Competição e modelagem florestal em fragmento de floresta estacional semidecidual submontana, MG

The Atlantic Forest natural resources have been largely explored, regardless; the forest is used in an unsustainable way, provoking an elevated level of damage in the Biome. Knowing the arboreal behavior, in a certain period, allows providing scientific grants for conservation and forestry management policies. The forestry competition is one of the factors that influence on the growth and longevity of the specimens, and is due the demand for essential resources, such as, water, light and nutrients. Therefore, this research aimed to analyze how the forestry competition affects the dendrometric measurements, by using distance independent and distance semi-independent competition indexes (radius at 2, 3, 4, 5 and 6 meters), and to aggregate the competitive aspects on the projection, through traditional models and neural artificial networks, in a Semi-Deciduous Seasonal Forest, in Minas Gerais (Brazil). In this study was carried out a classification of individuals according to their ecological groups and their light demand, in order to evaluate whether the calculation of the competition is more significant when evaluated for each group, rather than the generalized analysis to all species of the fragment under study. To evaluate the competition per ecologic group was more efficient, since it presented superior correlations. The BAL independent index obtained the best performance to analyze the influence of the competition on the specimens’ growth on the fragment studied. Projecting the dendrometric measurement using Neural Artificial Networks was more efficient than the traditional means. Author: Ilvan Medeiros Lustosa Junior Advisor: Prof. Dr Renato Vinícius Oliveira Castro Postgraduate program in Forest Sciences Brasília, 26 february of 2016