P> The evolution of many floral traits is thought to have been shaped by pollinator-mediated selection for increased attractiveness or an improved mechanical fit of flowers to pollinators. Yet, few studies have examined experimentally the independent and interactive effects of different aspects of flower morphology on plant reproductive success. In the orchid Platanthera bifolia, perianth size and spur length are positively correlated within and among populations. To test the hypothesis that pollination success and seed output increases with increasing perianth size and spur length, we manipulated the two traits in a factorial design in a long-spurred population of P. bifolia pollinated by long-tongued hawkmoths. Additionally, to determine whether differences in selfing rate can explain variation in fruit set and fruit size, we performed controlled self- and cross-pollination. Plants with long spurs had more flowers pollinated, more pollen removed and produced more and larger fruits compared to plants with short spurs. In contrast, perianth size did not affect the pollination success or fruit production of P. bifolia. Fruit production and fruit size did not differ among flowers pollinated with self- and cross pollen, respectively. This indicates that reduced pollen deposition rather than pollinator-mediated self-pollination caused the reduction in fruit set and fruit volume observed after shortening of the spur. The results demonstrate that spur length, but not perianth size, is critical for reproductive success in P. bifolia, and suggest that among-population differentiation in perianth size may reflect a correlated response to selection on spur length. The results are consistent with the hypothesis that visual display is less important than other cues for the reproductive success of P. bifolia, and underscore the necessity to experimentally examine the functional significance of putatively adaptive traits.
