GRIDDED MASS SELECTION FOR SEED NUMBER AND SEED WEIGHT IN GRAIN SORGHUM POPULATION NP21R (MILO, RANDOM MATING, RECURRENT)

Gridded mass selection of open-pollinated (OP) and self-pollinated (SP) heads for seed weight and seeds/head was applied to NP21R to form eight improved populations each. Genotype and genotype by year variances were estimated by testing 160 half-sib (HS) and 160 S1 families from the base population in a replications-in-blocks design in 1981 and 1982. The base and improved populations were tested for yield and other traits in two environments in a randomized complete-block design in 1983. Individual-plant data also were taken in the 1983 yield trial for estimation of plot to plot and within-plot variances. Investigating the usefulness of gridded mass selection for increasing seeds/head and weight was the primary objective. Results indicate that gridded mass selection of OP and SP heads significantly improved seed weight 16 and 9%, respectively, at seeds/head\u27s expense. Yield increased significantly. Selection of OP heads for increased seeds/head was ineffective, but yield increased significantly. SP head selection significantly increased seeds/head 11% at seed weight\u27s expense. Yield increased slightly. Estimates of genetic variance revealed tremendous variability for most agronomic traits. Heritabilities estimated from components of variance of both HS and S1 families were high on a family-mean basis and low on an individual-plant basis for all traits. Predicted gains ranged from 3 to 16% and from 1 to 10% for S1 and HS families, respectively, on a family mean basis, and from 2 to 7% and from 1 to 3% for mass selection of SP and OP heads, respectively, on an individual-plant basis. Realized heritabilities and realized gains from selection approached those predicted on an individual-plant basis. Yield was always positively correlated with seeds/head but had inconsistent correlations with seed weight. Gridded mass selection effectively improved seed weight and seeds/head in five of eight improved populations. Selecting SP and HS heads for seed weight should improve yield equivalent to direct selection for yield alone while selection for seeds/head may or may not improve yield