INDUCED GENETIC VARIABILITY FOR GRAIN AND FODDER YIELD ARCHITECTURE IN M₅ YELLOW PERICARP SORGHUM MUTANTS

Abstract

Yellow-pericarp sorghum landraces have value for food, feed and dryland production, but their breeding progress is constrained by limited selectable variability. This study quantified induced variability for 11 yield and yield contributing traits in 39 M₅ mutant derivatives of the rabi landrace Udgir Piwali, together with four commercial checks and the parental control. The entries originated from gamma irradiation (100–500 Gy), ethyl methanesulphonate (0.1–0.3% EMS) and selected combined treatments. They were evaluated during rabi 2025 26 at Parbhani, India, in a randomized block design with two replications. Analysis of variance detected genotype effects for every trait: nine traits were significant at P < 0.01, while days to 50% flowering and 100-seed weight were significant at P < 0.05. Grain yield ranged from 44.15 to 66.70 g plant⁻¹ and fodder yield from 72.85 to 139.50 g plant⁻¹. The leading mutant, 200Gy+0.2%EMS-1, yielded 66.70 g grain plant⁻¹, representing a 17.64% advantage over Udgir Piwali (56.70 g plant⁻¹) and a 0.23% numerical advantage over CSV-29R (66.55 g plant⁻¹). Phenotypic coefficients of variation exceeded genotypic coefficients for all traits. Moderate GCV occurred for panicle weight (15.67%), fodder yield (15.59%), panicle length (13.38%), plant height (12.71%) and panicle width (10.04%). High broad-sense heritability coupled with high genetic advance as a percentage of the mean was expressed by fodder yield (83.96%, 29.47%), plant height (79.76%, 23.42%), panicle weight (75.62%, 28.12%) and panicle length (70.60%, 23.19%). Grain yield showed moderate heritability (47.59%) and moderate genetic advance (11.18%). The results demonstrate useful, trait-specific induced variability in the M₅ population. Direct phenotypic selection should be most effective for panicle weight, fodder yield, panicle length and plant height, whereas grain-yield gains should be confirmed across environments and supported by component-trait selection.