Structure and dynamics of ecological networks with multiple interaction types

Organisms survive and reproduce by interacting with individuals of their own and of other species. Biotic interactions are extremely diverse in type, magnitude, or scale, and give rise to ecological networks with complex topologies. Such biotic networks have been shown to possess structural properties that enhance their resilience and robustness to perturbations, and thus are key elements for understanding community responses to disturbances such as environmental change or habitat loss. Despite the importance of interaction networks in ecological studies, and due in part to their variability, ecological interactions are notoriously difficult to document and quantify. Therefore, studies of ecological networks have focused on the most easily observable interactions, those between predators and their prey. In the last decades, studies of mutualistic networks have also risen to prominence and have demonstrated, for example, that food webs and mutualistic networks have markedly different topologies and this has ecological and evolutionary consequences for the species involved. One of the main challenges of contemporary community ecology is to expand our understanding of networks of a single interaction type to a more realistic view of ecological communities, by considering how different interactions mutually influence community structure and functioning. In order to tackle this challenge, a first step is to lay down overarching theoretical hypotheses about such complex networks. In this thesis I approach this general objective and analyse a series of fundamental questions about ecological networks. First I synthesise current methodologies for developing theoretical network models. I find that three main conceptual approaches have been used, and discuss their relative strengths, weaknesses, and potential uses. Second, I study whether species persistence in model communities is influenced by the frequency and distribution of the different interaction types. The prevalence of positive interactions within a community is shown to be key for species-poor communities, whereas in more speciose communities, different combinations of interactions can occur without affecting species persistence in a significant way. Furthermore, networks with randomly distributed interactions show less persistence than structured networks. In the fourth chapter, I focus on Species Abundance Distributions (SADs), one of the most studied patterns in community ecology, and ask whether their shape varies consistently for the different trophic guilds of a community. I compare theoretical expectations with SADs from empirical datasets, and find that SADs of plant communities are significantly less even and more skewed than SADs from mammal ones. Among mammal trophic guilds, there are no significant differences in the evenness or skewness of their SADs. I also find that increasing species degree significantly increases SAD evenness in model communities. In the fifth chapter, I incorporate another degree of complexity, namely the spatial perspective. Specifically, I analyse how interaction effects are propagated in space, such that interactions occurring in a local community may influence other communities connected to it by dispersing or foraging individuals. In this chapter, I find that the distribution of net effects of a species over another across the metacommunity is significantly different if the local communities are connected by dispersal, foraging, or a mixture of both. In the sixth chapter, I tackle the long-standing question of the variability of species interactions across environmental gradients. For approaching this question, I propose to differentiate non-resource and resource environmental factors. I analyse the prevalence of positive and negative interactions in model communities across a two-dimensional environmental gradient with one resource and one non-resource factor, and find that, according to the expectations, positive interactions respond to the non-resource factor, whereas negative interactions vary across the two axis of the gradient, with consequences for average persistence time and species diversity across the combined gradient.