Uncertainty and variability in risk from trophic transfer of contaminants in dredged sediments

Abstract

The risks associated with bioaccumulative contaminants must be considered when evaluating dredged material disposal alternatives. The bioaccumulation of organochlorines and other contaminants by higher trophic level organisms represents one of the most significant sources of uncertainty in risk assessment. Both population variability (e.g. true population heterogeneity in body weight, lipid content, etc.) and uncertainty (e.g. measurement error) in trophic transfer can lead to large errors in predicted risk values for ecological receptors. This paper describes and quantitatively evaluates sources of uncertainty and variability in estimating the risk to an ecological receptor (osprey) from the trophic transfer of polychlorinated biphenyls (PCBs) in sediments from the New York–New Jersey (NY–NJ) Harbor. The distribution of toxicity quotients is obtained using a food chain model for the osprey and specifying distributions for input parameters, which are disaggregated to represent either uncertainty or variability. PCB concentrations in sediment and water are treated as predominantly uncertain, whereas lipid content in fish, feeding preferences, and fish weight are assumed to contribute primarily to population variability in PCB accumulation. The analysis shows that point estimates of reasonable maximum exposure (RME) exceed the uncertainty bounds on the 95th percentile of variability. The analysis also shows that uncertainties in the sediment and water contaminant concentrations contribute more to the range of risk estimates than does the variability in the population exposure parameters. The separation of uncertainty and variability in food chain models can help to support management decisions regarding dredged material disposal by providing a quantitative expression of the confidence in ecological risk estimates. A rationale is provided for the distinction between uncertain and variable parameters based on management goals and data availability.