Outcrossing Versus Ecological Constraints in the Evolution of Dioecy

Abstract

Thomson and Barrett (1981) argue that the frequency of dioecy in self-incompatible plant lineages is a measure of the relative efficacy of ecological mechanisms versus selection for outcrossing in promoting dioecy; that there are few groups in which dioecy and self-incompatibility co-occur; and that selection for outcrossing is thus of primary importance in the evolution of dioecy. Here, I show that their argument contains several disabling assumptions, and that actual data on the distribution of angiosperm breeding systems do not support their empirical claim. The observed numbers of families and genera in which dioecy and self-incompatibility are known to co-occur do not differ significantly from those expected on the basis of the frequency of each trait. Furthermore, the incidence of dioecy in taxa displaying heterostyly is nearly identical to that seen in taxa having only homomorphic self-incompatibility, contradicting Thomson and Barrett's (1981) view that style-morph selection may be more important than other ecological factors favoring dioecy in self-incompatible plants. Based on a survey of 205 angiosperm families that are relatively uniform in breeding system, floral syndrome, and mode of seed dispersal, I propose two tests to distinguish between the ecological correlates expected from selection for out-crossing versus the ecological constraints outlined by Givnish (1980) and Bawa (1980). Both outcrossing and ecological models predict the observed association of dioecy with small, inconspicuous flowers, so that comparisons based on floral syndrome are inconclusive. However, only the Bawa-Givnish model predicts the observed association with fleshy disseminules. Ecological constraints appear important even for the evolution of self-incompatibility, based on the confirmation of a genic-ecological model predicting an association of self-incompatibility with large or specialized flowers and passively dispersed seeds. I outline potential pitfalls in the application of outcrossing arguments, and indicate specific means by which selection for outcrossing may favor dioecy directly or indirectly.

Keywords

Related Publications