Eucalipto, madeira serrada, metodologia de amostragem, regressão multivariada. Log sampling of Eucalyptus grandis trees submitted to selection for sawn timber and energy purposes

Abstract

By the assessment of ten technological traits of eucalypt wood for sawn timber and energy purposes,
it was developed a multivariate statistical procedure in order to determine the sequence of logs to be sampled, in such a way to represent all statistical variation contained within the tree and, accordingly, to establish the appropriate sampling intensity. In the present work, it was used a total of 40 logs from four trees of Eucalyptus grandis provenance Concórdia-SC aged 18 years. By using principal components regression analysis and stepwise selection techniques, it was showed that only two logs, corresponding to the first (0.05 m to 2.60 m) and fourth (8.85 m to 11.40 m) positions into the tree, contained 99.2 % of the total variation detected originally. In the case of adopting a single log, the recommendation was over the fourth log, which represented 97.5 % of the total
amount of the original variation. For the referred  population, the statistical procedure contributed substantially to reduce the high time-consuming and financial costs that are normally associated to studies oriented to this goal, without affecting the original statistical information exhibited by the whole group of logs that would be usually sampled.