Overview
Ecosystems are composed of tens to thousands of species whose physiology and interactions are mediated by habitat structure, fluctuating environments, and individual trait distributions. We rarely have data on all of the relevant state variables and generally lack first principles sufficient to strongly constrain predictions of future states or inform management actions. As a consequence, I argue that there is a clear need for data-driven approaches to predicting sparsely observed ecological systems. Empirical dynamic modeling (EDM) based on time-delay embedding offers one possible solution. Here, I will describe EDM, recent extensions to non-stationary systems, and applications to prediction and management in marine ecology.
Bio
Stephan B. Munch is an evolutionary ecologist and mathematical biologist who has worked on contemporary evolution in response to harvesting, transgenerational effects of temperature on life histories, and most recently on forecasting, inference, and control in complex, sparsely observed ecosystems. He received a PhD in Coastal Oceanography from Stony Brook University, was a faculty member at Stony Brook from 2005 to 2010, and moved to Santa Cruz in 2010 where he remains a Fisheries Ecologist with the National Marine Fisheries Service and Adjunct Professor in Applied Mathematics.