Selection on variance in flowering time within and among individuals.

We model the impact of pollinator visitation rate and behavior on the short-term evolution of population flowering phenologies determined by the distributions of flowering times within and among individual plants. Evolution of population flowering phenologies depends on the phenotypic variances and heritabilities of the within-individual mean and variance of flowering time. In the ecological scenarios we investigate selection does not produce a correlation of the mean and variance of individual flowering time. Self-incompatibility causes weak stabilizing selection on flowering time that acts to reduce the within-individual variance in flowering time. Disruptive selection due to pollinator limitation acts mostly to increase the among-individual variance in flowering time. Stabilizing selection due to pollinator attraction, or short reproductive season, acts mostly to decrease the within-individual variance in flowering time. Temporal autocorrelation of environmental stochasticity in pollinator visitation rate strongly selects to increase the within-individual variance in flowering time. These predictions can be tested by measuring the causal factors described above, partitioning the variance in population phenology within and among individuals, and estimating the inheritance of, and selection on, within-individual mean and variance of flowering time.