Stable isotopes and trace elements as indicators of diet and habitat use in cetaceans: predicting errors related to preservation, lipid extraction, and lipid normalization

Accurately predicting errors related to preservation, lipid extraction, and lipid normalization on chemical tracers would enable the use of archived samples in long-term studies of trophic ecology and habitat use of aquatic species. We determined whether stable carbon and nitrogen isotope ratios and concentrations of 14 trace elements can be accurately predicted from dimethyl sulfoxide (DMSO)-preserved mammal skin, which would provide equivalent estimates to that from unpreserved tissue. We tested 3 lipid-correction approaches for applicability to cetacean skin, a largely unexplored taxon and tissue, and provide a model for evaluating impacts of errors from lipid extraction or normalization on diet composition estimated using isotopic mixing models. DMSO had unpredictable effects on trace element concentrations, rendering DMSO-preserved samples inefficient for retrospective studies. However, lipid extraction and DMSO preservation resulted in predictable and similar, although not identical, effects on isotopic signatures across 4 cetacean species with different skin structure and thickness, making correction for these effects a potentially viable alternative to lipid and DMSO extraction. Generally, lipid-normalization models were reliable when applied to cetacean skin, as errors were similar to those from other species or tissues. Because model fit generally improved with data specificity, developing tissue- and species-specific parameters and equations is probably more important than model choice, although the mass-balance model was considered the most robust across aquatic vertebrates and tissues. The effects of errors associated with the various treatments and lipid normalization on isotopic mixing results increased as the isotopic distance among prey sources decreased, suggesting that empirical corrections as an alternative to δ13C determination from lipid-extracted duplicate samples need to be evaluateda priorirelative to study objectives and anticipated results.