摘要：

This work was carried out to estimate the repeatability coefficient, to compare the efficiency of methodologies used in the estimation process, as well as to determine the number of evaluations for selecting dwarf cashew (Anacardium occidentale L.) clones. The methodologies used to estimate the coefficient of repeatability were: analysis of variance in which the temporary effect is removed from the error (ANOVA); analysis of the principal components obtained from both the correlation matrix (PCCOR) and the covariance and phenotypic variance matrix (PCCOV); and the structural analysis based on the theoretic eigenvalue of the correlation matrix or mean correlation (EVCOR). Characters such as plant height (PH), canopy diameter (CD) and nut yield (NY) from 30 clones were evaluated during five years. The r value for PH and CD varied from 0.85 (ANOVA) to 0.96 (PCCOV). The yield r values varied from 0.51 (ANOVA) to 0.88 (PCCOV). Therefore, it was admitted that the principal components method employing the covariance and phenotypic variance matrix is the most efficient to dwarf cashew clonal selection, mainly in relation to yield. It was also concluded that two evaluations of both characters PH and CD, and three evaluations of NY are sufficient in the selective process.

展开