Comparison of Statistical Approaches Applied to the Ceriodaphnia dubia Chronic Toxicity Method
We compare the performance of two hypothesis test statistical approaches currently used in U.S. water regulatory programs to interpret the reproduction endpoint for the Ceriodaphnia dubia survival and reproduction aquatic toxicity test. These statistical approaches are (1) the well-known NOEC approach and (2) the Test of Significant Toxicity approach, which compares a control to the Instream Waste Concentration (IWC) that is of direct regulatory concern and identifies absence of toxicity by rejecting a hypothesis that the IWC and control responses differ by more than 25%. We summarize Ceriodaphnia test performance at eight California laboratories for control reproduction, and relate control coefficient of variation (CV) to the expected probability of declaring toxicity. As expected, this probability is closely related to control CV. Laboratories differed substantially in CVs and resulting probabilities of declaring toxicity depending on the statistical approach used. When a sample (at IWC) has only a small effect on the endpoint, it is highly desirable to avoid declaring the sample toxic. For TST, the probability of this outcome can be reduced by increasing the number of replicates used in the WET test. We estimate the replication needed to avoid declaring toxicity in most samples that differ less than 10% from control, in relation to average control CV.
Impact/Purpose
Research describes error rates of statistical procedures used to analyze data from the Ceriodaphnia dubia WET test in relation to intra-test (among-replicates) variability achieved by various laboratories. Error rates for TST and NOEC statistical approaches are compared using mathematical calculations and using resampling methods. Intra-test variability achieved by various California WET testing laboratories is summarized. The paper shows how probability of declaring a water sample to be toxic depends on intra-test variability (quantified by the coefficient of variation) and number of replicates used. This paper will be of interest to permitted dischargers having a WET monitoring requirement, to WET testing laboratories, and to regulatory agencies (USEPA and States0 in Region 9.Citation
Fox, J., J. Diamond, D. Denton, AND R. Stuber. Comparison of Statistical Approaches Applied to the Ceriodaphnia dubia Chronic Toxicity Method. Society of Environmental Toxicology and Chemistry, Pensacola, FL511-523, (2019). [DOI: 10.1002/etc.4347]Download(s)
- FOX-ETAL-2019-PREPUB.PDF (PDF) (NA pp, 588.9 KB, about PDF)