Genetic variation in seed production and its components in four cultivars of the pasture grass Setaria sphacelata

Variation in seed production and its components was investigated between and within four cultivars of the pasture grass Setaria sphacelata, in two experiments over two years, as a basis for future cultivar improvement. The study sought to determine the basis for the considerable differences in seed production of the four cultivars Nandi, Narok, Solander and Kazungula, to determine the extent of genetic variation in the four cultivars in attributes which contribute to seed yield, and to investigate genotypic consistency in seed production over years and seasons. Each experiment comprised 50 genotypes of each cultivar. In the first experiment, plants were harvested a set number of days after median flowering date whereas in the second experiment, which was unreplicated, each genotype was harvested a set number of days after it had flowered. In the first experiment, seed yields were generally highest for Kazungula, lowest for Narok and intermediate for Nandi and Solander. All measured attributes contributing to seed yield exhibited a high order of variation between and within cultivars, but the basis for the large difference in seed yield per plant between cultivars was tiller fertility rather that total tiller number. Averaged over the four harvests, there was a six fold to > 100–fold intra-cultivar genetic range in seed production, associated with differences in tiller fertility, which were associated with differences in date of first flowering. Broad sense heritability for seed yield averaged 0.68 for the four cultivars and showed little change over the four harvests. Genotypes which produced high seed yields in summer were also more productive of seed in autumn and the 0ore productive genotypes in the first year were also more productive in the second year. Cultivars differed in the relative importance of factors which contributed to the high seed yield of high-yielding genotypes. In the second experiment, genotypes with a high seed yield also generally had the highest tiller fertility, even though all genotypes were harvested the same number of days after first flowering. Within-cultivar correlations in seed yield between the two experiments were generally significant and the elite 20% of genotypes from this experiment had 1.2–2.9 times the seed yield of the same genotypes with a very different harvesting regime in the first experiment. It is concluded that opportunities exist in all four cultivars for improvement in seed production and that the selection criterion offering the best opportunity for advance would be fertile tiller number. In Narok, Solander and Nandi, this would result in increased tiller fertility, whereas in Kazungula, it would result in an increase in total tiller number. This revised version was published online in July 2006 with corrections to the Cover Date.