Studies involving additional descriptors has been important for soybean genetic improvement, so this study aimed to estimate the repeatability the length coefficient of hypocotyl, epicotyl, the internode between the unifoliated leaf nodes and the first trifoliate leaf, from the petiole and from the rachis of the first trifoliate leaf stage and determine the minimum number of evaluations needed to predict the real value of the genotypes. The study was conducted at the Universidade Federal de Viçosa, analyzing 85 soybean genotypes in four experiments in a completely randomized design. The numbers of necessary evaluations were calculated from the estimates of repeatability and coefficient of determination obtained by the methods of analysis of variance of the principal component and by the structural analysis. It was concluded that the lengths of the hypocotyl, epicotyl and the internode required fewer measurements compared with the length of the petiole and rachis to the same level of reliability, and with 15 measurements 85% of reliability was obtained for hypocotyl length by ANOVA and AE(correl) methods and 90% by CP(correl) and CP(cov) methods; 90% reliability was obtained for epicotyl and internode length by ANOVA, CP(correl), CP(cov) and AE(correl) method; 80% reliability was obtained for rachis length by ANOVA, CP(correl), CP(cov) and AE(correl) methods; and for petiole length, 21 measurements were needed for reliability by ANOVA and AE(correl) methods and 85% for CP(correl) and CP(cov) methods.
Glycine max; soybean breeding; number of evaluation; distinguishability
