Avaliação da riqueza de espécies arbóreas através dos métodos bootstrap e regressão quadrática de platô

Abstract

Tropical forests in Brazil have a high diversity of tree species and have been under great pressure due to agricultural expansion, livestock and other anthropic economic activities. Given these factors, phytosociological studies involving species richness are becoming increasingly important in order to know the existing biological potential, as well as justify the conservation of native forest remnants. Thus, the objective of this study was to verify the sufficiency of samples needed to quantify the tree richness of the Rio Grande River Basin (BHRG), in southern Minas Gerais, by assessing the level of fragments (plots per fragment) and for the whole area. of the basin (fragments in the total area), this analysis was extended to the punctual analysis of the Phytophysiognomy Semideciduous Seasonal Forest, which is the main physiognomy of occurrence in the watershed, also verifying the use of modeling for the total estimation of tree species. The method used was plateau regression, for this we used bootstrap resampling without replacement for the removal of subjectivity in data entry in the model adjustment. Plateau regression indicated the point where sample sufficiency would be found. It was concluded by the quadratic plateau response model that the sampling performed for the basin as a whole reached the sample sufficiency with 26 fragments. At the fragment level, sample sufficiency was achieved in 55% of the sampled fragments. The results presented here can be used for the purpose of indicating species destined for the recovery of degraded ecosystems, among other forest conservation applications.