Dissecting the yield architecture and selecting superior seedling-derived half-sib progenies of turmeric (Curcuma longa L.) through multi-trait selection indices

Turmeric (Curcuma longa L.), propagated vegetatively, harbors limited genetic variability for breeding, prompting exploration of rare seed-derived half-sib progenies to generate novel recombinants. The present study investigated 263 first-generation seedling-derived half-sib progenies from five maternal accessions alongside three checks in an augmented randomized complete block design over two years (2024–2025) at Kozhikode, India. Significant genotypic variation was observed across 17 vegetative, rhizome, yield, and percentage dry recovery, with high broad-sense heritability (60–97%) for most, including total yield per plant (80.73%), primary rhizome weight (64.3%), and dry recovery (95.73%). Genotypic coefficients of variation exceeded 20% for key yield components, indicating strong selection potential. Pearson correlations highlighted strong positive associations of total yield with primary rhizome weight (r = 0.81, p< 0.01), mother rhizome weight (r = 0.60), and primary rhizome number (r = 0.57), alongside a yield–dry recovery trade-off (r = -0.20). Stepwise regression identified primary rhizome weight as the strongest predictor (adjusted R² = 0.791). Structural equation modeling confirmed vegetative growth (measured by leaf width, leaf number, petiole length) strongly influences rhizome sink strength (β = 0.860, p = 0.027), mediating indirect effects on total yield per plant. Multi-trait indices—multi-trait genotype-ideotype distance (MGIDI), multi-trait stability (MTSI), Smith-Hazel (SH), and factor analysis-ideotype (FAI)—identified 53 superior progenies (20% intensity), with 14 common selections across methods, prioritizing balanced yield, sink capacity, and stability. These findings validate seedling progenies as a practicable resource for turmeric improvement, offering observable vegetative proxies for subterranean yield traits and ideotype-based frameworks to accelerate recombination breeding.