Beyond bulk δN: Combining a suite of stable isotopic measures improves the resolution of the food webs mediating contaminant signals across space, time and communities.

Top predators are used as indicators of contaminant trends across space and time. However, signals are integrated over complex food webs, and variation in diet may confound such signals. Trophic position, assessed by bulk δN, is widely used to infer the variation in diet relevant to contamination, yet a single variable cannot completely describe complex food webs. Thus, we examined relationships across three aquatic systems varying from a single species to a small food web using bulk values from four isotopes and 21 amino acid-specific values. Because variation in baseline ('source') δN can confound estimates of trophic position , we calculated trophic position from the difference between δN (δN for amino acids that change with trophic position) and δN (δN for amino acids that do not change with trophic position). Across all three systems, variation in δN explained over half of the variation in bulk δN, and stable isotope values that reflected the base of the food web (δC, δO, δS) predicted contaminants as well or better than δN-which was supported by a meta-analysis of other studies. In ospreys feeding in lakes, variation in δN across space created a spurious relationship between ΣDDT and apparent trophic position, and masked a relationship between ΣPCB and trophic position. In a seabird guild, changes in diet over time obscured temporal variation in contaminants over five decades. In Arctic fish and invertebrates, more accurate trophic magnification factors were calculated using δN. Thus, (1) using δN, instead of bulk δN, avoided incorrect conclusions and improved accuracy of trophic magnification factors necessary to assess risk to top predators; and (2) diet assessed with multiple spatial isotopes, rather than δN alone, was essential to understand patterns in contaminants across space, time and biological communities. Trophic position was most important for lipophilic 'legacy' contaminants (ΣDDT, ΣPCB) and habitat was most important for other contaminants (ΣPBDE, ΣPFAS, mercury). We argue that the use of amino acid-specific analysis of δN alongside 'non-trophic' isotopes should be a core feature of any study that examines the influence of trophic position on chemical pollution, as required for a chemical to be added to international conventions such as the Stockholm Convention.