Moving beyond honey bees as a surrogate: modeling approaches to expand the pesticide risk assessment landscape

Honey bees (Apis mellifera) and native North American bees provide crucial pollination services that support both agricultural production and wild plant diversity. Yet they are threatened by a multitude of stressors including exposure to pesticides, and as a result many species face range declines and extinction risk. As part of the pesticide risk assessment process, the US EPA considers risk to pollinators using individual bee toxicity tests and semi-field colony feeding studies. But because these methods are poorly developed for native bee species, honey bees are often used as a surrogate species. However, the diverse life histories of native solitary and eusocial bees may lead to key differences in pesticide exposures and impacts at the individual, colony, or population levels. To bridge this gap, EPA’s Office of Research and Development is developing modeling methods that allow for extrapolation from honey bee experimental data to other bee species in ecologically relevant scenarios and habitat areas. Here, we review past, current, and future efforts at EPA that use modeling to expand the pesticide risk assessment landscape such as: agent-based models that simulate bee colony and/or population dynamics in response to stressors, Bayesian model inference methods for fitting models to data, species sensitivity distributions, and spatiotemporal modeling of pesticide exposures in pollinator landscapes. Collectively, this research advances the Agency’s ability to protect a wide range of critical pollinator species and the ecosystem services they provide.