The evolution of geographic variation in flower size in Nemophila menziesii (Hydrophyllaceae): Tests of random and deterministic causes

The geographic pattern and the potential causes of geographic variation in flower (corolla) size were studied in Nemophila menziesii Hook. & Arn, an annual wildflower native to California. Using a data-set of flower lengths collected from herbarium specimens it was determined that flower size varies among three large regions in California. Four possible mechanisms that could account for the population differences in flower size were tested: phenotypic plasticity, mating system evolution, random evolutionary processes (genetic drift and population founder events), and natural selection. Common garden experiments revealed strong genetic determination to mean flower size in two populations, and multilocus estimates of the outcrossing rate (using isozyme markers) showed that six populations were primarily outcrossed. Neither phenotypic plasticity nor mating system evolution, then, can be the primary cause of population divergence in flower size because (1) there is strong genetic determination to mean population flower size, and (2) mating systems do not vary greatly. Matrix correlations between geographic distance and flower size distance were used to test for the presence of isolation-by-distance patterns (expected to be generated by random genetic drift and/or population founder events). The presence of significant isolation-by-distance patterns varies according to the spatial scale of the analysis. Random genetic drift and population founder events could have been important in promoting population differences in flower size in some areas of California. Natural selection for flower size was studied in two field populations which differ phenotypically and genetically in mean flower size. Selection gradient analysis (using lifetime seed production as the fitness estimate) detected positive, directional selection favoring large flower size in both populations. Similarly, the effective pollinators showed significant preferences for visiting larger flowers in both populations. These measurements of natural selection do not indicate that selection has played a prominent role in promoting population differences in flower size. Nevertheless, a comparative study of 10 populations of N. menziesii revealed that selection may be promoting divergence in flower size in another way, through the alteration of allometric growth curves (which describe the allocation of resources to flower size as a function of vegetative size).