Estimating the acute toxicity of waterborne chemicals in trout from measurements of median lethal dose and the octanol-water partition coefficient

Measurements of the median lethal dose (LD50) of chemicals to fish provide a simpler, faster and less expensive alternative to median lethal concentrations (LC50s). Since dose is controlled, LD50s may be a better basis for structure-activity relationships. However, LC50s are used extensively in aquatic hazard assessments. Therefore, LC50s and LD50s for rainbow trout of 48 organic chemicals were measured to determine whether LC50 could be predicted from LD50. A simple linear regression of log-transformed data provided the best fit (r = 0.70), but it accounted for only 50% of the variability in LC50s. LC50s were also strongly correlated with measures of molecular size [molecular weight (MW; r = −0.90) and Parachor (r = −0.89)] and of water-lipid partitioning [octanol-water partition coefficient (Kow; r = −0.73)]. Hence, inclusion of these factors in a multiple regression analysis was examined. Since MW and Parachor were negatively correlated with LD50 (r = −0.61 and −0.60, respectively), they were not true independent variables and were not used. However, there was no relationship of Kow with LD50, and a multiple regression analysis using both LD50 and Kow predicted LC50 with greater certainty (r = 0.92) than did either factor alone. Therefore, LC50 can be predicted from LD50 when partitioning of a chemical from water to lipid is recognized.