Quantifying patterns and processes of community assembly with co-occurrence analysis

Quantifying community assembly processes and understanding how communities respond to disturbance is a critical goal of ecology. Co-occurrence analysis is an increasingly popular tool that has been used to approach this problem, and quantifies how often a pair of species is found together with respect to a random baseline, usually calculated using possible pairs of species in an assemblage. Despite its potential to provide nuanced pairwise information, practical applications of this approach remain elusive. Associations calculated with cooccurrence do not demonstrate direct biotic interactions. Instead, relationships are often driven by combinations of several mechanisms, which can make associations appear statistically insignificant on the surface and difficult to interpret biologically. In the first chapter of this work, I begin with a short introduction to co-occurrence analysis and outline the challenges and potential advantages associated with it. In the second chapter, I confront these challenges by suggesting that co-occurrence data should be treated as a continuous probabilistic variable. I also demonstrate several ways co-occurrence analysis can untangle biological mechanisms when used in targeted comparative analyses. The third chapter provides an observational study of probabilistic co-occurrence networks, describing their structures and link weights using simulated and empirical examples. In the following chapters, I use two applications of the propose probabilistic framework to answer ecological questions about community responses to disturbance such as extinction and habitat alteration. In chapter 4, I study the end-Pleistocene North American mammal extinction to show how biotic and abiotic regulatory factors can be isolated as components of each co-occurrence. Moreover, I show that community reassembly of the surviving mammals after the extinction was driven by biotic interactions rather than concurrent climate change. In chapter 5, I demonstrate how co-occurrence analysis can highlight changes in functional trait distribution due to habitat alteration using site inventories of Neotropical birds and bats. The methods presented here can be extended in the future to improve our understanding of multi-taxon association networks and eventually entire ecosystems. Most of the investigated taxa showed a decrease in positive co-occurrences and an increase in negative co-occurrences in response to disturbances. These patterns suggest that anthropogenic disturbance decreases the ability of species to coexist and may reduce the ability of ecosystems to self-regulate via biotic interactions. Thus, the restoration of species interactions may require more attention in conservation and management.