The hypermodular structure of tripartite ecological networks.

Modularity in species interaction networks usually emerges from constraints imposed by species traits, phylogenies and spatial distributions. Though extensively studied in bipartite networks (BNs) (e.g. plant-herbivore), we do not know whether these drivers connect the dynamics and structures of different interaction modes. Here, we focus on tripartite networks (TNs) (e.g. plant-herbivore-parasitoid) and formalize the concept of hypermodules-cohesive groups of species resulting from module congruence across interlinked BNs. Using an original methodology with a new algorithm-HyperMod-and appropriate null models, we found that hypermodules are widespread in a diverse database of TNs. Additionally, assessing common drivers of bipartite modularity in TNs, we show that species taxonomy (as a proxy for phylogeny) and spatial segregation are also drivers of hypermodules. Our results are evidence for structural integration between interaction modes, improving our grasp of potential cascading effects in interaction networks and their resilience to external disturbances.