Evaluation of honey bee larvae data: sensitivity to PPPs and impact analysis of EFSA Bee GD

In addition to other assessments, the EFSA bee guidance document (2013) requires the risk assessment of plant protection products on honey bee larvae. At the time the EFSA GD was finalized, no data on honey bee larvae were available due to absence of suitable methods. That is why in 2013 the European Crop Protection Association (ECPA) perfomed an impact analysis of the new EFSA risk assessment, using extrapolated endpoints derived from acute oral honey bee endpoints. Today, a number of honey bee larvae toxicity studies (138 active substances or formulated products) have been conducted according to the newly developed testing methods for single exposure (OECD TG 237) repeated exposure studies until the end of the larval development (D7/D8) and repeated exposure testing (OECD GD 239) until adult hatch (D22). These experimental data have been used to determine the ‘pass rates’ for 215 worst case uses (72 fungicide spray and solid uses, 91 herbicide spray uses, incl. 8 PGR uses and in total 52 insecticide spray and solid uses, incl. 2 nematicide and 3 IGR uses) according to the EFSA Bee GD and to compare with the original ECPA impact analysis. As standardized test methods for non- Apis bees larvae were not available, risk assessment according to EFSA for bumblebees and solitary bees based on the honey bee endpoint as surrogate corrected by a safety factor of 10. Morevoer, the sensitivity of the NOEDs at D8 and D22 in repeated exposure (D 22) studies were analysed. Overall, the toxicity of fungicides and herbicides to honey bee larvae (expressed as means and medians of NOED and LD50 values) was moderate to low, while insecticides as expected displayed stronger toxicity. Moreover, the endpoints for herbicides were on average a factor of 2 higher than fungicides which ranges within the normal biological variability (factor of 3). In addition, it is unclear, if the difference is related to a slightly higher toxicity or other factors like different physical chemical properties (e.g. lower solubility). For insecticides, toxicity was about 125 (based on medians) and 6 to 8 (based on means) times higher than herbicides. In the screening risk assessment according to EFSA Bee GD the majority of fungicide (83.3%) and herbicide (95.6%) uses passed the risk assessment for larvae; whereas, for all insecticide uses thr pass rate was about 29%. In the Tier 1 risk assessment, these pass rates slightly increased and were even higher in the ‘treated crop’ and ‘weed in the field’ scenarios for fungicide and herbicide uses, almost being 100%. Pass rates for insecticide uses did not improve very much and amounted to be about 42% for both scenarios. When basing the risk assessment of bumblebee and solitary bee larvae on 1/10th of the honey bee larval endpoint, the majority of active substances and their respective products will fail the screening (overall about 96%) and Tier 1 risk assessment (overall about 90%). Alternative risk assessment approaches proposed by ECPA (e.g. following the EPPO approach; ECPA Option 1 using refinement options and more representative assumptions) or comparing an assummed exposure concentration to the NOEC (ECPA Option 2) led to a slight increase (Option 1) or even no differences in the pass rates (Option 2a) compared to EFSA Tier 1 risk assessment. Thus both, the standard risk assessment according to the EFSA Bee GD as well as the alternative ECPA Option 1 and 2 result in a clear distinction between products with high toxicity (insecticides) vs. non-toxic products (herbicides and fungicides) for the honey bee risk assessment. The sensitivity analysis of repeated exposure studies according OECD GD 239 indicated that in most cases toxicity did not increase during the pupation period between D8 and D22. Thus, the larval growing period between D3 and D8 represents the most sensitive period of the pre-imaginal development.