Genetics of total seed yield and its components in Phalaris aquatica L

Abstract

Eight plants within each of 55 half-sib families in a diverse breeding population of this perennial pasture grass were assessed for total seed yield in bagged inflorescences, numbers of panicles and spikelets per plant, weight per seed, and for important agronomic characters. Total seed yield was moderately highly heritable (h2N = s.e. = 0.52 ¦ 0.21), and was phenotypically variable. The mean seed yield of the progeny of the highest yielding 10% of the plants should be 44% higher than the mean of the original population. Seed yield was closely correlated genetically with the number of seeds per panicle or per plant, but not with weight per seed or number of panicles per plant. There was little variation for ovule fertility in the population. Visual estimates of spikelet number per plant were very highly correlated genetically with numbers of seeds per plant and total seed yield, but had a non-significant heritability (h2N = 0.30 ¦ 0.20). Scoring for spikelet number per plant before anthesis would enable one generation of mass selection and recombination to be completed each year. Methods for increasing the accuracy of spikelet number estimation are discussed. Provided the true heritability value is greater than a quarter of the heritability of total seed yield, annual cycles of selection on spikelet number score would give a faster rate of genetic advance than direct selection for seed yield on seed parents only each year, or on seed and pollen parents every second year. Seed yields greatly exceeding those of current cultivars should be attainable by recurrent selection. There should be no deleterious correlated responses in flowering time, herbage yield or plant height; mature plant weight and harvest index would increase.