You are here:
Comparison of honeybee colony model structure and parameter sensitivity for regulatory application
Citation:
Purucker, S., J. Minucci, R. TorneroVelez, AND D. Dawson. Comparison of honeybee colony model structure and parameter sensitivity for regulatory application. 2019 International Society for Ecological Modelling Global Conference, Salzburg, AUSTRIA, October 01 - 05, 2019.
Impact/Purpose:
Presented at the International Society for Ecological Modelling Global Conference 2019
Description:
Pesticide chemicals are increasingly engineered to have insect-specific activity. A side effect of this targeting is that some classes of modern pesticides are implicated in impacting beneficial insects, such as honey bees and wild pollinators. When testing new pesticides, traditional toxicity approaches focus on short-term experiments that yield concentrations protective of mortality endpoints. However, sublethal effects expressed at lower concentrations may be responsible for bee colony losses greater than those caused by direct mortality. Therefore, regulatory agencies increasingly request higher tier field and semi-field studies that can capture colony level effects. Interpretation of this data can be confounded by interactions with other environmental influences and stressors. To assist with these complications, two colony models have been separately developed to estimate effects at the colony level. The model BEEHAVE has implemented the conceptual honey bee colony model from the European Food Safety Authority. In addition, the USDA and USEPA have collaborated to add a pesticide exposure and effects component to the VarroaPop colony model. We perform a side-by-side comparison of the two sets of algorithms, highlighting similarities and differences in model structure and how they impact simulation results. We then challenge the models with a published higher-tier data set and implement a rejection-based Approximate Bayesian Computation as a simulation method for inference of common parameters of both models, including toxicity parameters. We then compare the posteriors for these known sensitive parameters and evaluate how the models perform versus the observations. This direct comparison yields opportunities for harmonization of emerging approaches for modeling higher-tier field data and inferring pesticide impacts on pollinators.