This study used data from six experiments with half-sib families of Eucalyptus grandis, carried out at six localities, to evaluate the following traits: circumference at breast height (CAP), plant height (ALT), rust incidence (FER), canker incidence (CAN) and number of normal trees per plot (NAR). The direct and indirect selection methodologies were compared with the selection indices (classic and Pesek and Baker method) using all the traits. Neither selection method yielded the expected distribution of genetic gains. Both the classic and Pesek and Baker's indices produced a more suitable distribution of expected genetic gains.
Forest tree improvement; selection indices
