Effects of number of flowers per raceme and number of racemes per plant on bumblebee visits and female reproductive success in Salvia nipponica (Labiatae)

We studied the adaptive arrangement of raceme flowers in dense and sparse habitats of Salvia nipponica. We recorded bumblebee visits and collected mature seeds to estimate outcrossing rates from allozyme genotypes. To examine the resource investment in flowering stalks, we measured their length and dry mass. We found that a greater number of open flowers in a raceme enhanced bumblebee visits and successive probings in both density plots. However, greater height of a raceme did not enhance these in plots of either density. In the high-density plot, both a greater mean number of open flowers in a raceme and a greater number of flowering racemes on a plant enhanced bumblebee visits, successive probings, and also enhanced outcrossing rate in spite of an increase in successive probings. Although the number of flowers per raceme had a greater positive effect on seed-to-ovule ratios than did the number of racemes, the number of flowers per raceme may be constrained because a longer flowering stalk needs greater buckling strength. Hence, it may be advantageous for plants in high-density plots to increase both number of flowers per raceme and number of racemes. In the low-density plot, a greater number of flowering racemes on a plant enhanced pollinator visits, but a greater mean number of open flowers in a raceme did not. Hence, it may be advantageous for plants to increase the number of racemes while keeping the numbers of flowers per raceme small. Thus, the adaptive number of flowers per raceme and number of racemes per plant may differ between high- and low-density plots.