Comparison of models that utilize fatty acids to provide estimates of diet composition

Diet composition data provide insights into how food webs respond to disturbances through changes in trophic interactions among constituents. Biochemical methods have recently gained popularity as alternative means to quantify diet compositions over multiple meals. Due to perceived power of included many variables, models tracing predator-prey relationships with fatty acid profiles have garnered much interest. Currently, a mixing model operating in a Bayesian framework (FASTAR), and a model that reduces statistical distances between predator and prey fatty acid profiles (QFASA) have been established. Herein we evaluate estimates provided by both models, using a novel controlled feeding experiment. Juvenile lake trout (Salvelinus namaycush) were fed diets that composed of various mixtures of bloodworm, daphnia, and shrimp for a period of 12 weeks. Diets included each prey in isolation, 50:50 mixtures of each pair of prey, and a diet composed of even mixtures of all three based on dry weights. Fatty acids of lake trout fed prey items in isolation were used as references, and each model was then challenged to estimate components of the mixed diets. Our results illustrate how fatty acids may be used to estimate diet components of wild individuals, providing data on long term foraging habits of wild individuals.