Simulation-based modeling of wild blueberry pollination

Abstract

The high variability of wild (lowbush) blueberry plants in spatial and genetic structure, in combination with bee foraging behavior varying between species, and the complexity of these factors interacting over time and space, are major obstacles to understanding of pollination dynamics subject to environmental change. The bottom-up modeling paradigm provides an ideal approach to bridging the gap between known mechanisms of individual organisms and unknown spatial–temporal dynamics of pollination at the field scale. By linking empirical data to stochastically-based ecological process modeling, we present a spatially-explicit agent-based simulation model that enables exploration of how various factors, including plant spatial arrangements, outcrossing and self-pollination, bee species compositions and weather conditions, in isolation and combination, affect pollination efficiency throughout a blueberry bloom season. The firmly validated open-source model is a useful tool for hypothesis testing and theory development for wild blueberry pollination researches. Sensitivity analysis suggested that fruit set and resulting measures of productivity such as fruit mass and viable seeds per fruit were sensitive to parameterization of blueberry genotype or clone size and the amount of blueberry plant cover in a field. Fruit set due to pollen compatibility was sensitive to ovule number per flower and foraging bee density. Simulation experiments allowed us to compare bee pollination efficiencies at the bee taxon population level (honey bees, bumble bees, and native solitary bees), the effect of foraging distance from bee nest or colony site on fruit set, and test whether the mechanism of gametophytic self-incompatibility (pre- vs. post-zygotic decision making by the plant) in wild blueberry pollination at the field level matters in estimating yield.