The aim of this work was to evaluate the seed yield repeatability of 27 guarana clones. Ten assays were carried out for clone evaluation using a completely randomized block design, with two replicates and three plants per plot with a 5x5 m spacing. The seed yield was measured during four years (1998, 1999, 2000 and 2001) after the beginning of the production phase, two years after planting. Four methods were used for repeatability coefficient estimation: analysis of variance (ANOVA), principal components analysis based on correlation matrix (CPCOR), principal components analysis based on phenotypical variance and covariance matrix (CPCOV), and structural analysis based on correlation matrix (AECOR). Minimum observations number and determination coefficient were also evaluated. The repeatability coefficient estimates based on these four years, in all ten environments, were relatively low - 0.2618 (AECOR), 0.3064 (CPCOR), 0.3183 (ANOVA) and 0.5308 (CPCOV) - and varied according to method and environment. At least four years of evaluation are necessary for obtaining the real genotypical value of the clones for an 80% determination coefficient.
Paullinia cupana; analysis of variance; structural analysis; principal component; clone selection
