2010 年 26 巻 p. 107-114
The traditional two-source linear mixing model, based on distinct 13C/12C signatures among sources and samples, has been widely used in a number of ecological dietary studies. This model is especially useful in the case that the elemental compositions (i.e. H, C, N, and O) are similar among sources (i.e. diets) and samples (i.e. animals). However, elemental compositions (e.g. C/N ratio) are commonly widely variable, which often leads to large errors in observed results and subsequent interpretations. In this study, we propose a modified mixing model, employing the 13C/12C ratio combined with the C/N ratio, to better evaluate the dietary properties of animals. As a case study, we investigate the relative contributions of riverine algae and terrestrial plants to the diets of aquatic insects, including five grazers and four filterers, in the Chikuma River, Japan. The traditional mixing model indicates that the algal contribution to aquatic insects ranges from 45% to 92%, whereas the modified mixing model yields values ranging from 64% to 97%. These results are in agreement with the C/N ratios of algae, plants, and aquatic insects (5.4±0.7, 13.4±3.3, and 4.9±0.5, respectively), indicating that the traditional mixing model may underestimate the algal contribution to aquatic insects. We suggest that the modified mixing model, using 13C/12C in combination with C/N, is better than traditional mixing models in terms of evaluating the contribution of various sources in animals, even in the case that elemental compositions are markedly different between animals and their diets.