Using a bottom-up model of plant-pollinator interaction networks to investigate the role of foraging decisions v.s. topological constraints in determining network structure

Background/Question/Methods

Network approaches to describing trophic and mutualistic interactions within ecological communities have emphasized the importance of network topology and adaptive foraging in determining species persistence, stability, and coevolution. Most models of ecological networks have taken a top-down approach, where network topology is generated using simple constraints such as total connectance, species richness, and some mechanism form niche hierarchy; species persistence with and without adaptive foraging is then modelled within this topology. This approach implicitly assumes that network topology is entirely the product of initial topological constraints, particularly niche hierarchy. For many communities, especially mutualists such as plants and their animal pollinators, mechanisms for niche hierarchy are complex or unclear, and foraging biology provides the most grounded approach to modelling species interactions. In contrast to previous top-down approaches, I ask 1) whether ‘realistic’ network topologies emerge from a simple foraging model without topological constraints, and 2) whether topological constraints influence network structure more strongly than foraging behaviour when they are included? I present a bottom-up simulation model of mutualistic plant-pollinator interaction networks where pollinator foraging decisions are modelled using the diet choice algorithm for optimal foraging. Pollinators’ foraging decisions respond to daily and seasonal changes in plant species’ flowering dynamics, which are modelled using a quantitative genetic framework for flowering phenology. Topological constraints on network structure are introduced via the attack rate parameter in the foraging model, which serves as a phenomenological limit to network connectance, and through species specific parameterizations of the foraging algorithm.

Results/Conclusions

The model successfully reproduces ‘realistic’ nested and compartmentalized network structures without incorporating topological constraints. As network topology is constrained to be increasingly sparse through the attack rate parameter, foraging decisions become less important and eventually trivial in determining network structure. Introducing topological constraints through species specific parameterization of the foraging algorithm result in more strongly nested topologies, confirming that when when niche hierarchy is enforced, nested topologies follow. Resulting network structures were sensitive to parameter values for species pairwise energy conversion efficiency and handling time, regardless of whether topological constraints were included. Specifically, strongly right skewed distributions for these parameters yield more nested and compartmentalized networks. Future studies exploring mechanisms for foraging decisions that yield ‘realistic’ network structures will be informative and complementary to those employing increasingly complex topological constraints.