Pesticides and phytoseiid mites: strategies for risk assessment

Laboratory toxicity trials may predict effects of chemicals under field conditions, but errors are inevitable. A chemical may be presumed harmless when in fact it has a detrimental effect, or it may appear highly toxic in the laboratory, but not in the field. Error rates depend on experimental setups, evaluation criteria, and ecological attributes, such as dietary range, of the organisms under study. The authors analyze results of standardized toxicity studies of pesticides on four species of predatory mites and assess the feasibility of drawing accurate conclusions from laboratory trials alone. This is by contrasting laboratory and field data, while varying interpretation criteria. At a 5% critical error rate, it was found that correspondence between lab and field experiments is only obtained for products harmless to Typhlodromus pyri. For this species these constitute only 30% of the total number of products in our database. Outcomes from lab tests with Amblyseius andersoni correspond with field results (for A. andersoni and A. finlandicus) either for products yielding harmless or for products yielding harmful side effects. The decision rules required to reach either classification are not compatible and hinge on field thresholds that may be unrealistic. For Phytoseiulus persimilis only harmful insecticides and harmless fungicides enabled the setting of decision rules that resulted in correspondence between lab and field trials of more than 95%. Why these species require different interpretation criteria is discussed together with suggestions for improvement of existing test protocols and the feasibility of using indicator species.