Modelagem de nicho e crescimento das principais espécies arbóreas da floresta estacional semidecidual de Minas Gerais

Abstract

The impacts of degradation and deforestation of forests at different scales throughout the tropics are notable. Thus, the correct allocation and growth potential of forest species using climatic variables to restore and reforest the anthropized regions of the widest forest typology of Minas Gerais becomes important. Thereby, this work was divided into two scientific articles. In the first, the main tree species in the Seasonal Semideciduous Forest (FES) of MG were defined, according to their features for forest restoration programs and modeling their niche according to climatic variables related to temperature variation and drought through Logistic Ridge Regression, with an analysis of its accuracy. The influence of the predictors on the adaptive success of the selected species and their distribution in the forests was also verified. Finally, the uncertainties of the species databases were assessed to accurately infer the locations that the species are adapted. In the second article, the objective was to identify forest species, among the main ones in the FES of Minas Gerais, with the greatest potential of growth rate in carbon for monospecific plantations and forest restoration in them. Thus, this study resulted in a discussion of important characteristics for some of the main species of the FES, about niche modeling and its growth on a macro scale. Thus, this study resulted in a discussion of important features for some of the main species of FES, for niche modeling and its growth on a macro scale. The selected main species were Copaifera langsdorffii, Inga vera, Tapirira obtusa, Myrcia splendens, Protium spruceanum, Myracrodruon urundeuva, Tapirira guianensis, Anadenanthera colubrina, Eremanthus incanus, Triplaris gardneriana, Cecropia pachystachya, Mabea fistulifera, Astronium fraxinifolium, Ocotea corymbosa, Machaerium villosum, Casearia sylvestris, Lithraea molleoides, Cupania vernalis, Croton urucurana and Guarea guidonia. In addition, this study demonstrated that iterative methods (Ridge Regression, LASSO, and bootstrap) could be used in predicting the niche and in growth of forest species.