AN EVOLUTIONARY FRAMEWORK FOR IDENTIFYING DETERMINANTS OF COLONIZATION IN HUMAN MICROBIOMES Rational probiotic therapies for our gut, skin, and other microbiomes have the potential to treat a wide range of diseases and promote wellness. However, we still lack basic knowledge about how bacterial strains colonize in the microbiome, such that it is difficult to predict if a particular bacterial strain will colonize an individual. Engineering human microbiomes will require human-specific knowledge that is difficult to infer from in vitro and animal models. My central vision is to use modeling and reconstruction of within-person evolution to infer recent bacterial natural history in human microbiomes with high precision. We will develop computational tools for high-through analysis of single-organism genomes and strain- level inference in metagenomic data. Using a combination of these tools and in-human studies, we will understand the colonization dynamics and on-person spreading of many commensal strains, determine the drivers and molecular strategies of within-person bacterial adaptation, and determine the degree to which within-person selection is person-specific. We will also investigate the connection between within-person evolution and community stability. This novel population-genetics paradigm will bound expectations for interpreting metagenomic data and offer clues on how to precisely manipulate human microbiomes, with the potential to lay the groundwork for the selection of probiotic strains with the highest chance of engraftment.