RESUMO
RATIONALE: Lipid correction models use elemental carbon-to-nitrogen ratios to estimate the effect of lipids on δ13 C values and provide a fast and inexpensive alternative to chemically removing lipids. However, the performance of these models varies, especially in whole-body invertebrate samples. The generation of tissue-specific lipid correction models for American lobsters, both an ecologically and an economically important species in eastern North America, will aid ecological research of this species and our understanding of the function of these models in invertebrates. METHOD: We determined the δ13 C and δ15 N values before and after lipid extraction in muscle and digestive glands of juvenile and adult lobster. We assessed the performance of four commonly used models (nonlinear, linear, natural logarithm (LN) and generalized linear model (GLM)) at estimating lipid-free δ13 C values based on the non-lipid-extracted δ13 C values and elemental C:N ratios. The accuracy of model predictions was tested using paired t-tests, and the performance of the different models was compared using the Akaike information criterion score. RESULTS: Lipid correction models accurately estimated post-lipid-extraction δ13 C values in both tissues. The nonlinear model was the least accurate for both tissues. In muscle, the three other models performed well, and in digestive glands, the LN model provided the most accurate estimates throughout the range of C:N values. In both tissues, the GLM estimates were not independent of the post-lipid-extraction δ13 C values, thus reducing their transferability to other datasets. CONCLUSIONS: Whereas previous work found that whole-body models poorly estimated the effect of lipids in invertebrates, we show that tissue-specific lipid correction models can generate accurate and precise estimates of lipid-free δ13 C values in lobster. We suggest that the tissue-specific logarithmic models presented here are the preferred models for accounting for the effect of lipid on lobster isotope ratios.