Regional and species specific bioaccumulation of major and trace elements in Arctic seabirds

Twenty-five essential and nonessential elements were analyzed in Arctic seabirds to study the influence of phylogeny, tissue, Arctic region, and diet on avian element accumulation and to identify co-occurrence among metals. Muscle and liver concentrations were positively correlated, generally being higher in liver than in muscle, and generally did not differ by sex. Zinc showed the highest absolute concentrations in all samples (mean, 11.2-26.7 mu g/g in muscle, depending on species and area), followed by copper (5.2-7.5 mu g/g), arsenic (0.5-5.4 mu g/g), selenium (1.0-5.8 mu g/g), rubidium (1.4-2.2 mu g/g), and cadmium (0.04-1.2 mu g/g). Mercury levels ranged from 0.05 to 0.8 mu g/g in muscle. The concentrations varied among species (dovekie (Alle alle], black guillemot [Cepphus grylle], thick-billed murre [Uria lomvia], black-legged kittiwake [Rissa tridactyla], northern fulmar [Fulmaris glacialis], ivory gull [Pagophila eburnean], Thayer's gull [Larus thayeri], and glaucous gull [Larus hyperboreus]), and between the northern Baffin Bay (Canada) and the Barents Sea, depending on the element. Whereas some elements (e.g., mercury and zinc) increased in absolute and standardized concentrations with trophic level in the northern Baffin Bay, most elements showed no relationship with trophic level or other dietary descriptors. In absolute concentrations, nonessential elements differed between regions, whereas essential elements differed among species but not within a species across the two regions. Standardized concentrations (element pattern) of both essential elements and nonessential elements generally did not differ between regions but was highly species specific and, thus, determined by the phylogenetic element regulation capacity. The usefulness of multivariate ordination in element wildlife studies is illustrated, which provides additional insight regarding element co-occurrence in wildlife, allows inclusion of species with low sample number, and reduces the possibility of type II1 errors created by low sample size.