Abstract
In animal-pollinated angiosperms, the "male-function" hypothesis claims that male reproductive success should benefit from large floral displays, through pollinator attraction, while female reproductive success is expected to be mainly limited by resource availability. As appealing as this theory might be, studies comparing selection strength on flower number in both sexes rarely document the expected asymmetry. This discrepancy could arise because flower number impacts both pollinator attraction and overall gamete number. In this study, we artificially manipulate floral displays to disentangle the fertility versus pollinator attraction components of selection, both in terms of mating and reproductive success. In females, flower number was under strong fertility selection, as predicted in the absence of pollen limitation. In contrast, in males, flower number was mainly under sexual selection, which in turn increased male reproductive success. However, these selection patterns were not different in males with artificially increased floral displays. This suggests that sexual selection acting on flower number in males do not occur because flower number increases pollinator attraction, but rather because more pollen is available to disperse on more mates. Our study illustrates the power of disentangling various components of selection with potentially sex-specific effects for understanding the evolution of sexual dimorphism.