Four approaches for estimating isotope discrimination factors produce contrasting dietary estimates for bears

Abstract: Isotope diet-tissue discrimination factors can significantly influence dietary proportions estimated with stable isotope mixing models. For bears, for which discrimination factors from controlled dietary experiments are currently lacking for most tissues and diets, literature reports 3 approaches have so far been used to compile carbon and nitrogen discrimination factors, and recently one additional approach has been proposed. We used these 4 approaches to compile carbon and nitrogen discrimination factors for the brown bear (Ursus arctos) population in Slovenia (southeastern Europe). By performing a sensitivity analysis of the stable isotope mixing models, we then tested how these different discrimination factors affected dietary estimates for bears. We found substantial differences in the dietary estimates among the applied approaches, with the approach that uses discrimination factors from different species producing most distinct results. Our results suggested that these differences were more pronounced among approaches with larger differences in discrimination factors and for isotopically less distinct food sources. We show that the 4 approaches can lead to contrasting conclusions about bear diets, which can have important ecological and management consequences. Therefore, when using stable isotope mixing models without experimentally obtained species-, diet-, and tissue-specific discrimination factors, the choice of the discrimination-factor compilation approach is vital. All 4 approaches have limitations that must be considered because they can result in erroneous discrimination factors when certain species, diets, and tissues are used. We review these limitations and provide recommendations about the use of these approaches in stable isotope dietary studies for bears.