The Marginal Ice Zone as a dominant source region of atmospheric mercury during central Arctic summertime

Mercury (Hg) is a toxic global pollutant with long atmospheric transport. In the atmosphere, Hg exists predominantly as gaseous elemental mercury (GEM), which has a long lifetime and comprises over 90% of surface air Hg. In the Arctic, atmospheric Hg shows a distinctive seasonality with springtime depletion events (AMDEs) and a summertime GEM maximum. The origin of the summertime maximum remains debated: proposed drivers include long-range transport of anthropogenic emissions from lower latitudes, oceanic Hg evasion (potentially fueled by terrestrial inputs), and re-emissions from the Arctic cryosphere. Most understanding to date is based on coastal station observations, leaving uncertainties for the central Arctic Ocean. This study analyzes continuous summertime (June 9–September 30, 2020) GEM observations from the MOSAIC expedition over the central Arctic to (1) characterize GEM levels and variability relative to coastal sites and (2) identify and quantify the dominant sources contributing to central Arctic summertime GEM using statistical modeling and trajectory-based source analyses.

Prior studies have established Arctic springtime AMDEs and subsequent summertime GEM enhancements at coastal stations (Alert, Zeppelin, Villum). Modeling suggested Asian long-range transport as a key primary source year-round (Durnford et al., 2010), while other work attributed summertime GEM to oceanic Hg evasion fed by terrestrial inputs from rivers and coastal erosion (Sommar et al., 2010; Fisher et al., 2012). More recent Hg isotopic analyses indicate dominant summertime re-emissions from the Arctic cryosphere, with a minor role for terrestrial inputs (Araujo et al., 2022), though the fraction of AMDE-deposited Hg retained until melt is uncertain and appears site-dependent. Reliance on coastal observations may bias understanding, given expected regional differences in sea-ice properties and upper ocean stratification that influence Hg redox, re-emission, and air–sea exchange across the central Arctic. Hence, central Arctic observational constraints are needed to resolve the origin of the summertime GEM maximum.

Study site and period: Summertime legs of the MOSAIC expedition, June 9–September 30, 2020, covering 78.34–90°N and 40.93°W–175.7°E. GEM measurements: Continuous GEM using a Tekran 2537B CVAFS analyzer with dual gold cartridges alternating every 5 minutes (flow 0.7 L min−1). Cartridges thermally desorbed at 550 °C. Inlet at the bow of RV Polarstern to minimize ship exhaust; inlet system included soda lime tubes and 0.45 µm Teflon filters. Data screened for local contamination. Calibration via internal permeation source and external injections with a Tekran 2505 before/after cruise; accuracy >95%. Detection limit <0.10 ng m−3. Independent concurrent GEM measurements in the University of Colorado container (Tekran 2537B, 15-min samples) were used for intercomparison and calibration adjustment. Ancillary observations: DMS (APIMS-ILS, 10 Hz averaged to 10 s, DL <5 ppt), CO (merged hourly data from two containers), SO2 (Thermo 43i). Daily surface ocean chlorophyll a (Chla) from ship’s underway system (11 m depth): filtration onto GF/F, extraction in 90% acetone at 4 °C overnight, fluorometric analysis with acidification step. Meteorology and sea-ice/ocean fractions: Navigation and meteorological data (wind speed WS, pressure P, air temperature Temp) from ship systems. Shortwave radiation (RD), ocean fraction, and sea-ice fraction from GEOS-FP (hourly, 2°×2.5°). Open-water fraction computed as ocean fraction minus sea-ice fraction. Transport analyses: HYSPLIT 168-h hourly back-trajectories (arrival altitude 50 m) driven by GDAS 1°×1° meteorology. For PSCF, the domain 57–90°N, 180°W–180°E was gridded at 0.5°×0.5°. PSCF_ij = (M_ij/N_ij)×W_ij, where M_ij counts endpoints associated with GEM above the overall mean, N_ij is total endpoints per cell, and W_ij downweights low-count cells. Most trajectories were in the low troposphere (57.3% <200 m; 81.1% <600 m; 91.2% <1000 m). Generalized Additive Model (GAM): A semi-parametric GAM (mgcv in R) with Gaussian identity link modeled GEM variability. Predictor selection based on AIC, F statistics, and R² retained six variables in three categories: (1) long-range anthropogenic influence: CO mixing ratio and 48-h endpoint distance from observation location (Traj48h); (2) local oceanic emissions: open-water fraction; (3) meteorology: WS, P, Temp. Smooths were penalized cubic regression splines. Model performance: adjusted R²=0.63, explaining 63.3% of GEM variance; 5-fold cross-validation yielded slope=1.00 and R²=0.99. Diagnostics (Q-Q plot, residuals vs. predictor, residual histogram) supported homogeneity, normality, and independence assumptions. Statistical relationships: Monthly summaries, comparisons to coastal stations, and correlations (e.g., GEM–Chla).

- Central Arctic summertime GEM ranged 1.02–2.99 ng m−3 with mean 1.54 ± 0.27 ng m−3 (n=31,176 five-minute points). June–August mean 1.60 ± 0.28 ng m−3 (78.34–90°N) was comparable to coastal Arctic stations (e.g., Alert 1.63 ± 0.37; Zeppelin 1.60 ± 0.23; Villum 1.63 ± 0.37 ng m−3). - Monthly means: July 1.80 ± 0.32 ng m−3 (n=8264) > June 1.59 ± 0.17 (n=5908) > August 1.42 ± 0.13 (n=8639) > September 1.35 ± 0.097 (n=8368). The highest hourly GEM (2.99 ng m−3) occurred in July under high solar radiation and low CO and SO2, indicating non-anthropogenic drivers. - No AMDEs were observed from June to September during MOSAIC (consistent with relatively warm June air temperatures, −0.86 ± 1.12 °C), though AMDEs were common in MOSAIC spring. - GAM source apportionment: Long-range transport of anthropogenic/terrestrial Hg from lower latitudes contributed minimally (<2%). Oceanic evasion accounted for over 50% of the explained GEM variability. Meteorological predictors contributed up to 37% (Temp 20%, P 9%, WS 8%). - PSCF identified the Marginal Ice Zone (MIZ) as the dominant source region for elevated GEM, rather than the fully ice-covered central pack. - Biological linkage: Significant positive correlation between GEM and surface ocean chlorophyll a (Chla) across the study, R²=0.43, p<0.001, with a phytoplankton bloom observed when in the MIZ. - Mechanistic interpretation for strong MIZ evasion: (1) spring AMDE-deposited Hg(II) loaded into surface waters during melt; (2) high phytoplankton biomass enhancing Hg(II) reduction to Hg(0) via photoreactive organics and enzymatic processes; (3) melting sea-ice and effective upper-ocean mixing in the MIZ removing barriers to air–sea gas exchange. In contrast, north of the MIZ, persistent meltwater stratification suppresses evasion. - Estimated Hg evasion flux in the MIZ: ~56 ng m−2 d−1 (back-of-the-envelope), exceeding Arctic open-ocean fluxes (<24 ng m−2 d−1) and an order of magnitude higher than the Arctic Ocean average evasion flux in a recent budget (3.7–7.3 ng m−2 d−1), but lower than Canadian Arctic Archipelago coastal flux (~130 ng m−2 d−1).

The observations and analyses resolve the origin of the Arctic summertime GEM maximum by identifying the Marginal Ice Zone as the dominant source region for atmospheric Hg via oceanic evasion. The GAM indicates minimal influence from long-range anthropogenic transport during summer, while PSCF localizes source contributions to the MIZ. The strong GEM–Chla correlation and timing with high solar radiation support photochemical and biological reduction of Hg(II) to Hg(0) in MIZ surface waters. Physical processes in the MIZ (melting, reduced sea-ice cover, vigorous mixing) enhance air–sea exchange, in contrast to the stratified, less biologically active central pack where evasion is limited. Meteorological influences (notably temperature) further modulate GEM via increased evasion and potentially reduced halogen-driven oxidation, consistent with trajectories implicating the MIZ. Together, these findings explain the summertime shift from a spring Hg sink to a summer source and underscore the central Arctic Ocean—especially the expanding MIZ—as a significant seasonal Hg emitter.

Continuous central Arctic summertime GEM measurements from MOSAIC, combined with GAM and PSCF analyses, demonstrate that oceanic evasion in the Marginal Ice Zone, rather than long-range anthropogenic transport, dominates the observed summertime GEM variability and explains the Arctic GEM maximum. Mechanistically, springtime Hg loading to surface waters, enhanced biological and photochemical reduction in the productive MIZ, and efficient air–sea gas exchange drive strong evasion. A preliminary flux estimate (~56 ng m−2 d−1) suggests MIZ evasion exceeds open-ocean values. In the context of rapid Arctic warming and widening summer MIZ, the central Arctic’s role as a summertime Hg source is expected to strengthen, with potential global implications given GEM’s long lifetime. Future work should include dedicated field campaigns and modeling to better constrain MIZ evasion fluxes and their interannual variability and to integrate evolving sea-ice and biological changes into Arctic Hg budgets.

- The Hg evasion flux from the MIZ is a back-of-the-envelope estimate and requires dedicated field and modeling studies for improved quantification. - The GAM explains 63.3% of GEM variance, indicating remaining unexplained variability and potential unrepresented processes. - PSCF results reflect mainly low-tropospheric transport and are subject to trajectory and gridding uncertainties; they indicate potential source regions rather than definitive source strengths. - Observations are limited to a single summer season (June–September 2020) and may not capture full interannual variability.