Summary1. Lipids are more depleted in 13C than proteins. Variable lipid contents in tissues affect therefore the measurements of stable carbon isotope ratios. Model based (also called mathematical) normalization has been suggested to correct δ13C values using the ratio of carbon to nitrogen (C/N) as a proxy for lipid content. This approach has not been thoroughly validated for terrestrial animals and it is not clear to what extent it is generally applicable or species/tissue specific.2. Ratios of stable carbon isotopes (δ13C) were obtained for muscle samples of 22 mainly terrestrial arctic mammal and bird species and for egg samples of 32 bird species from nine sites in the circumpolar Arctic. We used linear and nonlinear equations to model the difference in δ13C between samples from which lipids had been extracted chemically and bulk tissue samples. Models were compared on the basis of a model selection criterion (AIC) and of prediction error as estimated by cross‐validation.3. For muscle samples, a linear and a nonlinear equation performed equally well. The observed values were also well predicted by a previously published general equation for aquatic organisms. For egg samples, a nonlinear equation fitted separately to waterfowl and non waterfowl bird species fitted the data best. Prediction errors were, however, larger than for muscle samples.4. The generality of the inferred normalization equations was assessed by applying them to a second data set from a similar ecosystem, but produced in the frame of another study. The predicted lean δ13C values were within 0·5‰ of the observed values for 73% of the muscle samples, but only for 27% of the egg samples.5. Based on our results, we recommend model based normalization of δ13C values as an economic way to deal with varying lipid contents in muscle samples of mammals and birds. For egg samples, on the contrary, model based predictions had large errors. Therefore, we recommend chemical lipid extraction in order to estimate lipid‐free δ13C values for egg content.
Read full abstract