Premise of the study: The evolution of multiple floral traits often underlies the transition from outcrossing to selfing. Such traits can influence the ability to self, and the timing at which selfing occurs, which in turn affects the costs of selfing. Species that display variation in autonomous selfing provide an opportunity to dissect the phenotypic changes that contribute to variability in the mating system. 

Methods: In a common garden, we measured dichogamy and herkogamy in 24 populations of the protandrous mixed-mating herb Campanula americana, and related these to autonomous fruit set (autonomy). We then measured the timing of self-pollen deposition and fruit production in populations with high and low autonomy, and determined whether pollen germinability across floral development contributes to variation in autonomy. 

Key Results: Populations that transitioned more rapidly to female phase displayed elevated autonomous selfing, but herkogamy was unassociated with autonomous selfing. Selfing occurred more rapidly in highly-autonomous populations due to greater self-pollen deposition early in female phase. Pollen germinability in low-autonomy populations remained constant across floral development, but in high-autonomy populations it increased after floral anthesis and was highest near the onset of female phase. 

Conclusions: Reduced dichogamy, elevated self-pollen deposition, and higher pollen germination late in male phase contribute to both earlier selfing and greater selfing. These traits vary among populations, likely reflecting past selection on the mating system. While delayed selfing bears fewer fitness costs, the evolution of earlier selfing may be favored if self-pollen availability decreases over floral development.

Data was collected from Campanula americana plants growing in the greenhouse. Plant growth and data collection methods are detailed in the paper. The files include raw data. 

Each excel file has one sheet that contains the data and another sheet that contains the metadata, describing the data under each header.
In each data set, each ID is a different plant. The files "TimingOfSelfing_FOR UPLOAD" and "Individ_GHData_AutHerkDich FOR UPLOAD" contain missing data. Missing data resulted because not all variables were measured on each plant. Missing data is indicated by 'n/a'.