The aim of this work was to estimate the adequate size of experimental plots, for the evaluation of phenotypic descriptors in banana, from a trial of uniformity with the cultivar Tropical, in the spacing of 3x2 m, in a useful area with nine rows of 40 plants. The following variables were evaluated: plant height; pseudostem perimeter; number of emitted suckers and number live leaves, during flowering and at harvest; weight of the bunch and the hands; number of hands and fruits; weight of the second hand; weight, length and diameter of the fruit, in two cycles of production. The plants, considered the basic units, were combined to form various sizes of plots. The data were submitted to the analysis of variance in the hierarchical model. Plot size was estimated by the methods of maximum curvature, modified maximum curvature and comparison of variances. Heterogeneity index of the soil and the detectable difference among treatments mean were determined. The variability increased between cycles with reflexes in the estimates of the plot sizes, which varied with the method used, the variable evaluated and the production cycle. The modified maximum curvature method presents more adjusted estimates. Plots with six basic units (36 m²) are appropriate to evaluate with precision the phenotypic descriptors in the bananas.
Musa spp.; agricultural statistics; morphology; production
