Transgenic insecticidal crops have the potential to pose risks to non-target organisms. These risks need to be addressed as part of the environmental risk assessment that precedes the commercialization of any novel transgenic crop. An international initiative has been launched to develop a scientifically-sound, generic, and pragma tic approach to assess the risks to terrestrial non-target arthropods. The basis for this work is the widely-established and effective tiered testing approach from regulatory toxicology. The basic principles of this approach are described. These may provide guidance to countries that are currently developing their own non-target risk assessment guidelines and help to harmonize regulatory requirements in different regions.

A faunistic study investigating the potential side-effects of corn (Zea mays) genetically modified to express a truncated Cry1Ab protein derived from Bacillus thuringiensis subsp. kurstaki, on non-target arthropods was carried out under field conditions. The communities of non-target arthropods in the soil, on the leaves and flying in the crop area were monitored throughout the growing season. Water-treated, untransformed corn served as a control, and a spray application of a bacterial Bt insecticide (Delfin WG) and a synthetic insecticide (Karate Xpress) used to control the European corn borer (Ostrinia nubilalis; Lepidoptera: Pyralidae) acted as positive reference treatments. Results were analyzed using a principal response curve. Significantly lower infestations by the lepidopteran target species O. nubilalis were observed in the Bt-corn plots compared to the control. No effects of Bt-corn on the communities of soil dwelling and non-target plant dwelling arthropods were observed. A trend towards a community effect on flying arthropods was observed with lower abundance of adult Lepidoptera, flies in the families Lonchopteridae, Mycetophilidae and Syrphidae, and the hymenopteran parasitoids Ceraphronidae. Effects were weak and restricted to two sampling dates corresponding to anthesis. A short but statistically significant effect of Karate Xpress and Delfin was observed on the community of plant dwellers and a prolonged effect of Karate Xpress on the soil dwellers.

Rate-response toxicity tests on Aphidius rhopalosiphi were carried out with seven plant protection products using three different test systems. The first type of test system conformed to the standard laboratory testing guidelines and consisted of two treated glass plates fitted into a metal frame, which created an enclosure for the wasps. In the second type of test system, the plant protection products were applied to two bean-leaf disks mounted on agar filled dishes, which were fitted to a transparent plastic frame. The third type of test system consisted of potted barley plants, which were treated and covered with an acrylic cylinder. Adult wasps were exposed to the dried residues of the products for 48 h before wasp mortality was assessed. For each product and test system, the LR50 value (application rate at which 50% mortality of the wasps occurs) was determined with a Bayesian Probit analysis. Technically, rate-response testing was feasible with all three test systems, and rate-response relationships could be established. The results support a sequential testing scheme, as the LR50 values increased from 'glass plate test' to 'excised leaf test' to 'whole plant test' with all tested products. The LR50 values were 7.8-340 times higher on whole plants than on glass plates. Because of the variability of this factor, a numerical safety factor cannot be used to substitute extended laboratory testing for regulatory purposes.

The EU Plant Protection Product Directive 91/414/EEC recommends the EPPO/CoE Arthropod Natural Enemies Risk Assessment Scheme for guidance on how to conduct risk assessments for terrestrial non-target arthropods. This scheme is currently in the process of being revised by EPPO/ CoE. A major change will be the recommendation for the generation and use of ’Dose Response’ toxicity data instead of limit test data. In addition, the revised EPPO/CoE Non-target Arthropods Risk Assessment Scheme will replace the current arbitrary 30% threshold trigger value applied to limit test data, with a Hazard Quotient (HQ; = Ratio Application Rate/LC50 on Glass)), comparable to the successful approach adopted in the EPPO/CoE ’Honeybee Risk Assessment Scheme’. However, in order for this new approach to be implemented under 91/414/EEC, an appropriate regulatory HQ trigger value needs to be derived. Such an HQ trigger value has been established by calculating HQ values for the 2 recommended sensitive indicator species (T pyri andAphidius) for a wide range of products and validating opposite robust semi-field/field data. This validation indicated that an HQ trigger value of ≥ 12 forT pyri and ≥ 8 forAphidius spp., should be used to trigger higher-tier risk assessment and/or higher-tier testing for non-target arthropods. As these trigger values were validated with realistic semi-field/ field data they apply for both lethal and sub-lethal effects as well as single and multiple application scenarios. Due to the worst case assumptions used in this HQ validation analysis, no further uncertainty factors need to be applied for in-crop risk assessment. Whilst a small amount of uncertainty exists regarding the comparative sensitivity ofT pyri andAphidius spp. for off-crop non-target arthropod guilds of arthropods, this is balanced by the fact that the off-crop exposure assessment used in the HQ derivation, is at least an order of magnitude higher than that realistically likely in the field. This HQ approach and trigger value is an appropriate and conservative tool for tier 1 risk assessment, which should reduce the number of false positive results leading to unnecessary higher-tier testing.