Environmental predictors of dispersal traits across a species’ geographic range

AbstractVariation in habitat quality and quantity drive selection on dispersal traits in heterogeneous environments, but the extent to which environmental conditions predict geographic variation in dispersal is rarely evaluated. We assessed dispersal trait variation across the range of Cakile edentula var. lacustris, an annual herb that occupies beaches of the Great Lakes. Cakile edentula has dimorphic fruits that each contain one dispersive and one non‐dispersive seed. Previous work showed that plant height, branching density, and dispersive fruit wing‐loading can determine the distance that seeds disperse locally by wind, while pericarp thickness influences the distance they disperse by water. We tested if these traits vary predictably with latitude across the species’ geographic range, and if variation in dispersal characteristics can be predicted by the quality and quantity of habitat available at a site. We observed that the dispersive fruits from northern and southern populations had thinner pericarps than those from the interior of the species’ range, reflecting reduced long‐distance dispersal by water at both range limits. Plants at the northern range limit were shorter with less dense branching and lower wing‐loading than populations elsewhere in the range, suggesting that these populations have enhanced local wind dispersal. In contrast, southern populations exhibited traits with inconsistent effects on wind dispersal: plants tended to be short, which facilitates wind dispersal in C. edentula, but also had relatively higher branching density and distal segment wing‐loading that reduce wind dispersal. Geographic variation in maternal plant height and branching density was partially explained by variation in habitat quality, which declined at the species’ range limits. In addition, population differences in branching density, fruit wing‐loading, and pericarp thickness were predicted by the abundance and distribution of beach habitat. Finally, a common garden analysis recovered latitudinal patterns for the dispersal traits associated with fruits, but not those associated with maternal architecture. Thus, the geographic patterns of dispersal trait variation that we observed likely reflect responses to past selection by the distribution, abundance, and quality of habitat, strong plasticity in dispersal traits, and the effects dispersal itself has in shaping local adaptation by driving gene flow among populations.