Geoestatística aplicada a caracterização da distribuição espacial de espécies arbóreas em floresta ombrófila mista

Abstract

This work aimed to apply geostatistical analysis to verify the existence of spatial dependence in the distribution of arboreal species, and variables related on parcels installed in the National Forest of São Francisco de Paula, RS, in the Rio Grande do Sul Continuous Forest Inventory, in 2000. For this research, the plots 1537, 1538, 1539 and 1540 were select. To analyze the existence of spatial dependence of arboreal species and related variables, the "adjustment to feeling" semivariograms were conducted, obtaining the parameters of nugget effect (C0), level (C0 + C), range (a), coefficient of determination (R ²), Degree of spatial dependence (C/C0 + C) and sum of squared residuals (SSR). To check the quality of the adjustments, the cross-validation has been used. The geostatistical software GS + was used to select the models, describe the spatial dependence and timely realization of ordinary kriging interpolation of arboreal species and other variables. The results showed an elevated degree of spatial dependence on the species Podocarpus lambertii Klotzsch ex Endl (1538); Vernonia discolor (Spreng.) Less (plot 1538); Cryptocarya moschata Nees (1539); Matayba elaeagnoides Radlk (plot 1539); Shannon index - Weaner; and moderate to Myrciaria floribunda (West ex Willd.) (1537); Number of individuals from the plot 1540 (plot 1540). The other variables had no structured variance, indicating that the variables have a fully random distribution and it is not possible to obtain surfaces interpolated by kriging. This event is probably bound to the spatial dependence break, resulting from the distortion of the original forest formation due to the history of forest exploitation process.