Modelagem preditiva de distribuição de espécies pioneiras no estado de Minas Gerais

Abstract

The present work was conducted with the objective of performing a probability prediction of twenty-three pioneer species in the state of Minas Gerais, Brazil. This prediction will be done using the maximum entropy algorithm (Maxent) and genetic algorithm (GARP). The study was motivated by the demand for information to choose species for reforestation and recovery of disturbed areas programs. Therefore, the study was conducted to indicate the environmental suitability of each region of the state of Minas Gerais for each studied species. For each pioneer, we elaborated a map with the degree of environmental similarity to the location of its natural occurrence, thus indicating the species to be introduced in all study area with great chances of success. The produced maps presented similar distributions to that described in the literature. We found differences between the methods used for constructing these maps. In general, the genetic algorithm presented more inclusive results, that is, a larger area of potential distribution was predicted. The Maximum entropy algorithm showed a smaller area of potential distribution and was considered a more restrictive model. This allow us to determine that the Maxent mapped only areas with high environmental similarity, thus, there is less chance of error in the case of species introduction or field of work that intend to find new populations.