Use of a Food Web Bioaccumulation Model to Uncover Spatially Integrated Polychlorinated Biphenyl Exposures in Detroit...

We applied and tested a bioenergetic-based, steady-state food web bioaccumulation model to predict polychlorinated biphenyl (PCB) exposures in sport fish of the Detroit River (USA-Canada), which is a Great Lakes area of concern. The PCB concentrations in the sediment and water of the river were found to exhibit high spatial variation. The previously contained areas of high contamination may have spread to adjacent food webs as a result of fish movements. This process may cause biased predictions in single-compartment bioaccumulation models. We performed multiple simulations and contrasted model predictions against a database of 1152 fish sample records comprising 19 sport fish species. The simulations evaluated 4 spatial scales (river-wide, 2-nation, 4-zone, and 6-zone models) to reveal how the spatial heterogeneity of contamination and species-specific movements contribute to variation in fish PCB exposures. The model testing demonstrated that the 2-nation model provided the most accurate global prediction of fish contamination. However, these improvements were not equally observed across all species. The model was subsequently calibrated for poorly performing species, by allowing cross-zone exposures, demonstrating the importance of accounting for specific ecological factors, such as fish movement, to improve PCB bioaccumulation prediction, especially in highly heterogeneous water systems. Environ Toxicol Chem 2019;38:2771-2784. © 2019 SETAC.