Seawater samples were collected between April and June 2021 in the NGA, Bering Sea, and Chukchi Sea aboard the R/V Sikuliaq during three oceanographic cruises: Northern Gulf of Alaska Long Term Ecological Research (NGA LTER), Bering and Aleutian Internal Tide Mixing, and Hg Cycling in the Arctic (Supplementary Table S1). All sampling stations were on the Alaskan continental shelf, except KOD10, which was on the slope in the NGA. The water column depth at the sampling locations in the Bering and Chukchi Seas were shallow (48 ± 6 m) compared to the NGA (610 ± 477 m) (Supplementary Table S2). Water samples were collected at a single depth, the chlorophyll a maximum (Figure 1; Supplementary Table S2), which was selected because previous studies have shown in situ Hg methylation at that depth (Munson et al., 2018; Starr et al., 2022). The ship’s SeaBird 911+ CTD rosette was used to assess the chlorophyll a maximum during the down cast, which determined the sampling depth. During the downcast, other parameters were measured such as salinity, dissolved oxygen, photosynthetically active radiation (PAR), temperature, and pressure. Due to limited water allowances, Hg methylation, nitrification, and genetic samples were collected from the ship’s regular CTD rosette (non-trace metal clean) in the NGA. Mercury methylation, nitrification, and genetic samples from the other 13 stations throughout the Bering and Chukchi Seas were collected using a trace metal clean SBE19plus V2 SeaCAT Profiler (SeaBird) CTD rosette equipped with Teflon-coated, external spring Niskin bottles (General Oceanics). Samples from the trace metal CTD rosette were transferred to acid washed sample bottles (Hammerschmidt et al., 2011) inside a positive-pressure, plastic enclosure (the bubble), constructed at the beginning of the NGA LTER cruise. Seawater samples for genetic analysis was collected in an acid washed 5-gallon collapsible polyethylene carboy. Using an Eppendorf peristaltic pump, 2 L of sample water was filtered through a 0.22 μM Sterivex filter pack (Millipore). The Sterivex was filled with RNAlater (ThermoFisher), stabilizing the nucleic acids, and stored at −80°C until extraction (Bowman et al., 2020). Seawater for Hg methylation rate experiments was collected in 4 2-L acid washed plastic bottles. Seawater samples for nitrification pool dilution experiments were collected into an acid washed 1-L and 3 125-mL polycarbonate incubation bottles. Incubation experiments occurred upon sample collection.

Onboard dark incubation experiments were carried out with stable Hg isotope additions to quantify Hg methylation and demethylation rates. Four replicate 2-L unfiltered water samples were collected and promptly amended with 7.5 pM ²⁰⁰Hg^II (Oak Ridge National Laboratory) and 7.7 pM CH₃ ²⁰¹Hg by volume. MeHg synthesis was completed prior to the cruise with methyl-cobalamin and inorganic Hg isotopes (Oak Ridge National Laboratory). Initial sample volume was measured, and all samples were placed in a dark incubation chamber at in situ temperature (4°C) without further oxygenation. After 24 h, samples were removed, reduced to 1.5 L and immediately acidified to 1% H₂SO₄ by volume to stop biological transformations. Samples were stored unrefrigerated in a dark container until analysis.

Mercury methylation rate samples were analyzed at Wright State University (WSU) within 8 months of collection. Prior to analysis, these samples were neutralized to pH 4.9 ± 0.3 with 14 M potassium hydroxide, buffered with 14 mM acetate buffer by volume and amended with 0.1 mL of 2.5% ascorbic acid (Munson et al., 2014). Neutralized samples were derivatized with sodium tetraethylborate and purged with Hg-free N₂ gas. During the bubbling process, methylethylmercury was concentrated onto Tenax traps and analyzed on an optimized gas chromatography inductively coupled plasma mass spectrometer (GC-ICPMS; Perkin Elmer Elan 9000; Hintelmann and Evans, 1997). Ambient Hg standards and enriched isotope standards were analyzed on both the Tekran and ICP-MS to quantify both concentration and isotope abundance. Mercury methylation and demethylation rates were calculated from replicates at one time point using matrix methodology (Hintelmann and Evans, 1997; Ouerdane et al., 2009), as CH₃ ²⁰⁰Hg created and CH₃ ²⁰¹Hg lost since amendment respectively: Hg methylation rate (pM day⁻¹) = k_m * [Hg^II] and MMHg demethylation rate (pM day⁻¹) = k_d * [MMHg]. For Hg methylation rate potentials, first order specific rates (k_m) were also calculated as specific rate (day⁻¹) = Hg methylation rate * [Hg^II]⁻¹.

The collected 1 L seawater samples were amended with 20 mM ¹⁵NH₄Cl and inverted three times. The 20 μM spiked sample was transferred into three 125-mL polycarbonate incubation bottles. An initial sample was taken immediately from each incubation replicate via a 60 mL plastic syringe and filtered through a 0.22 μm filter into one 15-mL polypropylene tube and two 20-mL scintillation vials. An initial control was sampled from the 125 mL unamended seawater sample following the same methodology. Initial samples were frozen upon collection and stored until analysis. The four 125-mL incubation bottles were incubated for 24 h in the onboard dark incubation chamber at in situ temperature. After the 24-h incubation, the four bottles were removed, final samples from the control and replicates were collected using the same methodology as the initial samples, and samples were frozen until analysis at WSU.

Cadmium (Cd) reductions (to reduce NO₃ ⁻ to NO₂ ⁻) followed by sodium azide (NaN₃) reductions were completed on each sample to quantify the production of nitrous oxide (N₂O) from both nitrite and nitrate (NO_x ⁻; Heiss and Fulweiler, 2016; McIlvin and Altabet, 2005). To each sample, 100 mg and 6.6 g of dried magnesium and sodium chloride was added respectively and mixed. To complete the Cd reductions, 1 g of activated wet Cd powder was added to each sample and placed on a shaker table for 17 h. Samples were centrifuged for 15 min at 1,000 rpm to decant the reduced solution. 7.5 mL of supernatant was removed from the centrifuge tubes, transferred to a 12-mL Labco exetainer, and sealed with a single-wadded septa cap. Using a glass syringe with a needle, 0.25 mL of a solution containing 1:1 2 M sodium azide: 20% acetic acid purged with Argon gas was injected into the exetainer septa. Samples were shaken and incubated at 30°C for 1 h. Following incubation, the reaction was stopped by injecting 0.15 mL of 10 M NaOH through the exetainer septa with a clean needle and syringe (McIlvin and Altabet, 2005). Exetainers containing the reduced sample were stored upside down and shipped to the University of California—Davis Stable Isotope Facility for analysis of labeled ^45,46N₂O gas production using an isotope-ratio mass spectrometer (Hamilton and Ostrom, 2007). Incubation samples collected in falcon tubes were thawed and quantified for total NH₄ ⁺ using colorimetric flow-injection analysis (Lachat Quikchem 8500) at WSU. Results from both analyses were used to calculate the ¹⁵NH₄ ⁺ transformation and were corrected for NaN₃ reductions using Eq. 1, modified from Heiss and Fulweiler (2016), Beman et al. (2008), and Hampel et al. (2020). Nitrification rate experiments were only completed in the NGA, Bering Sea, Bering Strait, and Chukchi Sea. N T R = N 15 / N 14 ∗ N O x − t f − N 15 / N 14 ∗ N O x − t 0 α   ∗ t (1) w h e r e   α = N 15 H 4 + / N H 4 + 15 + N H 4 + 14

DNA was extracted using the DNeasy PowerWater Sterivex Kit (Qiagen) and assessed for quality using 260/280 and 260/230 ratios, with a NanoDrop Microvolume Spectrophotometer (ThermoFisher), verified by Qubit. Extractions were stored at −18°C until assay. Nitrospina-specific hgcA primers (Supplementary Table S3), Nitro_SP14_1F and Nitro_SP14_1R, designed from sequences found in Antarctic Sea ice were used for amplification (Gionfriddo et al., 2016). The SP8 sequence, 504 bp, from Gionfriddo et al. (2016) was used as a positive control for the real-time polymerase chain reaction (qPCR) assay. A set of seven serial dilutions (10^–4–10^–10 ng μL⁻¹) were prepared from the SP8 sequence and used to form a standard curve (R ² > 0.90). With a reaction volume of 20 μL, the following chemical volumes were used for amplification: 10 μL Luna Universal qPCR Master Mix (New England BioLabs), 1 μL Nitro_SP14_1F and Nitro_SP14_1R, 1 μL of sample (5 ng DNA), diluted SP8, or Nuclease free water, and 7 μL of nuclease free water. Samples were amplified on a Mastercycler ep Realplex2 Real-Time PCR system (Eppendorf) with thermal cycling parameters (outlined in Gionfriddo et al., 2016) and a melting curve (Supplementary Table S3). Amplification efficiency was 95.6%, and sample amplicon concentration was calculated. Gene abundance was reported as Nitrospina-specific hgcA gene copies mL⁻¹ and calculated using Eq. 2 and normalized to sample volume (mL). G e n e   A b u n d a n c e = 5   n g ∗ 6.0221 ∗ 10 23   m o l e c u l e s / m o l 504   b p ∗   10 9   n g / g   ∗   650   g / m o l (2)

The NGA is heavily influenced by the Alaskan Coastal Current (ACC), which is driven by alongshore winds and characterized by freshwater inputs derived from coastal runoff and the Copper River outflow (Stabeno et al., 2004). From fall to spring the NGA is dominated by downwelling currents and Ekman transport from off-shelf surface waters (Stabeno et al., 2004). The downwelling dissipates as winds decrease throughout the spring, allowing for the resuspension of key nutrients (such as nitrate, phosphate, and silicate) promoting primary production (Stabeno et al., 2004). During the 2021 Spring NGA LTER cruise there was a large early season diatom bloom from late April to early May 2021, which was thought to be induced by the increased sunlight and reduced winds. Chlorophyll a concentrations at the sampling depth (chlorophyll a maximum) across the NGA ranged from 0.3–10.1 mg m⁻³ (Supplementary Table S2). The highest value (10.1 mg m⁻³) was found at MID2, within the mixed waters under the Copper River plume. A small Copper River plume can be identified by a thin layer of nutrient deplete freshwater extending from the Copper River outflow. In the NGA, salinity increased with distance from the coastline, as the influence from runoff, freshwater inputs, and ACC wanes (Stabeno et al., 2016).

Water flows north across the Bering Sea shelf and Bering Strait, into the Chukchi Sea via the Bering Shelf Current and Chukchi Shelf current (Danielson et al., 2022). Higher concentrations of chlorophyll a and DO were seen in the surface waters in the southern Bering Sea near Unimak Pass (Supplementary Table S2). This can be attributed to the Aleutian North Slope current and deep mixing that resuspends key limited nutrients into the surface waters of this high nutrient low chlorophyll system (Aguilar-Islas et al., 2007; Danielson et al., 2022).

In situ Hg methylation occurred at all stations (0.003–0.008 d⁻¹). Specific (Hg-normalized) Hg methylation rate potentials (k_m) across all regions were comparable with previous measurements from the chlorophyll a maxima across the Arctic (0.004–0.010 d⁻¹; Lehnherr et al., 2011), Mediterranean Sea (0.004–0.038 d⁻¹; Monperrus et al., 2007) and the central Pacific Ocean (0.004–0.017 d⁻¹; Munson et al., 2018). Variability of k_m within each region is attributed to temporal and spatial heterogeneity between sampling stations impacting particle and nutrient distribution as a result of differing current dynamics (Stabeno et al., 2004; Strom et al., 2006). Differing assemblages of Hg methylating bacteria, Hg speciation, and demethylation rates also contribute to the variability of k_m within each sampled polar region. The highest k_m was measured at Unimak Pass (one-way ANOVA: p < 0.05; Table 1), which can be attributed to deep mixing, providing a substrate and microhabitat for Hg methylating anaerobic bacteria (Swift and Aagaard, 1976; Ortiz et al., 2015; Motta et al., 2022). The samples for incubation experiments in the NGA were collected in the early spring during a diatom bloom. This increase in productivity could decrease the amount of Hg^II available for Hg methylation. This could potentially explain the lower k_m across the NGA, when compared to other regions. Changes in water chemistry across the regions could result in differing contributions of instantaneous Hg methylation to the measured Hg methylation rate potentials. This could explain the variability seen in Hg methylation rate potentials within and across regions.

Across all regions, correlations between k_m and oxygen, ambient Hg, ambient NH₄ ⁺, salinity, and temperature were weak and not significant. The difference between predicted and measured MeHg concentrations is attributed to MeHg demethylation in the surface waters and mixing throughout the Bering Strait and Chukchi Sea (DiMento and Mason, 2017; Munson et al., 2018; Wang et al., 2018).

Methylmercury demethylation rate potentials are not reported as the demethylation rates were unrealistically high. Instantaneous demethylation could explain the low CH₃ ²⁰¹Hg recovered in the samples (Munson et al., 2018). For future Hg rate experiments, adding additional time points, including a time zero, should be done as suggested by others (Munson et al., 2018).

Potential nitrification rates ranged 0.5–2.0 nM d⁻¹ across all polar regions. These measurements are concurrent with in situ nitrification rates measured at deeper depths, 100–200 m, in oligotrophic regions such as the Sargasso Sea (2.0 ± 0.1 nM d⁻¹; Newell et al., 2013). The mean specific nitrification rate across all stations, 7.2 × 10⁻⁵ d⁻¹ is lower than the global ocean estimate 0.195 d⁻¹ (Yool et al., 2007), which is thought to be driven by high ¹⁵NH⁴Cl amendment. Nitrite and nitrate produced during nitrification pool dilution experiments are reported as potential nitrification rates to account for any bias from the high amendment, but the rates are low, suggesting that the nitrifiers become ammonium-saturated at low concentrations (Horak et al., 2013; Newell et al., 2013). In the Chukchi Sea (1.9 ± 0.8 nM d⁻¹), nitrification rate potentials were higher than previous measurements by Shiozaki et al. (2019) in the surface waters within the Chukchi and Beaufort Seas (<0.27 nM d⁻¹), which is not surprising as those rates are low for the open ocean (Yool et al., 2007). Nitrification rate experiments were not conducted at Unimak Pass, but it is hypothesized that Unimak Pass would have a higher apparent nitrification rate as deep mixing in this region would provide remineralized NH₄ ⁺ and metals (Cu, Mo, and Fe) for enzymes involved in both steps of nitrification, ammonia monooxygenase and nitrite oxidoreductase (Musiani et al., 2020; Chicano et al., 2021). It is likely that at the chlorophyll a maximum, competition between microbes and plankton taking up various forms of N are driving the lower nitrification rates in this study. The nitrite maximum is typically seen at the base of the euphotic zone as many known ammonia oxidizing bacteria have been shown to be photo-inhibited, resulting in higher nitrification rates with reduced light (Horrigan and Springer, 1990; Lomas and Lipschultz, 2006; Merbt et al., 2012). Previous studies in the North Pacific and Arctic Ocean show nitrification maximum occurring below the chlorophyll a maximum, around the nitrite maximum, under <1% PAR (Grundle et al., 2013; Shiozaki et al., 2016; Shiozaki et al., 2019). The potential nitrification rates measured at the chlorophyll a maximum across all subpolar marine regions reported in this study are therefore not expected to reflect the highest nitrification rates in the water column.

A positive linear relationship (R ² = 0.34, p < 0.05; Figure 2) was seen between nitrification and Hg methylation rate potentials across all stations, suggesting a possible link between these two processes in surface waters surrounding Alaska (NGA, Bering Sea, Bering Strait and Chukchi Sea). The linear regression slope between nitrification and Hg methylation rate potentials across all regions (m = 0.0087 pM MeHg nM⁻¹ NO_x ⁻; Figure 2) is the same order of magnitude as previous findings in the Amazon River Plume (m = 0.0028 pM MeHg nM⁻¹ NO_x ⁻; Starr et al., 2022). This difference in slope can likely be attributed to differing ammonia oxidizing bacteria and archaea activities, substrate availability, and/or microbial assemblages including nitrifier community structure.

Station STN13 in the Bering Strait had a higher relative Hg methylation rate potential compared to nitrification rate potential (Figure 2). Chlorophyll a concentrations at STN13 were four times higher than those of STN3, the other station within the Bering Strait sampled 12 days prior to STN13 (Table 1; Supplementary Table S1). A higher chlorophyll a concentration suggests a general increase in microbial activity or decreased grazing pressure. Data from the USGS Yukon River Pilot Station show little change in Yukon River discharge throughout the 12-day period between sampling, suggesting riverine inputs were not the main driver in differing rate potentials or the high chlorophyll a concentrations. First year ice melt could influence the rate potentials between the two stations in the Bering Strait. A reduction in ice cover could change current dynamics driven by winds promoting water column mixing and resuspension of particles inhabited by Hg methylators. Although these are potential reasons for why STN13 had a higher Hg methylation rate potential compared to nitrification rate potentials, further investigation is needed to conclusively determine the decoupling of these rate potentials in the Bering Sea.

Station MID2, in the NGA near the Copper River outflow, had a higher relative nitrification rate potential compared to the Hg methylation rate potential. This higher relative nitrification rate potential can be attributed to nutrient resuspension, such as NH₄ ⁺ and Cu, and differing communities of nitrifiers that can withstand freshwater plume dynamics (Bernhard et al., 2010). Lower substrate availability for Hg methylation associated the NGA diatom bloom could result from increased binding of Hg^II to particulate material, potentially accounting for the lower Hg methylation relative to nitrification rate potentials in the NGA. Overall, it is likely that the larger temporal and spatial differences among stations accounts for the higher variability in this study compared with the previous measurements in the Amazon River Plume (R ² = 0.78; Starr et al., 2022).

The stations within the Chukchi Sea had less temporal and spatial variability than the other regions, as these stations were sampled within the shortest time period (6-day) and across the smallest distance (110 miles; Supplementary Table S1). The correlation between Hg methylation and nitrification rate potentials in the Chukchi Sea was highly significant (R ² = 0.99, p < 0.001; Figure 3). Although the correlation between nitrification and Hg methylation rate potentials were strong, it cannot be concluded that nitrification rates alone are driving Hg methylation rates throughout these polar regions, as the functionality of Nitrospina-specific hgcA-like genes have not yet been confirmed. It is possible that the measured correlation between Hg methylation and nitrification may also be a result of other biogeochemical processes, such as general bacterial productivity, acting as a proxy (although nitrifiers are autotrophs, while most known Hg methylators are heterotrophs; Yool et al., 2007; Monperrus et al., 2007).

Nitrospina-specific hgcA-like genes were present across all stations. The Bering Strait had the highest average gene copy numbers (2,914 copies mL⁻¹) compared to the NGA (2023 copies mL⁻¹), Bering Sea (1895 copies mL⁻¹), and Chukchi Sea (1879 copies mL⁻¹). Unimak Pass had the lowest gene abundance (236 copies mL⁻¹), which was surprising as Unimak Pass had the highest k_m (Table 1). It is important to note that Hg methylation and nitrification rate potentials reported are a composite of the entire Hg methylator and nitrifier community, not just Nitrospina. Hg methylating microbes at Unimak Pass may be primarily anaerobic and particle-associated, as the station is characterized by deep water column mixing. We hypothesize that anaerobic and microaerophilic Hg methylating microbes inhabit settling or suspended particles, derived from sediment resuspension in the well mixed, shallow Bering Strait and Chukchi Sea, the well mixed waters of Unimak Pass, and the particle-rich Copper River plume. While this may include microaerophilic Nitrospina (Lücker et al., 2013), such particle microenvironments are likely dominated by anaerobic Hg methylating microbes such as Desulfobacteraceae and Methanomassiliicoccus, which have been found throughout the Bering Sea, Chukchi Sea, and Arctic Ocean water column (Bowman et al., 2020).

Across all sampled regions there was not a correlation between Nitrospina-specific hgcA-like genes and Hg methylation rate potentials. However, these Nitrospina hgcA-like amplicons only show gene presence, not enzyme function, so the lack of correlation may indicate that not all of these genes were active at the time of sampling. Despite a lack of correlation, the presence of Nitrospina-specific hgcA-like genes provides support for the potential for these organisms to be the primary methylators, given the relationship between Hg methylation and nitrification.