Frost flowers (FF) have been studied for their potential influence on ice-surface reflectivity and roles in atmospheric chemistry, but not as microbial habitats. We examined FF grown in a freezer laboratory from a bacteria-containing saline solution and FF formed naturally in the coastal (April) and central Arctic Ocean (September). All FF contained bacteria (up to 3.46 × 106 ml−1 in natural FF) with densities 3–6-fold higher than in underlying ice. Bacterial abundance correlated strongly with salinity in FF (p values ≤ 0.001), a correlation that held for all components of the surface-ice environment (p < 0.0001, coastal samples). Concentrations of extracellular polysaccharides were also elevated in FF and brine skim relative to underlying ice (up to 74-fold higher). Here we consider implications of finding microbes and exopolymers within the chemically reactive surface-ice environment to the photolytic production of oxidants and long-range transport of potential ice-nucleating particles in the atmosphere.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
 Frost flowers (FF), centimeter-scale structures composed of atmospheric hoar and liquid derived from sea ice brines, grow abundantly on newly formed sea ice [Perovich and Richter-Menge, 1994; Alvarez-Aviles et al., 2008]. They have been the focus of numerous recent investigations due to their ability to transport salts (and possibly other materials) concentrated in sea ice brines to the atmosphere. This transport mechanism, combined with the high specific surface area of FF (but see Domine et al. ), may play a significant role in tropospheric ozone depletion events common during spring polar sunrise [Rankin et al., 2002; Alvarez-Aviles et al., 2008]. Elevated mercury concentrations have also been measured in FF, exceeding that of coastal snow 9-fold [Douglas et al., 2005].
 FF are ephemeral, existing exposed on the ice surface for hours to days. Their fate is burial by snow or erosion by wind [Perovich and Richter-Menge, 1994]. Aerosol material can be sourced to FF by a higher ratio of Na to SO4 than found in (unfrozen) seawater. This signature is the result of mirabilite (Na2SO4·10 H2O) precipitation at −8°C, near the upper temperature limit of FF growth. Using this tracer, FF have been identified as a significant source of salt to glacial ice [Rankin et al., 2002].
 Despite their location at the frozen interface between ocean and atmosphere and their demonstrated role in concentrating and transporting sea salts, the biology of FF has not been investigated. We reasoned that FF, in wicking brine from the underlying ice or surface brine skim (a thin film of brine expelled upwards from the ice matrix), must also bring sea ice bacteria (Bacteria and Archaea [Collins et al., 2010]) into their structures, given that bacteria concentrate in the liquid brine phase [Junge et al., 2001]. A growing body of work on the microbiology of winter sea ice suggests that bacteria can remain active in carbon cycling [Junge et al., 2004; Wells and Deming, 2006] to temperatures as low as −26°C [Collins et al., 2008] in part due to the presence of cryoprotectant exopolymers (EPS), organic exudates produced by bacteria and phytoplankton [Krembs et al., 2002; Collins et al., 2008; Marx et al., 2009]. Experiments with microorganisms from other frozen environments suggest the persistence of metabolism to −35°C [Panikov and Sizova, 2007], leaving open the possibility for microbial activity in very cold FF. In the atmosphere, when temperatures are low enough to suppress metabolic activity, bacteria from a variety of environments can still influence physical processes by functioning as ice nucleators [Jayaweera and Flanagan, 1982; Christner et al., 2008].
 Motivated by preliminary observations of FF during the overwintering CFL-IPY 2007–2008 expedition [Deming, 2010], we established a system for growing FF in the laboratory and collected natural FF from coastal and central regions of the Arctic Ocean, analyzing for bacterial content and salinity. By collecting other features of the ice-surface environment, we sought to broaden the emerging relationship between bacterial and salt content in FF. At our coastal site we also analyzed for EPS, anticipating that biogenic materials other than intact microbial cells are wicked into FF. Finding elevated concentrations of bacteria and exopolymers in FF and brine skim has important implications for the transport of biogenic material from sea ice to atmosphere and for possible chemical reactivity.
 FF were grown in a laboratory freezer room at −21°C in a manner similar to that described by Style . A NaCl solution of 35 ppt, amended with bacteria (to 2 × 106 ml−1), was used instead of seawater. To ensure passive transport of the bacteria, a mesophilic bacterium (Halomonas pacifica) incapable of growth or motility under the test conditions was used. Individual FF and brine skim were harvested from the ice surface using an ethanol-sterilized spatula. Ice was sampled by removing a small section (the top 1–2 cm) with an ethanol-sterilized chisel. The samples were monitored while melting at room temperature (1–2 min); as the last ice crystals melted, subsamples were immediately fixed to 2% v/v with paraformaldehyde.
 Salinity was determined on unfixed aliquots of the melted samples using a handheld refractometer. Although an apparent departure from the convention for seawater, reporting in these units is made necessary by use of a refractometer (calibrated against ppt) and by sample salinities above the linear range for which the practical salinity scale is defined [Williams and Sherwood, 1994]. Bacteria were enumerated by epifluorescence microscopy using the DNA-specific stain, 4′,6-diamidino-2-phenylindole, counterstained with acridine orange. A minimum of 20 fields and 200 cells were counted for each sample. FF age was determined by time-lapse photography using a Nikon D70s camera controlled by a laptop running the Nikon Capture software outside the cold room. Photographs were taken at 1-min intervals throughout a typical experiment of 24–48 h. A bacterial enrichment index (IB) [Gradinger and Ikävalko, 1998] was calculated to determine the transport efficiency of bacteria into ice and FF (or back into solution) relative to salt. Significance of the mean deviation from IB = 1 (indicating equal incorporation of salt and bacteria) was determined using a Student's T-test.
 Individual FF, brine skim, sea ice, and other components of the surface ice environment were collected from shore-fast sea ice near Barrow, Alaska, on April 6 and 7, 2009. Respective daily air temperatures during sampling were −23°C and −32°C. Tidal action replenished surface ponds over this ice surface, allowing growth of new FF each night on a saline surface ice layer. The FF and all other surface ice samples were collected using an ethanol-sterilized spatula. Brine skim was sampled beneath each FF, followed by the upper 1–2 cm of underlying sea ice. Other samples included older FF and brine skim, buried by 5 cm snow (estimated age of 1 week), FF from atop a pressure ridge, and brine (collected in a sterile bottle) from a naturally occurring fissure in sea ice. Samples for bacterial counts were treated as for the laboratory grown FF, except that formaldehyde was used to fix the cells. Salinity was determined in the same manner as for laboratory grown FF. Samples for particulate EPS (pEPS) were transported frozen to the University of Washington in Seattle, where they were melted as before and filtered through a 0.4 μm polycarbonate filter. The concentration of pEPS was determined using the phenol-sulfuric acid (PSA) assay previously described [Marx et al., 2009].
 Individual FF were also collected from lead ice in the central Arctic Ocean during the LOMROG II expedition of the Swedish icebreaker Oden on September 2, 2009, at 84.84°N latitude and 14.67°E longitude. The air temperature was −8°C. Samples were melted (1:1 v/v) in 0.2 μm-filtered artificial seawater comprised of the major sea salts (35 ppt) and containing 4% formaldehyde (post-melting concentration of 2%) to minimize cell loss to osmotic shock during melting. A portion of each FF was melted separately for determining salinity by refractometer. Bacterial counts were conducted as for laboratory-grown FF and corrected for the artificial seawater dilution.
 In laboratory experiments, the abundance of H. pacifica cells in the saline ice that formed ranged from 7.64 × 105 to 1.13 × 106 ml−1, while in FF growing on top of the ice it was higher, ranging from 1.15 × 106 to 5.56 × 106 ml−1 (Figure 1a). In field-collected FF (Figures 1b and 1c), the natural bacterial abundance was also elevated relative to underlying sea ice, ranging from 1.28 × 105 ml−1 (Barrow) to 3.46 × 106 ml−1 (central Arctic Ocean) in FF compared to 0.28–3.83 × 105 ml−1 in underlying sea ice (Barrow (Figure 1b)). In all cases, bacterial abundance in FF correlated strongly with salinity (Figures 1a–1c). The correlation remained strong when other components of the surface sea ice environment were included (Figure 1b). For cases where comparable data were obtained, pEPS concentration though variable was always higher in FF and brine skim, reaching maxima of 725 and 1,420 μg glucose equivalents (gluceq) ml−1, respectively, than in the underlying sea ice with its maximum of 36.5 μg gluceq ml−1 (Figure 2). The concentration of pEPS did not correlate significantly with either salinity or bacterial abundance.
 Although prior work has shown that FF become increasingly saline as they age [Perovich and Richter-Menge, 1994; Martin et al., 1995], we did not find significant relationships between FF age and salinity (df = 12, r = 0.10, p = 0.73) or bacterial abundance (df = 12, r = 0.37, p = 0.19) in our relatively short laboratory experiments. Correlations between FF age, salinity and bacterial abundance appeared likely in our field data (Figure 2, sample set B), but age-based sampling was insufficient to test for these relationships. Calculated IB for laboratory-grown FF showed variability in the efficiency of bacterial transport into FF relative to salt transport, but departure of the mean of these values from IB = 1.0 was not statistically significant (mean = 0.99, df = 22, t = 0.306, p = 0.76), suggesting that bacteria were not preferentially enriched over salt in FF or brine skim.
 Finding elevated bacterial abundance and high concentrations of pEPS in saline FF identifies a new icy habitat to explore biologically, with potentially important ramifications for atmospheric chemistry. For example, bacterial cells and pEPS may serve as substrate for the photolytic production of oxidants and simple organic compounds, including formaldehyde and hydrogen peroxide. The production of these compounds in snow has been well studied due to the role they play in oxidation reactions within the troposphere [Jacobi et al., 2002]. Concentrations of both formaldehyde [Barret et al., 2009] and hydrogen peroxide [Beine et al., 2009] are elevated within FF. The mean value for pEPS in FF at Barrow (133 μg gluceq ml−1) represents 4.44 × 10−6 mol C ml−1, sufficient carbon to support the production (assuming complete conversion) of 133 μg ml−1 of HCHO (typically measured in the pg ml−1 range in snow and marine waters). Other types of organic compounds beside pEPS can also be expected within saline FF, given high levels of DOC [Thomas et al., 1995] and the potential for viral lysis of bacterial cells [Wells and Deming, 2006] in winter sea ice brines. Though not yet pursued, our preliminary observations of winter FF have revealed the presence of viruses [Deming, 2010].
 Microbial activity was not measured in this study, but recent work has demonstrated aerobic metabolism within the top centimeter of early spring sea ice (similar to our Barrow ice) correlated to the presence of pEPS [Meiners et al., 2008]. Whether this relationship holds for FF remains to be determined, but the presence of both pEPS and bacteria in elevated concentrations in FF is promising. Depending on type of activity, measurements of microbial activity in FF under severe winter conditions could define new (lower) temperature limits for the activity.
 Considering the ultimate fate of bacteria within FF raises other important issues. If buried and insulated by snow, the microbial community originating from FF may be more dynamic than when exposed to the atmosphere, undergoing microbial succession (species-specific growth and mortality) within a saline snow layer on the surface of the ice. This succession may resemble that understood to occur within the ice [Collins et al., 2010; Deming, 2010] and even influence underlying sea ice biology and chemistry via infiltration during warming [Brinkmeyer et al., 2004].
 A different fate awaits microbes contained in FF that erode from the ice surface by wind. Biological materials in the atmosphere, collectively known as biogenic aerosols, have been investigated increasingly for their roles in atmospheric chemistry and cloud properties. Biogenic aerosols over the central Arctic Ocean contain bacteria and EPS (and associated viruses), considered to be cloud condensation nuclei [Leck and Bigg, 2005]. Ice-nucleating bacteria have been ascribed an important role in some snowfall events [Christner et al., 2008]. Although the impact of biological ice nucleation increases with temperature (decreasing the importance of ice nucleators derived from FF during Arctic and Antarctic winters), the upper temperature for FF growth (about −8°C) is warm enough to allow a role during early fall and late spring. Aerosol collections by aircraft over the Arctic Ocean suggest that biogenic aerosols can be transported substantial distances [Jayaweera and Flanagan, 1982]. Wind erosion of FF filled with microbes and EPS may represent a unique dispersal mechanism for the organisms and their byproducts. If FF bacteria can function as ice nuclei, their long range dispersal becomes a means of delivering ice nucleators to lower latitudes, where warmer temperatures increase the significance of these particles to precipitation events.
 A more detailed analysis of the possible fate of bacteria in wind-eroded FF can be made where ion concentrations have been analyzed within glacial ice. In Antarctic glacial ice as in Arctic snow, saline FF have been recognized as a significant source of sea salt through their characteristic signature of SO4 depletion relative to Na [Wolff et al., 2003; Beaudon and Moore, 2009]. Using an average value for FF salinity, the abundance of Na within the Dome C ice core gives an approximation of the flux of FF material to the interior of the Antarctic continent. With our measured bacterial abundances in FF and assuming bacterial transport coincident with salt, the flux of FF bacteria into glacial ice can be estimated. We take 100 ppt as a typical salinity for saline FF (mean for Barrow FF = 99.2, n = 12, SD = 25.1), which agrees well with previous studies [Perovich and Richter-Menge, 1994; Martin et al., 1995, 1996], and 5 × 105 bacteria ml−1 as a typical value for bacterial abundance (mean for Barrow FF = 4.8 × 105 ml−1, n = 12, SD = 2.8 × 105). Na flux varies with FF production, approaching 500 μg m−2 yr−1 in recent times [Wolff et al., 2003]. In a FF entirely depleted in SO4, Na accounts for 27% of the mass of all ions. Thus a Na flux of 500 μg m−2 yr−1 is equivalent to 1,850 μg m−2 yr−1 total ions. For FF of 100 this represents a mass of 2.0 × 10−5 kg, or 0.018 ml. This volume of FF would contain 8,900 bacteria for a flux on the order of 8,900 FF bacteria m−2 yr−1 to Dome C.
 Modern molecular techniques may be able to detect this flux of bacteria from the marine environment to the continental interior. Although the bulk salinity of glacial ice is substantially lower than that of sea ice (and the interior brine volume substantially smaller), eutectic freezing does allow the presence of liquid films in glacial ice. Furthermore, the water activity (Aw) of these veins will reflect the Aw of brine inclusions in sea ice of the same temperature (though the concentration of solutes and pH will differ). This similarity suggests that saline FF bacteria may be able to resist lysis due to osmotic stress within glacial ice. Total bacterial abundance in bulk glacial ice is on the order of 102–103 bacteria ml−1 [Mader et al., 2006], the majority of which are assumed to be delivered with aeolian dust in snow. Given an average snow accumulation rate of 3.6 g cm−2 yr−1 at Dome C [Petit et al., 1982] and our estimate of FF bacterial flux to the site, bacteria from FF would account for one cell in every 5 ml of bulk ice. If, as in sea ice, bacteria partition into the liquid phase of glacial ice, a concentration factor between 104 and 105 would pertain [Mader et al., 2006] resulting in between 2 × 103 and 2 × 104 FF bacteria ml−1 of glacial brine.
 FF represent a critical link between materials contained within sea ice brines and the atmosphere. Although previous studies have investigated the importance of interactions between inorganic and presumably abiogenic materials concentrated within FF and the atmosphere, the biology of FF has been overlooked until now. We have demonstrated that bacteria and exopolymers are concentrated in saline FF relative to other components of the surface sea ice environment. This concentration within FF may allow bacteria to be involved directly in a number of atmospheric processes, including photochemistry and long-range transport by wind. A biogeochemical role for FF bacteria will depend in part on the ability of these organisms to metabolize at very low temperatures and low Aw, providing incentive to characterize the microbial community and assess metabolic activity within FF.
 We thank Hans-Werner Jacobi for stimulating input and the invitation to sample FF at Barrow, Alaska, Soeren Rysgaard and Christian Marcussen for the invitation to join LOMROGII, Matthew Barret and Matthias Wietz for assistance in the field, Shelly Carpenter for laboratory support, and Seelye Martin, Eric Collins, and Marcela Ewert for helpful discussion. This research was supported by NSF-OPP award 0908724 to JWD, NSF-IGERT support to JSB, and the UW Astrobiology Program.