An analysis of potential bias in the sensitivity of toxicity data used to construct sensitivity distributions for...

Since the mid-1970s, thousands of studies have evaluated the toxicity of various chemicals to aquatic organisms. Results from many of these studies have been used to develop species sensitivity distributions (SSDs) or genus sensitivity distributions (GSDs) for deriving water quality guidelines. Recently, there has been more emphasis on evaluating the toxicity of chemicals to sensitive organisms rather than the entire range of sensitivities. The SSD approach is intended to inform the derivation of guidelines for the protection of all species, not just those that were included in the SSD. The overemphasis of the more sensitive end of the SSD can contribute to a skew in the observed distribution such that the shape of the distribution is distorted from what it would be if all species could be tested, which ultimately affects the derived guideline value. The freshwater acute Cu GSD derived by the US Environmental Protection Agency (USEPA) is one that exemplifies this trend, with one-third of the genera in the GSD belonging to only 3 taxonomic families, all of which are nearer to the sensitive end of the distribution. The stronger representation of the more sensitive families does not seem to mirror the overall abundance of species within those families in nature. This tendency toward testing sensitive organisms is not seen in the chronic Cu SSD. In the present study, Cu toxicity literature is reviewed and long-term trends in the availability of toxicity information for species of varying sensitivity are examined. As part of the present review, the apparent bias that favors the publication of toxicity data for sensitive taxa is demonstrated, and implications for the representativeness of SSDs and their use in developing water quality guidelines are discussed. Integr Environ Assess Manag 2019;00:000-000. © 2019 SETAC.