STATISTICAL CONSIDERATIONS IN THE EVALUATION OF CHRONIC AQUATIC TOXICITY STUDIES

Statistical methodological issues for estimating the no observable effects concentration (NOEC) in chronic aquatic toxicity tests are addressed. The proposed approach utilizes a two-stage estimation procedure which first examines survival and then evaluates sublethal effects on those concentrations which do not affect survival. At each stage, a one-sided stepwise trend test is employed to test for progressiveness of response with increasing concentration. The log rank test is employed for the analysis of censored survival and time to maturation variables. Adjustments are made for the period of reproductive potential, and transformations are applied to the reproductive variables. Theoretical arguments and empirical results from four previously described Daphnia magna assays show that the proposed procedure should provide more meaningful and sensitive NOEC estimates than presently used methods.