ABSTRACT
211Bees are important pollinators, but both honey bees and wild bees are rapidly declining. One of the drivers for the decline of wild bees is land-use change, which affects the field mosaic and the fragmentation of the landscape. There is no general consensus in the literature about how wild bees respond to landscape configuration. Up to now, there are no simulation models that compare the performances of different solitary bees at the landscape scale. Body-size-related traits affect individual foraging behavior. We have therefore developed a spatially explicit individual-based simulation model, SOLBEE, with different bee traits and behavioral rules, which mimics the behavior and movement of pollen-collecting solitary bees. The model landscape is a square kilometer in size and consists of many patches with foraging habitat (varying amount) separated by inhospitable matrix (varying fragmentation levels). The foraging habitat has patch attributes such as flower density and minimum patch size. The bees differ in size (and in derived foraging traits such as velocity), nesting preference, and nest location. The system is further characterized by a timeframe of a single foraging day (with time steps of 1 s) in which bees forage and compete with each other for pollen. Bee numbers (and density within the landscape) are scaled with present foraging resources and body size. During the day, these central-place foragers displace pollen from flowers (distributed over a mosaic of patches) to their nest. The main goal is to compare how bees perform in terms of fitness (brood cells) and pollination services (number of flowers visited, foraging habitat visitation, and foraging distance) within a foraging day. Parameterization of the model input is mainly based on literature review, and the model’s rule behavior was improved with a pattern-oriented approach.
In several examples, we present four simulation experiments to investigate parameter effects and parameter sensitivity (Section 10.4). We show that the model 212produces realistic foraging behavior progressively during the foraging day and that the responses overlap well with values from the literature. This can be considered as a validation with the exception that model bees were found to be somewhat more efficient than real bees, yielding higher numbers of brood cells. A global sensitivity analysis of the parameters in their biological range revealed that the amount of pollen per flower (i.e., flower size) most influenced the number of brood cells and foraging habitat visitation. Body size was the dominant parameter for the number of visited flowers and the mean flown distance. The remaining two simulation experiments showed that the model can be considered robust against small changes (10% change in parameter value). The parameters that affected the responses here most were handling time per flower from the bee-related parameters and the amount of pollen per flower and pollen limitation (dispensing mechanism of the flower) from the landscape-related parameters. The simulation experiments yielded basic understanding of the model. Time constraints are more important for solitary bees than foraging resources, as they affect wood-nesting bees, which nest at field edges, more than soil-nesting bees.
