Inferring species interactions and predicting coexistence from community endpoints

An important goal in ecology, with many potential applications, is learning patterns of species interactions from data. Most traditional approaches to this problem rely on time-series data, which can be difficult to obtain and analyze. I work on alternative statistical methods that use “snapshots” of community abundances, taken from different species assemblages formed from a community of interest. This approach can be very tractable and scalable for experimental ecological communities, and performs well for predicting coexistence or abundances of unobserved species combinations. I am currently working to extend these methods for applications with relative abundance data and using biologically-inspired constraints to analyze communities where data is scarce.

Some relevant publications:
Predicting coexistence in experimental ecological communities (2020) Nature Ecology & Evolution
Modeling ecological communities when composition is experimentally manipulated (2022) Methods in Ecology and Evolution
Phylogeny structures species’ interactions in experimental ecological communities (pre-print)