Seabirds reveal mercury distribution across the North Atlantic

Abstract

Mercury (Hg) is a heterogeneously distributed toxicant affecting wildlife and human health. Yet, the spatial distribution of Hg remains poorly documented, especially in food webs, even though this knowledge is essential to assess large-scale risk of toxicity for the biota and human populations. Here, we used seabirds to assess, at an unprecedented population and geographic magnitude and high resolution, the spatial distribution of Hg in North Atlantic marine food webs. To this end, we combined tracking data of 837 seabirds from seven different species and 27 breeding colonies located across the North Atlantic and Atlantic Arctic together with Hg analyses in feathers representing individual seabird contamination based on their winter distribution. Our results highlight an east-west gradient in Hg concentrations with hot spots around southern Greenland and the east coast of Canada and a cold spot in the Barents and Kara Seas. We hypothesize that those gradients are influenced by eastern (Norwegian Atlantic Current and West Spitsbergen Current) and western (East Greenland Current) oceanic currents and melting of the Greenland Ice Sheet. By tracking spatial Hg contamination in marine ecosystems and through the identification of areas at risk of Hg toxicity, this study provides essential knowledge for international decisions about where the regulation of pollutants should be prioritized.

Keywords: ecotoxicology; mercury; spatial distribution.

Fig. 1.

(A) Winter distribution (weekly medoid locations from November to January for Atlantic puffins, black-legged kittiwakes, Brünnich’s guillemots, common eiders, common guillemots, and little auks and from November to December for Northern fulmars) colored by species (colored points) and breeding colonies (black points). (B) Predictions of the estimated Hg concentrations (in log) for 1,000 iterations (Methods) with highest values in dark blue and lowest predictions in yellow.

Fig. 2.

Schematic description of the statistical analyses with 1) data preparation, 2) regression kriging with step 1 (regression) and step 2 (ordinary kriging), and 3) mapping of model predictions (estimated Hg concentrations).