Análise uni e multivariada para progênies de Eucalyptus benthamii em várias idades

Abstract

The wide progress made by Brazilian production forests today is a reality due to genetic improvement, which comes to complement silvicultural and forest management actions, offering more productive, uniform populations and, mainly, adapted to different regions. Eucalyptus benthamii is a species of great importance in southern Brazil due to its high tolerance to cold combined with good adaptation and volumetric growth. Several studies have demonstrated the superiority of the BLUP procedure by the multivariate method when compared to the univariate BLUP in results where the multivariate selection resulted in more efficient and accurate selections. Thus, the objective of this work was to compare the genetic estimates by the univariate BLUP procedure and by the multivariate BLUP procedure for the test of E. benthamii half brother progenies. The population used in this work consisted of a progeny test implanted in a randomized block design, composed of 81 families with 20 replicates and one plant per plot, totaling 1620 individuals. The characters of growth DAP, height and volume, were evaluated in four ages, being, at 3 years of age, at 4, at 6 and at 7 years of age. Analyzes were performed using mixed univariate and multivariate approaches. For the multivariate approach, all age combinations for each variable were considered. The characters showed high variation and amplitude. The experimental precision, estimated by the accuracy, was classified as high to very high and the heritability, by the univariate procedure, classified from median to high. Multivariate analysis provided better estimates, even if in a small magnitude, of heritability in years 3, 4 and 7. At six years of age, univariate analysis showed better estimates of heritability. The multivariate analysis also resulted in estimates of superior genetic gains, and here, this superiority occurred in all years of measurement. Therefore, the use of variable measurement ages, as an aid in genetic estimates, through the multivariate procedure, proved to be of great value for experiments evaluated over several years.