Using Bayesian methods to gain mechanistic insight from ecological data
Ecological systems are notoriously complex and yet are often analyzed with relatively simple models. While this approach is often adequate to understand the phenomenological effect of a given treatment or environmental condition, more complex models can be necessary to unravel the underlying mechanisms resulting in the data. This can be further complicated as both biological and experimental sources can contribute to the variation seen in ecological data (e.g. competitive and environmental effects vs. observational errors). In our lab, we develop Bayesian models to elucidate underlying biological processes and make predictions of future dynamics for populations and communities. As part of the NSF funded Modelscape consortium, we are developing novel methods of using large datasets to parameterize mechanistic models of evolutionary and community dynamics.