Graduation date: 2007Pollination ecology may play an important role in the maintenance of selfing in populations of self-compatible hermaphroditic plants where both selfing and outcrossing occur (mixed mating). Behavior and abundance of pollinators can influence the two major modes of selfing; autogamy (selfing within a flower) and geitonogamy (selfing between flowers on the same plant). Autogamy may be selected for as a method of reproductive assurance during times of reduced or inefficient pollinator service. In contrast, geitonogamy is a negative and non-adaptive consequence of pollinators visiting multiple flowers per plant. Pollinator behavior and abundance, which can both influence levels of selfing, varies among pollinator types. Evidence suggests that of the two major pollinators of the blue columbine, Aquilegia coerulea, hawkmoths may decrease levels of selfing compared to bumblebees. The goals of this study are to 1) quantify the contribution of autogamy and geitonogamy to the overall selfing rate, 2) determine whether hawkmoths decrease levels of selfing compared to bumblebees and other floral visitors, and 3) examine how pollinator behavior influences levels of selfing in a Colorado population of A. coerulea. First, we estimated levels of selfing from groups of emasculated and control flowers using allozyme data to measure the genetic contribution of geitonogamy and autogamy to the population selfing rate. Second, contribution of autogamy to seed set was quantified as the difference in seed set between control and emasculated flowers. Third, we compared the realized levels of autogamy measured from allozyme data to the potential for autogamy, measured as seed set in absence of pollinators. Fourth, we compared the realized levels of geitonogamy measured from allozyme data to the potential for geitonogamy, calculated from the flowering phenology. Fifth, we conducted pollinator observations to document the number of different pollinators, their behavior, and their variation in abundance over three years of study. Finally, we determined if there was a relationship between yearly abundance of each pollinator type and yearly selfing rate. The allozyme data showed that geitonogamy was the primary contributor to the intermediate levels of selfing found in this population, suggesting that selfing is due to the negative consequence of pollinators visiting multiple flowers per plant and has no adaptive explanation. The realized level of geitonogamy was greater than the potential for geitonogamy, and resulted from pollinator preferences for a floral gender. In contrast, the realized level of autogamy was negligible, which suggested that reproductive assurance was not being selected for in this population. However, seed set from control flowers were significantly greater than that of emasculated flowers, indicating that autogamy contributed to an increase in seed set. The discrepancy between genetic and seed set data could be explained by bumblebee preference for control relative to emasculated flowers. Although the realized level of autogamy was minimal, plants in the population retained a high potential for autogamy, which may be selected for as a mechanism of reproductive assurance during years of low pollinator abundance. Pollinators in this population of A. coerulea included bumblebees, hawkmoths, solitary bees, wasps and flies. The abundance of all pollinators varied significantly among years except for solitary bees and wasps. In addition, we documented a significant variation in selfing rate (0.23-0.59) over the three years of study. A relationship was noted between years with increased hawkmoth abundance and a decrease in selfing rate. No relationships were observed between yearly selfing rate and abundance of the other pollinator types. The behavior most likely responsible for the decrease in selfing is that, unlike other pollinators, hawkmoths prefer female-phase flowers, which reduces levels of geitonogamy. Thus, the abundance and behavior of distinct pollinator types can differentially influence levels of selfing.