Diversity and characterization of mercury-resistant bacteria in snow, freshwater and sea-ice brine from the High Arctic


  • Editor: Max Häggblom

Correspondence: Niels Kroer, National Environmental Research Institute, Aarhus University, Frederiksborgvej 399, DK-4000 Roskilde, Denmark. Tel.:+45 46 30 13 88; fax: +45 46 30 12 16; e-mail: nk@dmu.dk


It is well-established that atmospheric deposition transports mercury from lower latitudes to the Arctic. The role of bacteria in the dynamics of the deposited mercury, however, is unknown. We characterized mercury-resistant bacteria from High Arctic snow, freshwater and sea-ice brine. Bacterial densities were 9.4 × 105, 5 × 105 and 0.9–3.1 × 103 cells mL−1 in freshwater, brine and snow, respectively. Highest cultivability was observed in snow (11.9%), followed by freshwater (0.3%) and brine (0.03%). In snow, the mercury-resistant bacteria accounted for up to 31% of the culturable bacteria, but <2% in freshwater and brine. The resistant bacteria belonged to the Alpha-, Beta- and Gammaproteobacteria, Firmicutes, Actinobacteria, and Bacteriodetes. Resistance levels of most isolates were not temperature dependent. Of the resistant isolates, 25% reduced Hg(II) to Hg(0). No relation between resistance level, ability to reduce Hg(II) and phylogenetic group was observed. An estimation of the potential bacterial reduction of Hg(II) in snow suggested that it was important in the deeper snow layers where light attenuation inhibited photoreduction. Thus, by reducing Hg(II) to Hg(0), mercury-resistant bacteria may limit the supply of substrate for methylation processes and, hence, contribute to lowering the risk that methylmercury is being incorporated into the Arctic food chains.


A growing body of evidence implies atmospheric mercury to be the main source of mercury entering the Arctic environment (Schroeder et al., 1998; Skov et al., 2004; Steffen et al., 2008). Because mercury may stay in the atmosphere as elemental mercury Hg(0) for a prolonged period of time, mercury emitted in other parts of the world finds its way to the Arctic (Steffen et al., 2008). During atmospheric mercury depletion events (AMDEs), Hg(0) in the atmosphere is oxidized to Hg(II) through reaction with Br atoms, released from refreezing leads (stretches of open water within fields of sea ice), when exposed to solar radiation (Skov et al., 2004; Steffen et al., 2008). Hg(II) is then deposited onto the snow and sea ice, thereby entering the ecosystem. Although atmospheric mercury is deposited year round in the Arctic, the deposition of mercury is doubled in the spring due to AMDE (Skov et al., 2004).

Studies have shown that part of the deposited mercury is reduced within the snow and emitted back to the atmosphere as volatile mercury (St Louis et al., 2005), and that the half-life of Hg(II) in snow is short (Dommergue et al., 2007). Evidence suggests that this reduction is light dependent (Poulain et al., 2004; Dommergue et al., 2007), but it is not known as to what extent biological reduction by microorganisms plays a role.

In temperate environments, it is well-established that microorganisms are important players in the biogeochemical cycling of mercury. Loci conferring bacterial mercury resistance typically encode a mercuric reductase (MerA) that reduces reactive ionic Hg(II) to volatile and less toxic Hg(0) vapor. Some resistance loci encode an additional enzyme, an organomercury lyase (MerB), that degrades organomercurials into Hg(II) (Barkay et al., 2003). This means that resistant bacteria carrying both the merA and the merB genes are capable of detoxifying organic and ionic mercury by transformation into gaseous Hg(0). Sulfate- and iron-reducing bacteria, on the other hand, may methylate Hg(II) under anaerobic conditions in sediments and wetlands (Loseto et al., 2004; Fleming et al., 2006). Thus, in temperate environments, bacteria may impact the levels of methylmercury (MeHg) by both methylating and demethylating processes. Whether this is also the case in High Arctic environments remains to be elucidated.

Mindlin et al. (2005) isolated mercury-resistant bacteria from sediments in permafrost, and Poulain et al. (2007) showed that the merA gene, specifying the reduction of Hg(II) to Hg(0), was present and actively transcribed in coastal lagoons and sea-ice leads in the Canadian High Arctic. Mercury-resistant bacteria from snow, freshwater and brine, however, have not been enumerated and characterized.

Here, we investigated the potential for biological mercury reduction in High Arctic environments during AMDEs. We isolated and identified mercury-resistant microorganisms from three different habitats in the High Arctic: snow, freshwater and sea-ice brine water. The isolates were characterized by the level of resistance to mercury at different temperatures and by their ability to reduce Hg(II) to Hg(0). In addition, we estimated potential microbial mercury reduction rates and found that the microbiological reduction may potentially contribute up to 2% of the total reduction.

Materials and methods

Study area

Samples of snow, freshwater and sea-ice brine water were collected in Spring 2007 at Station Nord in Northeastern Greenland. Snow and brine water were collected at two sites in Dagmar Sund (Site 1: 81°36.58′N; 16°42.83′W; Site 2: 81°35.46′N; 16°45.91′W) between Station Nord and Prinsesse Dagmar Island, while the freshwater samples were taken from a small ice-covered lake 2 km south of Station Nord (81°34.48′N; 16°37.46′W). The lake had been ice covered for at least 22 months before sampling.

Measurement of atmospheric ozone

Ozone was measured with a UV absorption monitor (Teledyne Technologies Inc., CA) with a detection limit of 1 part per billion volume (p.p.b.v.) and an uncertainty of 3% for concentrations >10 p.p.b.v. and 6% for concentrations <10 p.p.b.v. (all uncertainties are at 95% confidence interval) (Skov et al., 2004).

Sampling of snow, brine and freshwater

Sampling of snow and brine at Site 1 was carried out on May 12 and at Site 2 on May 25. Total mercury concentrations were measured at both sites, while microbiology sampling was carried out only at Site 1.

The snow depth from the surface to the sea ice was 120 cm at Site 1 and 105 cm at Site 2. A vertical snow profile of the snow pack was created by digging with a sterile shovel. Immediately before sampling, the outermost 1 cm snow of the profile was removed with a sterile knife. Snow for determination of dissolved organic carbon (DOC), bacterial numbers and enumeration and isolation of culturable bacteria was collected using a sterile Plexiglas corer (internal diameter: 14 cm, length: 25 cm) that was inserted horizontally into the snow pack. Six to seven cores were taken at 31–52 cm, 75–90 cm and 96–112 cm depths and transferred to sterile plastic buckets covered with a lid. Sampling was based on observations of the snow texture to insure that different snow layers were sampled. The snow was melted slowly at 5–7 °C (up to 48 h) to avoid stressing of the bacteria during melting. Depending on the snow texture, one core of snow resulted in 850–1250 mL melt water [average snow density of 0.28 g mL−1, which is comparable to previous observations at Station Nord (Ferrari et al., 2004)]. Sampling for determination of the total mercury concentration was performed using a glass tube (length: 11.5 cm, internal diameter: 2.1 cm) that was inserted horizontally into the snow at the same depths as described above. Samples were transferred to triplicate 100-mL glass bottles, containing 200 μL concentrated HNO3 (final pH 2), and sealed with a lid with a Teflon insert. All glassware was prerinsed with 4 M H2SO4. Samples were stored a 4–5 °C until analysis.

Brine water was collected by removing the snow cover and drilling three replicate holes (40-cm deep and 20-cm wide) with an ethanol-sterilized ice drill. Slush ice in the bore holes was removed with a sterile spoon and brine water seeping into the holes was collected using a sterile pipette and pooled. The salinity of the brine was measured with a refractometer (Atago). Subsamples (50 mL) for total mercury determination were treated as described for the snow samples.

Freshwater samples were collected on June 13 by pumping from approximately 70 cm depth below the ice. The hose from the pump was flushed with 3000 L of lake water before 2 L were collected in a sterile glass bottle. Subsamples (50 mL) for total mercury determination were treated as described for the snow samples.

Total mercury concentration

Samples (50 mL) were treated for 24 h with 500 μL of 2 mM BrCl2 to convert all mercury species to Hg(II). Excess BrCl2 was neutralized with the addition of 500 μL of 30% NH2OH. The volumes were adjusted to a total of 100 mL and poured into 1-L Teflon flasks containing 1 mL of 20% SnCl2. The flasks were immediately capped, shaken vigorously for 30 s and mercury in the head space was measured on a Tekran Mercury Vapor Analyzer 2537A. Three independent samples from each site were analyzed and the concentration of Hg(II) was determined against a six-point standard curve (0–100 ng HgCl2 L−1). Measurements that were not significantly different from the blanks were not included in the study (Students' t-test, P<0.05). All glassware used for the mercury measurements was rinsed three times in 4 M H2SO4 and rinsed several times in MilliQ water before use.


Triplicate samples of 10–15 mL melt water and brine water were filtered through 0.2-μm Sartorius Minisart syringe filters into acid-rinsed (10% HCl) glass scintillations vials and stored frozen at −15 to −18 °C. The syringe filters were rinsed in 5 mL sample water before collecting the sample. The concentration of organic carbon was measured on a Shimadzu TOC5000 analyzer (Kroer, 1993). Before analysis, samples were acidified and purged with O2 for 5 min to remove inorganic carbon.

Total bacterial abundance

Total bacterial abundance was determined by direct counts with an Olympus BH2 microscope. Bacteria were collected on 0.1-μm (brine samples) or 0.2-μm pore-size black polycarbonate membranes (Osmonics/Nucleopore) and stained with a 1 : 1000 × dilution of SYBR Gold (Invitrogen). Samples were analyzed immediately after sampling or as soon as snow samples were melted.

Enumeration and isolation of culturable bacteria

Bacteria were isolated by two different procedures: (1) direct plating and (2) preincubation under simulated natural conditions using polycarbonate membranes as a growth support before plating on standard medium (Rasmussen et al., 2008).

(1) Direct plating: 5 × 100 μL of melted snow, brine and freshwater were plated onto 10% strength of tryptic soy agar (TSA) (Tryptic Soy Broth, Sharlauf Microbiology, Denmark and Noble agar, Difco, Denmark) plates. The medium was prepared using autoclaved water from the appropriate sampling location. Plates were incubated at 4–10 °C and the CFUs were counted at successive intervals until a constant count was obtained. All colonies appearing on the plates were restreaked for purity on 10% TSA plates prepared with distilled water (for isolates originating from snow and freshwater) or 10% TSA supplemented with sea salts (Sigma-Aldrich) adjusted to the same salinity as the sea-ice brine (4.5%).

The isolates were streaked on 10% TSA plates supplemented with 10 μM HgCl2 to test for mercury resistance. Mercury resistance was scored as positive if single colonies grew on the plates. Isolates growing on the mercury plates were restreaked on fresh plates of an appropriate mercury-containing medium at least three times to confirm purity and mercury resistance.

(2) Preincubation on polycarbonate membranes: 5 mL brine water, 50 mL melted snow and 5 mL freshwater were filtered through 0.1-μm (brine) or 0.2-μm pore-size polycarbonate membranes (25 mm diameter, Nucleopore). The polycarbonate membranes were placed on fixed 0.22-μm Anopore discs of 25-mm Nunc tissue culture inserts (Nunc A/S Roskilde, Denmark) with the bacterial cells facing upward. Each of the assembled membrane-tissue culture inserts was placed in a separate well of six-well Costar plates, each well containing 1 mL of sample water supplemented with tryptic soy broth (TSB) at concentrations of 0%, 0.01% or 0.1% as growth medium. The membranes were incubated at 4–10 °C. The growth medium was replaced with fresh medium every 7 days and the formation of microcolonies was followed by microscopy. After 77 days of incubation, microcolonies on three replicate filter membranes from each location were dislodged by placing them in an Eppendorf tube with 1 mL salt buffer [KH2PO4 (0.25 g L−1), MgSO4·7H2O (0.125 g L−1), NaCl (0.125 g L−1), (NH4)2SO4 (0.2 g g L−1)] and vortexed vigorously for 1 min. Appropriate dilutions were prepared in salt buffer and 100 μL was plated on 10% TSA plates (prepared as described above). Plates were incubated at 6 °C. Between 140 and 200 colonies from each location were randomly picked and the colonies were restreaked for purity on 10% TSA plates prepared with distilled water (for snow melt water and freshwater samples) or 10% TSA supplemented with sea salts (Sigma-Aldrich) adjusted to the same salinity as the brine water (4.5%). Mercury-resistant isolates were identified as described above.

After 90 days, cells from one filter from each sampling location were extracted and plated onto 10% TSA plates supplemented with 10 μM HgCl2. All isolates growing on mercury plates were restreaked on fresh mercury plates at least three times to confirm their purity and mercury resistance.

Identification of isolated mercury-resistant bacteria

DNA was extracted from the isolated mercury-resistant bacteria by boiling lysis (Fricker et al., 2007). For some of the isolates (8D5s, 8D7, 8D12, 8D12b, SOK1b, SOK 15, SOK17a, SOK17b, SOK19, SOK19y, SOK27, SOK32, SOK33, SOK35, SOK38, SOK43, SOK48, SOK52, SOK57), extraction by boiling was not applicable. Instead the PowerMax DNA soil kit (MoBio Laboratories Inc., Carlsbad, CA) was used following the manufacturer's instructions.

The 16S rRNA gene of the bacteria was amplified by PCR with universal bacterial primer sets GM3F (AGA GTT TGA TCM TGG) and GM4R (TAC CTT GTT ACG ACT T) (Muyzer et al., 1995) using the following thermocycling conditions: 5 min at 95 °C followed by 35 cycles of 1 min at 95 °C, 1 min at 46 °C and 3 min at 72 °C; the last cycle was followed by 10 min at 72 °C. For a few isolates it was not possible to get PCR bands with primers GM3F and GM4R. Instead, primers 27f (AGA GTT TGA TCM TGG CTC AG) and 519r (GWA TTA CCG CGG CKG CTG) (Lane, 1991) were applied using 2 min at 95 °C followed by 35 cycles of 30 s at 95 °C, 30 s at 55 °C and 1 min at 72 °C followed by a final extension for 5 min at 72 °C. PCR mixtures (25 μL) consisted of 2 μL DNA template (10-fold dilutions were used to ensure optimal DNA concentration), 1 U Taq polymerase, 0.4 μM of each primer, 400 μM dNTPs and 2 mM MgCl2 (0.5 mM MgCl2 for 27f and 519r).

No PCR products were obtained from three of the isolates using the bacterial primers. Microscopic examination suggested that these isolates were not bacteria but putative microfungi. To verify this, a new PCR (2 min at 95 °C followed by 35 cycles of 30 s at 95 °C, 30 s at 55 °C and 1 min at 72 °C s followed by a final extension for 5 min at 72 °C) was set up targeting the internal transcribed spacer region ITS1–5.8S rRNA gene–ITS2 between the fungal 18S and 28S rRNA genes using primers ITS1 (TCC GTA GGT GAA CCT GCG G) and ITS4 (TCC TCC GCT TAT TGA TAT GC) (White et al., 1990). The PCR mixture consisted of 2 μL DNA template, 1 U Taq polymerase, 0.4 μM of each primer, 400 μM dNTPs and 0.5 mM MgCl2.

PCR products were gel purified using a Qiagen Gel Extraction Kit (Qiagen, Denmark). Sequencing of the purified PCR products in one direction with primer GM3F (or ITF1 or 27f) was performed by GENEWIZ (South Plainfields, NJ). The quality of the sequences was manually checked and the closest known relatives were determined using blast 2.2.1 (http://www.ncbi.nlm.hih.gov/BLAST/). The closest related sequences in GenBank together with one representative sequence from each subgroup (99% sequence similarity) were aligned and a phylogenetic tree was constructed using the neighbor-joining method using mega4 (Tamura et al., 2007).

Determination of the minimal inhibitory concentration (MIC) of mercury

The MIC was defined for at least one representative isolate from each of the 16S rRNA gene groups. Cell material of freshly grown isolates on 10% TSA plates supplemented with 5 μM HgCl2 was resuspended in 2 mL TBS and the OD (A600 nm) was adjusted to 0.100. From appropriate dilutions, 100 μL was plated on 10% TSA supplemented with 0, 10, 25, 50, 100 and 200 μM HgCl2. Plates were incubated at 3, 12 and 20 °C or at 4 and 15 °C, and numbers of CFUs were counted. The MIC was determined as the lowest concentration of HgCl2 that resulted in a >80% reduction in numbers of CFUs relative to control plates with no added HgCl2.

Mercury volatilization assay

The ability of the isolates to volatilize Hg(II) to Hg(0) was determined using the assay of Nakamura & Nakahara (1988). Isolates were grown on 10% TSA supplemented with 5 μM HgCl2 and resuspended into assay buffer [0.07 M phosphate buffer (pH 7.0) containing 0.5 mM EDTA, 0.2 mM magnesium acetate and 5 mM sodium thioglycolate]. HgCl2 was added to the cell suspensions to a final concentration of 250 μM. Microtiter plates containing the cell suspensions were covered with an X-ray film (Kodak Scientific Imaging, Ready Pack, Rochester, NY) and incubated at room temperature for 2 h in the dark. Alternatively, for cells producing a negative result, cells were resuspended in 10% TSB supplemented with 5 μM HgCl2 to insure induction of the mercuric reductase. After 20 min incubation at room temperature, the cells were washed and resuspended in assay buffer and set up in microtiter plates with 250 μM HgCl2. Isolate SOK32 showed very strong mercury reduction and was routinely used as positive control. Assay buffer with no cells served as the negative control.

Accession numbers

The 71 bacterial 16S rRNA gene sequences and the three fungal ITS regions have been submitted to GenBank and given the accession numbers GU932907GU932980.


Physical/chemical characterization of the sampling site

The atmospheric temperature during the sampling period ranged between −22.5 and −2.9 °C in May and between −12.8 and 7.8 °C in June. The temperature range on the two sampling days was −15.9 to −6.2 °C (May 12) and −10.7 to −6.3 °C (May 25), respectively. Continuous ozone measurements showed that atmospheric ozone was highly variable with periods close to 0 p.p.b. (Fig. 1a). Fluctuating and low ozone concentrations are usually observed during AMDEs (Fig. 1b) and, therefore, AMDEs were likely to have occurred during our sampling period.

Figure 1.

 Atmospheric ozone concentrations (grey line) and atmospheric mercury concentrations (black line). (a) The ozone concentration between March and December 2007. The sampling period is indicated by the rectangle. No mercury measurements were available in 2007. (b) Ozone and mercury concentrations in 2008–2009. As seen on the graphs in (b), on days when the ozone concentration was low, so was the mercury concentration. Thus, the ozone concentration is indicative of the mercury concentration, i.e. of AMDEs.

Measured mercury concentrations in brine and snow were in the range of 70–80 ng L−1 (Table 1). Several of the samples had mercury concentrations that were not significantly different from the blanks. The DOC concentration in brine was almost 40 mg L−1. In snow, the concentration was in the range of 1–4 mg L−1 (Table 1).

Table 1.   DOC and total mercury concentration
SampleDOC (mg C L−1)Total Hg (ng L−1)
  1. SEs (n=3).

  2. NS, not significantly different from blanks (Student's t-test, P<0.05); ND, not determined.

Site 1
 Brine37.5 ± 11.1NS
 Snow (31–52 cm)4.0 ± 0.7NS
 Snow (75–90 cm)3.8 ± 0.872.2 ± 4.2
 Snow (96–112 cm)1.3 ± 0.02NS
Site 2
 BrineND79.6 ± 26.8
 Snow (7–24 cm)ND76.5 ± 6.3
 Snow (42–58 cm)ND69.2 ± 13.1
 Snow (81–95 cm)NDNS

Bacterial abundance

The bacterial abundance in snow ranged from 8 × 102 to 3 × 103 cells mL−1 (Table 2). In freshwater and brine, densities were up to two orders of magnitude higher. The cultivability of the bacteria was highly variable, ranging from <0.1% in brine to 12% in snow (Table 2). The highest cultivability was observed in the two deepest snow layers. It should be noted that only one colony was isolated from the uppermost snow layer, making the estimate of the cultivability somewhat uncertain. Among the culturable bacteria, the highest percentage of mercury resistance (31.2%) was observed for the deepest snow layer, where the fraction of mercury-resistant bacteria was >15 times higher than at the other snow depths (0–1.7%) (Table 2).

Table 2.   Bacterial densities, cultivability and proportion of mercury-resistant isolates
SampleTotal counts (cells mL−1)*Direct plating (CFU mL−1)*Cultivability (%)Hg resistance (%)
  • *

    SEs (n=3–5).

  • Percentage mercury resistance calculated on the basis of numbers of CFUs from the direct plating.

Site 1
 Brine5.0 × 105± 8.3 × 1041.3 × 102± 4.0 × 1010.031.7
 Snow (31–52 cm)3.1 × 103± 1.5 × 1032.0 × 100± 2.0 × 1000.10
 Snow (75–90 cm)1.4 × 103± 1.1 × 1031.7 × 102± 1.2 × 10211.91.7
 Snow (96–112 cm)8.5 × 102± 9.0 × 1016.8 × 101± 3.0 × 1018.031.2
Freshwater9.4 × 105± 3.8 × 1042.5 × 103± 7.1 × 1010.31.6

Diversity of isolated mercury-resistant isolates

A total of 1100 bacteria were isolated. Among these, 71 were mercury resistant as indicated by growth on 10% TSA plates amended with 10 μM HgCl2. To ensure true mercury resistance, isolates were restreaked on TSA plates supplemented with 10 μM HgCl2 at least three times. Isolates were classified as mercury resistant only if growth on the plates was similar to growth on control plates without added mercury. Most of the mercury-resistant isolates originated from freshwater (45 isolates) and the deepest snow layer (23 isolates), while only one isolate originated from brine and one from each of the upper snow layers.

Among the mercury-resistant bacterial isolates, 14 different partial 16S rRNA gene sequences were identified. These were distributed within seven different phyla/classes: Alpha-, Beta- and Gammaproteobacteria, Actinobacteria, Sphingobacteria, Flavobacteria and Firmicutes. Within each phylum/class, several subgroups were observed with identical (99% sequence similarity) partial 16S rRNA gene sequences. A phylogenetic tree with representatives from each of these subgroups and their closest relatives is shown in Fig. 2. It should be noted that the closest relative for nine of the isolates was either a psychrophile or had been isolated from a cold environment.

Figure 2.

 Neighbor-joining tree showing the phylogeny of representative 16S rRNA gene sequences of the mercury-resistant isolates. The tree has arbitrarily been rooted by the archaeon Methanocaldococcus jannashii. Values to the left of the nodes reflect the bootstrap values (% out of 1000 replicates). Numbers in parentheses indicate the number of isolates within each subgroup having 99% sequence similarity. Closest relatives were identified as blastn hits with the lowest E-values. *Closest relatives from cold environments. The tree is drawn to scale.

In addition to bacterial isolates, 150 mercury-resistant fungal colonies were obtained from snow (depth 75–90 cm) after preincubation on polycarbonate membranes. The colony color (pink) and morphology of all these isolates were identical. Sequencing of the ITS1–5.8S rRNA gene–ITS2 region of two representative isolates showed that they were identical and that the closest relative was an Antarctic yeast belonging to the order Leucosporidiales. Another isolate from snow (96–112 cm) with white colony color was found to be most closely related to a psychrophilic species belonging to the genus Geomyces.

The diversity of the isolated mercury-resistant bacteria from freshwater and snow was generally similar. Both habitats were dominated by Gammaproteobacteria, which constituted 56% and 42% of the isolates in freshwater and snow, respectively. There were, however, differences between the two habitats as Flavobacteria and Firmicutes were found only in freshwater while Actinobacteria were found only in snow (31%). In brine, only one mercury-resistant bacterial isolate, belonging to the Firmicutes, was identified.

All 71 mercury-resistant isolates were psychrotrophs as they grew at temperatures ranging from 4 to 20–25 °C (room temperature). The maximal temperature for growth of the three isolates representing Flavobacterium group II was 20 °C. Hence, they could be considered true psychrophiles, defined as bacteria with a maximum growth temperature <20 °C (Gounot, 1986).

Mercury resistance of the bacterial isolates

MICs generally ranged from 5 to 50 μM (Table 3). One isolate belonging to Gammaproteobacteria subgroup I, however, showed an MIC value of 100 μM. Although some isolates belonging to the Beta- and Gammaproteobacteria, and Flavobacteria, showed higher MICs at higher temperatures, the resistance of most isolates did not show temperature dependence (Table 3). Isolates SOK79 and SOK70s, belonging to Flavobacteria and Sphingobacteria, showed highest MIC values at middle temperatures (Table 3). This could be explained by impaired growth at 20 °C, which would also affect mercury resistance.

Table 3.   Mercury inhibition (MIC) of bacterial isolates and their mercury-volatilizing ability
IsolateMIC (μM)Hg
  • *

    Volatilization assay: +, positive; −, negative.

  • Nomenclature of the isolates indicate their origin: SOK, freshwater; 8D, snow depth 96–112 cm; 30D, snow depth 75–90 cm; 68F, snow depth 31–52 cm; BD, brine.

Alpha ISOK55050
SOK19, SOK19y
Alpha II30D125555
Beta ISOK35505050
SOK17a,SOK17b, SOK18t, SOK48, SOK51, SOK57
Beta IISOK15121225
Beta III8D36121225+
Gamma ISOK32505050+
SOK33, SOK43, SOK52+
SOK61, SOK67525+
8D32, 8D4555
8D48b, 8D48s
Gamma IISOK19w+
SOK755050 50
SOK41, SOK44, SOK50, SOK54, SOK59, SOK65, SOK68, SOK73, SOK84, SOK85, SOK89
SOK84s, 8D26
8D21, 8D41
8D55s, 8D55t55+
Gamma IIISOK13505050
Flavo ISOK18b51212+
Flavo IISOK79122512+
Actino I8D5s252525
8D5b, 8D7, 8D8, 8D10, 8D11, 8D12, 8D13
Actino II68F56555
FirmicutesSOK1b, BD41505050

MIC was determined for several isolates of Gammaproteobacteria subgroups I and II and Flavobacteria subgroup II. Values varied considerably (from 5 to 100 μM, 5–50 μM and 12–50 μM, respectively) even though these isolates were 99% identical with respect to their 16S rRNA gene sequence.

Volatilization of mercury

All the Flavobacteria, several of the Gammaproteobacteria, and single isolates of the Alpha- and Betaproteobacteria reduced Hg(II) to Hg(0) (Table 3 and Fig. 3). Within the individual subgroups, there were discrepancies as not all isolates within the groups were positive with the exception of Flavobacteria (Table 3).

Figure 3.

 Mercury volatilization assay. Microtiter plates with cell suspensions in the presence of Hg(II) were covered with an X-ray film. If the bacteria in the suspension were able to reduce Hg(II) to Hg(0), Hg(0) would diffuse into the headspace of the well and react with the silver particles in the X-ray film. Hence, darkening of the X-ray film is an indication of the formation of Hg(0). 1, Negative control (mercury-sensitive bacterium). 2, Positive control (SOK32; Gammaproteobacteria group I). 3, SOK33 (freshwater isolate, Gammaproteobacteria group I). 4, SOK79 (freshwater isolate, Flavobacteria group II). 5, SOK5 (freshwater isolate, Alphaproteobacteria group I).


We isolated and identified 71 mercury-resistant microorganisms representing seven bacterial phyla/classes from the High Arctic during AMDEs. The isolates were highly resistant to mercury and some were demonstrated to be able to reduce Hg(II) to Hg(0).

Microorganisms were isolated by two different methods: direct plating and preincubation on filters before plating. Direct plating often selects for only a very small fraction of the community, while preincubation under conditions that simulate the environment has been shown to increase the cultivability of soil bacteria up to 2800 times (Rasmussen et al., 2008). The majority of the mercury-resistant isolates from snow originated from direct plating (24 isolates) while only one isolate was obtained using the preincubation approach. However, 44 out of 45 of our isolates from freshwater were obtained by the preincubation method. Thus, preincubation on filters before plating enabled us to isolate more resistant bacteria than would otherwise have been possible by direct plating on rich medium.

The bacterial density in snow was on the average 1.8 × 103 cells mL−1 melt water. This is comparable to concentrations in snow covers of glaciers in the Tibetan plateau (Liu et al., 2009) but about one order of magnitude lower than concentrations found in the snow cover at Spitsbergen, Svalbard (Norway) (Amato et al., 2007), and at alpine sites (Sattler et al., 2001; Bauer et al., 2002). The observed bacterial density in brine (5 × 105 cell mL−1) was comparable to densities in sea ice in the Arctic and Antarctic (Brinkmeyer et al., 2003).

The mercury-resistant isolates belonged to phyla and classes commonly found in cold environments. Alpha-, and Gammaproteobacteria, Actinobacteria and isolates belonging to Bacteriodetes are common in sea ice both in the Arctic and Antarctica (Brinkmeyer et al., 2003), and Betaproteobacteria have been shown to be present in snow covers of an alpine lake (Alfreider et al., 1996). Amato et al. (2007) isolated Alpha-, Beta- and Gammaproteobacteria, Firmicutes and Actinobacteria in snow covers at Spitzbergen, and Larose et al. (2010) identified DNA belonging to Alpha- and Betaproteobacteria, Sphingobacteria, Flavobacteria and Acidobacteria from the same area. We did not, however, identify any mercury-resistant isolates belonging to the Acidobacteria.

MICs of mercury ranged from 5 to 100 μM and were comparable to values of mercury-resistant isolates of temperate soils (de Lipthay et al., 2008). MIC values varied within some of the taxonomic groups, suggesting that the level of mercury resistance was isolate specific. The best described mechanism of mercury resistance in bacteria is the reduction of Hg(II) to Hg(0) by the mercuric reductase, encoded by the merA gene (reviewed in Barkay et al., 2003). This system is commonly found in a wide range of bacteria often encoded by plasmids or associated with transposable elements (Olson et al., 1979; Barkay et al., 2003). It is therefore likely that the isolate-specific mercury resistance patterns of our Arctic isolates may be the result of horizontal transfer of a mercury resistance gene. This is further supported by the fact that no correlation between the ability to reduce Hg(II) to Hg(0) and the taxonomic groups was observed. It should be noted, however, that the X-ray film darkening method is a relatively rough method for detecting mercury reduction. Negative results should, consequently, be interpreted with caution.

Total mercury concentrations were measured at different snow depths, in brine and in freshwater. Concentrations were within previously reported values (60–600 ng L−1) during AMDE (Lu et al., 2001; Lindberg et al., 2002; Steffen et al., 2008). To investigate whether the concentrations in snow and brine were high enough to maintain a selective pressure for mercury resistance, we compared our data with other studies reporting both mercury concentration and percent mercury-resistant bacteria in a wide variety of environments (Timoney et al., 1978; Barkay, 1987; Ranjard et al., 2000; Rasmussen & Sørensen, 2001; Mindlin et al., 2005; De Souza et al., 2006; Ball et al., 2007; de Lipthay et al., 2008; Rasmussen et al., 2008). A correlation between total or bioavailable mercury with percent mercury resistance was not apparent for the compiled data set (not shown). For example, in a coastal marine sample, the total mercury concentration was 1.7 × 10−4 μg mL−1 and 23% of the cultured bacteria were mercury resistant (de Lipthay et al., 2008), whereas only 2.9% of the bacterial isolates were resistant in a soil with 7.6 μg Hg g−1 or 8 × 10−4 μg bioavailable Hg g−1 (Barkay, 1987). Thus, the mercury concentration does not seem to be a sensitive predictor of the population of resistant bacteria. The levels of total mercury in our study were relatively low (70–80 ng L−1) when compared with other environments, but nevertheless up to 31% of our isolates were mercury resistant. Others have reported mercury-resistant bacteria from cold environments. For instance, Petrova et al. (2002) found up to 2.9% mercury-resistant bacteria in permafrost sediments, and frequencies of 1.5–4.7% (Miller et al., 2009) and 68% (De Souza et al., 2006) were observed in Antarctic seawater. Although mercury concentrations in polar environments are relatively low, mercury is likely to be highly bioavailable (Lindberg et al., 2002), thus selecting and maintaining mercury resistance in the microbial communities.

Hg(0) measured in snow fluctuates during AMDE (Dommergue et al., 2003; Kirk et al., 2006) largely due to volatilization when Hg(II) is reduced to Hg(0). Emissions from snow take place simultaneously with increasing solar radiation, and reduction rates decline drastically with snow depth and when snow samples are placed in the dark. Thus, the emission of Hg(0) from light-exposed snow is considered to be largely driven by photoreduction (Lalonde et al., 2002; Dommergue et al., 2003; Kirk et al., 2006).

Biological mercury reduction in situ is difficult to measure as Hg(II) may be reduced by chemical redox processes under dark conditions (Ferrari et al., 2004). Hence, measurements of formation of Hg(0) in the dark would not be indicative of biological reduction. To assess the potential for merA-mediated reduction in snow, we made an estimate based on our own data and on literature values. The average total mercury concentration in snow was 3.6 × 10−4 μM, and assuming that 15% was bioavailable (Lindberg et al., 2002), the available mercury concentration can be estimated to have been 5.4 × 10−5 μM. MerA-mediated reduction is known to follow Michaelis–Menten kinetics (Philippidis et al., 1991), i.e. the reduction rate can be calculated as V=(Vmax× [Hg])/([Hg] ×Km). Philippidis et al. (1991) determined Vmax and Km of the mercuric reductase to be 8.2 nmol min−1 mg−1 protein and 3.8 μM, respectively. Thus, assuming a per-cell protein content of 2.4 × 10−11 mg per cell (Zubkov et al., 1999), the per-cell reduction rate can be calculated (1.7 × 10−13 nmol Hg h−1 per cell). Numbers of resistant culturable bacteria in snow varied between 1.7% and 31.2% depending on the depth (Table 2). If the culturable bacteria were representative for the total bacterial communities, numbers of resistant bacteria can be calculated to have been 6.4 × 106 cells m−3 (75–90 cm) and 1.2 × 108 cells m−3 (96–112 cm) using a snow-to-melt water correction factor of 0.28. Based on these numbers and the per-cell reduction rate, bacterial reduction rates per cubic meter snow were calculated (Table 4).

Table 4.   Estimation of bacterial mercury reduction rates and their contribution to observed total reduction rates in snow
Station NordKuujjuarapik, Canada*Bacterial/total
Snow depth (cm)Bacterial reduction rate (nmol h−1 m−3)Snow
depth (cm)
Total reduction rate
(nmol h−1 m−3)
100% of
cells active
15% of
cells active
31–5200541.0 × 10−30
75–901.1 × 10−61.6 × 10−7802.6 × 10−40.1–0.4
96–1122.0 × 10−53.0 × 10−61021.7 × 10−41.8–11.8
Average1.1 × 10−51.6 × 10−6Average4.9 × 10−40.3–2.2

Estimated bacterial reduction rates increased with snow depth (Table 4). In the uppermost layer, no resistant cells were observed, and hence, no reduction could be estimated. However, from ∼83 to ∼105 cm depth, reduction appeared to increase by a factor of almost 20. Very little information is available in the literature on reduction rates of Hg(II) in snow. Dommergue et al. (2003) measured total mercury reduction in interstitial air at different depths in the snow pack in Kuujjuarapik, QC, Canada. They found that daytime release rates of Hg(0) decreased with increasing depth and suggested photoreduction to be the primary mechanism for the mercury reduction. A comparison with our data indicates that an average of up to 2% of the total reduction may potentially be bacterial and that bacterial volatilization of mercury becomes relatively more important in the deeper snow layers (Table 4). This agrees with a study of coastal waters in the Canadian High Arctic (Poulain et al., 2007), in which the relative contribution of bacterial reduction appeared to increase with depth, accounting for up to 94% of the total production of Hg(0) at the greater depths. Ferrari et al. (2004) estimated the night-time production of elemental gaseous mercury in snow at Station Nord and found fluxes that were 6–25 times lower than those reported by Dommergue et al. (2003). A comparison with our estimated potential bacterial reduction rates implies that most of the night-time production of Hg(0) may be attributed to bacterial activity.

In summary, we isolated a diverse group of mercury-resistant bacteria from snow, freshwater and sea-ice brine from the High Arctic during AMDE. The isolates showed a high level of resistance to mercury and some were shown to be able to reduce Hg(II) to Hg(0). Although photoreduction probably is the major mechanism of mercury re-emission to the atmosphere, mercury-resistant bacteria have the potential to contribute to the reduction of mercury in snow, especially in deeper snow layers where light attenuation limits photoreduction. Bacteria may impact the production of the potent neurotoxic MeHg directly by methylation and demethylation processes, and indirectly by controlling the supply of Hg(II), the substrate for methylation, by reducing Hg(II) to Hg(0). Thus, by reducing Hg(II) to Hg(0), the mercury-resistant bacteria in Arctic environments may limit the supply of substrate for the methylation processes, and hence, reduce the risk of MeHg being incorporated into the food chains.


We thank the staff at Station Nord for assisting with the logistics of the sampling and Dr Aurélien Dommergue for providing data to calculate the mercury reduction rates for Kuujjuarapik. This work has been funded by the Danish Agency of Science (J. no. 645-06-0233), the U.S. National Science Foundation (EAR-0433793) and the U.S. Department of Energy (DE-FG02-05ER63969). This work was partly financially supported by the Danish Environmental Protection Agency with means for the MIKA/DANCEA funds for Environmental Support to the Arctic Region. The findings and conclusions presented here do not necessarily reflect the views of the Agency. The Royal Danish Airforce is gratefully acknowledged for providing free transport to Station Nord.