ABSTRACT
In an experiment involving 248 half-sib families of Eucalyptus grandis in randomized block with 4 repetitions and linear plots of 6 plants, in the City of Rio Doce, Minas Gerais, to the 46 months of age, were evaluated the methodology of selection indices (Classic Index and of Pesek and Baker) and the techniques of the canonical analysis and factors analysis as tool of order of candidates to the selection with relation to the traits breast height, height of plants, number of trees with rust incidence, number of trees with cancer incidence and number of normal trees per plot. The index of Pesek and Baker was the only one to present expected genetic gains in the directions longed for all the evaluated traits. The techniques of the canonical analysis and factors analysis had not been efficient in providing expected genetic gains balanced with the intentions of the work, but they had allowed the suggestion of its application in other works for better verification of its potentialities as tool of order of candidates to the selection.
Key words:
selections gieth; canonical analysis; factors analysis