Addressing declining pollinator populations through new mathematics

Original Announcement

Insect pollination is vital to terrestrial ecosystems and agriculture, with honeybee pollination in the United States alone valued at over \$12 billion annually. Yet, substantial evidence shows that pollinator populations are in sharp decline, with honeybee colony losses reaching unprecedented levels — threatening agricultural sustainability and food security. These declines result from complex interactions among environmental factors (e.g., shifting climate patterns) and biological processes (e.g., exposure to agrochemicals such as insecticides and fungicides, parasitic infections, diseases, and habitat degradation).

Traditional models often struggle to represent the multiple, interacting factors operating across diverse spatial and temporal scales, and they frequently lack integration with empirical data for robust validation and parameterization. Addressing this complexity requires new mathematical approaches — particularly in dynamical systems, optimal control, reinforcement learning, and hybrid symbolic–data-driven modeling — capable of capturing intricate feedbacks and guiding actionable solutions.

This workshop will bring together mathematicians, biologists, environmental scientists, and beekeepers to develop biologically realistic, predictive models that reflect the intricacies of pollinator health, especially honeybees. The program will focus on three main themes:

1. Mathematical Modeling of Complex Pollinator–Environment Dynamics
   • Develop high-dimensional, nonlinear, and non-smooth models capturing the interplay of environmental change, agrochemical exposure, disease, and habitat fragmentation.
   • Incorporate delay differential equations, state-dependent delays, and stochastic processes to represent life stage transitions and environmental variability.
   • Integrate spatially explicit PDEs and fractional diffusion models for pollinator movement in heterogeneous landscapes.
2. Data-Driven Optimization and AI-Integrated Approaches
   • Apply reinforcement learning, Bayesian inference, and deep neural networks to predict colony collapse thresholds and adaptive behaviors.
   • Design real-time, data-informed management strategies for agrochemical application, parasite/disease control, and overwintering practices.
   • Fuse multi-scale climate variability data, environmental metrics, and biological observations to improve model parameterization and predictive accuracy.
3. Decision Support and Sustainable Management Strategies
   • Develop optimal control frameworks to balance ecological resilience with agricultural productivity.
   • Model synergistic effects of stressors to inform policy, conservation strategies, and precision beekeeping.
   • Create transferable decision-support tools that connect model outcomes to stakeholder needs, from commercial beekeepers to policymakers.

Material from the workshop

The workshop schedule.