| dc.description.abstract | We carried out phenotypic selection analyses to assess the relative importance of several Horal traits, as well as microhabitat, in determining both male and female pollination success in a nonrewarding orchid (Cypripedium acaule Ait.) in two Nova Scotia populations. Fruit production in these one-flowered, cosexual plants was strongly pollen limited within a season: 100% of individuals set fruit following hand-pollination, but only 5% and 13% did so under natural circumstances in the two populations. Male and female reproductive success were highly correlated, with >90% of plants that received a pollinium also dispersing one. Despite very low visitation rates, the chance of a. removed pollinium fertilizing another ovary was 36% and 51%, respectively, in the two populations. There was no evidence for selection acting differently, either through the two sexual functions or in the two populations. In contrast, the strength of directional selection on several traits differed among microhabitats within each population. Univariate and multivariate regressions indicated statistically significant directional selection acting on flower-opening date through male, female, and total (combined) pollination success in one population. For female, male, and total success in this population, we also detected correlational selection acting on opening date and flower height. There was also negative correlational selection acting on flower height and labellum length through female success. Microhabitats within the populations had a larger effect on both female and male success than did floral traits In particular the presence of ericaceous shrubs and an open canopy appeared to be closely linked with higher pollination rates. | en_US |
