Biomathematics Seminar Series - Michael Yodzis (Sept. 22, 2026)

Speaker: Michael Yodzis (Department of Integrative Biology)

Title: Using Linear Inverse Methods to Study Food Web Energy Flows and Whole Community Responses to Perturbation

Abstract:

Humans depend on ecosystems for many of the food and resources that we consume. Food webs are networks of predator-prey interactions among species, and their interaction strengths quantify the flow of energy in the network. Generalized Lotka–Volterra (GLV) models are often used to formalize these dynamics by connecting interaction strengths to species’ vital rates of growth, mortality, and biomass accumulation. For resource managers, interaction strengths are critical for understanding food web stability and system-wide responses to perturbation, yet their empirical estimation remains difficult and highly uncertain across spatial and temporal scales. Recent work by Gellner et al. introduces a variation of the Linear Inverse Method (LIM) using Markov Chain Monte Carlo (MCMC) sampling to estimate distributions of interaction strengths within feasibly- and energetically-constrained GLV models [1]. Building on this approach, we use LIM-based sampling and press perturbations to quantify how uncertainty in interaction strengths propagates to whole-community responses under sustained environmental or anthropogenic change. This framework provides a probabilistic foundation for assessing ecosystem vulnerability and resilience, offering more robust guidance for ecological management under uncertainty. We present simulated examples motivated by policy-relevant questions in Great Lakes fisheries.

References:

[1] Gellner, G., McCann, K., and Hastings, A. (2023). Stable diverse food webs become more common when interactions are more biologically constrained. Proc. Natl. Acad. Sci. U.S.A., 120(31), e2212061120. https://www.pnas.org/doi/10.1073/pnas.2212061120