Investigating Interspecies Correlation Estimation (ICE) models to predict chronic toxicity of insecticides and...

Interspecies Correlation Estimations (ICEs) are typically limited to predicting acute toxicity values between species that live in the same habitats. In this paper, we investigate the relationship between an aquatic and a terrestrial insect with respect to their chronic sensitivity to insecticides and fungicides. Chronic toxicity values, i.e., No Observed Effect Concentration (NOEC in µg active substance/L) for the non-biting midge Chironomus riparius, and No Observed Effect Dose (NOED in µg active substance/larva) for the honey bee Apis mellifera, were obtained from various databases, regulatory documents, and the open literature. Most of the data originated from regulatory studies. The derived correlation equations for the two insect species are based on bound endpoints from a total of 52 substances (25 insecticides, 27 fungicides; unbound endpoints could not be used). Applying the statistical criteria for identifying robust ICE models, two models were identified: One for insecticides and one for insecticides and fungicides combined, while fungicides alone did not fulfil the criteria. The obtained models can be used to predict honey bee larval chronic toxicity (NOED) from non-biting midge chronic toxicity data (NOEC) and vice versa. The developed ICE models are unique for two reasons: i) They predict across different habitats (i.e., aquatic and terrestrial), and ii) they are applicable for chronic toxicity. The models can provide toxicity estimations without experimental testing, which is especially helpful in the case of honey bees, which cannot be tested all year round.