Dissecting Factors Responsible For Individual Variation in Plant Fecundity

This paper analyses, on an individual plant basis, the various components determining seed production in the evergreen shrub Lavandula latifolia (Labiatae), in order to estimate their relative contributions to individual differences in fecundity. The relationships between fecundity and age, size, and growth rate are also examined. Experimental pollination and watering were done in an attempt to alter natural patterns of partial fecundity components. Fecundity of individual plants is partitioned into its absolute (number of inflorescences, number of flowers per inflorescence) and relative (fruits/flower and seeds/fruit) partial components, and the following questions are addressed: (1) Are the individual differences in relative components of fecundity? (2) What are the proportional contributions of relative and absolute components of fecundity to individual variation in seed production? (3) Are relative measures of reproductive success reliable predictors of individual differences in fecundity? (4) When induced experimentally, do changes in partial fecundity components translate into modifications of the population's ranking of absolute fecundities? Fecundity depended more on plant size than age, was unrelated to growth rate, and was limited by water, but not pollen, availability. Individual plants differed in all components of fecundity examined, and differences were consistent among years. The proportional contributions of the various components to individual differences in fecundity were very unequal. Almost all variation among plants in seed production (°90% of variance) was explained by the variances and covariances of number of inflorescences and number of flowers per inflorescence. Variation among plants in the two relative components (fruit and seed set), even though statistically significant, had a negligible influence on individual differences in fecundity. These patterns remained invariant despite significant modification of flower production and fruit set induced by artificially watering a part of the study population. Fruit and seed set alone are unreliable indicators of individual differences in the maternal component of fitness in L. latifolia, and a review of relevant literature indicates that this probably applies to other polycarpic plants as well. Plant reproductive biology studies may make erroneous inferences if they consider exclusively data on relative fruit and seed set, and neglect the absolute components of individual differences in fecundity. It is proposed that, to circumvent these potential problems, demographic aspects and absolute estimates of seed output on a per individual basis should be examined in addition to the customarily employed fruit and seed set measures.