Variability in the reproductive performance of beneficial insects in standard laboratory toxicity assays ‐ Implications for hazard classification of pesticides

Summary Nowadays the development of pesticides is not only directed to design compounds with sufficient efficacy against resistant and non‐resistant strains of pests as well as with a high margin of safety for man and the environment, but also to allow their potential use in IPM (Integrated Pest Management) programmes. For identification of IPM suitable pesticides, a whole battery of standardised laboratory and semi‐field test protocols have been developed. Based on the observed lethal and sublethal effects in those tests, compounds are categorised into different hazard groups which should advise farmers on their suitability for IPM. For a valid classification of compounds on the basis of laboratory test results, threshold values for lethal and sublethal effects have to be developed which reliably differentiate between harmful and safe compounds without bearing the risk of erroneous labelling and/or too frequent requests for higher tier testing. As an evaluation criterion for sublethal effects of pesticides, the reproductive performance of beneficial insects is frequently considered in standard laboratory assays. A preliminary analysis of available data on the reproductive performance of several standard test species elaborated during regulatory testing indicates that this evaluation criterion is subject to high variability. As expected, individual test species differed in their reproductive performance. Aleochara bilineata (Coleop‐tera: Staphylinidae) and Orius insidiosus (Heteroptera: Anthocoridae) showed a fairly homogeneous reproductive performance within test series. Based on the observed variability of the reproductive performance in control groups, the average probability of an erroneous labelling of pesticides for these two species was only 4% and 6% (maximum probability: 13% and 19%), respectively, when an adverse effect threshold value of 30% (= actual value of the EU (European Union)) was applied. In contrast, Coccinella septempunctata (Coleoptera: Coccinellidae), Chryso‐perla carnea (Neuroptera: Chrysopidae), Aphidius rhopalosiphi (Hymenoptera: Aphidiidae) and Trichogramma cacoeciae (Hymenoptera: Trichogrammatidae) exhibited a high variability in their reproductive performance, giving a mean probability between 25% and 35% (maximum probability: 36–64%) to label a pesticide either false positive or false negative. Besides species‐inherent variability, there was an indication that test‐inherent factors including parent sex ratio and parent breeding density may have had an influence on the reproductive performance of these insect species. Seasonal influences on the reproductive performance of the beneficial insects in laboratory testing were not evident. Based on our data analyses, there is a significant risk of erroneous classification of pesticides when the reproductive performance is quantitatively assessed following the currently established test protocols and an adverse effects threshold value of 30% is applied. We propose therefore that either the testing procedure for assessing the reproductive performance, or the effect threshold value for this evaluation criterion is reconsidered in the light of the high “species‐inherent” variability in the reproductive performance of some beneficial insects.