Standardizing acute toxicity data for use in ecotoxicology models: influence of test type, life stage, and concentration reporting.

Ecotoxicological models generally have large data requirements and are frequently based on existing information from diverse sources. Standardizing data for toxicological models may be necessary to reduce extraneous variation and to ensure models reflect intrinsic relationships. However, the extent to which data standardization is necessary remains unclear, particularly when data transformations are used in model development. An extensive acute toxicity database was compiled for aquatic species to comprehensively assess the variation associated with acute toxicity test type (e.g., flow-through, static), reporting concentrations as nominal or measured, and organism life stage. Three approaches were used to assess the influence of these factors on log-transformed acute toxicity: toxicity ratios, log-linear models of factor groups, and comparison of interspecies correlation estimation (ICE) models developed using either standardized test types or reported concentration type. In general, median ratios were generally less than 2.0, the slopes of log-linear models were approximately one for well-represented comparisons, and ICE models developed using data from standardized test types or reported concentrations did not differ substantially. These results indicate that standardizing test data by acute test type, reported concentration type, or life stage may not be critical for developing ecotoxicological models using large datasets of log-transformed values.