The determination of the spatial distribution of soil properties is important to understand soil constraints to plant nutrition and appropriate management of soil resources in recently cultivated areas, particularly where only data from exploratory soil surveys are available . This study addresses spatial variability of clay content (At) and exchangeable potassium (Kt) in soil samples collected in July of 1994, located in Southern Mato Grosso, Brazil. Geostatistical methods were used to describe the spatial variability and predict these soil properties along two transects, crossing different soil mapping units and land use types. By ordinary kriging, values in unsampled areas were appropriately estimated and places where more intensive sampling is required were identified. Both soil properties studied were influenced mainly by different soil class and exhibited a moderate spatial dependence and anisotropy in the horizon studied. Surveys of soil fertility, plus data gathered from soil survey, can be analyzed by geostatistical methods by decision-makers, to support management planning and to predict indicators related to soil quality as a measure of sustainability. Parameters from semivariograms associated to soil classes can be used to reduce sampling efforts to estimate At and Kt distribution and patterns in the area studied. Soil survey information allowed to reduce the sampling effort for the At about seven times.
soil properties; spatial variability; interpolation of data; soil sampling density