INFLUENCE OF PLANT ABUNDANCE ON POLLINATION AND SELECTION ON FLORAL TRAITS OFIPOMOPSIS AGGREGATA

Abstract

Competition for pollination has been hypothesized to select for the diver- gence of floral traits between species and populations. A primary prediction of this hy- pothesis is that the strength of competition for pollination, mediated by variation in plant abundance, should directly influence the strength of selection on floral traits. To test this prediction, I examined the relationships between multiple components of plant abundance and pollination, reproductive success, and phenotypic selection via female fitness on four floral traits in artificial and natural populations of the hummingbird-pollinated Iponlopsis aggregate. In the artificial arrays, I manipulated the absolute density and interspersion of neighboring I. aggregate and a competitor for pollination (Castilleja linariaefolia). I also measured natural variation in the absolute and relative density of these two species within a 2.5-m radius of focal I. aggregate plants in three natural populations. The strength of competition for pollination in the I. aggregata-C. linariaefolia system was only weakly influenced by local plant abundance. Both the absolute density and in- terspersion of plants in the arrays significantly influenced at least one component of I. aggregate's pollination and reproductive success, but the effects were not consistent across these components. For example, the treatment groups that received more nonspecific pollen were not the same ones that set more seeds/fruit. Within the natural populations, variation in relative and absolute plant density influenced two components of I. aggregata's polli- nation and reproductive success, but only in one to two populations per component. As would be expected from these inconsistent effects of plant abundance on pollination and reproduction, the strength of selection on floral traits of I. aggregata in both the arrays and natural populations was also only weakly dependent on abundance. Previous studies have indicated (1) that self-incompatible species such as I. aggregate should experience strong, abundance-dependent effects on pollination, and (2) that variation in plant abundance at a local scale should have a stronger effect on pollination than variation at a larger spatial scale. In contrast, my results suggest that the effect of plant abundance on pollination cannot be easily predicted from simple diagnostic traits such as breeding system or spatial scale.