The objective of this work was to model the diametric growth of Melia azedarach trees, in the region of Campanha, state of Rio Grande do Sul. 31 trees that showed no signs of pruning were randomly selected. Three increment borers per tree were removed, the first being 30 cm from the ground for age development and the other two orthogonal at 1.3 m (dbh) for growth analysis. Linear models with and without intercept were fitted to estimate the diameters with bark (y) as a function of the diameter without bark (x), using several transformations of the independent variables (x). In the modeling, the stepwise method was used for variable selection, being kept in the models only as a variable with the highest adjusted R² and with a significant F value at 5%. Next, multicollinearity was resisted through the variance survival value (VSV), being kept in the models only as the most significant variables by the t-test and with VSV less than 10. The selection of the best ones considered was based on the Standard Error of Estimation in percentage (Syx%), Coefficient of Determination adjusted (adjusted R²) and, in the case of diametric growth lines, the Akaike Information Criterion (AIC) was also used. The species presents moderate to fast growth and reaches sufficient dimensions for the production of wooden parts for joinery and carpentry. Among the medications used to describe diametric growth, the Chapman-Richards medication presented the best result, with statistics that allow its safe use under the conditions of this study.
Read full abstract