ALLOMETRIC SCALING AND FLORAL SIZE VARIATION IN COLLINSIA by

Allometric scaling theory has previously been used to estimate the functional relationship between two biological variables. In addition to parameter estimation, deviations from the general scaling relationship can be used to create hypotheses. Here, I explore deviations from the allometric scaling pattern for plant and floral size within the genus Collinsia on three levels: among species, within species, and among populations of a single species. Collinsia species are self-compatible annual herbaceous plants that have been shown to vary in floral size, autonomous fruit production, and estimated mating system. I quantified the amount of variation in characteristics related to plant mating systems: floral size and autonomous fruit production in a pollinator-free environment and used variation and scaling deviations to generate expectations about environmental selection pressures. I found that the scaling relationships differed on each of the three levels and that deviation from the general floral size-plant size relationship is common within this genus. The among-species regression explained only 20% of the variation in floral size, and species-and population-level regressions explained even less. The four species for which I obtained controlled environment estimates of vegetative and floral trait in this study differed significantly in autonomous fruit production, floral size, and plant size, while populations of C. heterophylla differed in floral and plant characteristics, but not autonomous fruit iii production. In addition, variation in plant size characteristics was 50-66% greater than variation in floral size characteristics suggesting selection to reduce variation in floral size and flexibility in plant size. Autonomous fruit production was correlated with floral size in C. tinctoria, with floral number in C. verna, and uncorrelated in C. heterophylla suggesting that floral trait and autonomous selfing ability varies among species. Using a comparative method and investigating factors correlated with plant mating system, such as floral traits, across a group of closely related species provides new insights into factors affecting their variation. for immense help in data collection, April Randle and Chris Heckel for reviewing this thesis, and the Ashman lab graduate and undergraduate students for discussing the ideas incorporated in this project.