Statistical models for ecological data: Applications for animal behavior and species' distributions
ABSTRACT: One primary goal of researchers studying ecological data is to infer important biological features and mechanisms that help explain why plants and animals are where they are and do what they do. One way to achieve this goal is through the careful specification and implementation of flexible and interpretable parametric statistical models. Such models provide methods for incorporating existing scientific knowledge and interrogating proposed hypotheses. Specifying useful parametric models requires synthesizing concepts from Mathematics, Statistics, and Ecology, and must balance practical demands in model flexibility, computational feasibility, and parameter interpretability.
In this talk I will describe two examples of novel parametric statistical methods developed to help answer essential questions in Ecology. The first is motivated by the study of animal movement and offers critical insights into the social behavior of dolphins exposed to naval sonar through the implementation of a model for interacting trajectories. The second is motivated by mapping the distribution of multiple tree species in a region of New Mexico experiencing dramatic shifts in wildfire intensity and frequency. The research driving both of these applications has been conducted in close collaboration with domain scientists and MS Statistics students. I will close with a brief survey of new and planned research directions.