Heritability Estimates, Genetic Correlations, and Predicted Gains from S1 Progeny Tests in Three Grain Sorghum Random‐mating Populations1

Abstract

Means, variances, heritabilities, genetic correlations, and predicted responses were obtained for grain yield, other agronomicharacters, and grain protein by testing 200 S1 families from each of three random‐mating populations of Sorghum bicolor (L.) Moench. Means of the three populations differed significantly for all characters except kernel weight. The NP3R population, composed of adapted material, was higher yielding than NP5R, which contained exotic germplasm; however, NP5R had twice as much genetic variability for yield and most other characters. NP7BR, a narrow population, yielded the same as NP5R but had the lowest variability for plant height and days to bloom. The grain yield of NP7BR was decreased and its grain protein increased because of poorly pollinated antherless (al al) male‐sterile plants. Variances for yield‐related characters in NP7BR also were biased. Genotype × year interaction variances were low in all populations compared to genetic variances except threshing percentage in NP5R. Broad‐sense heritability estimates for bloom date, plant height, yield, and kernel weight were high in all populations. Heads per plant had the lowest heritability, especially in NP3R and NP7BR.Genetic correlations usually had the highest values in NP5R and the lowest in NP7BR. Grain yield per unit area generally was best correlated with grain yield per head, plant height, and threshing percentage. Days to bloom and grain protein percentage were negatively correlated with yield and had low values, except for NP5R.Predicted gains in yield from selecting the highest 20% of the families in each population were 16.3, 10.2, and 8.7 q/ha per cycle for NP5R, NP3R, and NP7BR, respectively. The smallest predicted gains were for tillers per plant and grain protein percentage. Both NP3R and NP5R appear suitable for recurrent selection studies.