Assessing the Effects of Chemical Mixtures using a Bayesian Network-Relative Risk Model (BN- RRM) Integrating Adverse Outcome Pathways (AOPs) in Four Watersheds

Chemical mixtures are difficult to assess at the individual scale and are more challenging at the population scale. I have conducted a regional-scale ecological risk assessment by evaluating the effects of chemical mixtures on populations with a Bayesian Network- Relative Risk Model (BN-RRM) in four Washington state watersheds (Skagit, Nooksack, Cedar and Yakima). Organophosphate pesticides (diazinon, malathion and chlorpyrifos) were chosen as the chemical stressors and the Puget Sound Chinook salmon (Oncorhynchus tshawytscha) Evolutionary Significant Unit (ESU) were chosen as the population endpoint. Laboratory tests found that organophosphate pesticide mixtures act synergistically and impair acetylcholinesterase activity. Exposure-response equations for binary mixtures of organophosphates were incorporated into the BN-RRM framework to predict risk to a population. Dissolved oxygen and water temperature were chosen as ecological stressors. The Puget Sound Partnership’s management goal of Puget Sound Chinook is no net loss. A generic ocean-type Chinook salmon population model was used in this risk assessment. Each of the population model simulations started with 500,000 fish. Any number below 500,000 was defined as a net loss. Risk was defined the probability of not achieving the management goal. Calculations indicate synergism does not occur with measured concentrations. This is because malathion, the known synergist, was not found in concentrations that induced a synergistic response. However, at malathion concentrations of 3-15 µg/L, synergism with the other OPs is predicted to occur and does increase risk. My research demonstrates that mixture toxicity can be incorporated into a probabilistic model that estimates risk on populations