Elucidating food web structure of the Poyang Lake ecosystem using amino acid nitrogen isotopes and Bayesian mixing model

Abstract Compound‐specific isotope analysis of amino acids (CSIA‐AA) is regarded as a more advanced method of unraveling food web connections than the traditional bulk isotope approach. One of the most important assumptions of the CSIA‐AA approach is that the initial offset ( β ) between glutamate and phenylalanine in primary producers should remain unchanged. However, the considerable difference in β values between algae (+3.4‰) and vascular plants (−8.0‰) raises concern regarding the application of the CSIA‐AA approach for complex ecosystems that depend on both resources. In the present study, a Bayesian mixing model was used to estimate the individual contribution of basal dietary resources to each consumer of Poyang Lake, and then, the β was reevaluated to adapt to the assimilation of the individual primary producer for each consumer specimen. In general, the contributions of vascular plant‐derived resources to most consumers are consistent with their feeding behaviors. The smaller contribution of vascular plants to planktivorous fish (usually less than 20%) is biologically expected because of their feeding ecology. Using this strategy, we successfully obtained realistic trophic positions (TP) of consumers in Poyang Lake, which are significantly different from traditional TP algae or TP vascular . For instance, the TP value of grass carp (2.04) was consistent with the feeding behavior of this fish, i.e., they primarily feed on vascular hydrophytes. Therefore, the combination of the CSIA‐AA and Bayesian mixing model provides a better understanding of food web structures, even in a complex freshwater ecosystem.