Low-Dimensional Organization of Metabolic Phenotypes Underlies Recurrent Evolutionary Diversification

Metabolic phenotypes vary within microbial species, yet how such variation is organized remains unclear. Diversification in carbon-source utilization often appears idiosyncratic, showing weak correspondence to phylogeny or simple gene content. Here, we combine quantitative growth phenotyping of natural Escherichia coli isolates across 32 carbon sources with diversification observed during de novo laboratory evolution, together with a reaction-level description of metabolic similarity. Despite deep phylogenetic divergence, growth-rate profiles varied independently of lineage. Instead, growth rates across carbon sources covaried in recurrent modular patterns aligned with flux-based metabolic similarity. Closely analogous modular relationships re-emerged during de novo evolution, indicating parallel diversification across evolutionary contexts. Growth-rate variation in natural and experimentally evolved datasets collapsed onto a shared low-dimensional variance structure. Together, our results indicate that quantitative metabolic phenotypes vary along a limited set of recurring, module-linked axes, providing insight into why similar metabolic states can repeatedly arise across independent evolutionary contexts.