Floristic mapping through bee pollen: an individual-based modelling approach linking the composition of bee pollen loads and the space-time distribution of floral resources

Background/Question/Methods

The composition of honey bees' corbicular pollen loads contains information about both the bees' foraging behavior and the surrounding floral landscape. Yet, despite a number of studies reporting frequencies of either species or genetic markers in bee-collected pollen, little attention has been given so far to the statistics of the space-time sampling performed by foraging bees. In particular, there is a need to establish quantitative relationships between the (co-)occurrence of various genetic types in bee pollen loads and the spatial distribution of corresponding floral resources. To this end, we propose a model that predicts the genetic differentiation of pollen loads (measured by F_ST) by stochastic simulation of individual foraging bouts, for a given spatial genetic structure and a given bee movement model.  

Results/Conclusions

We performed simulations using three representations of the spatial genetic structure: a continuous field with a global spatial autocorrelation function; a series of discrete patches with distinct genotype frequencies; or a combination of the two (discrete patches with autocorrelation within patches). Bee movement while foraging is approximated by a correlated random walk (CRW). Our results show a range of field conditions where the model outcome (predicted F_ST) depends on a single parameter of the CRW (average distance from origin). Our individual-based modelling approach also readily estimates the sampling effort (number of pollen loads) necessary to achieve a given variance in the F_ST estimate.