Maximum Unified Fatty Acid Signature Analysis: a new approach to QFASA

Accurately estimating predator diets is crucial for understanding and predicting ecosystem changes. The desire for non-lethal but accurate diet estimation techniques has led to new approaches such as quantitative fatty acid signature analysis (QFASA), a now widely accepted method for estimating the diets of marine predators. We propose a novel alternative to QFASA, namely maximum unified fatty acid signature analysis (MUFASA), to estimate dietary proportions using fatty acid signatures. MUFASA is based on maximum likelihood estimation principles and consequently offers several theoretical advantages over QFASA, including estimates that possess desirable properties when model assumptions hold. In addition, the availability of a likelihood function enables the use of a broad range of existing methodologies that may have the capacity to address current well-known challenges associated with diet estimation via fatty acids. MUFASA and QFASA are compared using simulations based on wide-ranging diets, as well as real-life data from a captive study of harbour seals, for which diets are known. Estimates derived from QFASA and MUFASA are similar, suggesting that diet estimation in this context can potentially be viewed through an MLE framework. While not the primary focus of this work, bootstrap confidence intervals are also developed and preliminary results yield high coverage probabilities when the diet proportions are not near 0 or 1.