The evolution of combined versus separate sexes in Wurmbea (Colchicaceae)

The evolution of separate sexes has occurred repeatedly throughout the history of the flowering plants in a wide variety of life-history contexts and ecological conditions. Yet despite its frequent origin, gender dimorphism in plants is relatively rare, occurring in less than 10% of angiosperm species. Theoretical models predict at least two factors are important in selection for gender dimorphism—resource allocation and the avoidance of inbreeding. Both of these factors can limit the seed fitness of cosexual plants relative to females, aiding their invasion and maintenance in populations of cosexuals, hence the establishment of dimorphic sexual systems. Stressful environmental conditions may favour (1) resource reallocation by limiting investment into expensive functions, such as fruits and seeds, or (2) the expression of inbreeding depression through effects on the mating system. Using Wurmbea (Colchicaceae) as a study system, I investigated recent proposals that aridity has favoured the evolution of gender dimorphism through altered patterns of resource allocation or changes in pollination and mating.; Phylogenetic analysis revealed multiple evolutionary transitions between monomorphic and dimorphic sexual systems of Wurmbea. I found no evidence for a causal role of aridity in selection for separate sexes, although some species probably invaded dry habitats after gender dimorphism was established. One transition involved W. dioica ssp. alba, a polymorphic subspecies in Western Australia containing populations, that are either monomorphic or dimorphic for gender. I found substantial morphological and ecological differentiation between the sexual systems of W. dioica . Gender dimorphism was associated with drier microsites, allocation patterns favouring resource acquisition and storage, and physiological traits related to higher photosynthetic capacity and lower water use efficiency. In contrast, gender monomorphism was associated with wetter microsites, greater investment into leaves and flowers with increased resource availability and greater water use efficiency. Smaller flower size was favoured by low precipitation in both sexual systems. I found evidence that flower size and style morphology have important consequences for pollen dispersal in W. dioica, and may have mediated a change in the mating system in resource-limited sites. Thus, gender dimorphism in W. dioica may have resulted from selection to enforce outcrossing in resource-limited conditions.