Background: Populations that have repeatedly colonized novel environments are useful for studying the role of ecology in adaptive divergence - particularly if some individuals persist in the ancestral habitat. Such "contemporary ancestors" can be used to demonstrate the effects of selection by comparing phenotypic and genetic divergence between the derived population and their extant ancestors. However, evolution and demography in these "contemporary ancestors" can complicate inferences about the source (standing genetic variation, de novo mutation) and pace of adaptive divergence. Marine threespine stickleback (Gasterosteus aculeatus) have colonized freshwater environments along the Pacific coast of North America, but have also persisted in the marine environment. To what extent are marine stickleback good proxies of the ancestral condition? Results: We sequenced >5800 variant loci in over 250 marine stickleback from eight locations extending from Alaska to California, and phenotyped them for platedness and body shape. Pairwise FST varied from 0.02 to 0.18. Stickleback were divided into five genetic clusters, with a single cluster comprising stickleback from Washington to Alaska. Plate number, Eda, body shape, and candidate loci showed evidence of being under selection in the marine environment. Comparisons to a freshwater population demonstrated that candidate loci for freshwater adaptation varied depending on the choice of marine populations. Conclusions: Marine stickleback are structured into phenotypically and genetically distinct populations that have been evolving as freshwater stickleback evolved. This variation complicates their usefulness as proxies of the ancestors of freshwater populations. Lessons from stickleback may be applied to other "contemporary ancestor"-derived population studies.