Historical contingency limits adaptive diversification in a spatially structured environment

Understanding how genotype-by-environment (G × E) interactions influence evolutionary trajectories and contribute to historical contingency is key to predicting adaptation. In structured environments, populations often diversify into ecotypes. This specialization depends on ecological opportunity and also hinges on the adaptive landscape, as early beneficial mutations may restrict access to new niches unless alternative trajectories or compensatory mutations arise. Previous studies demonstrated that <i>Escherichia coli</i> populations rapidly diversify into two coexisting ecotypes in nutrient-rich, spatially structured environments, mediated by first-step mutations that upregulate type 1 fimbriae, a pilus involved in biofilm formation that enables surface colonization. Here, we investigated how first-step mutations shape evolutionary trajectories by experimentally evolving wild-type and fimbrial-deficient (Δ<i>fimA</i>) <i>E. coli</i> in structured and unstructured environments. In structured environments, <i>ΔfimA</i> initially confers a fitness benefit by eliminating the energetic cost of weak biofilm formation, but ultimately prevents range expansion, constraining adaptation relative to wild-type populations. In unstructured environments, where biofilms provide no advantage, both genotypes evolved similarly, with sequencing revealing parallel early mutational trajectories. Our findings provide one of the first experimental demonstrations that a single, clinically relevant first-step mutation in a non-essential gene can create an evolutionary "dead end," constraining subsequent diversification. These results highlight the ruggedness of adaptive landscapes in structured environments and show how early beneficial mutations can trap lineages on local fitness peaks, underscoring the role of G × E interactions in shaping the predictability and contingency of evolution.