Something from nothing? Predicting interaction networks across space using graph embedding
The study of species interactions in space is extremely difficult as some locations have very little documented species interaction data. In this seminar, I will summarize methodological developments about network representation that allow to learn from known interactions in order to use phylogenetic relationships to predict networks in space. I will then illustrate how knowledge of species interactions can be instrumental in our ability to monitor biodiversity across different spatial scales, including for the surveillance of zoonotic diseases.
Timothée Poisot is a professor at the department of biological sciences, Université de Montréal. His research group, the Laboratoire d’Écologie Prédictive et Interprétable pour la Crise de la Biodiversité (ÉPIC Biodiversity) focuses on the development of quantitative and computational tools to formulate actionable and explainable predictions at the interface between biodiversity loss, climate change, and the emergence of zoonotic diseases.