Nitrous oxide (N2O) contributes ∼6% of the total radiative forcing from long-lived greenhouse gases. While tropospheric concentrations have increased by 20% since the beginning of the industrial revolution, sources and sinks of N2O are still poorly quantified. In the Arctic, N2O atmospheric concentrations vary seasonally, due mainly to vertical mixing. The contributions of local natural sources to this cycle are still unknown. Here we report on N2O measurements conducted in the bottom 10 cm of the sea ice and in the underlying surface water (USW) from late March to early May 2008 in the southeastern Beaufort Sea and Amundsen Gulf. Bulk N2O concentrations in ice were low (∼6 nM) and were consistently undersaturated with respect to the USW (∼40% saturation) and the atmosphere (∼30% saturation). Loss of N2O via brine rejection during sea ice formation in fall and winter can explain these low N2O ice concentrations. An unknown fraction of this rejected N2O is likely ventilated to the atmosphere either directly from the ice or through leads during ice formation, while in spring and early summer, melting of the N2O-depleted sea ice is expected to lower the partial pressure of N2O of newly open waters which could act as a sink for atmospheric N2O. These first measurements indicate that sea ice formation and melt has the potential to generate sea-air or air-sea fluxes of N2O, respectively.
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.
 With an average ice extent of 15.2 × 106 km2 in winter [Comiso et al., 2008], the Arctic sea ice represents one of the largest and most dynamic biomes on the planet. It is recognized that microbial activity occurs in Arctic sea ice [Legendre et al., 1992], with profound effects on brine chemistry, including its gaseous components [Gleitz et al., 1995]. For decades, sea ice has been assumed to be impermeable to gas and was consequently considered an inert barrier for ocean-atmosphere-biosphere interactions. However, there is a growing body of evidence that exchange of gaseous compounds occurs between seawater underneath the sea ice and the air above [Loose et al., 2009]. Sea ice could also act as an important sink or source for some atmospheric gases during its growth and decay [Delille et al., 2007; Rysgaard et al., 2009]. During fall and winter, gas rejected with brine generated by sea ice formation could be transported to the deep ocean or ventilated to the atmosphere, whereas during sea ice decay in spring, undersaturated sea ice meltwater could act as a sink for gas in the atmosphere [Anderson et al., 2004; Kitidis et al., 2010].
 Nitrous oxide (N2O) is a greenhouse gas with a relative radiative forcing nearly 300 times that of CO2 on a per molecule basis [Forster et al., 2007]. N2O is also implicated in stratospheric ozone depletion and is currently considered to be the dominant ozone-depleting substance emitted in the 21st century [Ravishankara et al., 2009]. Studies on ammonium oxidation and anaerobic bacterial cultures have demonstrated that production of N2O can occur in sea ice [Priscu et al., 1990; Kaartokallio, 2001]. Rysgaard and Glud  reported anoxic N2 production in O2-poor microenvironments in Arctic sea ice through denitrification, a metabolic pathway implicated in both N2O production and consumption in the ocean. In view of the ongoing rapid changes in the Arctic icescape and an anticipated increase in first-year sea ice production, it is important to determine if sea ice represents a source or a sink of N2O for the Arctic atmosphere.
 Here we report the first measurements of N2O concentrations in first-year Arctic sea ice. N2O concentrations in bulk sea ice, in brine, and in underlying surface water (USW) just underneath the sea ice cover in the Beaufort Sea are discussed in relation to physical, chemical and biological environmental conditions.
2. Materials and Methods
2.1. Study Site
 The sampling was conducted in the Amundsen Gulf and southeastern Beaufort Sea (Figure 1) during March–May 2008, aboard the research icebreaker CCGS Amundsen as part of the Circumpolar Flaw Lead (CFL) system study of the Canadian International Polar Year (IPY) program. Drifting sea ice was sampled, on average, every 3 days from 28 March (Day 88) to 5 May (Day 126) (Figure 1). Since sampling was conducted at several locations, our sampling protocol thus integrates both spatial and temporal variability. However, the spatial variability is partly mitigated by ice drift, since some ice floes were resampled several times as the ship drifted with the ice. Over 12 sampling dates between Day 88 and Day 126, five different ice floes were sampled. In the following list of sample day numbers, the parentheses indicate groups of samples from the same ice floes: (88, 91, 94), (97, 100), 102, (107, 110), (117, 120, 123, 126). Note that N2O was not sampled on Days 94 and 110, although all other variables were measured.
 On each sampling date, four ice cores were drilled approximately 500 m from the ship using an ice corer (MARK II manual coring system, 9 cm internal diameter, Kovacs Enterprises). All ice cores were collected in first-year ice (average thickness 132 ± 30 cm) under a thin snow cover (less than 5 cm). Samples for determination of salinity, temperature, nutrients, chlorophylla (chl a) and N2O were collected as soon as possible after core extraction according to the protocols described below. Technical considerations allowed us to sample only the bottom 10 cm of the ice core for N2O measurement, so all other ice core data will be discussed in relation to the bottom 10 cm only.
2.2. Sea Ice Temperature and Salinity
 One ice core was dedicated to in situ temperature and bulk salinity measurements. Immediately after extraction of the core, the temperature was measured in the center of the core at two places in the bottom 5 cm and the next 5 cm of the sea ice using a handheld drill and a temperature sensor coupled with a stainless steel NTC food probe (IP65, Testo). The sensor had a resolution of 0.001°C and an accuracy of ±0.05°C. The two measurements were averaged to obtain an estimate of ice temperature in the bottom 10 cm. The bottom of the ice core was then cut into two 5 cm sections and stored in plastic containers for melting. The salinity of the thawed ice sections was measured with a WTW 330i conductivity meter and the results averaged to obtain an estimate of bottom ice salinity.
 A second ice core was used to determine nutrient concentrations. The bottom 3 cm (B3) and next 7 cm (B7), together constituting the lower 10 cm of the ice core, were slowly melted for 24 h in the dark. Thawed ice cores were filtered through precombusted (450°C for 5 h) Whatman GF/F filters and the filtrates were analyzed aboard the ship for nitrate and nitrite concentrations with a Bran and Luebbe nutrient autoanalyzer 3 (methods adapted from Grasshoff et al. ). The average concentration in the bottom 10 cm of the sea ice was calculated using the following equation:
where C3 and C7 are the nutrient concentrations (μM) of the bottom 3 cm and next 7 cm of the ice core, respectively, and Vol3 and Vol7 are the respective volume of meltwater (l) of each ice section.
2.4. Chlorophyll a
 A third ice core was used for chl a determination. B3 and B7 of the ice core were slowly melted in 0.2 μm filtered surface seawater collected at the sampling site. Duplicate melted samples (25 ml) were filtered onto Whatman GF/F filters and pigments were extracted in 90% acetone for 18 h at 4°C in the dark [Parsons et al., 1984]. Concentrations of chl a were then measured aboard the ship with a Turner Designs 10 AU fluorometer and corrected for the dilution of each ice core section in the filtered seawater using the equation of Cota and Sullivan . Areal chl a concentrations (mg m−2) from both sections were pooled to get the total chl a concentration in the bottom 10 cm of the sea ice.
2.5. Bulk N2O Concentration
 A fourth ice core was used for N2O measurements. Upon collection, the bottom 10 cm of the ice core was cut and drilled in the center using an electric drill fitted with a bit specially designed for sea ice sampling [Song et al., 2011]. The bit is a custom-made saw, consisting of a steel cylinder with a toothed edge to penetrate sea ice, thus creating a cylindrical ice block which fits tightly inside a 200 ml all glass syringe (Figure 2). The piece of sea ice was immediately transferred into the glass syringe which had been preflushed with ambient air and fitted with a three-way nylon valve as described inXie and Gosselin . The plunger was then tightly inserted to expel as much ambient air as possible from the syringe. The syringe was transported in a cooler, protected from light and allowed to thaw in a bucket of tap water aboard the ship. Headspace created in the syringe as the ice melted was slowly expelled by pushing the plunger. Postsampling experiments showed that such headspace represents ∼9% of the volume of melted water. This headspace is largely comprised of atmospheric air in the cavity created between the glass syringe and the irregular surface of the sea ice sample. Since N2O would be exchanged between the headspace and the melted ice, if we assume equilibrium between the two phases in the syringe, the error (underestimate) in our measured N2O concentrations would be 6.6% under the average conditions of salinity and N2O in our samples, at a temperature of 0°C. Water from the syringe was then transferred slowly through a Teflon tube into replicate 20 ml serum bottles (triplicate until Day 102, and duplicate thereafter). Caution was used to avoid bubbles and air mixing. The serum bottles were then spiked with 200 μl of a saturated HgCl2 solution to inhibit any bacterial processes. Using a 20 mm hand crimper, the serum bottles were sealed with butyl rubber stoppers and aluminum caps immediately after adding the HgCl2. Samples were stored in the dark at ambient temperature for further analyses performed ashore at our home laboratory.
 At the onshore laboratory, N2O concentrations were determined using a gas chromatograph (Varian 3800) equipped with a Pulsed Flame Photometric Detector (PFPD) operating in nitrogen mode and a Varian CP-PoraBOND Q fused silica column 25 m × 0.32 mm (i.d.) in a modification of the method described inScarratt et al.  for dimethylsulfide analysis. Samples were sparged with helium at ∼60 ml min−1 for 3 min in a heated (70°C) purge vessel. The upper part (15 cm long) of the purge vessel was surrounded by a cold water jacket (5°C) followed by a Pyrex chamber containing small pellets of calcium chloride. These two steps removed water vapor from the gas stream without condensation of N2O. Thereafter, a Teflon loop submerged in liquid nitrogen cryotrapped the N2O. The Teflon loop was subsequently heated (∼70°C), releasing the extracted gas onto the GC column. The system was calibrated using a certified permeation tube source (KinTek Laboratories, model 57SA) delivering N2O at 312 nl min−1. The helium-diluted output from the permeation tube was delivered to the purge and trap apparatus by switching standard volume sample loops inline with the purge gas stream and processing in the same manner as the samples. An arithmetic mean with standard variation was used to obtain mean values from duplicates. The N2O detection limit of the system is estimated at 3 nM in a 25 mL sample, using the method of Corley . The lowest measured concentration (Day 100) was actually slightly below the system detection limit and has been arbitrarily set to 3 nM on the graphic. An example of a typical calibration curve is provided as auxiliary material.
2.6. Estimation of N2O Saturation and Concentration in Brine
 N2O saturation level with respect to the air above the sea surface was calculated assuming a 2008 atmospheric mixing ratio of 322 ppbv, consistent with data from Barrow, Alaska which are available online from the NOAA Earth System Research Laboratory (http://www.esrl.noaa.gov/). The atmospheric equilibrium concentration and bulk N2O saturations in the bottom sea ice and underlying seawater were calculated using the N2O solubility equation of Weiss and Price  corrected for in situ salinity and temperature. N2O saturations in brine were corrected using measured in situ temperature and derived brine salinity. It should be noted that the very low temperature and high salinity in brine require us to extrapolate the equation of Weiss and Price  beyond its range of −1°C–40°C and 0–40 salinity, which may introduce error into the calculated N2O solubility.
where Sb is the brine salinity, T is the sea ice temperature, Vb is the brine volume fraction, V is the bulk ice volume fraction, Va/V is the air volume fraction, ρi is the density for pure ice (g cm−1), Ssi is the bulk salinity, F1(T) and F2(T) are empirical polynomial functions Fi(T) = ai + bi T + ci × T2 + di × T3, based on the phase relations for different temperature intervals from Cox and Weeks .
 N2O concentrations in brine (N2Ob) were calculated using an adapted equation of the CO2 dissociated constants of Mehrbach et al. 
where N2Ot is the bulk N2O concentration (nM) and St is the bulk ice salinity.
2.7. Seawater Sampling
 At each ice sampling station, water just under the sea ice cover (USW) was collected for water temperature, salinity, nutrients (nitrate and nitrite) and N2O determination through ice holes using an electric submersible pump (12 V standard engineered plastic pump, model Cyclone, Proactive Environmental Products) connected to a Tygon tube. A 450 μm Nitex screen was installed at the inlet of the pump to exclude large zooplankton from the samples. Water temperature, salinity, nutrient and N2O concentrations were analyzed as described above. N2O concentrations were determined on triplicate samples.
 Our sampling covered 38 days of the biologically active vernal period. Since the ice samples were collected from a limited number of ice floes (some of which were resampled several times), we assume for convenience that a large part of the variation observed in the different variables reflects temporal changes and accordingly all data are presented against time, but we acknowledge that spatial variation also contributed to the measured signals. In Figures 3 and 4, groups of samples from the same ice floes are indicated using different symbol shapes.
 The sea ice thickness (Figure 3a) was generally between 121 and 142 cm, except for Days 97 and 100 when sampling was conducted on a thinner floe of between 57.5 and 67.5 cm. Snow depth on the surface of the ice (Figure 3a) was initially 2.5–3.0 cm and then increased to between 4.0 and 4.5 cm after Day 107. The thinnest snow cover (1 cm) was found on the thinnest ice. Salinity in the bottom 10 cm of the sea ice increased from 5.65 to 8.45 between Day 94 and Day 117 and then decreased abruptly to 5.55 on Day 120 (Figure 3b). The temperature of the bottom ice varied between −2.0 and −2.4°C during the sampling period (Figure 3b). Nitrate and nitrite concentrations varied from <0.05 to 3.2 μM and from 0.01 to 0.2 μM, respectively (Figure 3c), and increased gradually throughout the study. Chl a concentrations increased gradually from 2.7 mg m−2 on Day 88 to reach a maximum concentration of 21.7 mg m−2 on Day 110 and then decreased somewhat at the end of the sampling period (Figure 3d). Variations in chl a suggest that we captured the prebloom (Days 88–100), and early bloom periods (Days 102–126). Data from the same stations presented in Song et al.  indicate that chl a continued to increase after the end of our sampling period. Bulk N2O concentrations in the bottom 10 cm of sea ice vary from <3 to 7.9 ± 1.1 nM, and saturations with respect to the atmosphere ranged up to 39% (Figures 3e and 3f). The lowest N2O concentration (<3 nM) was measured in the thinnest ice floe (Day 100) and the highest (7.9 nM) during the period of rapid chl a growth, but otherwise the concentration was relatively stable near 6 nM during most of the study. There were no significant relationships (Pearson Product Moment Correlation) between N2O concentration and any of the other variables measured in the bulk sea ice (salinity, temperature, nitrate, nitrite and chl a).
 Brine salinity and brine volume fraction varied between 34.8 and 42.5 and between 13.2% and 20.6%, respectively (Figures 4a and 4b). Assuming that almost all the N2O measured in the bulk ice was concentrated in brine (see Materials and Methods), we calculated that N2O concentrations in brine varied between 23 and 40 nM (Figure 4c) and saturation with respect to the atmosphere ranged from 142 to 274% (Figure 4d).
3.3. Underlying Seawater
 Salinity in the USW fluctuated around 31.5, reaching maxima of 31.9 and 32.0 on Days 100 and 126, respectively (Figure 5a). Water temperature remained stable at −1.7°C throughout the study (Figure 5b). Nitrate concentrations were somewhat variable but reached a maximum of 4.8 μM on Day 126 (Figure 5c), while nitrite concentrations remained relatively stable near 0.1 μM for most of the study (Figure 5c), with peaks on Days 94 and 120. N2O concentrations in the USW were ∼11 nM between Day 97 and Day 107 (during the bottom ice prebloom and bloom periods), and increased slightly to reach 18.8 nM on Day 123 (Figure 5d). N2O saturation with respect to the atmosphere in the USW varied from 60 to 111%, considerably higher than the saturations measured in the bottom ice (Figure 5e). There were no significant relationships (Pearson Product Moment Correlation) between N2O concentration and any of the other variables measured in the USW (salinity, temperature, nitrate, nitrite and chl a).
4.1. The N2O Undersaturation State of the Sea Ice
 Quantifying gaseous compounds in sea ice remains a technical challenge. In this study, we opted for the glass syringe technique which was developed and successfully used for the determination of other gases such as CO [Xie and Gosselin, 2005; Song et al., 2011] and oxygen (O2) [Rysgaard and Glud, 2004] and in a modified form for laboratory experimentations on CO2 [Rysgaard et al., 2007]. Glass syringes have also been used successfully to measure N2O bacterial production from seawater, without contamination from ambient air [Punshon and Moore, 2004]. Our technique reveals bulk N2O concentrations in the bottom 10 cm of sea ice which were consistently only ∼30% of those in the atmosphere. Since any inadvertent contact of these undersaturated samples with the atmosphere would raise their N2O concentrations, our results should be considered as maximal estimates. The strong undersaturated state of N2O in sea ice observed throughout the whole sampling period suggests that N2O was lost from sea ice prior to our sampling.
 It is well known that gases and other impurities such as nutrients and salts dissolved in seawater are not incorporated into the ice crystal matrix during sea ice formation but instead are concentrated in brine mixtures in channels and pockets within the sea ice [Eicken, 2003]. Brine volume fraction represented, on average, 16% of the bottom sea ice matrix (Figure 4b) during our study. Accordingly, our salinity-based calculations show that N2O concentrations in brine mixtures (Figure 4c) were ∼6 times higher than those measured in bulk sea ice during the study period (Figure 3e), resulting in oversaturation of N2O concentrations in brine with respect to the atmosphere (Figure 4d).
 As has been observed for other gases in sea ice [Tison et al., 2002; Anderson et al., 2004; Delille et al., 2007; Rysgaard et al., 2007; Loose et al., 2009], a physical rejection of brine that carries N2O from the sea ice to the USW as the ice cover thickens could explain the observed low bulk N2O concentrations. When ice is forming during cold weather, temperature decreases upward through the ice column, resulting in an upward increase in the salinity and density of the brine inclusions leading to convective instability [Vancoppenolle et al., 2010]. The brine mixture, including any dissolved gases, is thus rejected from the ice cover by gravity drainage to the underlying seawater during the period of sea ice formation [Papadimitriou et al., 2004]. In the Arctic Ocean, the sea ice cover normally grows until the end of March [Comiso et al., 2008]. Maximum ice thickness in the study region was observed on Day 97 (6 April 2008) [Brown et al., 2011], indicating that our sampling began at the end of the ice growth period, when the bulk ice N2O concentration was probably lowest. The absence of correlation between bulk ice N2O concentrations and salinity can be explained by varying rates of rejection from sea ice for the two compounds [Loose et al., 2009]. In addition, any biological production or consumption of N2O which may be occurring in the ice would further obscure any correlation. Biological N2O production could explain, for example, why the average N2O saturation (∼30%) is higher than would be expected assuming the gas is rejected from the ice at the same rate as salt (bulk ice salinity ∼21% of seawater).
 The observed low levels of N2O measured in the bottom of the sea ice might be partially explained by bacterial denitrification. In a study conducted in the same region in April 2004, heterotrophic activity was shown to increase oxygen consumption in the lower layers of the sea ice, resulting in high microbial denitrification activity at the ice-water interface [Rysgaard et al., 2008]. N2O consumption during denitrification only takes place in stable suboxic environments, a condition not favored by ice algal growth, and not observed during the Rysgaard et al.  study. However, given the reported heterogeneous distribution of oxygen levels in sea ice [Rysgaard and Glud 2004], denitrification leading to N2O consumption may nevertheless be possible.
4.2. Vernal Sea Ice Algal Bloom
 Our sampling covered the development of the vernal ice algal bloom from late March until early May (Figure 3d). No significant differences in N2O concentrations were observed during this period in either sea ice or brine (Figures 3e and 4c). Moreover, N2O concentrations in sea ice were not correlated with either nutrients or salinity. A significant increase in nitrate concentrations in sea ice, from ∼0.50 to 3.22 μM, was observed at the end of our sampling period (Figure 3c), possibly due in part to the nutrient pumping mechanism described by Vancoppenolle et al. . Such an increase in nitrate could also be due to nitrification. However, if nitrification was taking place in sea ice at this time, it had little impact on the low N2O concentrations in the ice. Alternatively the lower chl a concentrations at the end of the time series may indicate low nitrate uptake, and spatial variability of nitrate may play a role, although it should be noted that the last four samples (Days 117–126) were all collected from the same ice floe.
 N2O concentrations in the USW remained low and mostly undersaturated (10.2–18.8 nM; 60–111% saturation) with respect to the atmosphere, and increased slightly to reach equilibrium with the atmosphere at the end of the study period (Figures 5d and 5e). This observation contrasts with the generally high concentrations of N2O (up to 33.5 nM or 181% saturation) measured underneath the sea ice cover by earlier authors, who attributed them to upwelling [Rees et al., 1997] and to bacterial nitrification [Kitidis et al., 2010]. It should be noted, however, that Kitidis et al. also observed some undersaturation, with values as low as 14.7 nM (82% saturation). Before the bloom, the sea ice cover was probably semi-impermeable to gases, limiting air-sea exchange and the reequilibration of the N2O partial pressure. However, relatively warm ice surface temperatures after about Day 97 suggest that the ice became permeable to gas transport [Song et al., 2011]. The increase in N2O concentrations measured in the USW near the end of the bloom (Days 107–123) most probably results from air-to-sea flux of N2O via gas permeation and the shrinking of the sea ice cover. Additionally, snow meltwater could percolate through the ice and expel N2O-rich brine into the USW [Vancoppenolle et al., 2010]. The relative stability of the physical characteristics (salinity and temperature) of the USW during that period precludes any important changes in water mass due to advection or upwelling.
4.3. Implications for Atmospheric N2O Seasonal Variations
 Seasonal cycles of atmospheric N2O are well documented over polar regions. While the underlying mechanisms remain unclear [Khalil et al., 2002; Nevison et al., 2007], these cycles have been attributed to a summer increase in atmospheric vertical mixing and the resulting transport of N2O from the troposphere to the stratosphere [Nevison et al., 2004]. In addition to vertical mixing, Jiang et al.  suggested the existence of a seasonal cycle in the surface sources of N2O. Our data suggest that during spring and early summer, the complete melting of N2O-depleted sea ice would decrease the partial pressure of N2O in surface waters. Such a mechanism would act as a sink for atmospheric N2O in newly open water, as has been suggested by Kitidis et al. . During fall and winter, sea ice formation and the associated expulsion of N2O-rich brines represent a potential source of N2O for the upper water column and the atmosphere through leads. A permeable ice cover [Golden et al., 2007] could also permit direct N2O flux to the atmosphere from supersaturated brines. Our measurements indicate that sea ice formation and melt has the potential to generate sea-air or air-sea fluxes of N2O, respectively. Exchanges of N2O between ice, water and atmosphere could thus contribute to the seasonal N2O cycle and may become more significant as the Arctic multiyear sea ice is replaced by seasonal ice due to climate change [Holland et al., 2006].
 This work is a contribution to the International Polar Year-Circumpolar Flaw Lead system study (IPY-CFL 2008) and to the Canadian Arctic SOLAS (Surface Ocean Lower Atmosphere Study). The work was supported through grants from the Canadian IPY Federal Program Office and the Natural Sciences and Engineering Research Council of Canada. We would like to extend our gratitude to the officers and crew of the CCGSAmundsenfor logistical support; to Yong Zhang, Benoît Philippe and Gauthier Carnat for sampling and post-processing support including the brine volume determinations; and to Jean-Éric Tremblay for providing the nutrient data. This is a contribution to the research program of Arctic-SOLAS and Québec-Océan.