Estimating responses of fish populations to toxic contaminants

Abstract

AbstractIn assessments of risks of toxic contaminants to fish populations, the endpoints of ultimate interest are the persistence, abundance, and production of populations. In this article we demonstrate a method for expressing toxicity test data obtained from full‐life‐cycle tests in terms of the same indices used to assess effects of harvesting and power‐plant cooling systems on fish populations. Our approach involves fitting the logistic concentration‐response function to chronic test data and coupling the functions obtained to a fish life‐cycle model. Confidence bands derived from the data quantify uncertainties inherent in predicting population‐level effects from (1) full‐life‐cycle test data for the species of interest, (2) extrapolation of a concentration‐response function from an acute LC50 for the species of interest, and (3) extrapolation from an acute LC50 for another species. Using rainbow trout and largemouth bass as representative species, we evaluate the relative importances of three sources of uncertainty contributing to prediction intervals for populationlevel effects. We also compare population‐level concentration‐response functions to predicted and experimentally derived maximum acceptable toxicant concentrations for five toxic chemicals.