The main purpose of this study was to apply the mixed-effects theory in modeling Eucalyptus stand growth. The database is from homogeneous stands located on Brazilian coastal region, Espírito Santo and Bahia States. We used two levels of sampling: sample units and individual trees inside of sample units. As an example, we considered the logarithm of the basal area as response variable and inverse of age, logarithm of the total height and their interaction as covariates. The parameter estimate quality of the linear model was significantly improved. Also, by modeling the heteroscedasticity and the autocorrelation (AR(1)), the log-likelyhood values were increased significantly.
Multilevel mixed-effects model; Eucalyptus; autocorrelation; heteroscedasticity
