Generalized nonlinear models applied to the prediction of basal area and volume of Eucalyptus sp.

This paper aims to propose the use of generalized nonlinear models for prediction of basal area growth and yield of total volume of the hybrid Eucalyptus urocamaldulensis, in a stand situation in a central region in state of Minas Gerais. The used methodology allows to work with data in its original form without the necessity of transformation of variables, and generate highly accurate models. To evaluate the fitting quality, it was proposed the Bayesian information criterion, of the Akaike, and test the maximum likelihood, beyond the standard error of estimate, and residual graphics. The models were used with a good performance, highly accurate and parsimonious estimates of the variables proposed, with errors reduced to 12% for basal area and 4% for prediction of the volume.