A simple mathematical observation that diverse, highly connected, and strongly interacting systems cannot be stable has ensnared the imagination of ecologists for decades. Fundamentally, this paradox is so intriguing because it is central to the most persistent question in ecology: how do so many interacting species stably coexist? Without a basic understanding of the factors that contribute to ecosystem stability, we are unable to anticipate how such diverse ecosystems respond to environmental perturbations. One driver of such unpredictability may be the failure of models and theory to account for the true diversity of interaction types between organisms, which include competition, predation, positive effects, and parasitism. In this study, I examine species interaction networks that incorporate these multiple interaction types, analogous to multiplex models in the complexity literature in which networks of different interaction types are layered and linked. I explore how network architecture determines the response of these ecological multiplex networks to realistic perturbations, contrasted with the response of more common ecological networks to perturbation (e.g., food webs, mutualistic networks) for both previously published empirical interaction networks and a diversity of simulated networks.
Results/Conclusions
In this study, I show how a variety of control parameters, including the number of species in the network, the connectance among them, the strength, signs and asymmetry of interactions, influence the stability of different types of interaction networks in fundamentally different ways. In particular, I use comprehensive simulation to show how merging multiple diverse ecological networks creates new pathways for indirect interactions, generating uncertainty about the response of network to the perturbation of even a single species. Because previous work has focused on static interaction links of a single type, our ability to predict the realistic consequences of environmental change on diverse networks remains a limiting frontier in ecology. In this study, my goal is to both advance conceptual models of network behavior and guide the collection of empirical data to test such models. Predictability under climate change remains an elusive topic in ecology but multiplex networks offer a framework to capture the reality of interacting, interdependent systems and understand how their architecture contributes to response to perturbation.