Emergent simplicity in microbial ecosystems

Microbial communities can be incredibly complex, harboring hundreds of interacting species. Modern sequencing-based technologies allow resolving their composition to strain-level detail; however, predictive modeling at this resolution appears almost hopeless. An exciting interface area between ecology and statistical physics is developing an alternative approach, investigating community-level properties that might be predictable despite this microscopic complexity. I will describe our recent results suggesting that, surprisingly, higher community diversity can render simple models more predictive, rather than less. This suggests new strategies for functional predictions and community design that work because of diversity, not despite it.