You are here:
Getting BeeHave to Behave: Extending and Calibrating an Agent-Based Model of Honey Bee Dynamics for Pesticide Exposure Data Using Approximate Bayesian Computation (ABC)
Citation:
Dawson, D., J. Minucci, AND Tom Purucker. Getting BeeHave to Behave: Extending and Calibrating an Agent-Based Model of Honey Bee Dynamics for Pesticide Exposure Data Using Approximate Bayesian Computation (ABC). Society of Environmental Toxicology and Chemistry (SETAC) Meeting 2021, Virtual, NC, November 14 - 18, 2021. https://doi.org/10.23645/epacomptox.17331212
Impact/Purpose:
Presentation to the Society of Environmental Toxicology and Chemistry (SETAC) Meeting November 2021
Description:
Pesticide impacts to colonies of honeybees (Apis mellifera) are of increasing concern due to the importance of honeybees as crop pollinators. Honeybees have a multi-stage life history driven by a complex social system and make use of both fresh and stored food resources. This complicates predicting the influence of pesticide exposures on honeybee colonies and makes development of regulatory guidance challenging. Because of the complexity of honeybee social structure, agent-based modeling has been an attractive way to model stressors on colony dynamics. The agent-based model BeeHave was previously extended to model exposure to pesticides in pollen. In this study, we further extended BeeHave to incorporate exposure to pesticides via the nectar pathway. In addition, we used data collected from a field-based colony feeding study with the pesticide clothianidin to calibrate the model using approximate Bayesian computation (ABC). By avoiding the need for a specific likelihood function, the ABC method is well suited to calibrating complex agent-based models. Our enhanced model reasonably reproduced colony population dynamics following pesticide exposure via nectar. The ABC process also provided posterior distributions of input parameters for the model, as well as dose-response function parameters, allowing for the prediction of several key toxicological endpoints (e.g., LD50, NOAEL, and LOAEL). This model will be useful for evaluating a range of potential exposure scenarios for honeybee colonies. Further, the model may serve as a useful comparison against ongoing regulatory colony model development efforts in the US and Europe. The views expressed here are those of the authors and do not necessarily represent the views or the policies of the U.S. Environmental Protection Agency.
URLs/Downloads:
DOI: Getting BeeHave to Behave: Extending and Calibrating an Agent-Based Model of Honey Bee Dynamics for Pesticide Exposure Data Using Approximate Bayesian Computation (ABC)
DAWSON_SETAC_2021_DED102621.PDF (PDF, NA pp, 1319.096 KB, about PDF)