Bacterial Activity in South Pole Snow Is Questionable

Carpenter, Lin, and Capone (6) collected surface snow at South Pole Station, warmed it in the laboratory to temperatures of −17 to −12°C, injected solutions of radioactive thymidine and leucine, and found evidence for bacterial metabolism. They argued that the bacteria would also be metabolizing in situ and thus over a vast area of the Antarctic continent. Here we question that conclusion, by considering the availability of liquid water.

South Pole Station has a continuous record of weather observations since 1956, including temperatures registered by a thermometer 2 m above the snow surface. The two warmest months, December and January, both have average temperatures of −28°C (15). The maximum temperature has exceeded −17°C on only 8 days in 46 years, and the all-time record high is −13.6°C (15). At the South Pole there is no daily cycle, but at more northerly locations on the Antarctic Plateau, under direct sunlight at noon the snow surface can briefly be as much as 3 degrees warmer than the near-surface air. However, the highest temperatures are usually experienced under overcast cloud, when snow and air temperatures are equal.

Three processes might produce liquid water in such cold snow, as follows.

(i) A curved ice surface has a slightly lower melting temperature than a flat surface. The surface snow grains at South Pole have radii of 30 to 100 μm (11), for which the melting temperature is lowered by only 0.001 to 0.002 degrees (14), making this effect negligible.

(ii) At temperatures close to 0°C, a quasiliquid layer (QLL) exists on the surface of a snow grain, because the surface energy of an ice-vapor interface slightly exceeds the sum of ice-liquid and liquid-vapor surface energies. The thickness of the QLL has been measured by several methods (4, 8-10) and also computed theoretically (7, 16). It can be 10 to 100 nm thick at −0.1°C but shrinks rapidly with decreasing temperature, to 0.1 to 1 nm at −10°C. Carpenter et al. misquoted Anderson (1) by a factor of 100, saying that he found a QLL of 50 nm at −10°C. In fact, the thickness shown in Anderson's figure is only 5 Å (i.e., 1 to 2 molecular layers). They also cited Yershov (18) for evidence of 0.5 to 3% unfrozen water in permafrost at −10°C. However, that water is in the form of monomolecular layers between the lamellae of expandable clay minerals (2, 18), so it would be inaccessible to bacteria. In any case, Antarctic snow is very different from permafrost; it contains only 15 ppb of mineral dust (12).

(iii) The one process that can produce nonnegligible quantities of liquid water is lowering of the freezing point by solutes. Solutes are rejected from the ice lattice, so they become concentrated on the surfaces of snow grains, where they may create a thin liquid layer. The major solutes in Antarctic snow are H₂SO₄, HNO₃, NaCl, and HCl (13). In their relative abundances at South Pole, a freezing-point depression of 13.6 degrees requires a concentration of 2.7 M. Their concentration in bulk snow, 4 μM (13), implies a mass ratio (liquid/ice) of 1.5 × 10⁻⁶ if all the ions are partitioned into the liquid phase, a generous assumption (3). In ice, this acidic brine (pH ≈ 0) is located in veins at three-grain junctions (17). In snow, it would be located in an annular groove around the neck where two grains have joined by sintering. In South Polar snow, the radius of this neck is approximately one-half the grain radius (see Fig. 1 of reference 11). Assuming each grain contacts four other grains with radii of 30 to 100 μm, the width of the brine channel is 30 to 100 nm at −13.6°C; the channel is smaller and saltier at lower temperatures.

Thus, liquid water in Antarctic snow is hidden in narrow crevices much smaller than the 500-nm-diameter cells shown in the scanning electron microscopy images (6), so the water is unlikely to be accessible to them, even if they could tolerate its high acidity and salinity. In the laboratory incubations, by contrast, a minimum of 0.2 ml of low-salinity liquid water was present, at least until the injected solutions froze. This is 50,000 times the amount of liquid water naturally present in the snow. The mass of snow in the 7-ml test tube, at the average density of surface snow at the South Pole, 0.34 g cm⁻³ (5), would be 2.4 g. The heat capacity of ice is 0.5 cal g⁻¹ deg⁻¹; the latent heat of freezing is 80 cal g⁻¹. So, injection of 0.2 ml of liquid at 0°C would (by freezing) warm the snow from −12°C to the melting point, resulting in a slush which would then eventually refreeze after returning to the cold room. We suggest that the metabolic activity shown for South Polar bacteria, in snow that was significantly altered in the laboratory, does not provide evidence for metabolism in situ.

Bacteria undoubtedly exist in South Polar snow; they can be carried by the wind, as are other atmospheric aerosols. But within 15 years they are buried to a depth of 3 m, where the temperature is close to −50°C year-round (5), and even during their brief sojourn at the surface, water is exceedingly scarce.