UC Santa Cruz

Patterns of species interactions are both the cause and consequence of ecosystem dy- namics. Understanding the origin and function of specific interaction patterns at the ecosystem scale has been a long-term goal in ecology. These efforts are often limited by the enormous size of biological systems, the temporal transience of ecological inter- actions, difficulties in obtaining reliable measurements [3], knowing what is important to measure in the first place, and the time-scale over which observations are made. In the following chapters, I first introduce a probabilistic framework to incorporate field- measured (rather than experimental) strengths of interactions between species using stable isotope data. This framework provides a means to examine whether different variance structures are predictive of specific interaction patterns, such as nestedness and modularity. Secondly, to assess the impact of global climatic perturbations on mammalian communities over long time scales, I use stable isotope ratios of predators and prey to examine six independent paleontological communities ranging from Europe to Beringia and spanning the Last Glacial Maximum. Both the temporal and spatial evolution of species-specific relationships, as well as community-scale structures, are investigated to understand how changes in climate and prey abundance [99] influenced trophic interactions in the late Pleistocene. Although Chapters 1 and 2 concern large- scale emergent properties of food-webs, I introduce in Chapter 3 a process-based model designed to investigate the effects of mechanical constraints on the foraging strategies of anthropoid primates, and early hominins in particular. Although this model serves primarily as a predictive tool, towards the end of Chapter 3 I discuss how it can be used instead to inform independent estimates of diet, which may be particularly useful in a paleontological context where data are limited. Finally, in Chapter 4, I extend upon this reasoning and introduce a method by which resource availability data and mixing space geometry can be used to update estimates of trophic interactions from Bayesian isotope mixing models, and demonstrate its utility using data from a New Zealand intertidal community. The four chapters presented here introduce techniques and frameworks by which stable isotopes, statistical mechanics of networks, and process-based models can be used to investigate both the formation and time evolution of the patterns of species interactions in ecological communities.