Toxicant- and response-specific comparisons of statistical methods for estimating effective concentrations

Abstract

Standard U.S. Environmental Protection Agency (U.S. EPA) statistical analyses of whole effluent toxicity tests involve the estimation of the concentration associated with a specified level of inhibition relative to control responses. Current U.S. EPA estimation methods (linear interpolation or probit-based methods) are compared to a recently developed parametric regression-based estimator, the relative inhibition concentration estimator RIp. The RIp estimation technique, with level of inhibition p = 25%, is applied to a series of chronic toxicity test data from a U.S. EPA Region 9 database of reference toxicity tests. Tests on marine species are conducted with one reference toxicant, while the freshwater tests are conducted with several reference toxicants. While the U.S. EPA estimators and the RIp estimator are highly correlated for red abalone larval shell development, the degree of correlation for fathead minnow responses varies with reference toxicant tested. The strength of the relationship between the RIp and the standard U.S. EPA estimators varies as a function of the reference toxicant. Correlations range between 0.67 and 0.99. For all biological responses included in this evaluation (fathead minnow growth or survival and red abalone larval development), experiments occurred where the RIp is estimable, while the standard U.S. EPA estimators are not. Nonestimability of the standard EPA methods appears to be related, in part, to the failure of models to account for enhanced responses, such as a hormesis effect prior to toxicity being manifest. The ability to account for such enhanced responses is a strength of the RIp method. Finally, a variance component analysis suggests that lab-to-lab variability is relatively low for the red abalone but relatively high for the fathead minnow.