Keywords

Apis mellifera; pesticides; neonicotinoids; sensitivity analysis; Markov Chain Monte Carlo

Start Date

25-6-2018 3:40 PM

End Date

25-6-2018 5:00 PM

Abstract

Honey bee (Apis mellifera) colony losses have increased in recent decades in both Europe and North America. While multiple stressors to honey bee colonies appear to be driving this decline (including disease, nutrition, genetics), direct exposure to pesticides has been identified as a factor leading to increased bee declines. The simulation model VarroaPop is currently being modified by the USDA and USEPA to predict honey bee hive dynamics in response to pesticide exposure. However, applying this model to pesticides is complicated due to a lack of parameterization information from the supporting literature for many variables, especially those related to in-hive pesticide dynamics. Here, we utilize data from a field study which measured residues of several neonicotinoid insecticides in pollen and tracked population dynamics of exposed hives to improve our estimation of colony simulation model parameters relevant to VarroaPop and the new model components related to pesticides. We use Markov Chain Monte Carlo methods to sample the probability distribution of model parameters and examine the likelihood of each parameter combination, given the field-derived population data. Through this procedure, we obtain posterior distributions which represent the most likely parameter values given a realistic neonicotinoid exposure scenario. We use these pesticide-optimized parameter distributions to run a global sensitivity analysis for the updated posteriors in order to contrast with a sensitivity analysis based on the priors. This helps determine what factors are most important in driving hive success or failure following exposure events.

Stream and Session

F3: Modelling and Decision Making Under Uncertainty

Download
COinS
 
Jun 25th, 3:40 PM Jun 25th, 5:00 PM

Parameterization and sensitivity analysis of a honey bee colony dynamics model for neonicotinoid exposure events using Markov Chain Monte Carlo methods

Honey bee (Apis mellifera) colony losses have increased in recent decades in both Europe and North America. While multiple stressors to honey bee colonies appear to be driving this decline (including disease, nutrition, genetics), direct exposure to pesticides has been identified as a factor leading to increased bee declines. The simulation model VarroaPop is currently being modified by the USDA and USEPA to predict honey bee hive dynamics in response to pesticide exposure. However, applying this model to pesticides is complicated due to a lack of parameterization information from the supporting literature for many variables, especially those related to in-hive pesticide dynamics. Here, we utilize data from a field study which measured residues of several neonicotinoid insecticides in pollen and tracked population dynamics of exposed hives to improve our estimation of colony simulation model parameters relevant to VarroaPop and the new model components related to pesticides. We use Markov Chain Monte Carlo methods to sample the probability distribution of model parameters and examine the likelihood of each parameter combination, given the field-derived population data. Through this procedure, we obtain posterior distributions which represent the most likely parameter values given a realistic neonicotinoid exposure scenario. We use these pesticide-optimized parameter distributions to run a global sensitivity analysis for the updated posteriors in order to contrast with a sensitivity analysis based on the priors. This helps determine what factors are most important in driving hive success or failure following exposure events.