Dealing with lipid effects and lipid-extraction biases in δ13C and δ15N isotopic studies: a solution based on 28 marine invertebrate, fish and mammal species

Lipids are naturally depleted in 13C isotope in relation to its C sources, causing a bias in {delta}13C in bulk samples that varies with lipid content. Failure to take this issue into account results in inaccurate conclusions in food web and habitat use studies. Two approaches to resolve this issue are 1) to extract lipids from samples prior to measurement, a resource-intensive process that also can alter {delta}15N or 2) estimating a lipid-free {delta}13C using one of several equations that differ in levels of sophistication and generalization across taxa. Here {delta}13C and {delta}15N were measured on bulk and lipid-extracted muscle samples of a dataset of over 2000 specimens of 28 species of marine invertebrates, fishes and mammals. Our objectives were to 1) compare the effect of lipid extraction on {delta}13C and {delta}15N across taxa; 2) compare the performance of five normalization models, overall and on subsets of species; 3) propose a model to revert lipid-extracted {delta}15N back to their bulk values; and 4) identify the most suitable approach for dealing with lipid biases in isotopic ratios. Extraction caused an uneven enrichment in {delta}13C and {delta}15N across species. Model taxonomic specificity increased estimate accuracy in both isotopes. Models from Logan et al. (2008) and McConnaughey and McRoy (1979) performed better than the other models tested. {delta}15Nbulk could be reliably estimated based on {delta}15Nlipid-extracted using a linear model. This study provides a way forward for obtaining reliable {delta}13C and {delta}15N values in muscle tissue without the costs of duplicate analyses and represents a major step toward the harmonization of datasets collected under the two different approaches.