Field experiments involving soils require previous verification of the soil spatial variability. The objective of this research was to study the spatial variability of soil physical properties of an experimental plot using geostatistics. The experiment was conducted in 1982 at the Experimental Center of the Instituto Agronômico in Campinas, state of São Paulo, Brazil, on a Red Latosol (Rhodic Ustox) under conventional tillage in a 30 x 30 m area, with sampling points arranged in a 5 m square grid. The analyzed properties were water content, porosity, bulk density, penetration resistance, and water retention. The spatial variability was evaluated by geostatistical analysis of semivariograms and kriging interpolation of the data for the construction of maps. Spatial dependence occurred mainly for the variables from the upper soil layer (0-25 cm), showing moderate and strong spatial dependence. A significant positive correlation was found between water retention and bulk density. The observed spatial dependence and similar behavior of the variables allowed the inference that random sampling would not be enough to characterize a field, as it would fail to show the variability. In cases where an experiment requires independent samples, the response to the treatments would be affected.
semivariogram; kriging; conventional tillage; Red Latosol
