Enhancing toxicity test performance by using a statistical criterion.

Abstract

Aquatic toxicity tests are laboratory experiments that measure the biological effect (e.g., growth, survival, reproduction) of effluents, receiving waters, or storm water on aquatic organisms. These toxicity tests must be performed using the best laboratory practices, and every effort must be made to enhance repeatability of the test method. We evaluated the generated reference toxicant test data for insurance of a level of quality assurance for tests over time within a laboratory and among laboratories. We recommend the reporting and evaluation of the percent minimum significant difference (PMSD) value for all toxicity test results. The minimum significant difference (MSD) represents the smallest difference between the control mean and a treatment mean that leads to the statistical rejection of the null hypothesis (i.e., no toxicity) at each concentration of the toxicity test dilution series. The MSD provides an indication of within-test variability, and smaller values of MSD are associated with increased power to detect a toxic effect. We recommend upper and lower PMSD bounds for each test method in order to minimize within-test variability and increase statistical power. To ensure that PMSD does not exceed an upper bound, testing laboratories may need to increase replication, decrease variability among replicates, or increase the control mean performance.