COS 117-10
A quantitative framework to understand and predict the spatial structure of ecological networks from various sources of information
Integrating network theory to biogeography is among the most important and exciting challenges that macroecologists are currently facing, yet the idea that species interactions have a biogeographical structure of their own is often overlooked. Achieving this integration is necessary to progress towards understanding species interactions through time and space, and doing so to predict species geographical distributions and community dynamics. The challenge is however quite high given the amount of work required to measure species interactions in the field. Moreover, traditional field-based approaches to reconstruct networks cannot be used to assess the interactions between species that actually never co-occurred but are likely in novel communities. We propose a quantitative framework to infer the structure of local interaction networks. We start from the observation that local interaction networks are never simple random samples of a regional and stationary meta-network. We then develop a probabilistic approach to infer interactions in this meta-network, accounting for species traits, co-occurrence and phylogenetic relationships
Results/Conclusions
We find, not surprisingly, that spatial variation in community composition is the first factor responsible for the spatial variation in network structure. This variation is not random however, it is strongly driven by species-specific responses to variation in the environment. Secondly, we propose a general method to infer interactions in the meta-network from species traits and incomplete information about interactions. The framework we propose has potential practical implications such as the proposition of sampling guidelines for the inference of network structure, development of species distribution models accounting for biotic interactions and scenarios of future community structure following global changes.