrations are similar to those from a Vostok core collected from 3603 m (15) and to those measured in the accretion ice of Lake Bonney, a permanently ice-covered lake in the McMurdo Dry Valleys, Antarctica (16). Assuming that the bacterial partitioning observed between the ice cover and the water column of Lake Bonney (9) is the same as the partitioning in the Vostok system, we estimate that the surface water of Lake Vostok had bacterial cell concentrations on the order of 105 to 106 ml−1 when the ice in core 3590 was accreted.

Figure 2

Scanning electron (A toE) and atomic force (F) micrographs of particles within the core. (B) is a magnification of the area outlined in (A). Arrows indicate bacterial cells.

Genomic DNA was extracted (17) and amplified with archael and bacterial primers. No products were obtained from archael amplifications; bacterial 16S ribosomal DNA (rDNA) fragments were retrieved for terminal restriction fragment length polymorphism (T-RFLP) (18) and sequence analyses. About 12 peaks occurred on the DNA fingerprint of the bacterial 16S rDNA population (Fig. 3). Five peaks also occurred in the negative control and were considered artifacts. Most of the remaining peaks were assigned to major bacterial groups within either the alpha- orbeta-Proteobacteria (19). Seven unique sequences were obtained from cloned polymerase chain reaction (PCR) fragments (20). Four clones were assigned to the Acidovoraxand one clone to the Comamonas; extant members of both subgroups of the beta-Proteobacteria are known to have diverse habitats and physiologies. One clone was assigned to theAfipia subgroup of the alpha-Proteobacteriamembers, which are most commonly associated with root nodules, and one clone to the Actinomyces (typically commensal but often found in soil and water samples) (Table 2). Actinomyces have also been observed in Vostok glacial ice (21).

Figure 3

DNA fingerprint derived from Sau3A digests of 46f(FAM-labeled)/519r amplified bacterial community DNA. Peaks have been designated as an artifact (A) or assigned a taxonomic identity.

Table 2

Designation of taxonomic affiliation of PCR fragment clones from Lake Vostok core 3590. “No peak” indicates that the PCR fragment was not restricted by Sau3A. “Unknown” indicates that the T-RFLP analysis program did not identify a corresponding peak forComamonas. “Artifact” indicates that it appeared in both the negative control and the sample. NA, not applicable.

View this table:

If microbes indeed grow within Lake Vostok, their metabolism may include heterotrophy. Dissolved organic carbon (DOC) concentration in core 3590 was 0.51 milligram of carbon per liter (mg C liter−1) (22), indicating an advected or internal biological source of organic carbon to the lake. The DOC concentration in core 3590 is similar to that measured in the bottom accretion ice of Lake Bonney (16). We know of no DOC data available for Antarctic glacial ice. Recent Greenland snow has a mean DOC concentration of 0.11 mg C liter−1(23), which may actually be on the order of 0.01 mg C liter−1 if corrected for anthropogenic contamination (24). Applying observed partitioning coefficients of DOC obtained from accretion ice and the water column of Lake Bonney to core 3590 (9), we estimate that Lake Vostok had a DOC concentration of 1.2 mg C liter−1 when the ice in core 3590 was accreted. This concentration, if biologically labile, is adequate to support the growth of microbial heterotrophs.

Experiments to examine the incorporation of radiolabeled mannitol, thymidine, amino acid hydrozylate, and bicarbonate into cellular material revealed no heterotrophic or chemoautotrophic growth (25). Our incubation time (52 hours) may have been insufficient to measure cell growth in a slow growing population. Incubations were at 1 atm pressure (about 400 times as low as that in Lake Vostok) and in air, which may contain an unrealistically high oxygen concentration relative to Lake Vostok. The pressure and temperature in the lake should produce an environment low in oxygen owing to oxygen sequestration in gas hydrates (26). The pressure and oxygen conditions in our experiments could have suppressed or inhibited biosynthetic activity. Attempts to culture the cells at 1 atm and in air were negative (27). Microbes in extreme growth conditions might also be using substrates for maintenance activities rather than growth. Hence, metabolic results from core 3590 remain equivocal with respect to the actual viability of the microbes.

Microbes within a liquid water habitat deep below a frozen surface provide an analog for possible life on Europa, one of the Galilean moons of Jupiter. Galileo spacecraft results imply that a subsurface ocean exists on Europa (28). Although the thickness of the overlying ice in Europa is unknown (29), ice would accrete to the bottom of the ice cover and would also form in cracks, possibly extending close to the surface (30). Similar to Lake Vostok accretion ice, this ice may retain evidence for life, if present, in the europan ocean.


Stay Connected to Science

Navigate This Article