Nitrogen stable isotopes analysis of individual bone collagen amino acids was applied to archeological samples as a new tool for assessing the composition of ancient human diets and calibrating radiocarbon dates. We used this technique to investigate human and faunal samples from the Kitakogane shell midden in Hokkaido, Japan (5,300-6,000 cal BP). Using compound-specific nitrogen isotope analysis of individual amino acids, we aimed to estimate i) the quantitative contribution of marine and terrestrial protein to the human diet, and ii) the mean trophic level (TL) from which dietary protein was derived from marine ecosystems. Data were interpreted with reference to the amino acid trophic level (TL(AA)) model, which uses empirical amino acid delta(15)N from modern marine fauna to construct mathematical equations that predict the trophic position of organisms. The TL(AA) model produced realistic TL estimates for the Kitakogane marine animals. However, this model was not appropriate for the interpretation of human amino acid delta(15)N, as dietary protein is derived from both marine and terrestrial environments. Hence, we developed a series of relevant equations that considered the consumption of dietary resources from both ecosystems. Using these equations, the mean percentage of marine protein in the Kitakogane human diet was estimated to be 74%. Although this study is one of the first systematic investigations of amino acid delta(15)N in archeological bone collagen, we believe that this technique is extremely useful for TL reconstruction, palaeodietary interpretation, and the correction of marine reservoir effects for radiocarbon dating.

We present bone collagen amino acid (AA) δ(13)C values for a range of archaeological samples representing four "benchmark" human diet groups (high marine protein consumers, high freshwater protein consumers, terrestrial C(3) consumers, and terrestrial C(4) consumers), a human population with an "unknown" diet, and ruminants. The aim is to establish an interpretive palaeodietary framework for bone collagen AA δ(13)C values, and to assess the extent to which AA δ(13)C values can provide additional dietary information to bulk collagen stable isotope analysis. Results are analyzed to determine the ability of those AAs for which we have a complete set, to discriminate between the diet groups. We show that very strong statistical discrimination is obtained for all interdiet group comparisons. This is often obvious from suitably chosen bivariate plots using δ(113)C values that have been normalized to compensate for interdiet group differences in bulk δ(13)C values. Bi-plots of non-normalized phenylalanine and valine δ(13)C values are useful for distinguishing aquatic diets (marine and freshwater) from terrestrial diets. Our interpretive framework uses multivariate statistics (e.g., discriminant analysis) to optimize the separation of the AA δ(13)C values of the "benchmark"' diet groups, and is capable of accurately assigning external samples to their expected diet groups. With a growing body of AA δ(13)C values, this method is likely to enhance palaeodietary research by allowing the "unknown" diets of populations under investigation to be statistically defined relative to the well-characterized or "known" diets of previously investigated populations.

No abstract