PARAMETRIC AND NON-PARAMETRIC ANALYSES TO THE DETERMINATION OF STABILITY AND ADAPTABILITY OF SOYBEAN GENOTYPES

Abstract

This work aimed to evaluate the yield stability and adaptability of 30 soybean genotypes [Glycine max (L.) Merr.] in six environments at the State of Mato Grosso, Brazil. A randomized complete blocks design with four replicates was used. The trials were carried out at 2004/05 experimental years at the counties of Rondonópolis, Campo Verde, Nova Brasilândia and Vera. To evaluate the adaptability and stability, the methods developed by Eberhart and Russell (1966), Lin and Binns (1988), and modified by Carneiro (1998), Annicchiarico (1992) and Centroid Method (Rocha et al., 2005) were used. The grain yield mean varied from 2837 kg ha-1 (Rondonópolis III) to 3701 kg ha-1 (Vera) with general mean among the environments of 3200 kg ha-1. According to the methodologies developed by Eberhart and Russel, the SL 2238, SL 1949, SL 31, SL 2382 lines and the standards Uirapuru and M-Soy 8411, presented the highest means, wide adaptability (= 1) and statistically significant, which means a low stability of such materials. The SL 2238, SL 712, SL 1949 lines and the Uirapuru cultivar were classified by having the highest grain yield, adaptability and stability, according to the methodology proposed by Lin and Binns, Annicchiarico and Centroid, having, therefore, consistency among such methodologies. The SL 2238, SL 1949 and SL 712 lines presents superiority in terms of adaptability, stability and performance for grain yield, according to Lin and Binns (1988), Annicchiarico (1992) and Centroid (Rocha et al., 2005) methodologies.