You are here:
Application of the Web-based Interspecies Correlation Estimation (Web-ICE) tool to assess risks of national pesticide registrations to federally listed (threatened and endangered) species
Citation:
Donovan, E., Sandy Raimondo, K. Garber, AND M. Barron. Application of the Web-based Interspecies Correlation Estimation (Web-ICE) tool to assess risks of national pesticide registrations to federally listed (threatened and endangered) species. SETAC North America 35th Annual Meeting, Vancouver, BC, CANADA, November 09 - 13, 2014.
Impact/Purpose:
Present the work of OPP/GED collaboration on using Web-ICE for endangered species risk assessments at the annual SETAC meeting.
Description:
The National Academy of Science (NAS) recently recommended exploration of predictive tools, such as interspecies correlation estimation (ICE), to estimate acute toxicity values for listed species and support development of species sensitivity distributions (SSDs). We explored the ability of the Web-based Interspecies Correlation Estimation (Web-ICE) tool to predict acute toxicity values for listed species based on available toxicity data for a suite of standard test species. We will present the results of this analysis for predicting listed species acute toxicity based on use of species, genus and family level models within Web-ICE. As part of this analysis, the Web-ICE Rules of Thumb from Raimondo et al. (2010), which are intended to be guidelines regarding the application of best professional judgement to help identify robust ICE models and predictions of acute toxicity, were re-evaluated. The Rules of Thumb were derived from trends or approximations based on data containing considerable variability, and were not intended to be definitive limits. Additionally, some of the Rules were derived from the results of a leave-one-out cross-validation analysis that used all models. In this method, each pair of acute toxicity values for surrogate and predicted taxa were systematically removed from the original model, while the remaining data were used to estimate the toxicity of the removed predicted taxa. The cross-validation approach has recently been demonstrated to under-represent robustness of models with small sample sizes (N< 10). We will present a re-evaluation of these Rules using a new analysis of model validation results. In summary, model and prediction properties (Mean square error, MSE; R2; confidence intervals) are the only endpoints for which we recommend limits be applied; all others endpoints should be used as weight-of evidence.