Origins of variance in seed number and mass: interaction of sex expression and herbivory in Lomatium salmoniflorum.

As in many plant species, Lomatium salmoniflorum (Umbelliferae) individuals produce many flowers, only a subset of which produce mature seeds that escape seed parasitism and enter the seed bank. The interrelationships between the timing and number of flowers produced, sex expression, seed set, and seed parasitism were studied for their direct and indirect effects on the numbers and masses of viable seeds produced by individual plants. In a sample population of 369 plants that produced 161 386 flowers, 76% of the plants produced some hermaphroditic flowers. The percentage of hermaphroditic flowers increased significantly with the total number of flowers produced by a plant. Seed set was 65-90% in plants producing >600 flowers, but was highly variable in plants producing fewer flowers. Hand-pollinated plants showed the same pattern of seed set, suggesting that variable seed set in small plants may result from insufficient resources for seed development. The majority of schizocarps was produced by only 12% of the plants. Parasites killed 24.5% of the seeds prior to dispersal. Another 14.5% of the seeds lacked endosperm. Hence, the initial 161 386 flowers, which included 25874 hermaphroditic flowers each capable of producing two seeds, produced 42 468 seeds of which an estimated 25906 entered the seed bank as undamaged seeds with fully developed endosperm. Path analysis indicated that the number of hermaphroditic flowers on a plant and the percentage of seeds attacked by seed parasites had the greatest direct effects on the number of viable seeds entering the seed bank. The date at which a plant began flowering and the percentage of flowers setting seed had smaller or only indirect effects on viable seed production. Mean seed mass for plants was not significantly related to any of the factors that affected seed number, but little of the variance in seed mass occurred among plants. Masses of intact seeds in the population ranged 9-fold in both 1987 and 1988. Thirty-five percent of the variance was among seeds within umbels, 46% was among umbels within plants, and only 19% was among plants. The large variation among umbels within plants resulted from a seasonal pattern in which seeds from umbels produced late in the spring had lower mean seed masses than seeds from umbels produced early in the spring. Overall, the results indicate that both direct and indirect interactions between number of flowers, the date of initiation of flowering, seed set, and seed parasitism affect the number of viable seeds entering the seed bank. These interactions strongly bias viable seed output to a small minority of plants that produce many seeds with a wide range of masses over the growing season.