Determination of the number of years in Arabic coffee progenies selection through repeatability

The purpose of this study was to estimate the repeatability coefficient of grain yield in Arabic coffee and todetermine the minimal number of annual harvests required for a satisfactory level of reliability at the plant level. Theexperiment was conducted in Campinas (SP) in a randomized complete block design with 21 F3RC2 progenies of Icatugermplasm, for nine years, in nine replications. The repeatability coefficients were estimated by the method of analyses ofvariance, principal components based on correlation and covariance matrices and structural analyses based on correlationmatrices. The selection of F3RC2 progenies of Icatu germplasm can be carried out with an accuracy of 80% in the three firstharvests. The repeatability coefficient estimated by the covariance matrix method was the most adequate methodology toestimate this coefficient, since it minimizes the effect of biennial bearing.