Assessment of whole effluent toxicity test variability: Partitioning sources of variability

In this article, we quantify the variability of toxicity tests used in whole effluent toxicity (WET) testing and ambient water testing and demonstrate how knowledge of this variability can be used in the interpretation of compliance with WET limits in National Pollutant Discharge Elimination System permits. Whole effluent toxicity test endpoint accuracy and precision are important factors in establishing the credibility of test results. Initially, we developed a national data set consisting of raw reference toxicant data from freshwater and marine tests. The data set consisted of the most commonly used test species, protocols, and laboratories and included results from multiple tests over time within single laboratories. Using a random‐effects model, we evaluate and estimate the following variance components: between‐laboratory variability, variability as a function of dilution concentration, variability of toxicity tests conducted over time, and random error. A variance components model was used to calculate the relative contribution of each variance component to the total variability in specific test endpoints. All analyses were conducted separately for specific reference toxicant, test species, and test protocol combinations. We demonstrate how to use the resulting variance estimates to calculate the minimum significant difference expected for specific test species and test protocols and present an application with WET test data. We present an application using actual WET test results and make recommendations for ensuring the quality of the information resulting from future WET testing.