The use of a Bayesian network to calculate the risks of mercury contamination to fish and birds of the South River, Virginia

The South River watershed in western Virginia has a history of mercury contamination from past industrial practices. My study demonstrates how Bayesian networks can be used to conduct an environmental risk assessment of aquatic and riparian environments to assess the overall effects of mercury contamination to target species in the South River. The risk assessment focused on two species of fish, one game-fish, smallmouth bass and one non-game fish, white sucker and two species of birds, one piscivorous, Belted Kingfisher and one insectivorous, Carolina Wren. By examining the exposure pathways through various habitats in the study area, I created a conceptual model that linked the sources of mercury through cause and effect pathways to the potentially impacted target species. From this conceptual model, a three-tiered Bayesian network was constructed to represent not only the impacts of mercury but also the complex interactions and overall risk from other common chemicals, ecological stressors and habitat stressors. The model outputs and sensitivity analysis showed that mercury is one of the two top factors determining risk to fish and birds in the South River for at least one of the six regions in the South River study area. The results of this study demonstrate that Bayesian networks can be used to calculate risk for multiple stressors, and that they are a powerful tool for informing future management strategies for the restoration of the South River watershed.