Sensitivity of stable isotope mixing models to variation in isotopic ratios: evaluating consequences of lipid extraction

1. Stable isotopes of carbon and nitrogen are increasingly used in studies of animal diet reconstruction via mixing models. However, isotope ratios of both consumer and source tissues can be altered by various amounts of lipids, potentially leading to biased estimates of diet composition when they are not taken into account. 2. We investigated the consequences of lipid correction on the estimation of diet composition with mixing models. Using empirical data from three northern terrestrial trophic systems, we illustrated the direct effects of lipid extraction (LE) on the delta C-13 and delta N-15 of source and consumer tissues and its ultimate effects on the reconstruction of the consumer's diet. 3. In parallel, we developed a simulation tool in R, called fatsim, to assess sensitivity of mixing models to variation in isotopic ratios of samples from source or consumer tissues. This tool can be used to assess the effect of shifts in isotopic ratios caused by LE, or other sources of variation, in any trophic system and thus aid in decision making regarding lipid removal. 4. Using fatsim, we showed that the potential effects of LE on estimates of diet composition cannot be predicted without simulations, even in relatively simple systems. The sensitivity of a mixing model isotopic shift depends on the complexity of the system (number of sources) and on the relative positions of sources and consumers within the isotopic mixing space. 5. Our study confirms that the presence of lipids in tissues can bias the interpretation of diet reconstruction results. In a given trophic system, testing the sensitivity of a mixing model to LE can help decide whether lipid removal is required in order to avoid this bias.