The Coevolution of Complex Microbial Phenotypes

Microbial systems offer a fruitful setting for integrating ecological mechanisms with evolutionary processes, as phenotypic innovation often unfolds on ecological timescales. Despite rapid advances in genomics and experimentation, however, a general theory linking community structure, selection, and diversification in high-diversity systems remains incomplete. This dissertation advances that integration through two lines of modeling, one grounded in empirically observed microbial phenotypes and another abstracted from general principles of specificity. In models of microbe-virus CRISPR-mediated coevolution, density-dependent competition coupled with heritable, sequence-specific immunity generates alternating regimes of sustained host control and major viral epidemics—dynamics yet to be observed in other host–pathogen models. Predictive quantities derived from bipartite infection and tripartite escape networks anticipate these regime shifts. Moreover, when negative frequency-dependent selection from immune memory operates alongside directional selection imposed by host competitive asymmetries, their combined action enhances viral persistence and adaptation, while also giving rise to "royal family" dynamics of the microbial host population consistent with experimental observations. These results broaden the perspective on coevolutionary patterns of evolving systems with high diversity and structures, while also underscoring open questions about which ecological mechanisms stabilize viral persistence. A second line of work introduces a general consumer–niche framework in which consumers are represented by multilocus phenotypes, with binary alleles defining mutable trait axes that can be gained or lost. In this setting, the balance between intra- and interspecific competition determines the strength of negative frequency-dependent selection, which in turn promotes the evolutionary accumulation of trait axes --- an increase in community dimensionality. This framework reveals punctuated directional selection favoring the expansion of trait space, alongside emergent patterns of limiting similarity and greater invasibility with rising dimensionality. Taken together, the findings suggest that negative frequency-dependent selection --- well-known for maintaining coexistence --- can also drive the long-term expansion of phenotypic spaces, raising new questions about how genome architecture, innovation processes such as recombination and lateral gene transfer, and spatial structure shape the stability and resilience of high-dimensional communities under perturbation.