| Abstract: | To investigate the microgeographic spatial structure of genetic variation for quantitative traits in a natural population of Impatiens capensis, we performed a common-garden greenhouse experiment. Seedlings were collected at 10-m intervals from a 40 times 40-m permanent grid in a natural population and grown to maturity in a greenhouse. From these parents, 3 self-fertilized seed families per grid point were then grown in a randomized design in the greenhouse and scored for a variety of morphological and life-history traits. Virtually all of the traits displayed significant variation among families, and many were significantly heterogeneous among grid points, indicating microgeographic genetic differentiation on a fine spatial scale. Overall morphological divergence, measured as Mahalanobis distances between grid points, increased with geographical distance. In general, spatial autocorrelation coefficients of grid point character means were positive at 11–20 m and negative beyond 40 m, although power for significance testing was low. The first factor in a principal component analysis of grid point means was positively loaded on height-related traits and negatively loaded on total reproduction at 50 days, accounting for 31% of the variation. This factor displayed significant positive spatial autocorrelation at 11–20 m and negative autocorrelation at >40 m. The remaining factors showed no detectable spatial structuring among grid points. These differences in spatial pattern among characters suggest that forces other than drift may have influenced the genetic structure of the population. There was no evidence for density-dependent selection; seedling density was not significantly correlated with the grid point mean of any trait. |
