Phenotypic divergence persists despite substantial gene flow in a Mimulus guttatus metapopulation.

Local adaptation at fine spatial scales depends on whether selection can maintain trait divergence despite ongoing gene flow, yet the spatial scale at which this occurs remains poorly resolved in natural populations. Heterogeneous landscapes containing geographically proximate populations provide natural experiments to test how gene flow and selection jointly shape adaptive divergence. We studied 12 annual populations of Mimulus guttatus distributed across two nearby ridges in the Cascade Mountains of Oregon, USA, and combined field fitness estimates, common garden phenotyping, and genome-wide SNP data to assess genetic variation, quantify phenotypic divergence, and test whether contemporary selection explains observed population differences. Populations had high levels of genetic diversity yet exhibited clear population structure between mountain ridges. Phenotypic divergence among populations exceeded expectations from genetic structure for several traits including flowering time, stem height, and trichome number. Earlier flowering time and larger flower sizes were associated with hotter and drier populations, matching patterns of temporally-fluctuating selection in a nearby population. Phenotypic selection analyses indicated that contemporary selection gradients were generally weak or non-significant, indicating that past selection likely produced variation between populations. Together, our results demonstrate that strong phenotypic differentiation can arise at microgeographic scales despite substantial gene flow and temporal heterogeneity in environments.