Probabilistic Classification of Tree and Shrub Vegetation on Phytogeographic System

The phytogeographical system for vegetation classification split the vegetation in first level types: Forest and Grasslands. The Forest type can be recognized in tropical rain forest and seasonal, depending on the environmental conditions. This determines the occurrence of deciduous species in big or small quantity. And the grasslands are Savannah in majority. This work proposed probabilistic methods to classify these vegetation types based on priori occurrence of species. The test was carried out with forest inventory data using ten vegetation fragments in farm of Embrapa and Cascata's park, in Sete Lagoas city, Minas Gerais, Brazil. The distribution of species with occurrence in different types was adapted to set theory and Bayes theorem. This way it is possible calculate to belonging of species on vegetation types. The results were compared with usual classification. The main contribution of probabilistic methods was the increase the information to classify tree and shrub vegetation inventoried. It is especially recommended for transition regions between vegetation types.