The aim of this paper was to evaluate the selection efficiency in the F4 generation of coffee for agronomic traits and disease resistance through the use of mixed model methodology of Henderson. The experiment was carried out in Machado (MG) Brazil, using F4 progenies from a cross between cultivars of Icatu group with Catuaí Amarelo IAC 62 and IAC 17 and with the progenies IAC 5002 and IAC 5010. The experimental design was a square lattice with four replications and the evaluations were taken during two years (2010 and 2011). The traits evaluated were: production (L plant- 1), reaction to rust, vegetative vigor and reaction to Cercospora. We used the REML/BLUP procedure to estimate genetic parameters, the Mulamba & Mock selection index to verify the gains for multiple characters, the accurary and efficiency and also the multivariate clustering progenies through the Tocher method. The analysis of genetic parameters using mixed models was successful for selecting families for all characteristics under consideration. The adoption of four crop harvests may increase to 80% the accuracy in selecting individuals for production characteristics, reaction to rust and vegetative vigor. The selection of the 12 best progenies for multiple characters by Mulamba and Mock index provided an average gain of 70.4%.
Coffea arabica; early selection; Icatu; genetic improvement
