Spatial variability of forest soil properties studied using multivariate and geostatistical analysis

The objectives of this work were to identify groups of physical and chemical properties of an Inceptisol in a stand of Eucalyptus benthamii in a commercial plantation for management purposes using multivariate analysis, and to analyze the spatial continuity of the groups identified. Soil samples were collected based on a grid of spaced points of 70 x 70 m. A descriptive study characterized the distribution of the edaphic variables followed by an exploratory factor analysis to reduce the dimensionality of the data set and geostatistical analysis to estimate the degree of spatial dependence. From the exploratory factor analysis, three axes were selected and explained about 70% of the total variability of the data set, identifying a group of variables that correlated with structure, fertility and acidity. For those factors that presented spatial dependence, thematic maps were used to identify regions with different levels of soil fertility. The combination of multivariate and geostatistical analysis is useful for analysis of the spatial variability of soil attributes and for generating recommendations for forest management.