Effect of Marine Reference on Inferred Evolutionary Patterns of Freshwater Stickleback

Threespine stickleback are a model system for studying rapid and parallel evolution. Studies characterizing freshwater evolution often use contemporary marines as an ancestral proxy, an approach that relies on untested assumptions about the lack of phenotypic variance in these marine fish. Here, we survey marine individuals collected from several sites, asking whether there is evidence of phenotypic variation. We identified considerable phenotypic variation among fish from different sites. Thus, we investigated the impact of this phenotypic variance on the inferred pattern of freshwater evolution. We tested whether estimates of the magnitude of phenotypic divergence or parallelism were affected by the choice of marine reference. We found that for freshwater populations, the magnitude of phenotypic divergence was dependent on marine sampling location-with divergence estimates differing by up to 65% with the substitution of marine reference site. Geographic distance and environmental similarity between marine and freshwater sites explained some of the variance in these divergence estimates. In contrast, across marine sites, neither geographic distance nor environmental similarity predicted morphological similarity, suggesting other factors drive morphological divergence among marine fish. The magnitude of phenotypic parallelism, estimated using a multivariate vector-based approach, also differed significantly depending on the marine reference used. Together these results suggest that the choice of marine reference population, particularly its geographic distance from the focal population, is an important consideration when trying to characterize patterns of evolution in freshwater stickleback.