Reconstructing Herbivore Diets: A Multivariate Statistical Approach To Interpreting Compound-Specific Isotope Values
Abstract Stable nitrogen (N) isotope analysis of bulk tissues is a technique for reconstructing the diets of organisms. However, bulk nitrogen isotope (δ15N) values can be influenced by a variety of metabolic and environmental factors that can confound accurate dietary reconstruction. Compound-specific isotope analyses of amino acids (CSIA-AA) have demonstrated the power of the approach in understanding how the δ15N values of bulk collagen are assembled from the constituent AAs. Furthermore, by connecting these AA δ15N values within a robust biochemical framework interpretation of diet and environment are greatly enhanced. Several new proxies have emerged, built around selected AAs; however, the interconnectedness of AA biosynthetic pathways means that patterning of δ15N values across a wider suite of collagen AAs will occur under different environmental or dietary influences. This work seeks to test this idea by situating CSIA-AA within a robust statistical framework using principal component analysis (PCA) and Bayesian statistics to increase the interpretability of a wider range of AA δ15N values in terms of reconstructing herbivore diet. The model was tested using wild and domestic herbivores from the Neolithic settlements of Çatalhöyük (Turkey), Makriyalos (Greece), and Vaihingen (Germany) as case studies. It was found that at Makriyalos there was a sharp separation between domesticated and wild herbivores, which was present to a lesser extent at Çatalhöyük and not observed at Vaihingen. The case studies presented in this work demonstrate that multivariate statistical treatment of CSIA-AA data can deliver new insights into herbivore diet, exceeding those achievable with the Bayesian model.