Parameter uncertainty, sensitivity, and sediment coupling in bioenergetics-based food web models

A bioenergetics-based food web model was developed and calibrated using measured PCB water and sediment concentrations in two Great Lakes food webs: Green Bay, Michigan and Lake Ontario. The model incorporated functional based trophic levels and sediment, water, and food chain exposures of PCBs to aquatic biota. Sensitivity analysis indicated the parameters with the greatest influence on PCBs in top predators were lipid content of plankton and benthos, planktivore assimilation efficiency, Kow, prey selection, and ambient temperature. Sediment-associated PCBs were estimated to contribute over 90% of PCBs in benthivores and less than 50% in piscivores. Ranges of PCB concentrations in top predators estimated by Monte Carlo simulation incorporating parameter uncertainty were within one order of magnitude of modal values. Model applications include estimation of exceedences of human and ecological thresholds. The results indicate that point estimates from bioenergetics-based food web models have substantial uncertainty that should be considered in regulatory and scientific applications.