POPULATION STRUCTURE ALONG A STEEP ENVIRONMENTAL GRADIENT: CONSEQUENCES OF FLOWERING TIME AND HABITAT VARIATION IN...

Few studies have determined how gene flow and selection interact to generate population genetic structure in heterogeneous environments. One way to identify the potential role played by natural selection is to compare patterns of spatial genetic structure between different life cycle stages and among microenvironments. We examined patterns of spatial structure in a population of the snow buttercup (Ranunculus adoneus), using both adult plants and newly emerged seedlings. The study population spans a steep environmental gradient caused by gradual melting of snow within a permanent snowbed. Early-melting sites are characterized by denser vegetation, more fertile soils, and a longer growing season than late-melting sites tens of meters away. The flowering time of R. adoneus is controlled entirely by time of snowmelt, so the contiguous population is phenologically substructured into a series of successively flowering cohorts, reducing the opportunity for direct pollen transfer between early- and late-melting sites. For four highly polymorphic enzyme loci in this tetraploid species, there was subtle, but statistically significant, genetic differentiation between early, middle, and late-melting cohorts; adults usually showed greater differentiation among snowmelt zones than did seedlings. At two loci in adults and one locus in seedlings, homozygotes were more common than predicted at Hardy-Weinberg equilibrium, even when assuming maximum levels of double reduction during meiosis. This pattern suggests the occurrence of self-fertilization and/or population substructure. To determine how spatial isolation and phenological separation each contribute to genetic substructure, we used bivariate regression models to predict the numbers of allele differences between randomly paired individuals as a function of meters separation in space and days separation in flowering time. For newly emerged seedlings, we found that spatial separation was positively associated with genetic difference, but that the additional contribution of phenological separation to genetic difference was not significant. This implies that seeds and/or pollen move effectively across the snowmelt gradient, despite differences in flowering time. As was true for seedlings, spatial separation between paired adults contributed to greater genetic difference, but for a given spatial separation, the genetic difference between adult plants was reduced by phenological separation. This result implies that postemergence selection is favoring at least some seeds that migrate across the snowmelt gradient. Directional gene flow across the snowmelt gradient probably results from a genetic source-sink interaction, that is, the colonization of ecologically marginal late-melting sites by high quality seeds produced by the larger subpopulation in early-melting sites. Effective gene flow from high to low quality microenvironments is likely to impede adaptation to late-melting locations.