Evaluating and explaining the variability of honey bee field studies across Europe using BEEHAVE.

To assess the effect of plant protection products on pollinator colonies, the higher tier of environmental risk assessment (ERA) for managed honey bee colonies and other pollinators, is in need of a mechanistic effect model. Such models are seen as a promising solution to the shortcomings, which empirical risk assessment can only overcome to a certain degree. A recent assessment of 40 models conducted by EFSA revealed that BEEHAVE is currently the only publicly available mechanistic honey bee model that has the potential to be accepted for ERA purposes. A concern towards the use of this model is a lack of model validation against empirical data, spanning field studies conducted in different regions of Europe and covering the variability in colony conditions and environmental conditions. Here, we fill this gap with a BEEHAVE validation study against 66 control colonies of field studies conducted across Germany, Hungary, and the UK. Our study implements realistic initial colony size and landscape structure to consider foraging options. Overall, the temporal pattern of colony strength is predicted well. Some discrepancies between experimental data and prediction outcomes are explained by assumptions made for model parameterisation. Complementary to the recent EFSA study using BEEHAVE our validation covers a large variability in colony conditions and environmental impacts representing the Northern and Central European regulatory zone. Thus, we believe that BEEHAVE can be used to serve the development of Specific Protection Goals and the development of simulation scenarios for the European Regulatory Zone. Subsequently, the model can be applied as a standard tool for higher tier ERA of managed honey bees using the mechanistic ecotoxicological module for BEEHAVE: BEEHAVE ecotox .