Classe de dominância como variável auxiliar na modelagem para povoamentos de teca

Abstract

The classification of a tree in relation to its sociological position is known and easily determined in the current period of growth. However, there is no guarantee that the selected tree will remain in the same dominance class over time. With the purpose of deepening the understanding of the individual development of Tectona grandis L.f. trees, the objective of the study was to verify whether, within the plots, the dominance class of the tree varies with age and with the application of selective thinning. Subsequently, classification was inserted as an additive categorical and interaction variable in the hypsometric relationship in a traditional height-diameter model. The data used are from continuous forest inventories in 53 sample units between the second and eighth year of age of a seminal plantation in the municipality of Nossa Senhora do Livramento-MT. For each plot and age, dominance class limits were established based on the average of the Diameter at Breast Height (DBH) with the addition and subtraction of one and two standard deviation values of the DBH. The classification of each tree's dominance as oppressed, suppressed, monetary, co-dominant or dominant with the comparison of its DBH to the defined intervals. The variation in the dominance class of each tree between the second and eighth year, and before and after thinning, was proven. Before thinning, 65.55% of the trees were grouped in the same dominance class, and after thinning the value was 52.69%. From the second to the eighth year and with the application of selective thinning to the fifth year, most trees were reclassified into different classes. Stability of dominance classification was observed in most trees in consecutive evaluations up to the eighth year of age. After confirming the stability of the tree's dominance class over time, the most appropriate linear fixed-effects model was adjusted and selected for the data. The dominance class was then added to the regression model as an additive and interaction categorical variable. The adjusted regression models were accurate using the validation test, analysis of variance, standard error of the estimate, adjusted determination coefficient and graphical analysis of the residuals, considering the regression conditions. The straight-line linear regression model was adequate, with satisfactory quality-of-fit measures and minimally biased residual distribution and was selected to represent the hypsometric relationship of the settlement. With the analysis of variance, differences in the levels and inclinations of the curves were identified, and different equations were applied for a better estimate. The introduction of the dominance class as an additive categorical and interaction variable in the model increases the efficiency in estimating height, resulting in higher percentages of determination coefficients and lower standard errors of the forecast. The classification by dominance expresses the individual development in relation to the sampled population, with potential for use in silvicultural analysis, being able to help in the planning of silvicultural practices, in the modeling of dendrometric variables and in the selection of individuals.