Projects

Architecture and control of stratified networks

In a stratified network, species sit in layers, or trophic levels, and interact mostly with the layers just above and below them. Microbial food webs are one example: viruses prey on microbial hosts, the hosts consume resources, and together they set fluxes of carbon and nutrients that matter from the ocean to the gut. I model these as a predator–consumer–resource system and analyze it with the cavity method from spin-glass physics, which predicts the statistics of very large random communities without simulating each species. This relates the architecture a network realizes as it assembles (viral host range, microbe–microbe interactions) to the fluxes it produces.

Building on that, the project asks how architecture relates to a network's capacity to be controlled, at which trophic level the driver species sit, which flux states are reachable with the actuators at hand, and how much a realistically noisy picture of the network reveals about the flux we want to steer.

Eco-evolutionary drivers of hyperdiversity

Some communities hold far more coexisting types than simple niche arguments allow. This project asks how frequency-dependent competition, where a phenotype's success depends on how common it and its close competitors are, can push communities to use more niche dimensions. As crowding along existing traits intensifies, selection favors diversifying along new ones, and the dimensionality of the niche space grows along with the diversity it supports.

Immune-mediated predator–prey coevolution

CRISPR systems let microbes store fragments of viral genomes (spacers) as a heritable immune memory. Theoretical models of CRISPR-mediated coevolution demonstrate punctuated dynamics not observed in previous models of host–pathogen systems. This work follows how that memory and viral escape drive coevolution in diverse communities, including spacer lineages, escape probabilities and the paths coevolution takes.