Comparative study on the stability and adaptability of different models to develop a high-yield inbred line from landrace rice varieties

Multilocation trials are important in plant breeding carried out in the field of agronomy. A combination of agronomy and plant breeding is needed to improve plant characteristics and stability testing before releasing new varieties. A single stability analysis method may not be sufficiently representative to determine the performance of genotypes across environments, which sometimes lead to wrong decisions about genotype stability. The objective of this study was to estimate yield potential, adaptability, and stability of superior lines resulting from inbred landraces based on parametric and nonparametric, and a genotype main effect G × E interaction (GGE) biplot. Fourteen inbred lines from Bengkulu landrace rice varieties were evaluated in five environments from January 2019 to November 2020. The experiment was conducted using a complete randomized block design with two replications. The results showed that the highest yield grouping based on the G × E heat-map, genotypes G13(BKL4-B1-268-10), G10(BKL2-B3-264-6), G7(BKL1-B3-261-3), and G5(BKL1-B1-259-1), was found in Sungai Serut. There was a strong positive correlation (r = 1.00) between the mean yield (Y_i) and YSi, S²_di and Di, and W_i² and StabVar. Meanwhile, the genotype occurred in the top third of the ranks (TOP) had a positive correlation of 0.78, indicating a suitable stability parameter to identify high-yield genotypes. There were four consistently stable lines based on parametric and nonparametric stability analyses: G10(BKL2-B3-264-6), G7(BKL1-B3-261-3), G13(BKL4-B1-268-10), and G5(BKL1-B1-259-1). The GGE approach methods showed consistent stability, and lines G13(BKL4-B1-268-10), G10(BKL2-B3-264-6), G5(BKL1-B1-259-1), G7(BKL1-B3-261-3), and G6(BKL1-B2-260-2) had high-yield potential, wide adaptability, and stability, and are recommended for further testing as candidates for new varieties.