Frequency-dependent fitness effects are ubiquitous in rapidly adapting microbial populations

In simple microbial populations, the fitness effects of selected mutations are typically assumed to be constant, regardless of mutant frequency. This assumption underpins predictions about evolutionary dynamics, epistasis, and the maintenance of genetic diversity. Here, we systematically test this assumption using beneficial mutations from early generations of the Escherichia coli long-term evolution experiment (LTEE). Using flow cytometry-based competition assays, we find that frequency-dependent fitness effects are the norm rather than the exception, occurring in ∼80% of strain pairs tested. Most competitions exhibit negative frequency dependence, where fitness advantages decline as mutant frequency increases. We demonstrate that the strength of frequency dependence is predictable from invasion fitness measurements, which explain approximately half of the biological variation in frequency-dependent slopes. We also observe violations of fitness transitivity in several strain combinations, indicating that competitive outcomes cannot always be predicted from fitness measured against a single reference strain. High-resolution measurements reveal that frequency dependence changes substantially within the growth cycle, with the net per-cycle effect reflecting the balance of opposing dynamics at different growth phases. Our results demonstrate that, even in a model system designed to minimize ecological complexity, subtle interactions between closely related genotypes create frequency-dependent selection that can fundamentally alter evolutionary trajectories.