ABSTRACT
Species sensitivity distributions (SSDs) are used globally to generate water quality guidelines (WQGs). In Canada, a suite of models has been endorsed for describing SSDs. However, these models may not be suitable for substances with multiple modes of toxic action such as pesticides. Pesticides can produce multimodal SSDs where sensitive target organisms comprise one mode of the SSD and non-target organisms comprise the remaining mode(s). Guidelines from this type of SSD might be estimated using only the most sensitive taxa or using a multimodal distribution. The multimodal method presented here uses all data meeting data quality criteria and is thus in keeping with the concept that data comprising an SSD are a random sample from the population of interest rather than a subset thereof. The bimodal method can simultaneously emphasize the more sensitive portion of the dataset by allowing estimation of WQGs using a statistical subset of the data. In the case of the atrazine dataset example, this allowed estimating a WQG emphasizing more sensitive taxa whereas no parametric models fit only the more sensitive data and the small sample size (5) precluded the use of nonparametric methods.
ACKNOWLEDGMENTS
This work was partially funded by Agriculture Canada through Environment Canada to estimate Ideal Performance Standards and by a research grant to the primary author by the Ontario Ministry of Environment. The authors also acknowledge Natalie Feisthauer and Roxanna Roshon of Stantec Consulting Ltd. for their efforts in dataset compilation and Glenn Suter II, USEPA, and two anonymous reviewers for helpful comments. All statements made herein are the sole responsibility of the authors.