Trait-based inference of ecological network assembly: A conceptual framework and methodological toolbox

The study of ecological networks has progressively evolved from a mostly descriptive science to one that attempts to elucidate the processes governing the emerging structure of multitrophic communities. To move forward, we propose a conceptual framework using trait-based inference of ecological processes to improve our understanding of network assembly and our ability to predict network reassembly amid global change. The framework formalizes the view that network assembly is governed by processes shaping the composition of resource and consumer communities within trophic levels and those dictating species’ interactions between trophic levels. To illustrate the framework and show its applicability, we (1) use simulations to explore network structures emerging from the interactions of these assembly processes, (2) develop a null model approach to infer the processes underlying network assembly from observational data, and (3) use the null model approach to quantify the relative influence of bottom-up (resource-driven) and top-down (consumer-driven) assembly modes on plant–frugivore networks along an elevational gradient. Simulations suggest that assembly processes governing the formation of pairwise interactions have a greater influence on network structure than those governing the composition of communities within trophic levels. Our case study further shows that the mode of network assembly along the gradient is mainly bottom-up controlled, suggesting that the filtering of plant traits has a larger effect on network structure relative to the filtering of frugivore traits. Combined with increasingly available trait and interaction data, the framework provides a timely toolbox to infer assembly processes operating within and between trophic levels and to test competing hypotheses about the assembly mode of resource–consumer networks along environmental gradients and among biogeographic regions. It is a step toward a more process-based network ecology and complete integration of multitrophic interactions in the prediction of future biodiversity.